Electron–Water Interactions and Implications for Liquid Cell Electron

Article pubs.acs.org/JPCC

Electron−Water Interactions and Implications for Liquid Cell Electron Microscopy Nicholas M. Schneider,† Michael M. Norton,† Brian J. Mendel,† Joseph M. Grogan,† Frances M. Ross,*,‡ and Haim H. Bau*,† †

Department of Mechanical Engineering and Applied Mechanics, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States ‡ IBM T. J. Watson Research Center, Yorktown Heights, New York 10598, United States S Supporting Information *

ABSTRACT: Liquid cell electron microscopy enables direct in situ imaging of processes in liquids and objects suspended in liquids with nanoscale resolution. However, the irradiating electrons affect the chemistry of the suspending medium, typically an aqueous solution, producing molecular and radical products such as hydrogen, oxygen, and hydrated (solvated) electrons. These may impact the imaged structures and phenomena. A good understanding of the interactions between the electrons and the irradiated medium is necessary to correctly interpret experiments, minimize artifacts, and take advantage of the irradiation. We predict the composition of water subjected to electron irradiation in the electron microscope. We reinterpret available experimental data, such as beam-induced variations in pH and colloid aggregation, in light of our predictions and show new observations of crystallization and etching as functions of dose rate, resolving conflicting reports in the scientific literature. We make our computer code available to readers. Our predictive model is useful for designing experiments that minimize unwanted beam effects, extending liquid cell microscopy to new applications, taking advantage of beam effects for nanomanufacturing such as the patterning of nanostructures, and correctly interpreting experimental observations. Additionally, our results indicate that liquid cells provide a new tool to study radiolysis effects on materials and processes.

■

INTRODUCTION Direct imaging of processes in liquids with the nanometer resolution of the electron microscope has become possible through recent advances in the fabrication of liquid cells.1−3 Liquid cells for transmission (TEM) and scanning transmission (STEM) electron microscopy are hermetically sealed from the vacuum environment of the microscope and contain a liquid layer, tens to hundreds of nanometers thick, at near atmospheric pressure, sandwiched between two electrontransparent thin membranes. Because the liquid layer is thin, electrons are transmitted with minimal scattering, enabling realtime imaging of objects suspended in the liquid. Liquid cell electron microscopy can be used, among other things, to image electroplating and electrostripping,4−7 nucleation and growth of nanoparticles and bubbles,1−3,8−13 interactions among nanoparticles and between nanoparticles and interfaces,2,4−7,14−19 undirected and directed formation of colloidal crystals,16,19 and macromolecular conformations and biological processes.20−23 In many of these systems, water is the suspending medium. Although a detailed understanding of the interactions between the electron beam and the irradiated medium is essential to interpret, suppress, and exploit radiation-induced phenomena, only a few studies address beam effects in liquid cell electron microscopy quantitatively. Radiation chemistry has, however, been studied intensively because of its © 2014 American Chemical Society

importance in diverse disciplines such as medicine, atmospheric science, food preservation, and nuclear energy production.24−29 Ionizing radiation (photons, γ-rays, neutrons, electrons, etc.) readily transfers energy to the irradiated medium with effects that are relatively independent of the type of radiation. This energy excites and dislodges orbital electrons, which results in the generation of heat and radical and molecular species. The immediate irradiation products continue to react, causing a cascade of chemical reactions and production of various additional species.28−31 In liquid cell electron microscopy experiments, beam-induced temperature changes are usually insignificant. Under typical operating conditions, energy transfer from the electron beam to water increases the water temperature by, at most, a few °C.3 In contrast, the radiationinduced chemical reactions lead to significant changes in the solution composition. Although radiolysis has been investigated extensively due to its importance in medicine and in nuclear reactors,24−29 the conditions encountered during electron microscopy are vastly different. For example, the dose rate associated with a 300 keV electron beam of 1 μm radius and 1 nA current is 7.5 × 107 Gy/s, which is 7 orders of magnitude Received: July 23, 2014 Revised: August 26, 2014 Published: August 28, 2014 22373

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

Article

variety of systems.28,33 The primary products then participate in a cascade of slower chemical reactions, creating additional species such as O2. At these longer times, it is necessary to consider large numbers of interrelated reactions between the different molecular and radical products. To account for the evolution that occurs on the time scales of interest in liquid cell electron microscopy (seconds to hours), we employed a kinetic model. For neat water alone, dozens of reactions are needed to model the complex chemistry occurring in the system.25,26We used the model set forth by Elliot and McCracken25 that accounts for the spur reactions via the Gvalues. Table S1 in the Supporting Information lists the 79 reactions and reaction rate constants included in this model. Since the concentration of water is much larger than that of the radiolysis products, we simplify the mathematical model by assuming that the water acts as a solvent. This allows us to model the system with a set of first-order reactions. Since only a small fraction of the water molecules decompose, the model can be further simplified by assuming that the mass of the water remains constant throughout the process. We verified the applicability of this approximation by demonstrating that the predictions of the simplified model favorably agree with those of a more rigorous model that does not assume constant water mass (see the Supporting Information). We determined the concentration fields of the various species as functions of space (x) and time (t), where x is a position vector. The evolution of the concentration Ci(x,t) of species i is described with the reaction-diffusion equation:

greater than the typical dose rate (∼1 Gy/s) generated by common radiation sources.24,26 Hence, much of the data available in the literature is not directly applicable to the conditions prevailing in the electron microscope. The objective of our study is to compute the concentrations of radiolysis products as functions of electron beam irradiation parameters, time, space, and solution composition, under conditions typical for electron microscopy. Although we cannot always carry out a direct comparison between theoretical predictions and experimental data, we will use these computations to explain qualitatively several phenomena observed during liquid cell imaging such as bubble nucleation and growth, precipitation of cations from solution, the dissolution of metals, and the aggregation of colloids. In this paper, we first introduce a mathematical model to determine the concentrations of various radiolysis products. We then compute the concentrations of the various species as functions of time, radiation dose rate, and initial solution composition, when the entire liquid medium is uniformly irradiated (homogeneous reactions). We next use the model to predict concentrations in the case of a finite-diameter electron beam when only a fraction of the liquid volume is exposed to irradiation and the radiolysis products diffuse away from the irradiated region. Lastly, we describe a few experimental observations that indirectly support our theoretical predictions and use the predictions to provide insights into other yet incompletely understood observations. The computer code is made available in a public repository32 with details provided in the Supporting Information. We provide these codes to enable other researchers to carry out computations for their own experimental conditions. The codes can also be augmented with additional processes, such as the interactions between the radiolysis products and solutes. Our goals are to facilitate the design of experiments that minimize unwanted artifacts, assist in the interpretation of experimental data, enable constructive use of beam effects, and suggest new opportunities for liquid cell electron microscopy in the study of radiolysis and its effect on materials.

