Thiol Coordination Softens Liquid Metal Particles To Improve On-Demand Conductivity

Abstract

Achieving tunable rupturing of eutectic gallium indium (EGaIn) particles holds great significance in flexible electronic applications, particularly pressure sensors. We tune the mechanosensitivity of EGaIn particles by preparing them in toluene with thiol surfactants and demonstrate an improvement over typical preparations in ethanol. We observe, across multiple length scales, that thiol surfactants and the nonpolar solvent synergistically reduce the applied stress requirements for electromechanical actuation. At the nanoscale, dodecanethiol and propanethiol in toluene suppress gallium oxide growth, as characterized by transmission electron microscopy and X-ray photoelectron spectroscopy. Quantitative AFM imaging produces force–indentation curves and height images, while conductive AFM measures current while probing individual EGaIn particles. As the applied force increases, thiolated particles demonstrate intensified softening, rupturing, and stress-induced electrical activation at forces 40% lower than those for bare particles in ethanol. To confirm that thiolation facilitates rupturing at the macroscale, a laser is used to ablate samples of EGaIn particles. Scanning electron microscopy and resistance measurements across macroscopic samples confirm that thiolated EGaIn particles coalesce to exhibit electrical activation at 0.1 W. Particles prepared in ethanol, however, require 3 times higher laser power to demonstrate a similar behavior. This unique collection of advanced techniques demonstrates that our particle synthesis conditions can facilitate on-demand functionality to benefit electronic applications.

Gallium alloys, such as eutectic gallium indium (EGaIn, Ga = 85.8% and In = 14.2%), are a special class of metal that is liquid around room temperature.¹ EGaIn possesses a unique combination of low viscosity, good biocompatibility, high deformability, and high electrical/thermal conductivity.¹⁻⁶ Due to these properties, EGaIn has broad applications in drug delivery, wearable electronics, and microfluidics.^1,3,7−11 Bulk liquid EGaIn largely behaves like a yield-stress fluid.¹¹ The yield threshold is governed by the presence of a thin, elastic gallium oxide skin surrounding a Newtonian liquid metal core.¹¹ The formation of the surface oxide has been modified to inhibit the flow of macroscopic EGaIn or act as an insulator before stress-induced electrical actuation.¹²

The relevance of EGaIn expands significantly when it is in particle form, facilitating numerous applications. For example, EGaIn colloidal suspensions can act as vehicles for surface-functionalized drug molecules or for printing electrically conductive circuits.^9,13 Embedded liquid metal particles in elastomers have especially helped advance flexible electronics.^12,14,15 However, the particle diameter is an important parameter to reduce since larger microparticle networks show lower dielectric breakdown strength.^12,14,15 Instead, EGaIn nanoparticles inside a polymer composite still preserve the elasticity of the polymer while enhancing the electrical permittivity.¹⁶⁻¹⁹ Similarly, printed circuits of EGaIn nanoparticles on flexible substrates have also demonstrated impressive dielectric and elastic properties for wearable pressure sensors.²⁰⁻²² Rigid metal nanoparticles, such as gold and silver, have been commonly used when printing conductive inks for high-resolution printing applications. However, when strained, these unyielding designs show high electrical resistance when inside elastomers or on stretchable material.^23,24 Unlike solid metal-embedded elastomers, those with EGaIn particles can tolerate higher strain because particle coalescence maintains the electrical continuity.

To make EGaIn particles, the liquid metal is dispersed in an immiscible fluid. This mixture does not make emulsion droplets, and this is due to the propensity of gallium to oxidize. When EGaIn particles are exposed to an oxidizing agent, such as an ambient atmosphere or a polar solvent, such as water, the particles spontaneously grow a stiff and electrically insulating oxide skin. This oxide provides structural stability for the liquid metal core and its thickness has been measured to be around 4 nm.^3,25,26 This oxide skin or shell presents both opportunities and challenges. However, to receive an electrical signal from any device made of EGaIn particles, the liquid metal particles need to coalesce. Therefore, control of this oxide skin and the ability to rupture it “on demand” is of the utmost importance.

The surface oxidation growth can be temporarily suppressed if particles are generated under an inert atmosphere, but this is not always practical. Instead, particles are generally prepared in the polar solvent ethanol, and different dispersing surfactants are used to control the oxide shell and impart stability against coalescence or particle aggregation.^25,26 Silanes, phenols, and nonionic molecules are all able to stabilize the oxide layer and the EGaIn particles themselves.^3,13,27

Less commonly, nonpolar solvents doped with thiols or ionic surfactants have also accomplished colloidal stabilization of EGaIn particles prepared in these solvents.^25,28,29 Thiolated ligands, such as 1-dodecanethiol, compete for surface sites alongside the gallium oxidation growth and have been shown to reduce the oxidation thickness of particles prepared in chlorobenzene.^1,25,30 Stability can also be imparted by ionic surfactants when EGaIn particles are prepared in toluene.²⁸ Interestingly, toluene is also used as a solvent when chemically modifying copper particles to inhibit oxidation for colloidal electronic applications.³¹

On-demand coalescence, or electromechanical actuation, occurs when there is sufficient applied stress to deform and rupture the oxide shells in a collection of particles to generate a conductive particle network. Many studies investigate the ability of the oxide shell to rupture by performing experiments on either an ensemble of particles or bulk samples made from EGaIn particles. Mechanical, thermal, or chemical stress is applied to the sample, either by stretching, compression, heating, laser ablation, or exposure to extreme pH.^27,29,32,33 Sufficiently high applied stresses result in successful rupture of the oxide layers within the bulk collection of particles, which is then assessed by electrical measurements. Any degree to which the applied force can be reduced to obtain an electrical signal represents an improvement in the electromechanical response of EGaIn.

