Active Liquids and
Soft Materials

To exploit the extraordinary mechanics of liquids and other soft matters
for their unconventional uses in robotic and intelligent systems.

Research Vision
Research Vision

About me
    Jiahe Liao
  • Jiahe Liao is a postdoctoral researcher in the Soft Machines Lab with Prof. Carmel Majidi at Carnegie Mellon University. His research focuses on active soft materials that leverage the physics of liquids, particularly liquid metals, to unlock advanced material architectures and mechanisms that mimic biological functions in robots.
  • Liao received his Ph.D. in robotics from Carnegie Mellon University in 2022 and completed a postdoctoral fellowship at the Max Planck Institute for Intelligent Systems in Germany from 2022 to 2024. Originally from Taiwan, his foundational research on soft actuators advances the fields of soft robotics, soft condensed matter physics, and bioinspired materials for next-generation robots.

Recent News

Liquid Metals:
The All-Purpose Soft Materials

With their unique combination of high surface tension, high fluidic deformability,
and stimuli responsivity, liquid metals offer unrivaled research potentials
from surface tension-driven artificial muscles to advanced robotic materials and structures.

Liao et al. In Preparation. / Liao et al., Advanced Science (2022) / Liao et al. Carnegie Mellon PhD Thesis (2022) / Zhao et al., Nat. Electron. (2023) / Zhang and Liao et al., Adv. Funct. Mater. (2023).

Liquid+Metal → Active mechanisms for robotic and intelligent systems

Soft materials, structures, robots, and systems can take advantage of their liquid-driven characteristics.

Liao et al., Advanced Materials (2023)
Liao et al., Advanced Materials (2023)
  • Active mechanisms: The concept of liquid metals as active liquids comes from their intrinsic potential to produce mechanical energy (i.e., in deforming or breaking a droplet) or electrical energy (i.e., in a battery through redox reactions) that is primarily driven by their fluidic and liquid-specific properties. For the goal of creating high-performance artificial muscles, nerves, and other robotic and computing devices, liquid metals arise as an excellent candidate for their excellent combination of high fluidic deformability, high electrical and thermal conductivities, and high surface tension. The development of active components driven by any combinations of these liquid metal characteristics will potentially impact a wide range of areas from robotic to computing systems.

Related Publication

Future Research Potentials

Liquid metals Soft actuators Active materials Surface tension Interface Instability Stretchable electronics Batteries Microrobotics Soft robotics Mechanical machine intelligence

Liquids as active materials can bridge the gaps in broader areas

Research and collaborative potentials in bringing autonomy, power-efficiency, mechanical compliance,
biocompatibility and other multifunctionalities to the next-generation robotic and computing devices.

Liao et al., In preparation Liu et al., Advanced Energy Materials (2019) Wissman et al., Advanced Science (2017) Zhang and Liao et al., Advanced Functional Materials (2023)
Liao et al. In Preparation. / Liu et al., Adv. Energy Mater. (2019) / Wissman et al., Adv. Sci. (2017) / Zhang and Liao et al., Adv. Funct. Mater. (2023).
  • Robotics: Liquid metal actuators are emerging artificial muscles that are energy-dense and require significantly less activation voltages (~100 vs. ~103 V) than most existing actuator technologies. Further structural enhancements such as miniaturization and encapsulation will make them suitable for practical deployment in systems such as legged mobile robots, industrial robot arms, and miniature medical robots, where capillary forces can outperform most types of forces, such as electrostatic and magnetic forces, that drive the existing robotic actuators.
  • Energy: Batteries may take advantage of liquid metals as active components to achieve stretchability, which may benefit applications such as wearable technologies and untethered soft robots. Particularly for the latter, the search for a power source that is mechanically compliant and electrochemically reliable has been of interest to the robotics communities. Liquid metals are a promising energy solution.
  • Computing and Electronics: Neuromorphic computing devices based on a network of liquid metal connections can potentially establish a new research paradigm, where the liquid properties are exploited to mimic the neural topology of the brain. While these artificial neurons, in their physical forms, have been primarily built on the traditional semiconductor-based transistor technologies, the remarkable ability of active liquid metals has shown promising potential.
  • Medical: Shape-morphing devices (mm-scale and sub-mm-scale) driven by liquid metals. While at an early stage of development, the ability of liquid metals to generate significant forces relative to their length scale and good biocompatibility hint at the possibility of self-deploying medical devices such as stents in the body for surgical procedures.