∂Ci = Di∇2 Ci − ∂t

j

kjkCiCj + R i (2)

j,k≠i

In eq 2, the first term on the right-hand side represents diffusion, with Di the diffusivity of species i in water. The second and third terms represent respectively destruction and production of species i through chemical reactions (Table S1), with reaction rate constants kij. The last term

■

Ri =

KINETIC MODEL When an incident electron interacts with an individual water molecule, it transfers energy to the molecule, exciting orbital electrons. At very early times (10 ps) after the energy transfer event, the water molecule decomposes into a few species, including hydrated (solvated) electrons eh−, hydrogen radical H•, hydroxyl radical OH•, and H2. These reaction products, known as the initial yield, are concentrated nonuniformly along the incident electron path in discrete volumes, dubbed spurs, surrounding the affected water molecule.3,24,25,28−31,33 Shortly ( 0 to represent the initial oxygen concentration. When other solutes are present, the initial conditions must be adjusted appropriately. To the first order of approximation, we assume that all confining surfaces are chemically inert and, therefore, impermeable to all species, i.e. ∇Ci·n̂ |wall = 0

(6)

where n̂ is a unit vector normal to the surface. In the following sections, we will solve the kinetic model, first when the entire liquid medium is irradiated uniformly (the homogeneous case) and then in the case when the liquid medium is irradiated by a finite radius beam and the species diffuse away from the irradiated region (the heterogeneous case). The homogeneous model was solved with Mathematica, MATLAB, and COMSOL’s reaction chemistry module. The Mathematica and MATLAB codes are available on the online source code repository GitHub.32 Implementation details for each software package are described in the Supporting Information. The heterogeneous model was solved with the multiphysics, finite element program COMSOL.

Figure 1. Concentrations of eh−, H•, H2, H2O2, OH•, and O2 as functions of time. Initially neat, deareated water is irradiated continuously at a dose rate of 7.5 × 107 (Gy/s).

We repeated calculations similar to those of Figure 1 for various dose rates and always observed the eventual establishment of a steady state. However, both the magnitude of the steady state concentration and the time that it takes to reach steady state depend on the dose rate. Figure 2 depicts the steady state concentrations of the species from Figure 1 as functions of the dose rate.

■

HOMOGENEOUS REACTIONS In this section, we focus on the case when the entire liquid volume is irradiated uniformly. This situation differs from the normal operating conditions of many liquid cells, in which the source term Ri is present only in the region irradiated by the electron beam but reflects the conditions of some thin graphene cells. The homogeneous model allows us to gain insights pertaining to reaction kinetics without the added complications of mass transfer. Since the concentration distribution is uniform throughout the domain, Ci(x,t) = Ci(t) and we can dispose of spatial dependence and eliminate the diffusion term from eq 2. We first consider the case of neat water and then investigate solute effects. Neat Water. We integrated numerically the homogeneous equations for neat, deaerated water with the initial conditions CH2O(0) = 55.56 (M) and CH+(0) = COH−(0) = 10−7 M to find the temporal evolution of the 16 species that are involved in the reactions tabulated in Table S1. In the calculations, we used the G-values for high-energy electrons in neat water33,35 and assumed, to the first order of approximation, the G-values to be independent of solution composition. For brevity, we present the temporal evolution of a subset of species that we deem most relevant to processes taking place during liquid cell electron microscopy. The concentrations of other species can be readily obtained with the programs provided in the Supporting Information. Figure 1 depicts the concentrations of eh−, H•, H2, H2O2, OH•, and O2 as functions of time. The simulation was carried out for the dose rate associated with a 300 keV beam of 1 μm radius and 1 nA current, 7.5 × 107 (Gy/s) (eq 4). Irradiation begins at time t = 0 and is maintained continuously thereafter. As t increases, the concentrations of H2, H2O2, OH, and O2 increase monotonically and rapidly from zero to achieve steady state values within about 1 ms after the start of the irradiation. The concentrations of eh− and H• initially increase, peak, and then decline to their steady state values within the same time frame. A key result in Figure 1 is that the concentrations of the radiolysis products do not grow unabated with time. Eventually, reverse reactions convert radiolysis products back to water, enabling the establishment of a steady state.

Figure 2. Steady state concentrations of eh−, H•, H2, H2O2, OH•, and O2 as functions of dose rate. Deaerated, neat water is subjected continuous irradiation.

The results of Figures 1 and 2 suggest that the concentrations of radiolysis products as functions of time and dose rate can be correlated using power laws. For the dose rates commonly associated with liquid cell electron microscopy, the dependence of the steady state concentration on dose rate can be approximated with a power law Css, i ∼ αiψ βi (M)

(106 < ψ < 1010 Gy/s)

(7)

Table 1 lists the values of αi and βi (0.37 < βi < 1.21) obtained by fitting eq 7 to the calculation results. We define the time to steady state tss (ignoring initial fluctuations) as the time needed for the concentration to achieve 95% of its steady state value, i.e., [Ci(tss) − Ci(∞)]/ Ci(∞) = 0.95. The dependence of tss on dose rate can be approximated with the power law tss, i ∼ aiψ −m (s)

(106 < ψ < 1010 Gy/s)

(8)

where the exponent m ∼ 0.50 ± 0.01 for all species. The coefficients ai are of order 1 and are listed in Table 1. In the absence of diffusion and solutes, eqs 7 and 8 and Table 1 provide rapid, approximate estimates of the steady state concentrations of radiolysis products as functions of dose rate and time in liquid cell experiments. Significantly, a steady state evolves in all cases examined, including the high dose rates common to electron microscopy, confirming our previous findings.3,36 Furthermore, the concentrations of almost all radiolysis products increase sublinearly (βi < 1) as the dose rate increases. Thus, doubling the dose rate leads to less than a 222375