The mechanical sensitivity of EGaIn particles formed in alcohol can be improved by coordinating surfactants like silanes and thiols while the particle network is strained or pressure-sintered.^{24,27,34−36} In these same studies, electrical resistance measurements corroborate the bulk mechanical actuation. For instance, the thiolation of the EGaIn particles in ethanol leads to a ∼50% reduction in the pressure needed to rupture a bed or film of the liquid metal capsules.³⁵ After thiolation up to concentrations of 1 mM, Boley and co-workers measure activation forces needed to mechanically sinter a film of EGaIn particles. Thiolation in ethanol leads to smaller particles and results in a lower activation force than anticipated based on control studies of bare particles.²⁴ More recently, stretchable EGaIn particle films prepared by thiolation after acidification in isopropanol have also shown tremendous electrical response at low and high strain percentages.³⁶

Probing the response of individual particles instead of bulk clusters may provide the resolution needed to observe the core–shell mechanical properties. By using atomic force microscopy (AFM), Lear and co-workers uncover that the force required to rupture individual thiolated EGaIn nanoparticles is up to 45% greater than unfunctionalized EGaIn prepared in ethanol.³⁴ Additionally, the rupture force increases linearly with particle diameter. However, this appears to be the only work that connects measurements of individual particle mechanical properties as a result of applied stress to both chemical pretreatment and electrical performance of macroscopic EGaIn under compression. Interestingly, electromechanical actuation can also be accomplished through particle rupture in response to applied strain as an alternative to applied stress.³⁶

Existing literature demonstrates that EGaIn particles can be used to impart bulk conductivity. Every length scale, from the nanoscale surfactant to the macroscale, plays a role in the EGaIn performance. Still, further work is needed to fully connect the particle preparation, suspension chemistry, oxide shell characteristics, individual particle properties, and bulk electromechanical performance of EGaIn. Table 1 in the SI summarizes the contributions of nine articles characterizing different length scales in EGaIn particle suspension preparations, including a few using surfactants and/or nonpolar solvents to prepare particles. Ideally, to formulate a complete picture of the behavior of a given particle preparation, all measurements would be performed on a single preparation. Furthermore, the colloidal stability revealed with toluene suggests that non-standard preparations of EGaIn particles may enhance actuation performance.

In this work, we investigate the impact of interfacial chemistry on both oxide shell thickness and electromechanical actuation of single particles and connect these results to macroscopic conductivity. We examine EGaIn particles prepared in ethanol only, the most common method, and we modify these particles with thiol surfactants. We choose both 1-dodecanethiol, which has been investigated in the literature, and the shorter molecule 1-propanethiol, which has not. We extend this thiol investigation to particles prepared in a nonpolar solvent, toluene, which has been minimally investigated.

Our results demonstrate that surface oxidation is suppressed in the nonpolar solvent, toluene, as compared to ethanol, and that thiolation in toluene even further inhibits the oxide shell. We use AFM to visualize stress softening, a phenomenon that has not been reported previously. This softening intensifies in particles thiolated in toluene. Our conductive AFM data reveals that the force needed to rupture 1 μm single thiolated particles is reduced by 40% compared to standard preparations. Further, the force to rupture micron-scale particles thiolated in toluene is ten times less than is needed to rupture particles prepared in ethanol alone.³⁴ To connect single-particle properties to the macroscale and potential applications, we demonstrate that laser ablation can actuate an electrical signal across a film of EGaIn particles. Thiolation of the particles reduces the laser power required for electrical actuation by threefold. This study reveals that thiolated EGaIn is observed to have a greater susceptibility to electrical actuation across both a single particle and a particle network. This correlative surface chemical and electromechanical study comprehensively addresses their interdependence and may help our understanding for lower-power printing of electronic circuits or acute pressure sensors.

Results and Discussion

Tuning EGaIn Oxide Thickness through Synthesis Conditions

In this study, multiple advanced analytical techniques comprehensively characterize the reduction of gallium oxide through the coordination of thiol surface capping agents. In Figure 1a, to demonstrate the core–shell particle anatomy, atomic resolution transmission electron microscopy (TEM) confirms the presence of both dodecanethiol and propanethiol on the surfaces of particles prepared in toluene. For simplicity, we abbreviate the nomenclature for the five formulations as ethanol, toluene, PPT-toluene, DDT-toluene, and PPT-ethanol, as needed.

Figure 1.

Spectroscopy and electron microscopy characterization of EGaIn particles. (a) Four EGaIn samples investigated by TEM (top row) and HAADF STEM (bottom row). From left to right, the samples imaged are bare EGaIn particles in ethanol and toluene, followed by surfactant coated particles of DDT-toluene and PPT-toluene at 0.5 mM each. The HAADF-STEM images are all inverted and subjected to median filtering and Gaussian blur to facilitate oxide shell identification. All scale bars represent 10 nm. (b) Cartoon illustrating the common gallium oxidation growth when sonicated without coordinated thiols. (c) Cartoon showing the chemistry of a thiolated particle in toluene: the head of the surfactant directly coordinates onto the pure metal, but its body is intercalated within the gallium-dominant oxide layer. (d) To corroborate the oxide thicknesses measured with HAADF STEM, XPS is used to evaluate the relative GaO intensity across the same samples. EGaIn nanoparticles made in the polar solvent (ethanol) show the greatest oxidation. The thiolated gallium in the nonpolar solvent (toluene) showed considerably lower GaO intensity. (e) Presence of both thiol species is then confirmed by the sulfur high-resolution scans. In this entire figure, to minimize the oxidation variability, all the data are collected in triplicate within a 24 h window.

In the TEM images (from left to right on the top row), the presence of thiol coordination is confirmed by the brush-like morphology, which is not present in the samples prepared in toluene or ethanol alone. High-angle annular dark-field scanning TEM (HAADF-STEM) measures the gallium-dominated oxide thickness. In the HAADF-STEM images, brighter regions correspond to higher atomic numbers. The thicker electron clouds from higher atomic numbers increase the scattering angles of transmitted electrons. EGaIn’s core, which contains the highest combination of atomic numbers, is the brightest part of the image: the core scatters electrons at the highest angles. The darkest region in the images corresponds to the carbon background and carbon-based thiol molecules. The medium brightness corresponds to the gallium oxide shell.

