Actively Triggerable Metals via Liquid Metal Embrittlement for Biomedical Applications

Actively triggerable materials, which break down upon introduction of an exogenous stimulus, enable precise control over the lifetime of biomedical technologies, as well as adaptation to unforeseen circumstances, such as changes to an established treatment plan. Yet, most actively triggerable materials are low‐strength polymers and hydrogels with limited long‐term durability. By contrast, metals possess advantageous functional properties, including high mechanical strength and conductivity, that are desirable across several applications within biomedicine. To realize actively triggerable metals, a mechanism called liquid metal embrittlement is leveraged, in which certain liquid metals penetrate the grain boundaries of certain solid metals and cause them to dramatically weaken or disintegrate. In this work, it is demonstrated that eutectic gallium indium (EGaIn), a biocompatible alloy of gallium, can be formulated to reproducibly trigger the breakdown of aluminum within different physiologically relevant environments. The breakdown behavior of aluminum after triggering can further be readily controlled by manipulating its grain structure. Finally, three possible use cases of biomedical devices constructed from actively triggerable metals are demonstrated.


Introduction
Devices that exhibit robust mechanical properties during their functional lifetimes while being able to break down at end-oflife are highly desired for various biomedical applications, including long-term drug delivery and sensor-integrated health monitoring. Such technologies can be realized using triggerable materials, which break down in response to external stimuli. [1][2][3][4][5][6][7] Compared to passively triggerable materials, which react with their environment to break down via mechanisms like hydrolysis or oxidation, actively triggerable materials break down upon introduction of an exogenous stimulus ( Figure 1A). [1] Actively triggerable materials therefore enable biomedical technologies to be adaptive and predictable, which both become especially important and challenging as functional lifetime increases.
To date, most actively triggerable materials are polymeric, [1][2][3][4][5][6][7][8] incorporating responsive chemical interactions into their designs using the toolkit of synthetic chemistry. However, functional properties necessary for specific biomedical applications are particularly challenging to program into responsive soft materials, including mechanical toughness and durability, which are important for interfacing with regions of the body that experience high mechanical loads like the gastrointestinal tract (e.g., biting and chewing forces, peristalsis); and high electrical and thermal conductivity, which are important properties for bioelectronics. Metals, by contrast, tend to be superior to polymeric materials in these respects, and therefore are currently used clinically where such properties are required. [9,10] However, responsive metals to date are either passively degradable under highly specific environmental conditions [10] or require significant electrical power to shift their electrochemical potential to drive dissolution. [11] To create actively triggerable metals, we hypothesized that biocompatible liquid metals could serve as exogenous stimuli to initiate the on-demand breakdown of solid metals via liquid metal embrittlement ( Figure 1B). In this mechanism, liquid metals penetrate the grain boundaries of certain solid metals, causing a dramatic reduction in fracture strength as cohesion between grains become destabilized by the presence of the liquid. [12] For biomedical applications, eutectic gallium indium (EGaIn) is an ideal stimulus because it is liquid at room temperature and has been explored extensively in biointerfacing applications like skin-integrated electronics and drug delivery, possessing demonstrated biocompatibility in various formats. [13][14][15][16][17][18] Gallium alloys like EGaIn are known to severely embrittle aluminum, which can be used to construct biomedical structures and devices that withstand significant mechanical loads during their functional lifetime while retaining the ability to be eliminated non-invasively at end-of-life ( Figure 1C). Aluminum is also regularly ingested from food consumption at an average rate of 10 mg day −1 , though normal consumption can reach levels as high as 1 g day −1 since aluminum is a common component of pharmaceutical products. [19,20] In this work, we demonstrate that EGaIn can be formulated to enable triggerable breakdown of aluminum under various physiologically relevant conditions ( Figure 1D). Additionally, we show that the breakdown behavior of aluminum after triggering can be readily tuned by manipulating its morphology using well-established metal processing techniques ( Figure 1E). We conclude by demonstrating three possible use cases of actively triggerable metals in biomedical applications that necessitate high mechanical strength during their functional lifetime combined with on-demand breakdown at end-of-life.

