Molecular electronic devices based on atomic manufacturing methods

Abstract

As semiconductor devices approach fundamental physical scaling limits, molecular electronics has emerged as a potential technological paradigm for sustaining Moore’s Law through the capabilities of single-molecule-scale functional manipulation and quantum modulation. At the foundational research level, the convergence of atomic-precision fabrication techniques with molecule–electrode interfaces and molecular orbital engineering has enabled the directional construction of electronically functional single-molecule devices, including molecular switches, rectifiers, and field-effect transistors, accompanied by preliminary validations of molecular device array integration. However, molecular electronics confronts multifaceted challenges spanning device-level bottlenecks in precise molecular assembly, accurate quantum charge transport characterizations, and performance reproducibility, coupled with integration-level limitations imposed by conventional two-dimensional planar architectures that fundamentally constrain functional density scaling, rendering the realization of high-density integrated molecular devices with operational logic capabilities exceptionally demanding. To address these critical issues, researchers have developed various device fabrication and characterization techniques in recent years, such as the integration of top-down micro/nano-fabrication technologies with bottom-up atomic manufacturing approaches, which have significantly enhanced the stability of molecular devices and data reproducibility. This review systematically summarizes recent advances in preparation methodologies for molecular electronic devices with high reproducibility and reliability, with prospective emphasis on an integrated architecture strategy combining atomic manufacturing technologies with three-dimensional (3D) integrated manufacturing technologies, offering a potential roadmap to transcend conventional two-dimensional integration paradigms and realize logical computing functionalities in molecular electronic devices.

Introduction

Since the beginning of 2020 s, the progress in the integrated circuits industry driven by Moore’s Law^1,2 has significantly decelerated when achieving mass production at the 3-nm technology node, where gate lengths have been reduced below 15 nm, while the International Technology Roadmap for Semiconductor (ITRS) has extended the mass production of sub-1 nm node by the year of 2035^3,4. As the process node is being shrunk closer to physical limits^5,6, R&D investments for advanced processes have exponentially grown: the construction cost of a 3-nm node production line exceeds $20 billion, while the investments for next-generation 2-nm node are expected to increase by 50%^7,8, resulting in a historic inflection point where the decrease of per-transistor cost exceeds the increase of investments. However, how to further extend Moore’s Law requires quite a challenge, not only from discovering new working principles but also from developing new manufacturing technologies. Molecular electronics has emerged as a potential post-Moore-era direction using single molecules as functional units, with its core value manifested in three aspects: first, new operating mechanisms through quantum tunneling transport at single-molecule scale⁹, distinct from classical transport in traditional semiconductor devices; second, the feature size of single molecules is at sub-5 nanometer scale, demonstrating potential high-density integration^10,11; third, the molecular orbital energy levels for tunneling properties and interface coupling effects can be chemically synthesized in atomic-level precision for tuning electrical properties of molecular devices¹².

In the field of molecular electronics, assuming that each molecule can be functioned as an individual device, the construction of molecular electronic device arrays via self-assembled monolayer (SAM) technology could theoretically achieve an integration density of up to 10¹⁴ molecular devices per square centimeter, surpassing the transistor integration density of approximately 10¹¹ devices/cm² in 1-nm node integrated circuit by 2-3 orders of magnitude. Recently, researchers have developed molecular electronic devices based on quantum tunneling effects, including conducting wires¹³, insulators¹⁴, switches^15,16, Fano resonance devices¹⁷, diodes/rectifiers^18,19,20, thermoelectric devices²¹, optoelectronic molecular units²², nonvolatile molecular memristors²³, and even single-molecule transistors^24,25,26,27. However, challenges persist in achieving high stability and reproducible characterization of molecular electronic devices, and unresolved issues remain in the heterogeneous integration of molecular devices with silicon-based microelectronic systems regarding multi-physical field coordination, such as interface coupling and thermodynamic stability^28,29, while systematic methodologies for high-density integration and logic interconnects of molecular devices have yet to be established³⁰, which are critical for practical applications of molecular electronics. Therefore, to achieve highly stable construction and highly reproducible operation of molecular electronic devices, numerous micro/nano-technologies for molecular junction fabrication and even atomic manufacturing methods for molecular device assembly have been developed. Future solutions may integrate bottom-up atomic-precision fabrication techniques such as molecular self-assembly³¹ with top-down silicon-based manufacturing technologies, alongside implementing advanced packaging technologies, potentially for future high-density integrations and inter-connections for logic operations and computing through molecular circuits^32,33,34 (Scheme 1).