From left to right, each HAADF-STEM image in Figure 1a is inverted with median and Gaussian blur filters to facilitate identification of the gallium oxide shell. The 2-D thickness measurements of each gallium oxide shell are annotated in red. The thickness is largest when EGaIn particles are prepared in the polar solvent, ethanol, and measures 4.02 nm. The oxide thickness slightly reduces in the nonpolar solvent, toluene, and measures 3.36 nm. The coordination of both thiols onto EGaIn, when particles are prepared in toluene, generates much thinner oxide skins, measuring 1.49 and 1.18 nm in DDT-toluene and PPT-toluene, respectively. See SI Table 2 for the oxidation thicknesses measured in triplicate. The cartoons in Figure 1b,c show the preparation-dependent changes in the morphology. The oxide shell is thicker without surface capping thiols suppressing further growth. For both propanethiol and dodecanethiol molecules, the head of the surfactant coordinates directly with the pure metal, as in SAM formation.^30,35,39 Thiol headgroups can coordinate with the EGaIn liquid metal directly because the oxide has been removed by acid before sonication, although it is worth noting that not all surfactants can coordinate with EGaIn in this way.³⁰

In Figure 1c, the surfactant tail is intercalated within the gallium-dominated oxide layer, with the oxide layer and thiol headgroups competing for sites on the liquid metal core. If the oxide layer is in place around the particle, it forms a physical barrier preventing coordination of the thiol with the metal. With the oxide layer removed, the thiol can coordinate with the metal. This has been seen in two separate situations. Upon allowing the oxide to form on micron-sized particles, dodecanethiol (DDT) does not coordinate with the metal, as suggested by the greater pressure requirement to impart conductivity.³⁵ However, after the oxide forms, and then is subsequently removed by acid, thiols indeed coordinate with the metal, reducing the pressure requirement to impart conductivity.³⁵ In another study, instead of treating EGaIn particles with acid after sonication, trace amounts of acid are introduced during sonication and the resulting particle films display high conductivity while maintaining a high surface area.³⁶ These two studies suggest that the addition of acid, whether during or after particle formation, is sufficient to remove the acid layer and allow for thiol complexation with the liquid metal itself. Interestingly, our experimental method, in which acid is introduced immediately before sonication, corroborates that the introduction of acid removes the oxide. In our method, care must be taken to maintain the sample in low oxidizing conditions since the acid treatment occurs before particle formation. We do this by purging with argon.

To corroborate oxidation suppression, X-ray photoelectron spectroscopy (XPS) is used to quantify gallium oxide’s relative abundance. The overlay in Figure 1d contains gallium high-resolution scans of the particles prepared in ethanol and toluene, and thiolated EGaIn prepared in toluene as a function of photoelectric binding energy. At 20.5 eV, the normalized gallium oxide or Ga 3d_5/2 peak is 0.85 in ethanol and 0.76 in toluene. With thiol coordination, the gallium oxide peak intensity decreases to 0.67 and 0.7 in DDT-toluene and PPT-toluene, respectively. Finally, to validate the thiol morphology imaged with TEM, high-resolution scans of sulfur show distinct S 2p_3/2 peaks from the headgroup of both thiol molecules (Figure 1d). The normalized peak intensity of the toluene sample displays a near zero signal. This indicates that the sulfur intensity in DDT-toluene and PPT-toluene originates from the thiol head groups.

After STEM and XPS were used to confirm the intended core–shell anatomies, atomic force microscopy (AFM) was used to visualize the morphology of each engineered particle. AFM height images (Figure 2) illustrate the particle integrity of thiolated EGaIn nanoparticles compared to those prepared in ethanol. From left to right, each row shows five height images obtained via the contact-based QI mode with an increasing force series, with F_Applied = 10–200 nN. The vertical color scale indicates the height in each image. All three particles show a maximum height of approximately 1000 nm. The scale bars in white are 300 nm.

Figure 2.

Mechanical response of single EGaIn particles with and without surface thiolation is probed across a force series (10–200 nN). As the AFM tip indents across the surface, the ethanol and thiolated particles demonstrate stress-induced softening. Images of the DDT-toluene and PPT-toluene particles suggest extreme softening and oxide rupturing at much lower forces than that seen in the ethanol sample. All inset scale bars equal 300 nm.

When F_Applied = 10 or 20 nN, all three particles maintain similar spherical geometries. The thiolated particles, made in nonpolar toluene, lose their original shape and exhibit global deformation when F_Applied = 50 and 100 nN. However, the EGaIn particle made in polar ethanol maintains its original roundness. At 200 nN, the image of the PPT-toluene particle indicates that the oxide layer is destabilized and the internal conductive liquid flows outward. At the same force, DDT-toluene also shows significant destabilization and the liquid metal flows outward in a more localized manner. Meanwhile, at 200 nN, the ethanol-based EGaIn particle remains intact, demonstrating only a localized deformation. Since the diameter of the particle made in ethanol remains nearly constant even at the highest applied force, the on-demand performance of these particles appears less responsive to stress.

To supplement the deformation observed in the xy-plane, the vertical color scales show deformation in the z-plane (Figure 2). Across this F_Applied series, the particle made in ethanol does not undergo significant deformation: the relative maximum height at F_Applied = 200 nN is reduced by <70 nm compared to the height measured at F_Applied = 10 nN. In comparison, DDT-toluene and PPT-toluene single particles undergo 282 and 431 nm of deformation, respectively, as F_Applied increases from 10 to 200 nN. To avoid interference from the liquid metal core, F_Applied does not exceed 200 nN.

The single-particle method is responsible for each contact-based image in Figure 2. Each of the 16,384 pixels in each image contains an associated force–indentation curve. SI Figure 1 represents one example image from this five-point force series to showcase the high-throughput nature of the single-particle method. SI Figure 1a,b displays an elastic modulus embedded image in which each pixel represents a raw force indentation curve, for a particle prepared with 1.0 mM propanethiol in toluene, and F_Applied = 50 nN. The center cropping circle in yellow removes the substrate and particle perimeter. SI Figure 1c shows a histogram of the modulus measurements calculated by the software.