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

Article

Table 1. Steady State Concentration Power Law Parametersa homogeneous αi

species −

eh H• H2 H2O2 HO2• HO2− HO3• H+ O2 O2− O3• O3− OH• OH− O−

2.49 6.31 8.88 2.20 2.39 4.34 2.32 2.07 2.11 8.21 7.96 2.10 1.45 1.20 4.56

× × × × × × × × × × × × × × ×

βi −10

10 10−9 10−8 10−7 10−11 10−12 10−18 10−9 10−10 10−10 10−16 10−15 10−8 10−8 10−13

0.51 0.38 0.50 0.44 0.71 0.72 1.21 0.51 0.67 0.51 0.84 0.83 0.46 0.37 0.69

axisymmetric can be handled in a similar way. Figure 3 (left column) depicts H2, eh−, O2, and H+ concentrations as

heterogeneous αi

ai 3.18 5.92 3.55 3.16 7.04 2.30 4.43 6.52 6.99 6.49 6.42 5.72 0.40 4.12 4.67

1.02 2.61 1.88 4.43 8.90 2.91 4.53 6.83 2.54 5.55 4.45 1.77 4.28 1.87 1.87

× × × × × × × × × × × × × × ×

βi −9

10 10−8 10−7 10−7 10−12 10−12 10−19 10−9 10−10 10−10 10−17 10−16 10−8 10−8 10−13

0.50 0.33 0.52 0.45 0.85 0.85 1.50 0.48 0.77 0.63 1.17 1.13 0.46 0.43 0.84

The coefficients αi and βi in the correlation for the steady state concentration Css,i ∼ αiψβi (M) of species i and the prefactor ai in the correlation tss,i ∼ aiψ−m for the time to steady state of species i for dose rates relevant to EM in the range from 106 to 1010 Gy/s. In the case of the heterogeneous model, the concentrations are at the beam’s center. The beam radius is 1 μm, and the liquid cell’s sidewalls are 50 μm away from the beam’s center. a

fold increase in concentration of most radiolysis products. The sole exception is HO3 whose steady state concentration increases superlinearly (βHO3 ∼ 1.2) with dose rate. Finally, the time it takes to establish a steady state in a continuously irradiated medium scales inversely as the dose rate to the power ∼1/2, meaning that at the high dose rates common to electron microscopy, the homogeneous system reaches steady state rapidly (Figure 1), within 1 ms, much shorter than the duration of most experiments. Although, as we shall see later in the paper, these approximations provide useful insights into beam effects, more elaborate models are needed to account for the effects of finite beam size and solutes.

Figure 3. Heterogeneous model predictions for the spatial and temporal evolutions of H2 (a, b), eh− (c, d), O2 (e, f), and H3O+ (H+) (g, h). The left column depicts the concentrations of the selected radiolysis products of neat water as functions of the radial distance from the center of the irradiated region at various times. The right column depicts the concentrations of the same products at the center and edge of the irradiated region and at the perimeter of the liquid cell. The beam and liquid cell radii are respectively 1 and 50 μm. The dose rate is 7.5 × 107 Gy/s.

■

HETEROGENEOUS REACTIONS In most electron microscopy experiments, only a fraction of the liquid in the cell is irradiated at any given time. Radiolysis products generated within the irradiated region diffuse out and continue to interact outside that region. To examine the effects of diffusion, we consider a model system composed of a cylindrical electron beam of radius a located concentrically in a liquid-filled circular chamber with radius W. In all the calculations presented in this section, we use a = 1 μm and W = 50 μm. In practice, the effective W may be as large as a few millimeters. We focus here on circumstances relevant to TEM, where the electron beam is maintained at a fixed position for a prescribed time interval. Scanning microscopy requires somewhat different treatment and is not addressed here. We coded the multiple reactions of Table S1 in COMSOL and accounted for mass transfer by diffusion. Since the bottom and top surfaces of the liquid cell are assumed inert, we consider only one-dimensional diffusion. We use cylindrical coordinates (r, θ) with the origin at the beam’s center and assume axisymmetry. We solve eqs 2−4 in two domains. In the irradiated domain 0 ≤ r ≤ a, the equations include the relevant source terms Ri. In the region outside the beam a < r ≤ W, ψ = 0 (and therefore Ri = 0). Liquid cell geometries other than

functions of the radial position r at various times. The irradiated region (r ≤ a) is shaded in gray. Figure 3 (right column) depicts the concentrations of the species at the beam’s center (r = 0) and edge (r = a) and at the device’s outer, impermeable surface (r = W) as functions of time. Importantly, all species reach equilibrium concentrations within secondsmuch longer than the equilibration time in the homogeneous case. This longer equilibration time results from the relatively slow diffusion process that takes place over time scales on the order of tD,i ∼ W2/Di. Typical diffusion coefficients of solutes in water are on the order of 10−9 m2/s (see Table S3). Accordingly, when W = 50 μm, the diffusion time constant (tD,i) is on the order of seconds, consistent with Figure 3. This time constant increases quadratically as the liquid cell size increases. Because of the chemical reactions between the radiolysis products, the actual equilibrium concentration depends on position within the liquid cell. 22376

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

Article

radiolysis effects. We first discuss the effect of the beam irradiation on a solution’s composition, in particular its pH. Then we interpret several phenomena, specifically aggregation, bubble formation, and metal deposition, within the framework of our kinetic model and explain some conflicting reports on crystal growth and etching during electron microscopy. Radiolysis of Aerated Water. Often, solutes dramatically alter the equilibrium concentrations of radiolysis products.26,28,29 Since in most EM experiments it is not practical to deaerate the solution prior to its introduction into the liquid cell, one will often encounter solutions with dissolved oxygen. We first, therefore, examine the effect of oxygen dissolved in the solution on the concentrations of the radiolysis products. Aerated water is defined here as neat water saturated with oxygen at atmospheric conditions without any dissolved CO2. The saturation concentration of oxygen is 0.255 mM as calculated with a Henry’s law constant of 0.0013 mol/(kg bar).37 Figure 4 depicts the ratios of the steady state