Scheme 1

Fabrication-Functionalization-Integration-Application process schematic of molecular electronic devices

In this review, we systematically summarize critical advancements in the field of molecular electronics, with a focus on device fabrication, functionalization strategies, and large-scale integration technologies, while providing insights into the future development of this area. First, we explore the fabrication technologies of molecular electronic devices, including molecular tunneling mechanisms, manufacturing methods for molecular junctions, atomically precise fabrication approaches, and functionalization of molecular electronic devices. Second, the preliminary attempts of the two-dimensional molecular devices integration are reviewed, further adopting from the conventional three-dimensional integration in the semiconductor industry by advanced package technologies, the strategy of 3D high-density integration of molecular electronic devices is proposed by combining bottom-up assembly and top-down manufacturing. Finally, from the dual perspectives of challenges and opportunities in molecular electronics, we outline future directions for integrated molecular electronic systems in potential applications such as storage and logic computing, as well as molecular sensing and characterization.

Fabrication of molecular electronics devices

Molecular Tunneling Mechanism

Molecular junctions, as the core building blocks of molecular electronic devices, exhibit distinct quantum characteristics in their charge transport behavior at the nanoscale³⁵. When the interelectrode distance is reduced to single-molecule dimensions ( < 3 nm), classical Ohm’s law fails, and quantum tunneling starts to dominate charge transport⁵. This section systematically elucidates the fundamental principles of molecular tunneling through two perspectives: charge transport mechanisms and quantum interference effects.

The construction of molecular junctions fundamentally relies on the formation of adjustable nanoscale gaps (0–3 nm) between metallic electrodes, with their electron transport characteristics governed by two key factors: the intrinsic electron transmission properties of the molecular backbone and the chemical coupling strength between terminal anchoring groups (e.g., thiol groups) and metal electrodes³⁶. In a prototypical metal-molecule-metal junction architecture, the conductance follows an exponential decay relationship with molecular length described by $G=A{{\rm{e}}}^{-\beta l}$, where $G$ represents the total junction conductance, $A$ corresponds to the vacuum tunneling conductance$\,{G}_{0}\,$(regarded as a quantum mechanical representation of contact resistance that directly correlates with molecule–electrode coupling strength), $l$ denotes the effective transport length of the junction, and $\beta$ represents the tunneling decay constant of the molecular framework. Depending on the nature of electron-molecule interactions, charge transport mechanisms in molecular junctions can be categorized into two fundamental regimes (Fig. 1a): coherent transport (preservation of quantum coherence in electron wavefunctions during molecular traversal) and incoherent transport (destruction of quantum coherence through energy dissipation processes)^7,37. From a theoretical perspective, the Non-Equilibrium Green’s Function (NEGF) methodology³⁸ provides a first-principle computational framework for precisely modeling quantum transport phenomena, enabling simultaneous resolution of molecular orbital energeis (static electronic structure) and dynamic characterization of quantum tunneling processes through molecular orbitals, including critical mechanisms such as quantum interference effects and inelastic scattering events, thereby achieving quantitative predictions of transport properties like I-V characteristics and conductance quantization behaviors. As a cornerstone of quantum transport theory, NEGF has become the standard computational approach in molecular electronics simulations^39,40,41, rigorously addressing the evolution of electron density matrices under non-equilibrium conditions to describe quantum transport processes in nanoscale molecular devices.