Oxide Shell Mechanics Measured by AFM

Figure 3a,b shows schematics of the AFM nanoindentation measurements, in which the AFM probe tip is pushed into the EGaIn particle with applied force F_Applied. Figure 3c shows the force–indentation curve, F(δ), revealing the mechanical response of the particle. The elastic modulus or Young’s Modulus, E, quantifies the relative stiffness of a material and determines the magnitude of the material response F(δ).⁴⁰⁻⁴² Multiple theoretical models are available to describe F(δ) depending on the nature of the material being probed.

Figure 3.

Mechanical response between a pure elastic and a core–shell particle is distinguishable. (a) To calculate the elastic modulus of an elastic particle, the Hertz model is often employed in AFM software. (b) However, there is a more appropriate linear, mechanical model for EGaIn which incorporates a fluid-filled core inside a stiff shell. (c) A 150 nN force curve from a 0.5 mM PPT-toluene EGaIn suspension is overlaid with both the Hertz model and the linear core–shell model.

For uniformly elastic materials, as shown in Figure 3a, the Hertz model is commonly used to measure E from the nanoindentation of a flat plane by a sphere of radius R:^43,44

1

where the Poisson ratio, υ, of EGaIn is estimated to be 0.25,⁴⁵ and R_tip = 10 nm corresponds to the AFM probe tip radius. Eq 1 holds under the assumptions that δ ≪ R_tip and also that the particle radius R ≫ R_tip. Thus, the AFM tip is approximated as a sphere indenting a material that is approximated as a plane. Typically, AFM software provides built-in models, including the Hertz model, to calculate the elastic modulus of a material. Note that, in eq 1, F ∼ δ^3/2.

The raw data from the AFM, seen in Figure 3c, however, show that F ∼ δ beyond contact is located at (0,0). This suggests that the built-in assumption of Hertzian deformation of a uniformly elastic particle is not appropriate. Instead, we apply a linear theory for the indentation of liquid-filled shells to estimate the effective elastic modulus of EGaIn’s outer oxide shell, E_shell.⁴⁰ The core–shell model presented by Taber incorporates shell thickness t:

2

where c is a constant of order O(1) that depends on the ratio of the core thickness to the particle size and has been calculated for a range of core–shell geometries.⁴⁰ The measured shell thickness of the EGaIn particles is on the order of O(1) nm. Given that R for the EGaIn particles is 500 nm, R/t ∼ 500, and thus c = 2.25.⁴⁰ The slope of eq 2 is often referred to as the stiffness k = cE_shellt²/R.^26,34

In Figure 3c, the differences between these two models are graphically clear. Raw data are shown as a solid blue line, representing data taken on a particle from a 0.5 mM PPT-toluene EGaIn suspension with F_Applied = 150 nN. Fits to the two models are plotted as dashed lines. The core–shell particle model, in blue, fits the raw, linear data much better than the uniformly elastic sphere model, in red, which fits poorly. The agreement of the core–shell particle model to the raw data suggests that the particle exhibits a linear elastic response up to nearly δ ∼ 100 nm or 10% of the particle diameter. Some measurements exhibit a nonlinear elastic response beyond the small deformation region (SI Figure 2).

Figure 4 shows measurements of the individual particle-scale mechanical response of particles prepared with and without propanethiol in both ethanol and toluene. Results for particles prepared in toluene with dodecanethiol are shown in SI Figure 3. For each particle preparation measured by both the single-particle and ensemble methods, we fit the linear portion of F(δ) using eq 2 to obtain the slope, k = cE_shellt²/R in N/m. Figure 4a,b,d,e shows aggregated results of the stiffness k, as a function of the applied force on the x-axis, with legends indicating surfactant concentration. As k depends on both E_shell and t, a decrease in the slope could indicate either a decrease in E_shell or a decrease in t, or some combination thereof. However, each data set shown in Figure 4a,b,d,e represents one suspension preparation. Therefore, t is constant within each trace and does not change with F_Applied. We then use the measurements of t to provide direct measurements of E_shell. For particle preparations without surfactant, HAADF-STEM average measurements provide t = 4.04 nm. For particles in toluene and 0.5 mM propanethiol, HAADF-STEM averages show t = 1.23 nm. Literature values similarly suggest t = 1.28 nm for EGaIn particles prepared in ethanol with a similar thiol, 0.5 mM dodecanethiol.²⁶ We use these three values of t to estimate E_shell for particles with 0 and 0.5 mM propanethiol in both ethanol and toluene and plot the results in Figure 4c,f.

Figure 4.

Particle-scale mechanical response shows a decrease in stiffness k and oxide shell modulus E_shell with increasing applied stress and thiolation in toluene. The top and bottom rows show results obtained for particles prepared with propanethiol in ethanol and toluene, respectively. The standard preparation of bare particles in toluene is seen in the black data set in the top row. The plots of k as a function of F_Applied are shown for both the ensemble (a,d) and single-particle (b,e) methods, where k is the slope of the F(δ) curves obtained by AFM. Estimates of E_shell for particles with 0 and 0.5 mM propanethiol are plotted for ethanol (c) and toluene (f).

Both the ensemble and single-particle methods show a significant decrease in stiffness k, the slope of F(δ), as F_Applied increases regardless of the amount of the surfactant. That is, while F(δ) is linear beyond contact, the slope decreases with subsequent measurements obtained at increasingly greater F_Applied (SI Figure 2). This behavior suggests that the oxide shell itself is becoming less stiff: both k and E_shell decrease with F_Applied. In most samples, both k and E_shell decrease by a factor of 2 or more as F_Applied increases from 20 to 200 nN. The decrease in stiffness may be akin to strain softening observed in crystalline materials. Metal oxides in bulk can exhibit strain softening in response to changes in microstructure.⁴⁷ However, since we observe this behavior in response to applied stress rather than applied strain, we refer to it as “stress-softening.” This phenomenon has not before been reported in EGaIn on the microscale nor in gallium oxide shells on the nanoscale.

