🔬 materials science

Dark Metastable Conduction Channels near a Metal-Insulator Transition

This study reveals that current pulses in 1T-TaS $_{2}$ generate previously undetected metastable filamentary conduction channels that define the material's hidden metallic state, enabling electrical control over their formation and positioning for potential neuromorphic computing applications.

Original authors: T. R. Devidas, Dror Yahav, Jonathan T. Reichanadter, Shannon C. Haley, Matan Sterenberg, Joel E. Moore, Jeffrey B. Neaton, James G. Analytis, Beena Kalisky, Eran Maniv

Published 2026-09-11

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: T. R. Devidas, Dror Yahav, Jonathan T. Reichanadter, Shannon C. Haley, Matan Sterenberg, Joel E. Moore, Jeffrey B. Neaton, James G. Analytis, Beena Kalisky, Eran Maniv

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Solid matter generally falls into two camps: metals, which let electricity flow freely, and insulators, which block it. Between these two extremes lies a chaotic frontier where materials can switch back and forth, often behaving in ways that defy simple explanation. In these strange zones, the usual rules of physics sometimes break down, and electrons stop acting like individual particles and start behaving as a tangled, collective mess. Scientists are deeply interested in these transitions because they often hide new states of matter that could revolutionize technology, from faster computers to more efficient energy storage. One such material, a layered crystal called 1T-TaS2, has long puzzled researchers. It can be an insulator, a metal, or something in between that scientists call a "hidden state," a condition that appears under specific conditions but vanishes when the material is heated or cooled in different ways. The mystery has been that no one could see exactly how electricity moved through this hidden state, leaving a gap in our understanding of how these materials work.

A team of researchers has now peered inside this hidden state and found that electricity does not flow evenly through the material as one might expect. Instead, when they applied a specific burst of electrical current to the crystal, the electricity was forced into narrow, invisible highways that ran along the physical edges of the sample. They call these pathways "dark metastable conduction channels." The term "dark" is used because these channels are invisible to standard measurements that look at the whole block of material; they only reveal themselves when the researchers used a highly sensitive magnetic camera to map the flow of current. The "metastable" part means these channels are stuck in a temporary but long-lasting state. Once created, they remain there indefinitely, even after the power is turned off, until the material is warmed up or hit with a different electrical pulse to erase them.

The researchers discovered that these channels are not just a random glitch but are deeply tied to the material's internal structure. The crystal 1T-TaS2 arranges its atoms in a specific pattern called a charge density wave, which looks like a grid of star-shaped distortions. When the material is in its insulating state, these stars are locked in place. The team found that to create the conducting channels, they had to start with this locked, insulating pattern. If they cooled the material too quickly, skipping the formation of this specific pattern, the channels would not form at all, even if they applied the same electrical pulses. This proved that the channels are a unique feature of the hidden state, distinct from other metallic phases the material can take. Furthermore, the channels always chose to travel along the physical boundaries of the crystal, such as the edges or the interfaces where the material meets the metal contacts used to measure it. By changing where they injected the electrical pulse, they could force the channel to switch to a different edge, effectively rewiring the path of the electricity.

This ability to write, erase, and move these channels suggests a new way to think about memory and computing. In a standard computer, a switch is either on or off, representing a one or a zero. Here, the researchers showed that they could create multiple different conducting paths within the same piece of material, each with its own resistance. By choosing which edge the channel follows, they could encode different states of information. Because these states are non-volatile, meaning they stay put without needing constant power, and because they can be manipulated with simple electrical pulses, they resemble the way biological neurons form connections in the brain. This opens the door to designing electronic components that mimic the flexible, adaptive nature of neural networks, potentially leading to computers that learn and process information more like living systems.

The team also ruled out a common explanation for such behavior. In many materials, when electricity flows, it heats up the sample, and that heat can melt the insulating structure to create a conductive path. However, the researchers found that the heat generated by their pulses was not enough to explain the effect. They showed that samples with similar resistance levels but different internal structures did not form these channels, even when heated in the same way. This indicates that the formation of the channels is a complex quantum process driven by the specific arrangement of the material's defects and boundaries, rather than simple thermal melting. The discovery confirms that the hidden state is a distinct phase of matter, governed by the movement of topological defects—tiny imperfections in the atomic grid—that align along the edges to create these persistent, controllable highways for electricity.

Technical Summary: Dark Metastable Conduction Channels near a Metal-Insulator Transition

Problem Statement
Materials undergoing metal-insulator transitions (MIT), such as the transition-metal dichalcogenide 1T-TaS2, exhibit complex behaviors at intermediate energy scales where standard quasiparticle descriptions often fail. In 1T-TaS2, the system transitions from a metallic Nearly Commensurate Charge Density Wave (N-CDW) phase to an insulating Commensurate CDW (C-CDW) phase below ~180 K, characterized by a $\sqrt{13}\times\sqrt{13}$ Star-of-David lattice distortion. While the C-CDW state is insulating, it can be driven into a metallic "hidden state" via laser pulsing or high electric fields. This hidden state is anomalous: it exhibits Ohmic resistance that decreases with increasing temperature, vanishes at a mysterious transition temperature ( $T_H \approx 70$ K) unrelated to known CDW phase transitions, and its fundamental nature remains debated. A central question is whether this hidden state is merely a manifestation of the supercooled N-CDW phase or a distinct state arising from the C-CDW ground state, and how it is spatially distributed within the material.