Initiating Embrittlement with Direct Contact
The high surface tension of bare EGaIn poses a challenge for reproducible triggering of aluminum, as it prevents EGaIn from establishing adhesive contact with solid metal surfaces. In air Adv. Mater. 2023, 35, 2208227 Figure 1. Actively triggerable metals for biomedical applications. A) Proposed lifecycle of a biomedical device built with actively triggerable metal components, which enable technologies to be endowed with advantageous properties like high mechanical strength during their functional lifetime (1). Upon addition of an exogenous stimulus (2), rapid breakdown of the device is initiated (3). B) Mechanism of liquid metal embrittlement that can be leveraged to create actively triggerable metals. C) To realize possible biomedical applications of actively triggerable metals, D) different liquid metal formulations enable reproducible metal triggering under different environmental conditions. E) The breakdown behavior of the triggered metal can also be controlled by manipulating metal morphology. © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH or polar solvents like ethanol, EGaIn oxidizes to form a 1-5 nmthick self-limiting layer of gallium oxide (Ga 2 O 3 ) on its surface, which lowers its surface tension significantly and improves its wettability. [17,21,22] We found that mechanical agitation of EGaIn lowered its surface tension (Figure 2A) and converted the liquid metal into a paste that could be painted and adhered onto aluminum in air with high precision ( Figure 2B,C). X-ray photoelectron spectroscopy (XPS) of the Ga2p transition confirmed that agitation increases the relative amount of gallium coordinated to oxygen as Ga 2 O 3 rather than coordinated to the bulk EGaIn ( Figure 2D). Agitation therefore can break through existing oxide layers to expose new EGaIn surfaces to air, increasing the overall ratio of Ga 2 O 3 to EGaIn. Upon wetting, embrittlement was observed despite the presence of oxides on both the EGaIn and aluminum surfaces, which could indicate that the oxide layers are thin enough for electrons from the bulk EGaIn to tunnel through. [23] Using the paintable formulation, the impact of EGaIn exposure time on fracture behavior was measured with tensile elongation testing of pure aluminum sheets. As exposure time  Oxidized EGaIn formulations for direct triggering of aluminum. A) While the surface tension of EGaIn is normally too high to allow wetting (inset), mechanical agitation converts the liquid metal into a smearable, and adhesive paste that can be adhered onto B) curvilinear and vertical surfaces and C) applied with high precision. D) XPS of the Ga2p transition of EGaIn before (dark green) and after (light green) mechanical agitation illustrates that agitation increases the relative amount of Ga coordinated to gallium oxide as Ga(III). E) Representative tensile elongation curves (left) and maximum tensile force (right, n = 3) of aluminum decrease as a function of exposure time to oxidized EGaIn. F) Fracture kinetics of preloaded aluminum increases as a function of relative humidity. Representative curve (left) and average time to reach 10% of initial force (right, n = 3). G) Proposed presence of interfacial water inhibiting triggered embrittlement of aluminum by EGaIn in neutral pH water. H) Prototype of a foam-based applicator for oxidized EGaIn to enable I) sufficient application of pressure to initiate embrittlement underwater.
increased, the maximum tensile force that could be withstood before fracture decreased, with aluminum samples becoming effectively incapable of sustaining any tensile load after 1 h ( Figure 2E). Additionally, relative humidity was found to impact embrittlement kinetics. Pure aluminum sheets were pre-loaded in the normal direction to 10 N-within their linear elastic range-then contacted on the opposite side with the paintable EGaIn after 5 min. As relative humidity increased from 17% to 85%, the time for pre-loaded samples to fracture after triggering decreased by over an order of magnitude ( Figure 2F). This is because, in addition to embrittling aluminum, EGaIn also prevents exposed aluminum grains from forming the passivating aluminum oxide layer that is normally present at the surface, and the de-passivated aluminum reacts with water in the following reactions: [24][25][26] ( ) Thus, EGaIn not only triggers aluminum breakdown via fracture, but also via an accelerated dissolution process when water is present.