The mechanical response seen in Figure 4 also shows a dependence on the surfactant concentration. With a concentration as small as 0.25 mM propanethiol, the slope of F(δ) decreases by as much as nearly a factor of 5. This indicates that the presence of surfactant coordinated with the metal EGaIn compromises and softens the oxide shell. However, as mentioned above, this decrease in the slope could be due to some combination of a decrease in E_shell or a decrease in t. Since the slope of F(δ) depends on t², the nearly fourfold decrease in oxide shell thickness would be sufficient to explain an up to a 10-fold decrease in stiffness k. As such, the addition of 0.5 mM propanethiol increases E_shell slightly, by ∼30–50% for all F_Applied, when prepared in ethanol. However, the addition of 0.5 mM propanethiol decreases E_shell for all F_Applied, when the mixture is prepared in toluene. This decrease in E_shell is slight, ∼10–30% when F_Applied < 100 nN. When F_Applied = 200 nN, however, the decrease is significant: E_shell drops an order of magnitude from 35.8 to 2.3 GPa. This decrease reflects the same disruption of the oxide shell and its mechanical integrity seen in the image in Figure 2.

Notably, comparison of the ensemble data in Figure 4c,f indicates that the values of E_shell obtained on particles prepared in toluene, regardless of thiolation, are softer when compared to those of particles prepared in ethanol. Thiolation in toluene further softens E_shell. We suspect there is reduced coordination of thiol with the metal when ethanol is the solvent. The polar nature of ethanol promotes oxide formation, which competes with the thiol in coordinating with the liquid metal surface. The nonpolar toluene suppresses oxide formation, allowing a greater degree of coordination with the thiol. Thiol coordination, in turn, can prevent oxide growth, hence comprising the mechanical strength of the shell. A greater degree of thiolation leads to more compromised mechanics in the particle response, which manifests as a lower stiffness.

Interestingly, the very low stiffness of the thiolated particles matches the surface tension of EGaIn particles when measured via pendant drop: <1 N/m.³³ The comparable stiffness of an EGaIn droplet and a thiolated EGaIn particle suggests that the thiolated EGaIn particle might also be interpreted as having a surface tension. A recent AFM study interprets the on-demand breaking of EGaIn onto self-assembled monolayers (SAMs) of thiols as being due to surface tension mechanics.³³ In this interpretation, stiffness is independent of the particle size. Indeed, when we probe thiolated particles of different sizes, we find k is relatively constant over a range of R from 150 to 500 nm (SI Figure 4), suggesting the possibility of interpreting k as a surface tension response. However, measurements of bare particles prepared in ethanol reveal that k does depend on particle size over the same range of R (SI Figure 4).

Studies of the interfacial rheology of the oxide skin also suggest that the oxide skin itself can be described as having both a viscous and a loss modulus.⁴⁶ The oxide skin has both fluid-like and solid-like properties. In the absence of thiols, the interfacial elasticity dominates surface tension effects by several orders of magnitude.⁴⁶ In our study, we interpret stiffness k in terms of a thin elastic shell due to this dominance of the interfacial elasticity. Further evidence is seen in the AFM images in Figure 2, in which bare particles, when contacted by 200 nN, start to show dimples on their surface. The addition of thiols certainly both softens and thins this shell, as seen by the reduction in stiffness k, approaching the value of pendant drop measurements.³³ However, if thiolation were to cause surface tension effects to dominate the mechanical response of the oxide layer, we might expect the particle to burst like a bubble when punctured. Indeed, the viscosity of the inner, pure liquid EGaIn is only slightly greater than water.¹¹ However, the rupture of thiolated particles with sufficient force occurs locally, with liquid metal flowing out, as seen in thiolated particles contacted by 200 nN (Figure 2). The rupture does not resemble a burst bubble. There are measurable ridges and other features on the surface of the particle that persist long enough to be captured in the ∼20 min required for the AFM scan.

When comparing the mechanical responses of our EGaIn particles to measurements on particle clusters, we find that the stiffness, k, of our particles is comparable to the measured stiffness of particle clusters.²⁶ That is, for F_Applied = 200 nN, on micron-sized particles, we find that k ranges from ∼0.02 to 2 N/m for all particle preparations in ethanol. The stiffness of micron-sized clusters of particles prepared in ethanol, similarly, exhibit stiffnesses ranging from k ∼ 0.3–3 N/m.²⁶ Interestingly, each cluster is made up of particles that are 50 nm in diameter, ∼20× smaller than our particles. Further, the nanoindentation geometry measuring the particle clusters uses a flat punch that is much larger than the particles and also larger than the clusters.²⁶ Nonetheless, our stiffness measurements on the micrometer scale are comparable. Another interesting comparison arises when considering bulk and interfacial rheology performed on the metal-air interface.⁴⁶ Measurements were performed using the du Nouy ring method. Even for bulk samples of EGaIn, either pure or doped with aluminum, the interfacial storage modulus of the thin oxide skin on the surface of the liquid metal is on the order of 1 N/m. Although the rheological characterization of bulk EGaIn in a rheometer suggests a viscoelastic interface, characterized by both storage (elastic response) and loss (viscous response) moduli, the storage modulus dominates over the loss modulus, indicating a solid-like response of the oxide skin.⁴⁶

The surfactant concentration dependence seen in Figure 4 suggests that there may be an optimal amount of surfactant to tune the mechanical properties of EGaIn nanoparticles. With propanethiol added to EGaIn nanoparticle suspensions in either ethanol or toluene, 0.25 and 0.5 mM tend to soften the mechanical response of the particles. When the surfactant is added at 1 mM, however, the mechanical response stiffens slightly. That is, the green (1 mM) data sets in Figure 4a,b,d,e generally have magnitudes slightly larger than those of the pink (0.25 mM) and blue (0.5 mM) traces. While the increase at higher concentrations is slight, the literature suggests it is possible excess surfactant may interfere with the ability of the surfactant to soften EGaIn particles. A somewhat similar effect is observed in measurements of the bulk compression force necessary to rupture a collection of thiolated EGaIn particles, and more force is needed for rupture when the thiol concentration exceeds 3 or 5 mM as compared to thiol concentrations below 1 mM.²⁴