Actuators:
Muscle-Like Performance by Exploiting Mechanics of Liquid Metals

By taking advantage of their high surface tension and interfacial electrochemistry,
liquid metal actuators are capable of generating muscle-like, energy-dense contractions at low voltages.
They offer a powerful alternative to the existing energy-intensive artificial muscles for mobile soft robots.

Liao et al. In Preparation. / Liao et al., Advanced Science (2022)

Liquid metal actuators: Towards high-performance robot muscles

High surface tension and fluidic deformability in electrochemically activated liquid metal bridges.

Liao et al., Advanced Materials (2023) Liao et al., Soft Matter (2021) Liao et al., Advanced Science (2022)
Liao et al., Advanced Materials (2023) / Liao et al., Soft Matter (2021) / Liao et al., Advanced Science (2022)
  • Surface Tension: When a droplet of liquid metal bridges between two solid bodies, the liquid surface alone no longer governs the overall kinematics and dynamics. Instead, a variable surface tension is linked to the force and deformation across the bridge. This provides the theoretical basis for a liquid metal actuator, whose surface tension can be electrically modulated across a very wide range.
  • Scaling Laws: As surface tension becomes relatively more significant at smaller length scales (i.e., force scales with length), it gives liquid metal actuators a unique scaling advantage over other actuators driven by electrostatic forces (i.e., scale with length2) and magnetic forces(i.e., scale with length3). By these laws, it is estimated that a µm-scale liquid metal droplet can produce mechanical work at ~106 J/m3 (compared to ~104 J/m3 in natural muscles).
  • Electrochemistry: While traditionally done by electrocapillarity or electrowetting at ~102 V, it is now known that surface tension of liquid metals can be reversibly controlled by depositing the surface oxide at ~1 V. These low-voltage, electrochemically driven mechanisms give liquid metal actuators an unrivaled advantage over most existing power-intensive artificial muscle technologies.
  • Structural Scalability: Liquid metal actuators have a unique structural advantage where it is theoretically possible to increase force production by splitting the liquid volume and increase surface area. This paves the way for a high-performance actuator with a large number of small-scale liquid metal droplets.

Related Publication

Future Research Potentials

Soft actuators Surface tension Interface Microrobotics Soft robotics Artificial muscles Robot locomotion Robot manipulation

Higher energy densities, Faster shape change: A powerful artificial muscle

Liquid-driven actuators have unique structural and scaling advantages over existing actuator technologies.

Liao et al., Advanced Materials (2023) Liao et al., Advanced Materials (2023)
Liao et al., Advanced Materials (2023)
  • High Performance: Mechanical work generated by liquid metal actuators increases when downscaled and is predicted to surpass most existing artificial muscles when made near microscale. Thanks to the fluidic deformability, liquid metal actuators are known to change their shape at a remarkable speed, which contributes to their ability to operate at higher frequencies (>20 Hz as currently known).
  • Low Operational Requirement: Liquid metal actuators operate at very low voltages and eliminate the need for high voltage power source as with many existing electrostatic and magnetic actuators, many of which run on hundreds of volts. This makes liquid metal actuators particularly favorable in certain constrained systems such as untethered mobile robots that do not have the luxury of carrying a bulky power source.

Beyond Artificial Muscles: Versatility for Robot Locomotion,
Manipulation, Logic and Computing Systems

Liquid metal actuators, with their ability to actively deform, to break and merge, offer exciting
engineering possibilities beyond the physical limitations of traditional mechatronic systems.