While embrittlement could be reproducibly initiated with the paintable formulation in air, no observable interaction occurred when the formulation was placed onto aluminum submerged in neutral saline solution, which is representative of many physiological environments. This can likely be attributed to the presence of an interfacial water barrier between the oxidized EGaIn and the aluminum surface ( Figure 2G), [27] analogous to how interfacial water renders many adhesion chemistries untenable underwater. [28] Accordingly, the interfacial water layer could be overcome through mild manual pressure, which we demonstrated by prototyping a custom applicator comprising oxidized EGaIn paste laminated onto the surface of a piece of melamine foam ( Figure 2H). Using this applicator, triggered aluminum dissolution, as evidenced by the formation of characteristic gas bubbles from the H 2 (g) produced from de-passivated aluminum reacting with water, was initiated specifically at the points of contact ( Figure 2I).

Initiating Embrittlement Remotely
Obviating the need for direct application of EGaIn onto the target metal, we additionally observed that remote triggering was possible within acidic aqueous environments such as simulated gastric fluid (SGF, pH 1.2). At pH values below 2, both gallium oxide and aluminum oxide are thermodynamically unstable. [25] We speculate that molecular-level contact between EGaIn and aluminum is possible in acidic water despite the presence of interfacial water by eliminating the oxide barriers that are present in neutral pH environments ( Figure 3A). Because aluminum oxide is thermodynamically unstable in acid, aluminum in these conditions will passively degrade over time; thus, in acidic environments, active triggering complements passively degrading systems to ensure an additional mechanism for adaptability and safety.
Using optical microscopy, we captured the behavior of pure aluminum samples cut into 5 mm-diameter circles and immersed in 200 µL of SGF upon addition of a 20 µL droplet of EGaIn. Triggering, as represented by accelerated dissolution of the aluminum, was only observed when the EGaIn droplet fell onto a region of aluminum that was already reactive, as indicated by the presence of gas bubbles ( Figure S1, Supporting Information).
To increase the likelihood of contact with a de-passivated region of aluminum, EGaIn was formulated into microparticles and nanoparticles to increase its surface coverage. Nanoparticles with an average size of 176 nm were generated by sonicating the EGaIn in ethanol; the mechanical energy from sonication forms new surfaces that immediately become stabilized by a surface layer of Ga 2 O 3 ( Figure 3B, Figure S2A, Supporting Information). [22] Microparticles with an average size of 2 µm were generated by adding 10 vol% of 0.3 m HCl to the nanoparticle/ethanol mixture for 30 min, during which time the particles aggregated and sedimented due to partial acid etching of the Ga 2 O 3 , after which they were transferred to pure ethanol to re-oxidize the particles' surfaces and stabilize them at the larger size ( Figure 3C, Figure S2B, Supporting Information). In contrast to a single droplet, 20 µL of EGaIn formulated into microparticles coated the entire surface of aluminum circles, resulting in reproducible triggering in SGF ( Figure 3D).
Unlike microparticles, EGaIn nanoparticles were small enough to remain stably suspended in ethanol. Once added to SGF, the EGaIn nanoparticles aggregated and sedimented over the course of 90 min as their Ga 2 O 3 surfaces were etched away ( Figure 3E,F). Because their sedimentation is not immediate, EGaIn nanoparticles could trigger aluminum suspended within SGF. Using string, 50 mg 5000-series aluminum rings, which are alloyed with magnesium, were suspended within 10 mL of SGF. In SGF, the aluminum rings passively degraded at a tunable rate ranging from 0.05 mg h −1 for pristine rings to a highly accelerated rate of 0.66 mg h −1 for rings subjected to a sensitization treatment, which promotes intergranular corrosion due to migration of magnesium to the alloy's grain boundaries ( Figure 3G). [29] To actively trigger the rings, 200 mg of EGaIn nanoparticles in 2 mL of ethanol were added to the containers, which were subsequently placed on a shaker at 150 rpm at 37 °C for 30 min. The samples exposed to the nanoparticle formulations lost mass at an average rate of 21.67 mg h −1 , a dramatic increase from even the sensitized passively degrading samples ( Figure 3H).