Figure 3 illustrates the spatial and temporal behavior of some of the radiolysis products. A slowly reacting species such as H2 is continuously produced within the irradiated region and diffuses away (Figure 3a). At early times (t < 10−4 s), the production rate exceeds the diffusion rate and the H 2 concentration at the center of the beam increases rapidly. At later times, the production is balanced by diffusion, which is reflected by the plateau in CH2(0,t) (Figure 3b). Eventually, as the H2 concentration outside the beam builds up and the diffusive flux decreases, the H2 concentration in the irradiated region resumes its growth to eventually approach its maximum, equilibrium value. The time it takes to reach equilibrium depends on the size of the device. When irradiation continues for a sufficiently long time, the entire liquid cell will achieve a steady state. H2O2 (not shown) behaves similarly to H2. The more reactive radiolysis products, such as eh−, H•, OH•, H3O+, and O2, exhibit more complex behaviors than H2. The highly reactive eh−, H, and OH exist mostly within and in close proximity to the irradiated region. As they diffuse away from the irradiated region, where they are produced, they are consumed through chemical reactions. As a result, outside the irradiated region, the concentrations of the highly active species drop rapidly to zero (Figure 3c). The somewhat complex behavior of eh− (Figure 3c,d) can be understood by considering its interactions with O2 (Figure 3e,f), a major scavenger of eh−. Since oxygen production is delayed somewhat, initially the eh− concentration in the irradiated region increases rapidly. Once the production of O2 ramps up (t > 10−5 s), the O2 scavenges eh− and reduces its concentration. As a result, the concentration of eh− as a function of time exhibits a peaks at t ∼ 5 μs. As time increases further, the eh− concentration declines to its equilibrium value. The peak in the spatial distribution of eh− next to the edge of the irradiated region at intermediate times (i.e., t ∼ 10−4 s) is attributable to the mass transfer of O2 by diffusion away from the irradiated region. The temporal and spatial distributions of the oxygen concentration exhibit opposing trends to that of eh−. H3O+ (or H+, Figure 3g,h) behaves similarly to the other radical products, but with somewhat higher concentration outside the irradiated region than eh−. After the onset of irradiation, the H3O+ concentration in the irradiated region quickly peaks and then drops to a steady state. Outside the irradiated region, the H3O+ concentration grows slowly by diffusion. Eventually the entire liquid cell will have an increased H3O+ concentration and reduced pH. For a solvent initially at pH 7 exposed to a 1 μm radius beam of 1 nA (a dose rate of 7.5 × 107 Gy/s), at steady state the pH will drop to 4.9 within the irradiated region and ∼6.1 outside the irradiated region. As in the homogeneous case, we can fit a power law similar to the one in eq 7 to specify the concentrations of radiolysis products at the beam’s center as functions of the dose rate. The calculations were carried out for a liquid cell with radius W = 50 μm, but the results should be approximately applicable for any sufficiently large W (i.e., W > 10 μm). The prefactor αi and the exponent βi (0.3 < βi < 1.5) are listed in Table 1.

Figure 4. Ratio of aerated to deareated steady state concentrations of eh−, H•, H2, H2O2, OH•, and O2 as functions of dose rate.

concentrations of eh−, H•, H2, H2O2, OH•, and O2 in irradiated water initially oxygen-saturated (aerated) and in oxygen-free (deareated) water as functions of the dose rate. Since molecular oxygen is a scavenger of hydrated electrons,26 the presence of oxygen in the solution reduces the eh− concentration at low dose rates. The initial presence of oxygen also decreases the steady state concentrations of the radical species OH• and H•, while increasing the concentrations of the molecular species H2, H2O2, and O2. Most importantly, the deviations of the steady state concentrations of radiolysis products in the aerated solution from the corresponding values in the oxygen-free solution are a strong function of the dose rate. When the dose rate is sufficiently high (>∼108 Gy/s), the steady state oxygen concentration in the deareated solution exceeds the initial saturated oxygen concentration. As a result in that case, the steady state concentrations of all species are nearly independent of the initial oxygen concentration. In other words, all the ratios in Figure 4 asymptotically approach the value of 1 as the dose rate increases. However, at lower dose rates, the presence of initial oxygen in solution does have a significant relative impact on the steady state concentrations of the radiolysis products. Effect of Irradiation on pH. Since the radiolysis products include H3O+ ions, e-beam irradiation alters the solution’s pH. Figure 5 depicts the steady state pH as a function of dose rate and initial pH of deaerated water. We have assumed that the conjugate pairs to the protons/hydroxides of the added acid/ base (e.g., SO42− in sulfuric acid and K+ in potassium hydroxide) do not significantly participate in the various reactions. This assumption may not always be valid, and it may be necessary to include additional kinetic reactions to account

■

IMPLICATIONS FOR ELECTRON MICROSCOPY Understanding beam−liquid interactions is essential for correctly interpreting experimental data, designing experiments that minimize unwanted effects, and taking advantage of beam effects. We and others have observed various phenomena during liquid cell electron microscopy that can be attributed to 22377

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

Article

of the initial pH value as reflected by the plateaus in Figure 6a. This is consistent with prior reports on radiolysis at low dose rates.24,26 As the solution becomes more acidic (pH < 5), the steady state concentrations of eh− and O2 decline while those of H, H2, and H2O2 increase compared to their corresponding values in a solution with preirradiation pH = 7. The concentration of OH• changes little, however. In alkaline solutions (pH > 10), the steady state concentrations of eh−, H•, and O2 are lower while those of H2, H2O2, and O2 are higher than their corresponding values in a solution with an initial pH = 7. In other words, the steady state concentrations of radiolysis products vary greatly, depending on whether the solution was strongly acidic or strongly basic preirradiation. Figure 6b compares the effect of initial pH in aerated and deaerated solutions. The radiolysis products approach their deaerated values at extreme pH values (pH < 5 and pH > 11). At moderate pH values, the radiolysis product concentrations in the aerated solution assume different steady state values within a factor of 2 of their deaerated counterparts. It is, therefore, most important to consider aeration effects when carrying out experiments at moderate initial pH values. Beam-Induced Aggregation. Grogan et al.,16 Li et al.,17 and Woehl et al.,39 among others, have shown that otherwise stable colloidal suspensions aggregate during electron microscope imaging. However, the causes of aggregation have not been clarified. A plausible hypothesis is that the ions produced by the irradiation increase the ionic strength of the solution and reduce the Debye screening length and the resulting repulsive force among particles. To examine this hypothesis, we calculate the effect of radiolytic ionization on the Debye screening length. For example, we estimate that the Debye screening length of the 5 nm diameter, charge-stabilized Au spheres (pH 7, 0.01 wt % AuCl3) examined by Grogan et al.16 is ∼7.3 nm. The ions produced at the dose rate used in the experiments (∼7.5 × 109 Gy/s) reduce the Debye screening length by less than 15%. This is a relatively small change that is unlikely to destabilize the suspension, although such effects must be considered when designing experiments with colloidal systems. An alternative hypothesis is that the effects of irradiation on pH, discussed in Figure 5, may alter the particles’ surface charge and zeta (ζ) potential. In the experiment discussed above,16 we estimate based on Figure 5 and the dose rate used that the pH of the irradiated solution decreases from 7 to 3.25. This is estimated to increase the ζ potential from ∼−33 to ∼−23 mV. In other words, the Au particles lose much of their surface charge and are driven toward the isoelectric point. Colloidal suspensions with |ζ| ≤ 25 mV are considered unstable.40 E-