Nanoscopic Conductivity: Improved Oxide Shell Penetration

To investigate the on-demand electrical functionality of the ethanol control compared to the nonpolar, thiolated EGaIn particles, we performed conductive AFM (CAFM), as shown in the schematic in Figure 5a. When probing the center of individual particles with a platinum and iridium-coated tip, an electrical response can be measured using a transimpedance amplifier once the oxide skin is penetrated. The CAFM instrument simultaneously measures force–indentation curves and current I as a function of indentation δ, with picoampere resolution. The force–indentation curves are used to identify the point of contact at δ = 0 nm. Positive values of δ indicate that the AFM tip has started indenting the particle. Figure 5b,c shows the simultaneous measurements of I. The two traces shown in each plot correspond to the two values of F_Applied, both immediately before and after the oxide shell is broken. The measured current is normalized to its maximum value, I_max so that the baseline current (in red) and the current spike upon penetration (in blue) can be seen simultaneously. Comparing Figure 5b,c shows the effect of thiolation in a nonpolar solvent. In Figure 5b,c, the spike in current occurs at F_Applied = 250 nN and F_Applied = 200 nN for particles prepared in ethanol and toluene without a surfactant, respectively. In Figure 5c, however, with 0.5 mM dodecanethiol in toluene, the current spike occurs at a lower force, F_Applied = 150 nN. This electrical response indicates that thiolated EGaIn is easier to penetrate. In short, we find a very nicely supported sequence of ‘best preparation’ when electromechanical actuation is needed: DDT-toluene > toluene > ethanol. Figure 5 demonstrates this on the particle level, and Figure 6 shows the bulk. The nonpolar solvent and thiolation have a synergistic effect in improving the mechanical response of EGaIn microparticles.

Figure 5.

Conductive AFM (CAFM) is utilized to better understand the on-demand functionality at the nanoscale. (a) There is a measurable current when the platinum–iridium (Pt–Ir)-coated AFM probe penetrates the oxide shell and then makes contact with the conductive liquid core. CAFM helped identify the threshold needed to penetrate the oxide of the thiolated and ethanol-only EGaIn particles. (d) At 150 nN, 0.5 mM DDT-toluene requires a lower force to penetrate and generate a current spike compared to the 200 and 250 nN needed to penetrate without thiolation (b,c). In all samples (b–d), the red curve indicates the baseline current (I_o) before the oxide layer is assumed to be punctured. To better connect these measurements with XPS and STEM, this actuation event is also collected within a 24 h window and observed in triplicate.

Figure 6.

To demonstrate the on-demand electromechanical performance on the macroscale, a laser power sequence is used to study the threshold before oxide ablation. (a) Schematic shows the sample layout during the laser power sequence. The whole rectangle contains various shades of gray. The darkest shade represents “as is” EGaIn particles without laser exposure. The nine rectangles outlined in red represent the regions of the sample ablated by the laser. Each is labeled with laser power. The laser beam sublimates the oxide shell, which turns the particles’ gray appearance closer to a chrome color with each higher power. Elemental mapping confirmed the presence of EGaIn after ablation. The 0.3 W laser beam established a complete circuit across the connected letters of “EGaIn” in series to batteries and a blue LED light. (b) SEM micrographs are collected to observe the morphological differences across all power values. The thiolated EGaIn shows coalescence at the lowest laser power, while the particles made in ethanol or toluene demonstrate this behavior only at the highest laser power. From the left to right, ethanol (top row), toluene (middle row), and DDT-toluene (bottom row) are compared across the power sequence. All scale bars are equal to 1000 nm. (c) Multimeter probes are used to measure the relative resistance immediately after laser exposure. The power sequence shows that thiolation reduces resistance values by magnitude when compared to EGaIn from just ethanol or toluene. To better connect these measurements with STEM, XPS, and CAFM, this laser power sequence is also collected within a 24 h window and observed in triplicate.

Macroscopic Conductivity: Thiolation Unlocks Lower Powers for Actuation

In real-world applications, any electromechanical performance of EGaIn is activated at the macroscale. In an effort to bridge this fundamental, particle-scale research with bulk properties of EGaIn, we dropcast particle suspensions onto a silicon wafer and investigate the ability of a laser to break the particles’ oxide shells to create a conductive platform.

Figure 6 shows the results of this investigation. Figure 6a displays a schematic of the coated silicon wafer. A high-powered laser is applied to each of 9 regions on the wafer, each indicated by red boxes in 6a. A laser power sequence is performed at 0.1, 0.2, and 0.3 W, with three iterations at each power. After each laser pulse is completed, it is important to confirm that the EGaIn particles are not ejected off the substrate. In Figure 6a, confirmation of EGaIn’s even surface coverage is evaluated using energy-dispersive spectrometry (EDS) mapping and high-resolution scanning electron microscopy (SEM). As shown in the Ga (pink) and In (yellow) element maps, the EGaIn particles remain fixed to the substrate even at 0.3 W (Figure 6a).

Figure 6b shows SEM micrographs taken shortly after the laser power sequence on EGaIn suspensions prepared in ethanol (top row) and 0.5 mM DDT-toluene (bottom row). Laser power increases from left to right in the images. The heat of the laser disrupts the oxide shell, ultimately vaporizing the oxide and causing particle sintering. At the lowest power used, 0.1 W, particles prepared in ethanol and toluene demonstrate only minor surface ablation. At the same laser power, thiolated EGaIn particles already exhibit regions of coalescence. In the images of particles prepared in ethanol and toluene, important morphological changes are not seen until the laser power reaches 0.2 W. Thiolation facilitates the destruction of EGaIn’s oxide across all laser powers (Figure 6b).