Fig. 1: Molecular tunneling mechanism and quantum interference phenomenon.

a Schematic diagram of direct tunneling and hopping tunneling, reproduced from ref.⁷, copyright 2022, with permission from Institute of Physics Publishing. b Quantum interference in benzene ring molecular junctions: Phase relationship of wave functions through two transport paths (R represents the anchoring group bound to the electrode). c Quantum interference-driven electron transmission probability (para: red, meta: green) vs energy. (b, c) reproduced from ref.⁴⁴, copyright 2017, with permission from Springer Nature

At the nanoscale, when multiple electron transport pathways coexist within a molecule, quantum state superposition between different orbitals induces quantum interference effects (QI) that substantially modulate conductance characteristics and govern charge transport behavior. This phenomenon not only establishes a crucial theoretical framework for understanding molecular-scale charge transport, but more importantly, opens new avenues for quantum manipulation of functional molecular devices^42,43. A paradigmatic example is observed in benzene-derived molecular systems: para-connected configurations create spatially symmetric dual-channel transport systems that maintain synchronized phase coherence in wavefunction superposition, generating constructive quantum interference (CQI) near the Fermi level manifested as broad electronic transmission plateaus, whereas meta-connected configurations exhibit π-phase differences in wavefunctions due to topological distinctions in transport pathways (Fig. 1b), triggering destructive quantum interference (DQI) that suppresses electron transmittance at specific energy levels (particularly around the Fermi level), resulting in dramatic conductance variations spanning orders of magnitude⁴⁴ (Fig. 1c). This quantum-interference-based conductance modulation mechanism demonstrates unique nonlinear response characteristics, with its pronounced quantum control effects providing essential physical implementation schemes for developing high on-off ratio and low-power molecular electronic devices.

Fabrication of molecular junctions

The construction of electrode–molecule–electrode junction systems stands as a central challenge in molecular electronics research. To achieve precise capture of target molecules within nanogaps, two primary fabrication strategies have been developed: static molecular junctions, fabricated via micro/nano-fabrication techniques with static electrode spacing, offering high structural stability and interfacial tunability; and dynamic molecular junctions, which enable repetitive formation-breakage cycles through controlled electrode displacement, generating extensive conductance datasets for statistical analysis to improve detection accuracy of molecular electric properties. The subsequent sections will systematically elaborate on the technical principles underlying these two approaches and conduct a comparative analysis of their strengths and limitations.

Static molecular junction

Static molecular junction devices, serving as pivotal platforms in molecular electronics, enable the exploration of charge transport mechanisms under quantum confinement effects and the development of nanoelectronic components surpassing silicon-based limitations by establishing precise conducting channels at single-molecule scale (typically <3 nm) between fixed electrodes. These devices are characterized by their non-volatile architecture: molecules are stably anchored within electrode gaps through covalent bond interactions^45,46 or non-covalent bond interactions^47,48, forming solid-state electronic interfaces with well-defined structure-property relationships. Compared to dynamic molecular junctions, static counterparts exhibit superior mechanical stability and electrical reproducibility. This section briefly introduces common methodologies for constructing static molecular junctions.

Electromigration junction

Electromigration refers to the phenomenon where momentum transfer between electrons and atoms under an applied electric field drives atomic migration against the current direction, resulting in void formation and eventual circuit failure in conductors. Long recognized as a major reliability threat to microelectronic devices⁴⁹, electromigration has recently been repurposed as a critical technique for fabricating nanogaps in molecular-scale devices^50,51,52. A common approach involves increasing the local current density at weak points of metallic nanowires to induce rupture, forming face-to-face source and drain electrodes (Fig. 2a, b). In 1999, Park et al.⁵³ first demonstrated electromigration-generated 1 nm nanogaps by progressively increasing the bias across the electron beam lithographic patterned gold nanowires. However, linear current ramping risks excessive Joule heating, which can melt nanowires and deposit metallic debris in gaps⁵⁴, interfering with molecular bridging and generating spurious signals. Additionally, the resulting nanogap sizes exhibit poor controllability and broad distributions, often yielding oversized gaps unsuitable for molecular junctions.