Figure 5. Steady state pH as a function of dose rate and initial pH prior to irradiation. Deaerated water.

for interactions between the radiolysis products and the conjugates (e.g., ref 38) When the dose rate is low ( 3 approach pH = 3 as the dose rate increases (Figure 5). Alkaline solutions are most strongly affected by radiolysis at the dose rates used in electron microscope imaging. Since the solution pH controls diverse processes, ranging from electrochemistry to aggregation, it is important to account for the effect of radiolysis on solution pH when interpreting data from liquid cell experiments. Effect of Initial pH on Radiolysis Product Concentrations. It is clear from the above that at any given dose rate the steady state pH (or H3O+ concentration) depends on the initial pH of the solution preirradiation. Since the concentrations of all radiolysis products are interrelated, the steady state concentrations of the other radiolysis products also depend on the initial pH. These effects are important to consider since strong acids and bases are frequently employed in liquid cell experiments to improve salt solubility. Figure 6a depicts the steady state concentrations of eh−, H•, H2, H2O2, OH•, and O2 as functions of the initial solution pH in deaerated water. Figure 6b depicts the ratios of radiolysis products in aerated and deaerated water as functions of the initial pH. As before, we assumed here that the conjugate pairs to the protons/hydroxides of the added acid or base do not participate in the reactions. At moderate pH (5 < pH < 9), the steady state concentrations of all radiolysis products is nearly independent

Figure 6. (a) Steady state concentrations of eh−, H•, H2, H2O2, OH•, and O2 as functions of the pH value prior to irradiation for deaerated water under uniform irradiation (homogeneous case). (b) Ratio of radiolytic products concentrations in initially oxygen-saturated water (CO2(0) = 0.255 mM) and in deaerated water as functions of initial pH. The dose rate is 7.5 × 107 Gy/s. 22378

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

Article

beam induced aggregation of charge-stabilized particles is thus likely due to the effect of radiolysis on the irradiated solution’s pH. Crystallization and Patterning. Many microscopists have reported electron-beam-induced metal crystallization and crystal growth when imaging platinum, silver, copper, and gold salt solutions in liquid cells.3,11−13,41,42 These observations are consistent with our calculations. When an aqueous salt solution is irradiated, radiolysis products that are reducing agents, such as hydrated electrons, may reduce cations and induce precipitation under certain conditions.35,43,44 This process has even been employed to purify various metals utilizing radiation sources other than electrons.36 The dose rate (Figure 2 and eq 7) controls the concentration of the reducing agents and thus the rate of reduction and, indirectly, the rate of mass transfer to a crystal (e.g., ref 41). Since the rate of mass transfer can, among other things, control the crystal growth habit,45 by controlling the dose rate, one can obtain different types of crystals and control whether materials are formed by single atom addition or by coalescence of clusters (oriented assembly).9,10,26,28,29,37 Because of their short lifetime, hydrated electrons are present at significant concentrations only within the irradiated region and are rapidly extinguished by reactions with other species outside the beam region (Figure 3c). Thus, precipitation will take place mainly in the illuminated region, allowing patterning of deposits using the electron beam as a “pencil”. Grogan et al. used the electron beam to precipitate gold nanowires from gold solution, writing the names of the authors’ institutions.3 An additional level of control can be achieved by applying electric fields.42,46 Etching and Growth of Nanoparticles. During imaging of Au nanorods in water, we observed particles decreasing in size (etching), increasing in size (growth) and remaining unaltered, depending on the dose rate with other parameters held constant. Examples are shown in Figure 7a and the Supporting Information video. The various columns in the figure correspond to different dose rates, and the rows correspond to different times. At low dose rates, growth occurred. By cycling through various zoom levels and microscope parameters in the STEM and hence dose rates, we were able to transition repeatedly between etching and growth regimes. We hypothesize that radiolysis products are responsible for the phenomena. But why does etching take place under certain circumstances and growth under others? Since the radiolysis products include both strong reducing and oxidizing agents, we suggest that the observed behavior reflects changes in the relative concentrations of the reducing and oxidizing agents. To determine whether growth or etching would take place, we need to consider the rates of the corresponding reactions. Unfortunately, the reaction rate constants are not known for many of the implicated reactions. The situation is further complicated by the presence of the oxidizer Br− (from the CTAB used in the particle synthesis) and products resulting from the interaction of radiation with Br− such as Br2•−, Br3−, and Br2 that react with gold and for which we do not account.44 Instead, since the system reaches steady state quickly, we use the ratio of the steady state concentration of the primary reducing agent eh− (standard reduction potential of −2.9 V)31 and the primary oxidizing agent OH• (standard reduction potential of 2.7 V)31 in aerated water at pH 7

Figure 7. Growth and etching of metal nanoparticles at various e-beam dose rates. Stills from video (Supporting Information) (a), growth and etching as a function of dose rate (Gy/s) (b), and steady state concentration ratio of eh− to OH• as a function of dose rate (Gy/s).