Controlling Breakdown Behavior
Because liquid metals propagate through the grain boundaries of the solid metals they embrittle, we hypothesized that the breakdown behavior of aluminum after EGaIn triggering could be tuned by modulating its grain structure. As confirmed by scanning electron microscopy with detail resolved both by polishing and focused ion beam, as-received 5000-series aluminum wire possessed an anisotropic morphology, with grains elongated in the lengthwise direction, parallel to the direction of extrusion used to shape aluminum into wire ( Figure 4A). Recrystallization occurred upon annealing at 400 °C for 20 min,  Figure 4A). These morphological differences manifested in different fracture kinetics upon triggering. Using a universal tensile testing machine, wire samples were pre-loaded in the normal direction to 3, 30, and 50 N, and then contacted from the opposite side with the paintable EGaIn formulation. For both annealed and untreated samples, the time to fracture after triggering increased with pre-load force, yet for all pre-load force values, the speed at which annealed samples fractured after triggering exceeded that of the pristine samples by an order of magnitude ( Figure 4B). This observation is consistent with the fact that the annealed isotropic samples possess more direct fracture paths via grain boundaries oriented in the normal direction compared to the untreated anisotropic samples.
Similarly, morphological differences of the samples were clearly observable in how they dissolved in water after triggering. Samples of equivalent length were contacted at their center points with paintable EGaIn and then submerged in water. In untreated samples, the EGaIn appeared to propagate in the lengthwise direction, and then exfoliated from the outside in ( Figure 4C, Movie S1, Supporting Information). By contrast, in annealed samples, because the EGaIn could diffuse in both the lengthwise and normal directions without preference, the dissolution stayed more localized to the contact point ( Figure 4D, Movie S2, Supporting Information), such that the sample fractured instead of uniformly thinning. These different dissolution behaviors were also observed with aluminum samples suspended in SGF and triggered using EGaIn nanoparticles; annealed rings fractured at the particle contact point, whereas untreated rings became uniformly thinner ( Figure 4E).

Demonstrations of Potential Biomedical Uses
Metals are currently used in several clinical applications that leverage their superior mechanical strength to withstand or support large mechanical loads. Replacing these materials with actively triggerable metals is a straightforward way to achieve systems that retain the strength required during their functional lifetime while additionally being able to break down ondemand at end-of-life. Moreover, the rapid breakdown rate after triggering shortens the time required to perform and validate successful completion of procedures, reducing burden on both clinicians and patients alike.
For instance, metal staples are used when large forces are needed to hold different pieces of tissue together. [30,31] However, mechanically removing staples after healing could cause additional tissue damage. We created aluminum fasteners out of annealed 5000-series wire and stapled them into ex vivo porcine skin tissue. In our design, the staples are secured by the fact that their two legs are pinched inward toward one another ( Figure 5A, Figure S3A, Supporting Information). To remove the staples on demand, paintable EGaIn was smeared onto the exposed surface using a piece of melamine foam and allowed to rest for 1 min, after which water was added onto the staples to initiate their dissolution ( Figure 5B, Figure S3B, Supporting Information). Upon dissolution of the bridge, the legs could no longer oppose one another, and so were easily removed in the process of wiping away the dissolution products ( Figure 5C, Figure S3C, Supporting Information).
Additionally, metal stents are used to hold open bodily cavities like the esophagus and colon. Actively triggerable esophageal stents were fabricated by laser cutting a stent design into a sheet of pure aluminum, the ends of which were then soldered together to form a 1 cm-diameter cylindrical shape ( Figure 5D,E). In vitro compression testing was performed to 50% of the initial stent diameter to mimic in vivo peristaltic forces. [4] Upon exposure to 20 µL of EGaIn for 20 min in air, the maximum force withstood by the stents within that range of compression dropped to 28% of the control value, while the  force at 50% strain dropped to 6% of the control ( Figure 5F). These strength reductions are conservative, considering that the in vivo environment is hydrated and patients can ingest water after triggering, both of which would further accelerate degradation. To reflect this scenario, stents were triggered within ex vivo esophageal tissue, after which the disintegrated stents could be flushed out of the esophagus with water ( Figure 5G).