Methodology
The authors fabricated cross-bar devices using 1T-TaS2 single crystals, employing both Focused Ion Beam (FIB) microstructuring and exfoliation onto sapphire substrates with pre-patterned contacts. They utilized a dual-mode electrical protocol:

Transport Measurements: Four-probe AC resistivity measurements tracked the global resistance changes during cooling and heating cycles.
Pulsing Protocols: High-density DC current pulses ( $I_{write}$ ) were applied to switch the device from the high-resistance insulating state to the low-resistance hidden state. Intermediate current pulses ( $I_{erase}$ ) were used to revert the system to the insulating state.
Scanning-SQUID Magnetometry: To resolve the spatial distribution of current flow, the authors employed scanning Superconducting Quantum Interference Device (SQUID) microscopy. By flowing a low-density AC probe current ( $I_{read}$ ) and mapping the resulting magnetic flux, they reconstructed local current density maps. This technique bridged the gap between bulk transport and atomic-scale probes, allowing visualization of current paths at the mesoscopic scale (~2 $\mu$ m resolution).
Control Experiments: The study compared slow-cooled samples (which undergo the N-CDW to C-CDW transition) with rapidly supercooled samples (which remain in the metallic N-CDW phase) to isolate the origin of the hidden state.

Key Results

Discovery of Dark Metastable Conduction Channels (DMCs): The study reveals that the "hidden state" is not a homogeneous metallic phase but is mediated by confined, filamentary conduction channels. In the insulating C-CDW state, current flows homogeneously. Upon application of a threshold write pulse ( $I_{write} \sim 10^4 - 10^5$ A/cm $^2$ ), the current path abruptly localizes to specific physical boundaries or interfaces within the device. These channels are "dark" because they are invisible to bulk transport measurements (which average over the entire sample) and atomic-resolution probes that do not capture the mesoscopic network.
Electrical Control and Metastability: The DMCs are non-volatile; they persist indefinitely at low temperatures until the system is warmed above $T_H$ (~70 K) or an erase pulse ( $I_{erase}$ ) is applied. The system can be reproducibly switched between the high-resistance insulating state and the low-resistance hidden state by toggling between specific current pulse amplitudes.
Distinct Origin from N-CDW: Crucially, supercooled N-CDW samples (which are metallic and have low resistance) do not exhibit DMC formation when subjected to write pulses. The current remains homogeneous, and the resistance does not change significantly. This proves that the hidden state and its associated DMCs are distinct from the supercooled N-CDW phase and originate specifically from the C-CDW ground state.
Role of Boundaries and Defects: The location of the DMCs is determined by physical boundaries (edges of the flake or device) and the point of charge injection. The authors observe that DMCs preferentially form along edges, suggesting they are aligned networks of topological defects inherent to the C-CDW phase. The choice of injection point can steer the DMC to different boundaries, demonstrating electrical control over the channel's geometry.
Joule Heating vs. Defect Dynamics: While Joule heating is present, the authors argue it is not the sole mechanism. Supercooled samples experience similar heating but do not form DMCs. Furthermore, the erasing process requires lower current densities than the writing process, inconsistent with a purely thermal melting mechanism. The data supports a model where current pulses manipulate the energy balance of bound topological defect pairs, aligning them to form conducting filaments.

Significance and Claims
The paper claims to resolve the nature of the hidden state in 1T-TaS2 by identifying it as a network of metastable, boundary-aligned conduction channels (DMCs) rather than a bulk phase transition. This finding:

Explains the "Hidden" Nature: It clarifies why the state is "dark" to previous measurements—the conduction is highly localized and filamentary, masking the insulating nature of the bulk material.
Links to Topological Defects: It provides a physical mechanism for the hidden state, linking it to the mobility and alignment of topological domain defects in the C-CDW lattice, offering an explanation for the absence of thermodynamic anomalies at $T_H$ .
Enables Electrical Design: The authors demonstrate that the location and existence of these channels can be electrically written, erased, and reconfigured. This capability allows for the creation of multi-state resistance devices where information is encoded in the specific path of the DMC.
Neuromorphic Potential: The authors suggest that these controllable, non-volatile filamentary paths could serve as "synaptic" components for neuromorphic computing, leveraging the ability to create complex, reconfigurable conductive networks within a single material.

The study concludes that the distinction between the metal and insulator states in 1T-TaS2 is a matter of the connectivity of this defect network, which can be distributed (insulating) or organized into "highways" (metallic/hidden state) through electrical manipulation.

Technical Summary: Dark Metastable Conduction Channels near a Metal-Insulator Transition

More like this