Ξ=

[e−h ] ψ 0.54 ∼ 0.47 = ψ 0.07 [OH]ss ψ

(10)

to gauge whether etching or growth will occur. Equation 10 suggests that Ξ increases monotonically (although weakly) with dose rate. We expect that etching will take place when Ξ is small, while growth will take place when it is large. This expectation is consistent with our experimental observations (Figure 7), in which etching predominates at lower dose rates and growth at higher dose rates. Figure 7b summarizes qualitatively the results of numerous experiments. In the figure, we assigned the values −1, 0, and 1 to etching, no change, and growth and depicted the observations as a function of Ξ. The figure illustrates that we experience etching when ψ < 109 and growth when ψ > 1010. The increased importance of reducing reactions compared to oxidizing reactions as the dose rate increases may help to explain the often complex behavior observed11 during experiments with solutions containing metal ions. Bubble Formation and the Existence of Steady State. Our model predicts that radiolysis product concentrations do not grow unabated but reach an steady state within milliseconds in the homogeneous case and at time scales on the order of the diffusion time for typical liquid cells in the heterogeneous case. When the liquid cell has a diameter W = 1000 μm, the time to steady state is hundreds of seconds. Two experimental observations support the existence of a steady state state. First, the lack of visible bubbles at moderate dose rates, even after prolonged irradiation, suggests the existence of reverse reactions that prevent the concentrations of the gaseous radiolysis products H2 and O2 from growing unabated and 22379

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

Article

therefore greatly exceeding their saturation concentrations.3 For example, irradiation of DI water at 1.4 × 109 Gy/s for over 50 min did not produce any bubbles. Second, at relatively high dose rates, we observed periodic, highly reproducible nucleation and growth of nanobubbles.3 The periodic behavior observed in this experiment strongly suggests the existence of a steady background concentration field far from the bubble, consistent with steady state conditions. Since in most circumstances, bubbles are undesired as they interfere with the imaged phenomena, it is important to calculate the expected steady state concentrations of the gaseous radiolysis products H2 and O2 under the experimental conditions used, so that one can estimate whether they will exceed their saturation concentrations by an amount sufficient to facilitate bubble nucleation. To form bubbles, one would need to exceed the saturation concentration by many times.47 Using Henry’s law, and assuming realistic values of the pressure in the liquid cell, we can calculate the steady state concentrations of the gaseous species and compare them with the saturation concentrations. Figure 8 depicts, for example, the

may form in these solutions under beam conditions that would not lead to bubbles in neat water. Bubble nucleation and growth is frequently observed during electron microscope imaging,3,16,20,38,48,49 occasionally with adverse effects. The ability to predict the conditions needed for bubble formation improves our capacity to design experiments that avoid bubble formation or that allow bubbles to form under control, enabling liquid cell electron microscopy to provide an effective means to study nanobubble nucleation and growth.

■

CONCLUSIONS We have utilized a kinetic model for water radiolysis and applied it to the high dose rate regime encountered during liquid cell electron microscopy. We calculated the time evolution of the concentrations of 16 radiolysis products in both the homogeneous case (uniformly irradiated liquid cell) and the heterogeneous case (only a small region of the liquid cell’s volume is irradiated). In the latter case, we also examined the spatial distribution of the various radiolysis products inside and outside the irradiated region. We established that radiolysis products reach steady state concentrations at the high dose rates associated with electron microscope imaging. As the dose rate increases, these steady state concentrations increase while the time to achieve the steady state decreases. We expressed the steady state concentrations and times to steady state as functions of the dose rate using power laws. In the heterogeneous case, the time to steady state depends on the liquid cell’s dimensions and in most cases is dictated by the diffusion time. Our model allows us to examine the effects of the solution composition before irradiation, such as its pH and oxygen concentration, on the steady state concentrations of the radiolysis products. In aerated solutions and low dose rates, aeration can change the steady state concentration of radiolysis species by as much as 4 orders of magnitude. But in solutions subject to high dose rates, aeration does not affect significantly the steady state concentrations. Similarly, the initial pH of the solution has a large effect on the steady state concentrations, particularly at high and low pH values. Significantly, irradiation has a strong effect on the local pH of a solution. When only part of the liquid volume is irradiated, the model reveals the effect of mass transfer by diffusion and the relationships between radiolysis species such as oxygen and the hydrated electron, both inside and outside the irradiated area. The complex interplay between radiolysis species and solution composition makes it challenging to make quantitative comparisons between model predictions and experimental data. However, we were able to use model predictions to carry out qualitative comparisons. The model predictions are consistent with available experimental observations of colloid aggregation, crystallization, and bubble nucleation. Model predictions also help us understand new observations, in which altering beam dose rate drives a transition from etching to growth of metallic particles. The model we have described here is useful for designing liquid cell experiments that minimize artifacts as well as in enabling one to anticipate artifacts and take advantage of beam effects to induce desired phenomena. Most importantly, the model assists in interpreting experimental data obtained with liquid cell electron microscopy. The liquid cell can also be used to study radiolysis and the behavior of materials under extreme conditions.

Figure 8. Concentration of H2 at the beam’s center in the heterogeneous case for device radii sizes of 50 μm (red line) and 1000 μm (blue line) and the saturation concentration of H2 in water at various pressures (horizontal lines). Initially neat, deaerated water is irradiated continuously at a dose rate of 7.5 × 107 Gy/s. The beam radius is 1 μm.

concentration of radiolytic H2 at the center of the irradiated region as a function of time in a liquid cell with an effective diameter of 50 and 1000 μm. The horizontal lines in Figure 8 mark the saturation concentrations at the various indicated pressures. The actual pressure in a liquid cell depends on the design of the cell, the loading conditions, and the extent of the bowing of the windows. Typical pressures are in the range of a tenth to a few atmospheres, and due to the deflection of the electron transparent membranes, the pressure inside the device will often drop below atmospheric once the liquid cell is inserted in the high vacuum of the electron microscope. Calculations similar to those shown in Figure 8 can be used to identify the maximum dose rate that allows for indefinite bubble-free imaging or, alternatively, the imaging time available for bubble-free operation at a given dose rate. For the moderate dose rate of Figure 8, prolonged, bubble-free imaging is attainable even at atmospheric pressure, as the supersaturation needed for bubble formation is likely much greater than the predicted steady state concentration. Since the initial solution composition affects the concentrations of radiolytic products, the dose rate needed for bubble formation also depends on the initial solution composition. For example, Figure 6a illustrates that extreme values of initial pH, such as pH < 5 and pH > 10, lead to substantially higher steady state concentrations of H2especially when pH > 12. Bubbles 22380