To corroborate the SEM observations, multimeter probes are used to measure the relative resistance immediately after each laser exposure (Figure 6c). Without laser ablation, the layer of EGaIn has a resistance Ω on the order of 10⁸ ohms. With the application of the laser, EGaIn prepared in DDT-toluene demonstrates superior electrical actuation: Ω drops by nearly 7 orders of magnitude at the lowest laser power used, 0.1 W, and drops slightly further when the power is 0.2 and 0.3 W. In stark contrast, for EGaIn prepared in ethanol and toluene, Ω drops by a few orders of magnitude when the laser power is at 0.1 and 0.2 W, but the sample does not become electrically functional until the laser power increases to 0.3 W (Figure 6c). However, at 0.2 W, there is a notable performance improvement in the toluene sample over the ethanol sample. Regardless, neither nonthiolated EGaIn samples match DDT-toluene’s electrical functionality until the laser power increases to 0.3 W. Given the log scale of the y-axis, the results suggest that toluene alone performs better than ethanol, while the addition of DDT-toluene greatly outperforms both bare particle preparations.

Conclusions

The surface chemistry of EGaIn particles plays an important, tunable role in targeting specific mechanical and electrical outcomes. Understanding the conditions needed to obtain electromechanical tunability can help in the development of reliable electronic devices. Our results show that controlling the gallium oxide interface with surface active molecules reduces the mechanical stress threshold for the electrical actuation of EGaIn particles. In particular, surface coordination of thiol, when EGaIn particles are prepared in toluene, thins and softens the oxide shell, improving its electromechanical response. EGaIn behavior that is softer and becoming electrically functional with ease may facilitate further surface chemical investigations. Before EGaIn particles can be deposited into circuit patterns, suspensions must demonstrate reliable flowability when confined. Therefore, softening the EGaIn surfaces could improve the transport of particles for inkjet printing applications. Maintaining reliable transport throughout liquid metal printing may accelerate the incorporation of liquid metals into the flexible electronics market. Ultimately, by gaining a better understanding of EGaIn surfaces, we can better tailor electromechanical performances with the appropriate applications and, in doing so, justify the use of EGaIn in a broader array of engineering systems.

Materials and Methods

Materials

EGaIn (75.5% Ga/24.5% In, PubChem SID 24 872 973) and surfactants 1-dodecanethiol (Pubchem CID 8195) and 1-propanethiol (Pubchem CID 7848) are all obtained from Sigma-Aldrich. Toluene (Pubchem CID 1140) and pure ethyl alcohol are obtained from Electron Microscopy Sciences. Trichloro(octadecyl)silane and trichloroethylene are obtained from Sigma-Aldrich. A multiprobe cantilever with spring constants of 0.2, 2.7, 7.4, and 40 N/m is fitted on the AFM system (All-In-One-Al, BudgetSensors); however, only the probe with 2.7 N/m is used to probe the EGaIn particles. Stock solutions of surfactants are prepared in 30 mL of the suspending fluid at concentrations 0, 0.25, 0.5, and 1.0 mM.

Methods

Preparation of EGaIn Suspensions

EGaIn nanoparticles are prepared in suspensions of either ethanol or toluene with or without surfactant. For all suspensions, 50 μL of EGaIn is dispensed in 3 mL of the stock solution. As received, EGaIn contains heavily oxidized gallium, and we break this down in two steps. First, the gallium oxide across the bulk EGaIn droplet is removed by using hydrochloric acid (HCl). Approximately 50 μL of 37% HCl is added, mixed, and then removed by pipet. Strong acids are commonly used to separate the native oxide from EGaIn.¹¹ Immediately after removal of the acid from the droplet, 3 mL of ethanol or toluene is added to the same vial, now being purged by argon to discourage the reformation of the oxide. The second time the oxide is disturbed is during ultraprobe sonication while purging in argon.

Our nanoparticles are formed using a QSonica Q55 probe sonicator at an amplitude of 30% for 30 min, with 10 s of rest each minute to dissipate the heat generated from sonication. A sonication time of 30 min is chosen to obtain an average particle diameter of approximately 898 nm (SI Figure 5). During characterization, this facilitates the identification of our intended particle size, which is 1 μm, because the biggest particles are easily distinguishable. Particle size varies inversely with sonication time: decreasing or increasing the sonication time increases or decreases the largest particle diameter above or below 1 μm, respectively.

To ensure spherical particles, all suspensions are staked in an ice bath.²³ To reduce the oxygen at EGaIn’s surface, 10 min before and during sonication, all samples are subjected to argon purging in a sealed box (Figure 7). After suspension preparation, a centrifugation step ensures the removal of excess surfactant. All samples undergo 500 relative centrifugal force (RCF) for 3 min three times. After each interval, the supernatant is removed and replenished to a total of 3 mL with the neat solvent that was used in the corresponding stock solution. This centrifugation is carried out for all characterization techniques.

Figure 7.

EGaIn particles are commonly fabricated using an ultraprobe sonicator. As the liquid metal breaks down into smaller particles, the growth of the surface oxide provides some stability.

(S)TEM Analysis

Before imaging, 50 μL of each EGaIn suspension is diluted approximately 50× with the corresponding solvent (toluene or ethanol). Both solvents used are volatile and are therefore sufficient for an evaporative process. Approximately 30 μL of this diluted suspension is dispensed onto Lacey carbon 200 mesh copper TEM grids. At this concentration, EGaIn nanoparticles maintain an acceptable isolation during imaging. The evaporation occurs in minutes, and this technique is called dropcasting. The core–shell structure and morphology of nanoparticles are imaged at 200 kV using a high-angle annular dark field scanning transmission electron microscope (HAADF-STEM) on the ARM200F STEM (JEOL, Akishima, Tokyo, Japan). In the HAADF STEM mode, the resolution at 200 kV is 78 picometers, and 8C is the probe size used. All HAADF-STEM images are inverted, followed by 2 pixels of median filtering and then 2 sigma of Gaussian blur filtering to help identify EGaIn’s oxide shell. Median filtering and Gaussian blurring are the commonly used techniques in presenting TEM images.^37,38