Fig. 2: Static molecular junction.

a Top and cross-sectional views of gold nanodot devices fabricated via electromigration and self-breaking techniques. b Current–voltage (I–V) characteristics of controlled electromigration in air. a, b reproduced from ref.³²⁵, copyright 2024, with permission from American Chemical Society. c Schematic of the EGaIn-exTTF-Au molecular junction. d Rectification mechanism of the EGaIn-exTTF-Au junction. c, d reproduced from ref.³²⁶, copyright 2024, with permission from Wiley. e Layered schematic of the single-walled carbon nanotube (SWCNT) OCGT device, reproduced from ref.³²⁷, copyright 2021, with permission from Wiley. f Graphene-based single-molecule transistor (SMT) architecture with a porphyrin molecule as the central component, reproduced from ref.²⁶, copyright 2024, with permission from Springer Nature. g DNA positioning principle: electron-beam or photolithographic patterning on HMDS-passivated substrates generates hydrophilic hydroxyl groups for selective DNA assembly. h SEM image of the hybrid SiO₂-AuNPs-DNA nanostructure assembly. g, h Reproduced from ref.¹²⁰, copyright 2023, with permission from Springer Nature

To address these limitations, researchers have optimized electromigration protocols for precise nanogap control. One notable strategy is self-breaking, where O’Neill et al.⁵⁵ pre-thinned gold nanowires to atomic-scale constrictions via programmed electromigration. Leveraging gold’s high room-temperature mobility, spontaneous rupture under zero bias produced nanogaps with minimal residual metal clusters. Another breakthrough is feedback-controlled electromigration (FBEM), pioneered by Strachan et al.⁵⁶ in 2005. By monitoring conductance as a real-time feedback signal to adjust applied bias, they achieved deterministic control over nanogap widths.

Advanced characterization techniques have elucidated electromigration dynamics. In 2007, Heersche et al.⁵⁷ performed in situ transmission electron microscopy (TEM) imaging, revealing asymmetric void nucleation near the cathode. Concurrently, Taychatanapat et al.⁵⁸ employed real-time scanning electron microscopy (SEM) to demonstrate that reducing series resistance suppresses metallic debris formation. Strachan⁵⁹ and Stöffler⁶⁰ further utilized TEM and scanning force microscopy (SFM) to monitor crystallographic plane evolution and energy profiles during electromigration, proposing mechanistic models to explain these observations.

SAM EGaln junction

Self-assembled monolayers (SAMs), highly ordered molecular layers spontaneously formed on solid substrates, typically require initial assembly on a bottom electrode followed by deposition of a top electrode to complete the molecular junction. Conventional methods involve evaporating solid metals as top electrodes⁶¹, however, gold atoms can permeate SAMs at room temperature, forming buried conductive layers that short-circuit devices⁶². Researchers have thus explored alternative strategies. Liquid mercury, compatible with diverse molecular systems and capable of forming non-damaging conformal contacts with molecular layers, emerged as a key solution. In 2001, Holmlin et al.⁶³ constructed Ag-SAM(1)-SAM(2)-Hg junctions using SAMs and liquid mercury to measure electron transport rates across organic films with varying molecular architectures. They demonstrated that junction conductance depends on SAM molecular structure and thickness, further comparing electron transport in SAMs bound via van der Waals interactions versus covalent, hydrogen, or ionic bonds⁶⁴.