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

■

Article

Liquid Transmission Electron Microscopy. Nano Lett. 2011, 11, 2809−2813. (10) Zheng, H.; Smith, R. K.; Jun, Y. W.; Kisielowski, C.; Dahmen, U.; Alivisatos, A. P. Observation of Single Colloidal Platinum Nanocrystal Growth Trajectories. Science 2009, 324, 1309−1312. (11) Noh, K. W.; Liu, Y.; Sun, L.; Dillon, S. J. Challenges Associated with in-Situ TEM in Environmental Systems: the Case of Silver in Aqueous Solutions. Ultramicroscopy 116, 34−38. (12) Lee, J.; Urban, A.; Li, X.; Su, D.; Hautier, G.; Ceder, G. Unlocking the Potential of Cation-Disordered Oxides for Rechargeable Lithium Batteries. Science 2014, 343, 519−522. (13) Bresin, M.; Nadimpally, B. R.; Nehru, N.; Singh, V. P.; Hastings, J. T. Site-Specific Growth of CdS Nanostructures. Nanotechnology 2013, 24, 505305. (14) Zheng, H.; Claridge, S. A.; Minor, A. M.; Alivisatos, A. P.; Dahmen, U. Nanocrystal Diffusion in a Liquid Thin Film Observed by in Situ Transmission Electron Microscopy. Nano Lett. 2009, 9, 2460− 2465. (15) Ring, E. A.; de Jonge, N. Microfluidic System for Transmission Electron Microscopy. Microsc. Microanal. 2010, 16, 622−629. (16) Grogan, J. M.; Rotkina, L.; Bau, H. H. In Situ Liquid-Cell Electron Microscopy of Colloid Aggregation and Growth Dynamics. Phys. Rev. E 2011, 83, 061405. (17) Li, D.; Nielsen, M. H.; Lee, J. R. I.; Frandsen, C.; Banfield, J. F.; De Yoreo, J. J. Direction-Specific Interactions Control Crystal Growth by Oriented Attachment. Science 2012, 336, 1014−1018. (18) White, E. R.; Mecklenburg, M.; Shevitski, B.; Singer, S. B.; Regan, B. C. Charged Nanoparticle Dynamics in Water Induced by Scanning Transmission Electron Microscopy. Langmuir 2012, 28, 3695−3698. (19) Park, J.; Zheng, H.; Lee, W. C.; Geissler, P. L.; Rabani, E.; Alivisatos, A. P. Direct Observation of Nanoparticle Superlattice Formation by Using Liquid Cell Transmission Electron Microscopy. ACS Nano 2012, 6, 2078−2085. (20) Mirsaidov, U.; Ohl, C.-D.; Matsudaira, P. A Direct Observation of Nanometer-Size Void Dynamics in an Ultra-Thin Water Film. Soft Matter 2012, 8, 7108−7111. (21) Jonge, N. D.; Peckys, D. B.; Kremers, G. J.; Piston, D. W. Electron Microscopy of Whole Cells in Liquid with Nanometer Resolution. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 2159−2164. (22) Peckys, D. B.; Veith, G. M.; Joy, D. C.; de Jonge, N. Nanoscale Imaging of Whole Cells Using a Liquid Enclosure and a Scanning Transmission Electron Microscope. PloS one 2009, 4, e8214. (23) Dukes, M. J.; Peckys, D. B.; de Jonge, N. Correlative Fluorescence Microscopy and Scanning Transmission Electron Microscopy of Quantum-Dot-Labeled Proteins in Whole Cells in Liquid. ACS Nano 2010, 4, 4110−4116. (24) Pastina, B.; LaVerne, J. A. Effect of Molecular Hydrogen on Hydrogen Peroxide in Water Radiolysis. J. Phys. Chem. A 2001, 105, 9316−9322. (25) Elliot, A. J.; McCracken, D. R. Computer Modelling of the Radiolysis in an Aqueous Lithium Salt Blanket: Suppression of Radiolysis by Addition of Hydrogen. Fusion Eng. Des. 1990, 13, 21−27. (26) Joseph, J. M.; Choi, B. S.; Yakabuskie, P.; Wren, J. C. A Combined Experimental and Model Analysis on the Effect of pH and O2(aq) on Γ-Radiolytically Produced H2 and H2O2. Radiat. Phys. Chem. 2008, 77, 1009−1020. (27) Spinks, J. T.; Woods, R. J. An Introduction to Radiation Chemistry; Wiley: New York, 1990. (28) Allen, A. O. The Radiation Chemistry of Water and Aqueous Solutions; Van Nostrand: New York, 1961. (29) Draganic, I. G.; Draganic, Z. D. The Radiation Chemistry of Water; Academic Press: New York, 1971. (30) Schwarz, H. A. Applications of the Spur Diffusion Model to the Radiation Chemistry of Aqueous Solutions. J. Phys. Chem. 1969, 73, 1928−1937. (31) Buxton, G. V.; Greenstock, C. L.; Helman, P. W.; Ross, A. B. Critical Review of Rate Constants for Reactions of Hydrated Electrons,

METHODS Gold nanorod experiments were carried out with our custommade liquid cell, the nanoaquarium,46 in an FEI Quanta 600 FEG Mark II with a transmission detector. The aqueous solution contained gold nanorods prepared by the Murray group50 with a trace amount of the surfactant cetrimonium bromide (CTAB) and measured pH ∼7. The devices were filled under atmospheric conditions, and the solution was aerated.

■

ASSOCIATED CONTENT

S Supporting Information *

Video of growth and etching of nanorods; kinetic model; Tables S1−S3; Figures S1 and S2. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail [email protected] (F.M.R.). *E-mail [email protected] (H.H.B.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Device fabrication was carried out at the Cornell NanoScale Facility (NSF Grant ECS-0335765), a member of the National Nanotechnology Infrastructure Network. Electron microscopy was performed at the Penn Regional Nanotechnology Facility and the IBM T. J. Watson Research Center with the valuable assistance of Mr. Peter Szczesniak of UPenn and Dr. Mark C. Reuter and Mr. Arthur Ellis of IBM. Gold nanorods were generously provided by Dr. Christopher Murray. The work was supported, in part, by Grants 1129722 and 1066573 from the National Science Foundation.