Liao et al., Advanced Materials (2023)
Liao et al., Advanced Materials (2023)
  • General Actuation: Liquid metal actuators are general-purpose actuators that support a wide range of shape change and forces. Besides conventional linear contraction, rotation, bending, torsion, the high degrees of freedom intrinsic to liquids allow them to be tailored to specific applications.
  • Locomotion: Liquid metal-based microrobots are capable of moving themselves in a medium by propulsion, undulation, rolling, jumping. Locomotion at larger scales can also be driven by general actuation.
  • Manipulation: The most well-known application is the self-alignment of soldered components, where the liquid solder moves the electronic components in place. Modern liquids and liquid metals are also known for their capabilities of adhesive gripping, lifting, and folding in self-assembly systems.
  • Logic and Computation: As liquid metals are good electrical conductors, their actuation can be used to create, modulate, and remove connections that are fluidic and electrically conductive at the same time. This enables broad research potentials for reconfigurable circuits that can potentially represent and compute any logic states.

Robotic Materials:
Liquid Properties Unlock Hidden Potentials in Materials

From stretchable electronics to robotic materials with energy-harvesting, self-healing, and neural reconfigurable
capabilities, materials innovations with liquids and other liquid-like soft matters open up new research potentials
in bringing novel material architectures to the next-generation robotic and biologically inspired systems.

Zadan et al., Adv Mater. (2021) / Zhao et al., Nat. Electron. (2023) / Ohm et al., Adv. Mater. (2023)

Liquids+Soft Materials: When liquids enable robotic material functionalities

From multi-scale droplets to core-shell structures, liquids introduce a rich set of remarkable
mechanical, electrical, and chemical properties overcome the physical limitations of solid materials.

Ohm et al., Nature Electronics (2023) Zadan et al., Advanced Materials (2021) Ohm et al., Advanced Materials (2023) Zhang and Liao et al., Advanced Functional Materials (2023)
Ohm et al., Nat. Electron. (2023) / Zadan et al., Adv Mater. (2021) / Ohm et al., Adv. Mater. (2023) / Zhang and Liao et al., Adv. Funct. Mater. (2023).
  • Stretchable Conductors: Soft hydrogels with predominantly water can achieve significantly higher electrical conductivity when embedded with silver fillers. This combination of high stretchability, conductivity, and biocompatibility benefits soft robots and bioelectronics.
  • Self-Healing Materials: Soft organogels can obtain electrical conductivity from liquid metal fillers, which also give them both mechanical and electrical self-healing capabilities.
  • Energy Harvesting: Liquid crystal elastomers can achieve two functionalities in a thermoelectric robot at the same time: They actuate the robot for locomotion towards a heat source, and they harvest thermal energy from it.

Related Publication

Future Research Potentials

Stretchable electronics Stretchable batteries Self-healing materials Shape-memory materials Conductive hydrogels Stimuli-responsive materials Robotic materials Bioinspired robotics

Metastructures:
When Many Droplets Work Together to Reshape a Structure

Reversible and reprogrammable shape transformation can be achieved
by structural arrangement of repeating liquid bridges,
which reveal remarkable characteristics unseen in current solid-state mechanical metamaterials.

Liao et al. In Preparation.

Shape-memory effects in liquid metal bridges by phase transformation

To harness the thermo-mechanical behavior when a deformed solid-state metal “melts” into liquid-state.

  • Pseudoelasticity in liquids: Elastic deformation is known to occur in solid structures (metals, elastomers, hydrogels, etc.). Similar elasticity-like behaviors are known in liquids for their tendency to minimize surface energy. Like the traditional shape-memory alloys, which have recoverable strains after large deformation, liquid metals can potentially mimic similar shape-restoring behaviors across the liquid-solid transition.

Future Research Potentials

Phase-change materials Solid-liquid phase transition Shape-memory materials Elastocapillarity Shape morphing Metastructures Metamaterials Liquid bridges Deployability Surgical devices

Metastructures: A network of shape-memory liquid metal bridges

Shape transformation can be constructed with a liquid metal network at very high degrees of freedom.