Lastly, actively triggerable metals can be used as highstrength connectors to enable the active disassembly of multimaterial, multicomponent biomedical devices. A compelling use case for such devices is in gastric resident systems that rely on geometric confinement within the stomach for longterm retention ( Figure 5H). Such systems have previously been demonstrated for weeks-long health monitoring (when integrated with electronics) [32] and drug delivery (when integrated with drug-loaded formulations). [33,34] However, for non-invasive removal, previously reported gastric resident devices relied on passively degrading connectors to break them down in a preprogrammed manner into pieces small enough to be eliminated from the gastrointestinal tract. [35] By contrast, gastric resident devices with actively triggerable metal connectors would allow for rapid, non-invasive, and on-demand elimination, enabling patients to respond quickly to unexpected complications like allergic reactions.
We fabricated star-shaped devices with all dimensions exceeding the 2 cm diameter of the pyloric sphincter for The staples are sufficiently secured so that they can be picked up by tweezers. B) Oxidized EGaIn was smeared onto the staple surface and then dissolved over 1 min by adding water and wiping away dissolution products. C) With the bridge of the staples dissolved, the rest of the staples are easily removed, and no residue remains in the cross-section of the tissue. D) Aluminum esophageal stent fabricated by laser cutting. E) An aluminum stent placed endoscopically in a pig esophagus was sufficient to keep the esophageal tract open. F) Comparison of compressive force for aluminum stents alone and stents after triggering (n = 3). G) After triggering, the stent was able to dissolve within ex vivo porcine esophagus (inset) and be rinsed out of the tissue cavity using water. H) Concept for a triggerable multimaterial gastric resident device. I) Prototype of a gastric resident device with polymeric components linked together by crimping through aluminum tubes. The central elastomer core enables the device to be folded into a swallowable 000 gelatin capsule (inset). J) Successful deployment of the gastric resident device in a pig stomach through a 000 gelatin capsule. K) Force required to pull out two PCL pieces crimped together through an aluminum tube, comparing control tubes to triggered tubes for different aluminum treatments (pristine vs annealed) and different number of contact points (number in parentheses) (n = 3). L) In vitro disassembly of gastric resident device into small polymeric components (inset) after triggering with EGaIn microparticles in SGF. M) In vivo disassembly of a triggered gastric resident device within porcine gastric cavity. retention in the stomach ( Figure 5I,J). [35] Individual polycaprolactone (PCL) arms, each shorter than 2 cm, were connected to a central elastomer core via crimping through a hollow aluminum tube ( Figure S5, Supporting Information). Each crimping connection was a strong friction fit that required over 100 N of force to disconnect ( Figure 5K). The strength of these connections was dramatically reduced in air upon contact with EGaIn, with the post-triggering strength decreasing as the number of contact points increased, and when annealed aluminum was used instead of pristine tubes ( Figure 5K). Active disassembly of a star-shaped device made with annealed aluminum tubes was demonstrated ex vivo in 20 mL of SGF within 30 min upon addition of 200 mg of EGaIn microparticles suspended in 2 mL of ethanol ( Figure 5L). Triggered devices were also endoscopically introduced into the gastric cavities of pigs in vivo in a terminal setting and were observed to disassemble within 30 min ( Figure 5M), illustrating that gastric peristalsis forces are sufficient for disassembly after triggering.

Biocompatibility
We performed acute toxicity studies of EGaIn via oral administration in rodent and found EGaIn be non-toxic by all histological and blood panel indications at a high dose of 300 mg kg −1 (Figures S6 and S7, Supporting Information). Additionally, all rats gained weight consistently throughout the two weeks following dosing ( Figure S6A, Supporting Information). Moving forward, future toxicity studies specific to different form factors and dosages are recommended to inform the continued development of actively triggerable metal-based biomedical technologies that utilize liquid metal stimuli. Overall, the promising biosafety of orally administered EGaIn at a high dose, coupled with previous literature that has also identified specific biocompatible forms and routes of administration for EGaIn, [15,17,21,36,37] suggest that EGaIn can be safe to use within the body for biomedical purposes.
Finally, thermal imaging using an IR camera showed that a maximum surface temperature of 28.6 °C was obtained 2 min after submerging a 1 cm-diameter piece of embrittled aluminum in room temperature (22.1 °C) water ( Figure S8, Supporting Information), which suggests that heat damage to the body is unlikely. [38] While the reaction between aluminum and water is exothermic, water also serves as a heat sink, [38] and in vivo environments are commonly either water-rich (e.g., in the stomach) or are readily accessible to be rinsed with water (e.g., on the skin). This measurement is conservative, since the aluminum sample was placed in contact with EGaIn for 2 h prior to testing, whereas in an actual triggering scenario the contact time would likely be much more limited.