■

REFERENCES

(1) de Jonge, N.; Ross, F. M. Electron Microscopy of Specimens in Liquid. Nat. Nanotechnol. 2011, 6, 695−704. (2) Grogan, J. M.; Schneider, N. M.; Ross, F. M.; Bau, H. H. The Nanoaquarium: a New Paradigm in Electron Microscopy. J. Indian Inst. Sci. 2012, 92, 295−308. (3) Grogan, J. M.; Schneider, N. M.; Ross, F. M.; Bau, H. H. Bubble and Pattern Formation in Liquid Induced by an Electron Beam. Nano Lett. 2014, 14, 359−364. (4) Williamson, M. J.; Tromp, R. M.; Vereecken, P. M.; Hull, R.; Ross, F. M. Dynamic Microscopy of Nanoscale Cluster Growth at the Solid−Liquid Interface. Nat. Mater. 2003, 2, 532−536. (5) Radisic, A.; Vereecken, P. M.; Hannon, J. B.; Searson, P. C.; Ross, F. M. Quantifying Electrochemical Nucleation and Growth of Nanoscale Clusters Using Real-Time Kinetic Data. Nano Lett. 2006, 6, 238−242. (6) Radisic, A.; Vereecken, P. M.; Searson, P. C.; Ross, F. M. The Morphology and Nucleation Kinetics of Copper Islands During Electrodeposition. Surf. Sci. 2006, 600, 1817−1826. (7) White, E. R.; Singer, S. B.; Augustyn, V.; Hubbard, W. A.; Mecklenburg, M.; Dunn, B.; Regan, B. C. In SituTransmission Electron Microscopy of Lead Dendrites and Lead Ions in Aqueous Solution. ACS Nano 2012, 6, 6308−6317. (8) White, E. R.; Mecklenburg, M.; Singer, S. B.; Aloni, S.; Regan, B. C. Imaging Nanobubbles in Water with Scanning Transmission Electron Microscopy. Appl. Phys. Express 2011, 4, 055201. (9) Evans, J. E.; Jungjohann, K. L.; Browning, N. D.; Arslan, I. Controlled Growth of Nanoparticles From Solution with in Situ 22381

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382

The Journal of Physical Chemistry C

Article

Hydrogen Atoms and Hydroxyl Radicals. J. Phys. Chem. Ref. Data 1988, 17, 513−886. (32) NMSchneider/Radiolysis, https://github.com/NMSchneider/ Radiolysis (accessed June 30, 2014). (33) Hill, M. A.; Smith, F. A. Calculation of Initial and Primary Yields in the Radiolysis of Water. Radiat. Phys. Chem. 1994, 43, 265−280. (34) Christensen, H. Remodeling of the Oxidant Species during Radiolysis of High-Temperature Water in a Pressurized Water Reactor. Nucl. Technol. 1995, 109. (35) Hart, E. J. The Hydrated Electron: Properties and Reactions of This Most Reactive and Elementary of Aqueous Negative Ions Are Discussed. Science 1964, 146, 19−25. (36) Chernyak, A. S.; Zhigunov, V. A.; Shepot’ko, M. L.; Smirnov, G. I.; Dolin, P. I.; Bobrova, A. S.; Khikin, G. I. Precipitation of Gold and Silver From Cyanide Solutions by Hydrated Electrons Generated by Ionizing Radiation. J. Appl. Chem. USSR (Engl. Transl.) 1981, 54, 6. (37) Speight, J. Lange’s Handbook of Chemistry; McGraw-Hill Professional: New York, 2004. (38) Atinault, E.; De Waele, V.; Schmidhammer, U.; Fattahi, M.; Mostafavi, M. Scavenging of and OH Radicals in Concentrated HCl and NaCl Aqueous Solutions. Chem. Phys. Lett. 2008, 460, 461−465. (39) Woehl, T. J.; Evans, J. E.; Arslan, I.; Ristenpart, W. D.; Browning, N. D. Direct in Situ Determination of the Mechanisms Controlling Nanoparticle Nucleation and Growth. ACS Nano 2012, 6, 8599−8610. (40) Alvarez-Puebla, R. A.; dos Santos, D. S., Jr.; Aroca, R. F. SERS Detection of Environmental Pollutants in Humic Acid−Gold Nanoparticle Composite Materials. Analyst 2007, 132, 1210. (41) Park, J.; Kodambaka, S.; Ross, F. M.; Grogan, J. M.; Bau, H. H. In Situ Liquid Cell Transmission Electron Microscopic Observation of Electron Beam Induced Au Crystal Growth in a Solution. Microsc. Microanal. 2012, 18, 1098−1099. (42) den Heijer, M.; Shao, I.; Radisic, A.; Reuter, M. C.; Ross, F. M. Patterned Electrochemical Deposition of Copper Using an Electron Beam. APL Mater. 2014, 2, 022101. (43) Remita, H.; Lampre, I.; Mostafavi, M.; Balanzat, E.; Bouffard, S. Comparative Study of Metal Clusters Induced in Aqueous Solutions by Γ-Rays, Electron or C6+ Ion Beam Irradiation. Radiat. Phys. Chem. 2005, 72, 575−586. (44) Abidi, W.; Remita, H. Gold Based Nanoparticles Generated by Radiolytic and Photolytic Methods. Rec. Pat. Eng. 2010, 4, 170−188. (45) Mullins, W. W.; Sekerka, R. F. Stability of a Planar Interface during Solidification of a Dilute Binary Alloy. J. Appl. Phys. 1964, 35, 444. (46) Grogan, J. M.; Bau, H. H. The Nanoaquarium: a Platform for in Situ Transmission Electron Microscopy in Liquid Media. J. Microelectromech. Syst. 2010, 19, 885−894. (47) Jones, S. Bubble Nucleation From Gas Cavities  a Review. Adv. Colloid Interface Sci. 1999, 80, 27−50. (48) Huang, T.-W.; Liu, S.-Y.; Chuang, Y.-J.; Hsieh, H.-Y.; Tsai, C.Y.; Wu, W.-J.; Tsai, C.-T.; Mirsaidov, U.; Matsudaira, P.; Chang, C.-S.; et al. Dynamics of Hydrogen Nanobubbles in KLH Protein Solution Studied with in Situ Wet-TEM. Soft Matter 2013, 9, 8856−8861. (49) Klein, K. L.; Anderson, I. M.; de Jonge, N. Transmission Electron Microscopy with a Liquid Flow Cell. J. Microsc. 2011, 242, 117−123. (50) Ye, X.; Jin, L.; Caglayan, H.; Chen, J.; Xing, G.; Zheng, C.; Doan-Nguyen, V.; Kang, Y.; Engheta, N.; Kagan, C. R.; et al. Improved Size-Tunable Synthesis of Monodisperse Gold Nanorods Through the Use of Aromatic Additives. ACS Nano 2012, 6, 2804−2817.

22382

dx.doi.org/10.1021/jp507400n | J. Phys. Chem. C 2014, 118, 22373−22382