SEM with EDS

Before imaging, 100 μL from each 3 mL EGaIn suspension is dispensed onto half of a 3 in. silicon wafer. For imaging after laser ablation, no additional dilution or centrifugation is employed so that the silicon wafer is not exposed. EGaIn particles are then imaged in high vacuum mode using the secondary electron detector on the FlexSEM TM-1000 II (Hitachi, Tokyo, Japan). An accelerating voltage of 4 kV offers good resolution while maintaining surface information. The spot intensity is set to 30 (unitless scale from 1 to 100) and approximately 6 mm as the imaging working distance. Due to their small sizes (≤100 nm), a field emission scanning electron microscope provides higher resolution images (Helios FIB-SEM, Thermo Fisher Scientific. Hillsboro, OR USA). The electron beam is set to 3 kV and 3 nA with a 5 mm working distance. The high vacuum allows us to use the secondary electron through the lens detector (TLD) while in immersion mode (high-resolution mode).

X-ray Photoelectron Spectroscopy

X-ray photoelectron spectrometry (XPS, Thermo Fisher Scientific K-Alpha, Hillsboro, OR, USA) is used to quantify the relative abundance of gallium (Ga³⁺), gallium oxide (GaO), and sulfur to demonstrate evidence of the thiol. To prepare the samples, 100 μL from each 3 mL EGaIn suspension is dispensed onto one-quarter of a 3 in. silicon wafer. A survey spectrum is collected at a pass energy of 200 eV from binding energies of 0–1350 eV. The dwell time is set to 10 ms with 5 total scans. High-resolution scans of gallium (Ga 2p3/2) and sulfur (S 2p3/2) photoelectron peaks are collected using a pass energy of 50 eV. The dwell time is set to 25 ms with 10 total scans. In the range of binding energies 1114–1122 eV, the intensity ratio of elemental gallium, located at 1116.7 eV, and the native oxide, located at 1118.7 eV, are used to determine the impact of sample preparation on EGaIn’s outer oxidation layer.

Atomic Force Microscopy

We use atomic force microscopy (AFM) to generate force–indentation curves while simultaneously imaging each particle’s morphology after forces are applied. Two different AFMs are used in this study. The first probes the center of a single particle (Cypher AFM, Asylum Research, Oxford Instruments). To gather statistics, we measure an ensemble of particles, ranging from 3 to 6 particles, for each condition studied. We refer to this collection of measurements as the “ensemble” method. The second AFM uses Quantitative Imaging (QI) mode to measure the mechanical response across an entire EGaIn particle (JPK NanoWizard 4 XP, Bruker). In this method, the AFM tip performs indentations while scanning across the surface of the particle, measuring thousands of force–distance curves, and thus providing high-throughput force–indentation data. Statistics using this method are therefore gathered on a single particle, and our collections include a range from 25 to 46 indentations per particle. As such, we refer to this collection method as the “single-particle” method. Before collecting ensemble or single-particle-based data, 50 μL from each 3 mL EGaIn suspension is diluted approximately 50× with the corresponding solvent (toluene or ethanol). Approximately 50 μL of this dilution is dispensed and evaporated on one-quarter of a 3-in. silicon wafer. At these concentrations and volumes, identifying individual particles for contact-based imaging is easier.

Beforehand, to reduce adhesion between surfaces, all AFM cantilevers are ozone-treated for 20 min and then soaked in a 3 mM solution of octadecyltrichlorosilane in trichloroethylene for 48 h. Toluene is used to clean all probes after the soak. For all applied forces F_Applied, force curves F are obtained as a function of indentation δ. The following values of F_Applied are used for both the ensemble and single-particle methods: 10, 20, 50, 100, and 200 nN. In the single-particle method, a 128 × 128 pixel image is collected for each particle, thus containing 16,384 force indentation curves. The image is cropped around the center of the particle, removing the substrate and the particle edges. In the ensemble method, F(δ) curves are collected on 5–10 particles per sample condition using at least the five values of F_Applied previously mentioned.

Conductive Atomic Force Microscopy

Nanoscopic electromechanical measurements are carried out using conductive atomic force microscopy (CAFM). A traditional AFM is used (Cypher AFM, Asylum Research, Oxford Instruments) but with a specialized cantilever and platinum–iridium-coated AFM probe to perform conductivity measurements (Dual Gain ORCA Cantilever Holder). This additional AFM tool uses current to measure localized changes in conductivity and is completed simultaneously in contact mode. The only sample preparation difference from the previous AFM section is that the exposed substrate is sputtered with approximately 40 nm of gold to close the circuit with the probe if the sample itself is also conductive. In this way, CAFM measures both whether a sample is conductive and its conductivity. The sensitivity of the first and second current sensors are set to 1 nA/V and 1 μA/V, respectively. The sample voltage is set to 200 mV, the amplitude is set to 2.0 V, and the frequency is set to 0.99 Hz.

Laser Ablation Conditions

Ethanol or toluene EGaIn suspensions are dropcast onto a 3 in. silicon wafer. A laser power sequence is conducted in increments of 0.01 W from 0.02 to 0.1 W using the ProtoLaser U4 (LPKF. Garbsen, Germany). The laser is set to a frequency of 100 kHz, a mark speed of 400 mm/s, and a hatching grid of 1 μm with 20% hatching contour overlap. Each power increment separately ablates an approximately 2 × 5 mm region. A multimeter with sharp probes measures the resistance (Ω) across all the evenly ablated regions of each evaporated suspension.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.4c01988.

• Comparative data summarizing literature measurements on EGaIn particle actuation to contextualize this study; oxidation thickness measurements for each sample using TEM; example of the high-throughput nature of the single-particle AFM measurement method; AFM force indentation curves for the full range of F_Applied on both bare and thiolated EGaIn particles; DDT-toluene EGaIn particle stiffness and elastic modulus calculations; size dependence of AFM stiffness measurements for both bare and thiolated EGaIn particles; and particle size distribution data as a function of sonication time and sample preparation, as obtained by dynamic light scattering (PDF)

The authors declare no competing financial interest.