Discussion
Actively triggerable metals combine the advantageous functional properties of metals with controllable end-of-life. We demonstrated the potential for actively triggerable aluminum to be used in biomedical applications where the high mechanical strength of metals is desired; moving forward, actively trig-gerable metals could also leverage other properties, like high electrical conductivity, which would enable bioelectronics that can be non-invasively eliminated from the body after use. Additionally, a major advantage of metals is the centuries of accumulated knowledge on their characterization and processing. Through our comparison between annealed and pristine aluminum, we demonstrated that morphology can be utilized to precisely program the breakdown behavior of the triggerable metal; moving forward, kinetics and regions of embrittlement can likely be even more fine-tuned through methods to control grain structure and size. These methods could conceivably enable other metals besides aluminum to be rendered more susceptible to robust embrittlement by biocompatible liquid metals as well. For example, under significant applied stress or multiple fatigue cycles, gallium has already been observed to embrittle metals like Zn, Cu, and ferrous alloys like steel. [39] By shifting the paradigm away from studying embrittlement as an undesirable failure mechanism, we anticipate that metal processing techniques can developed so that these metals can exhibit sufficiently fast and robust embrittlement responses at body temperature for use in biomedical applications as well.
Future applications of triggering modalities could also involve combination approaches wherein, for example, embrittlement amplifies a different exogenous stimulus, such as heat or light ( Figure S9A, Supporting Information), or is initiated by passive and active pumping mechanisms like capillary action and electronically controlled micropumps, respectively ( Figure S9B, Supporting Information). Such designs would further enable embrittlement to be initiated in hard-to-reach areas of the body, as in the case of gastric resident devices, without requiring direct endoscopic or operational access, while also serving as engineering controls to overcome potential biofouling barriers and mitigate safety concerns.

Conclusion
Predictability of biomedical technologies over their entire lifetimes is critical for translation and accessibility, especially those designed for long-term retention in the body. Devices that can be actively triggered to break down on demand provide maximal control over lifespan. Actively triggerable metals impart predictability without comprising on functional properties like mechanical strength. Using liquid metal stimuli, which break down solid metals in a grain boundary-mediated mechanism, predictability is further enhanced due to the relative ease with which the metal's breakdown behavior can be programmed by its morphology. Historically, liquid metal embrittlement has been viewed as a failure mechanism, studied so that it can be avoided. With the continued generation of mechanistic insights into embrittlement phenomena, [12] we envision that an explicit focus on designing metals to be susceptible to embrittlement will open the door to a new class of actively triggerable biomedical metals beyond aluminum.
Synthesizing Eutectic Gallium Indium Formulations: Paintable EGaIn paste was obtained by using a pipette tip to stir EGaIn until it could be drawn into a line that held its shape without re-aggregating. The underwater applicator was prototyped by inserting a piece of melamine foam into a 1 mL syringe, using the plunger to hold the foam in place at the syringe opening. EGaIn nanoparticles were synthesized using a QSonica Q55 probe sonicator, by ultrasonicating EGaIn at 100 mg mL −1 in ethanol at an amplitude of 50% for 30 min, with 5 s of rest each minute. Microparticles were obtained from nanoparticles by adding 0.3 m HCl to the ethanol suspension to a total of 10 vol% for 30 min, after which the solvent was replaced with pure ethanol.
Material Characterization: EGaIn micro-and nanoparticles were drop-cast onto silicon wafers and imaged using a FlexSEM TM-1000 II from Hitachi (Tokyo, Japan). Particle size analysis was performed using Mountain Maps software, Version 8.2, with 1071 microparticles and 404 nanoparticles. Aluminum microstructure was imaged using a Thermo Scientific Helios focused ion beam scanning electron microscope (FIB-SEM). Prior to imaging, aluminum samples were polished using a NANO 1000S Polishing Head from PACE Technologies Corporation. After loading on the FIB-SEM, samples were milled using a gallium ion beam set to 30 kV and 9.3 nA to image cross sections in orientations both parallel and perpendicular to the sample extrusion direction. X-ray photoelectron microscopy (XPS) was performed using the Thermo Scientific K-Alpha XPS System. A high-resolution scan of Ga2p was performed with a dwell time of 25 ms and 10 total scans, and the data was normalized to the elemental gallium peak at 1116.7 eV. Ultravioletvisible spectroscopy (UV-vis) was performed on 20 mg mL −1 EGaIn nanoparticle suspensions in SGF using a Cary 3500 Spectrophotometer from Agilent and a transmission wavelength of 300 nm.