  • Collective mechanical behaviors: Similar to snap-through buckling in solid materials, propagation characteristics can be theoretically achieved with liquid metal bridges through their breakup instabilities. Another collective behavior of interest is localized shape morphing by selectively actuating some liquid metal bridges, which are expected to transform the metastructure into a preprogrammed shape. With phase transition, the metastructure is further considered to have the shape-memory ability activated by heat.
  • Potential applications: By thermally activating and cooling the liquid metal bridges between the liquid and solid states, a network of these bridges can collectively form a shape-morphing, deployable metamaterial with higher degrees of freedom than the sum of its parts. These devices can potentially enhance or augment existing applications of shape-memory alloys such as medical stents. In addition to the traditional shape-memory properties, liquids also benefit from their versatile fluidic characteristics that allow them to break and merge freely, which may provide novel solutions to certain limitations of current in-vivo medical devices.

Neural Systems:
Artificial Brain with Liquid Metal Networks

By taking inspiration from neural connections in the brain,
we seek to create neuromorphic computing devices through reconfigurable liquid metal connections
for learning, encoding, and memory of information in a new liquid-driven computational architecture.

p
Liao et al. In Preparation.

Lesson from Nature: A neural network of electrical interconnections

Neural computing → Synaptic connections on which signals transmit. Liquid metal bridges, with their electrical conductivity and fluidic deformability, have unique potentials to mimic the neural system.

  • Structural insights: The biological inspiration from the brain, which processes information through a network of neurons, has become the foundation of modern artificial intelligence, neuromorphic computing and alike. For decades, researchers across all disciplines have created neural systems to capture different abstract and physical aspects of the biological neural networks. All are strongly influenced by the connectivity of neurons where electric signals propagate differently across different pathways. How to construct such a neuroplastic or rewirable machine? One solution is by creating materials that can mechanically form and organize electrical connections in a manner informed by these structural insights.
  • Neural network and signaling: The key to modern intelligent systems is the learning and encoding of complex information through adjustment of synaptic weights, which control how signals are transmitted. An electrical interpretation is that by varying resistance of every wire in a network, it will be able to learn and retain information. This neuromorphic approach bridges physical and computational intelligence.
  • Liquid metal bridges: Controllable shape change in liquid metals make them an excellent choice for building a neural network. Any liquid metal droplet can form a bridge of any shape from one location to another. It should also be able to adapt, grow, or disconnect as desired. One engineering question towards this goal is how to reshape these liquid metal connections so that the structure can be retained within a soft medium.

Related Publication

Future Research Potentials

Neuromorphic computing Mechanical machine intelligence Artificial neural networks Electrical reconfigurability Capillary bridge breakup and merging Instability Liquid metal circuits Hydrogels Neuroplasticity Mechanologic Mechanical memory Mechanical machine learning Brain-on-a-chip Neural interface Neural rehabilitation

A bridge in the gap: Promising potentials in a wide range of impactful fields

Neuromorphic computer. Brain-on-a-Chip. Liquid-State Integrated Circuits. Neural Repair and Rehabilitation.

  • Mechanical machine intelligence: A liquid metal neural network will be electrically reconfigurable through altering their mechanical connections. In this sense, this approach has some characteristics of machine intelligence by mechanical means, but it also resembles the biological nervous system by propagating signals electrically. Further investigation may create a new paradigm for neuromorphic computing.
  • Logic, memory, learning: From a computational perspective, for any machine to compute, it has to be able to process input and generate definite results. This should ideally be achieved by the liquid metal neural networks of various complexities.
  • Neuromorphic materials: The material versatility of liquid metals offer an interesting alternative to the modern transistor-based computing technologies. Will the development of these liquid circuits lead to capabilities unseen in their solid-state counterparts? Will they be able to interface with biological nervous systems to augment or rehabilitate for restoring impaired neural functions? The answers to these questions call for a foundational understanding of this material architecture and a collaborative effort to translate it into impactful areas.
Research Vision