Mechanical Characterizations: Mechanical characterization was performed using an Instron Universal Testing Machine. All tensile elongation testing was performed at a rate of 0.5 mm s −1 . The impact of EGaIn contact time on tensile elongation behavior was assessed using 0.12 mm-thick pure aluminum sheets cut into dogbone shapes per ASTM specifications. To maintain a consistent contact area between EGaIn and aluminum, Kapton tape was laser cut into 5 mm-diameter circles and placed at the center of the dogbones as a mask for EGaIn application. To test the fracture properties of pre-loaded samples, 5000series aluminum rods were held in place by two drill press vises and loaded from the center of the rods to the specified pre-load force using a custom piece attached on a compression plate. A custom setup using a humidifier and humidity monitor was built to assess the impact of relative humidity. Pull-out force measurements for crimping connections were performed by clamping tension grips onto two PCL arms connected by a 4 mm-diameter, 6 mm-length aluminum tube crimped at two points, and EGaIn was applied 30 min prior to testing.
Triggering Experiments: Microparticle-and macroscopic dropletinduced triggering experiments were performed using 20 µL total EGaIn. 99.6% pure aluminum sheet stock were punched into 5 mm-diameter circles and immersed within 200 µL SGF inside a 96 well plate. Videos of the experiments were obtained using a HAYEAR 4K UDH HDMI microscope camera. Nanoparticle-induced triggering experiments were performed by suspending aluminum parts with string within 20 mL SGF in a 50 mL Falcon tube. After adding particles, the Falcon tubes were incubated at 37 °C on a 150 rpm shaker for 30 min, after which they were transferred to deionized water. 30 min was chosen as a conservative estimate for particle exposure based on expected gastric emptying times. Temperature measurements from the reaction between triggered aluminum and water were obtained using a PerfectPrime Thermal Imaging Camera IR001851.
Fabricating Aluminum-Based Devices: Aluminum tissue staples were shaped manually from 5356 aluminum wire. Esophageal stents were laser cut from 99.6% pure aluminum sheets (Black Cat Labs, Somerville, MA, USA) and rolled into 1 cm-diameter cylinders that were sealed using aluminum solder wire (96.5 Sn/3.5Ag 0.062 flux core). Gastric resident star devices were constructed by crimping polymeric device elements to aluminum tubes, which served to connect different elements together. Tubes were cut to desired lengths using a pipe cutter. PCL arms were compression molded by melting PCL pellets at 90 °C in a custom split 3D-printed mold (FormLabs Form 2, Rigid 10K). Hollow PCL tubes were made by compression molding with the same 3D-printed mold and inserting a metal rod through the molten PCL before cooling. Elastic cores were made by injection molding Elastollan at 223 °C through hollow PCL tubes placed within a custom split 3D-printed mold.
Animal Studies: All animal experiments were approved by the Committee on Animal Care at the Massachusetts Institute of Technology. Toxicity studies were performed in outbred female rats (CD Sprague Dawley IGS rats) purchased from Charles River. EGaIn was dosed to rats at 300 mg kg −1 body weight via oral gavage. Animals were monitored daily over 2 weeks for signs of acute toxicity, after which they were euthanized and blood serum and tissue samples were collected for evaluation. Ex vivo pig tissue (skin, esophagus) was obtained from Blood Farm Slaughterhouse (West Groton, MA, USA). In vivo pig experiments were performed on Yorkshire pigs (80-100 kg) due to the anatomical similarity of their GI tract to that of humans. Pigs were sedated prior to endoscopy-assisted insertion of an overtube, as previously described, [5,35,40] which was used to administer devices.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.