Mercury’s high toxicity, volatility, and difficulty in forming submicron contacts spurred the search for safer alternatives. Whitesides et al.⁶⁵ firstly introduced an eutectic gallium-indium alloy (EGaIn) as a non-toxic substitute. EGaIn forms metastable non-spherical structures (e.g., conical/filamentous tips) at room temperature, exhibits high conductivity, and achieves low-contact-resistance interfaces with diverse materials⁶⁶. A typical protocol involves contacting a syringe-deposited EGaIn droplet with a silver film and retracting to form conical tips. The team⁶⁷ also studied EGaIn’s rheological behavior in microfluidic channels, demonstrating rapid channel filling under pressure and structural stability upon pressure release, providing critical insights for microfluidic integration. Wan et al.⁶⁸ later stabilized EGaIn in microfluidic devices to construct reproducible tunneling junctions with precise electrical characteristics. SAM-EGaIn junctions have become pivotal tools for studying molecular electronic properties⁶⁹ (Fig. 2c, d) such as rectification^70,71 and odd-even effects^72,73. Nijhuis et al.⁷⁰ tested rectification currents in thiol-based SAMs using SAM-EGaIn systems, modulating rectification mechanisms by spatially tuning conductive and insulating moieties within SAMs. Thuo et al.⁷² leveraged EGaIn’s conical tips to statistically compare charge transport in n-alkanethiols with odd versus even methylene groups, revealing higher current densities for even-numbered chains.

The native Ga₂O₃ layer on EGaIn surfaces critically influences functionality. In 2009, Nijhuis et al.⁷⁴ proposed that Ga₂O₃’s non-Newtonian properties stabilize conical tips for reliable SAM contacts. Subsequent studies by Cademartiri⁷⁵ and Reus⁷⁶ characterized Ga₂O₃’s properties, including an average thickness of ~0.7 nm and resistivity, confirming SAM-dominated charge transport with negligible contributions from the oxide layer.

This review concentrates on the origins and evolution of SAM EGaln junction technology. For extended discussions on its properties, applications, and recent advances, consult the 2023 review article by Zhao et al.⁷⁷.

Carbon nanotube junction

Metal electrodes, while pivotal in molecular electronics, face several critical challenges: (a) high interfacial contact barriers with molecules, leading to inefficient charge injection; (b) enhanced quantum tunneling effects in ultrathin metallic films, which obscure weak molecular signals; (c) lack of directional chemical bonding sites on surfaces, resulting in device performance variability⁷⁸. Carbon nanomaterials, including carbon nanotubes (CNTs) and graphene, have emerged as promising alternatives due to their unique properties. As low-dimensional molecular-scale systems, CNTs bridge fewer molecules, enabling single-molecule junction formation. Their π-π conjugated frameworks provide superior conductivity, while functionalizable termini allow covalent bonding for exceptional mechanical, thermal, and chemical stability^79,80.

Discovered by Iijima et al. in 1991⁸¹, CNTs exhibit metallic or semiconducting properties without passivated interfacial states, facilitating integration with high-κ dielectrics (Fig. 2e)^82,83. Three primary methods have been developed to create molecular-scale gaps in CNTs: electrical breakdown⁸⁴, oxygen plasma cutting⁸⁵, and focused ion beam (FIB) etching. In 2001, Collins et al. discovered that multi-walled CNTs (MWNTs) undergo stepwise failure through rapid oxidation of outer carbon shells when biased beyond power thresholds^86,87. Building on this, Tsukagoshi et al.⁸⁸ applied high currents to MWNTs bridging Pt/Au electrodes, generating sub-50 nm gaps for pentacene-based organic nanotransistors via thermal evaporation. While electrical breakdown enables rapid nanogap fabrication, its poor controllability results in random gap sizes. In 2008, Wei et al.⁸⁹ combined electrical breakdown with electron-beam-induced deposition (EBID), using SEM beams to refine CNT gaps while performing in situ characterization for controlled nanogap formation.

Oxygen plasma cutting, first reported by Guo in 2006⁹⁰, involves e-beam lithography to pattern PMMA-coated single-walled CNTs (SWNTs), followed by oxygen plasma exposure to create sub-10 nm gaps with carboxyl-terminated electrodes. These carboxyl groups capture amine-functionalized molecules via stable amide bonds, yielding robust molecular junctions resilient to environmental fluctuations. FIB etching, demonstrated by Roy et al.⁹¹, enables precision gap control in SWNTs through ion beam current and exposure time optimization, with subsequent nitric acid treatment for carboxylation.

Comparative analysis reveals tradeoffs: electrical breakdown offers simplicity but poor control; plasma cutting suits scalable integration yet requires e-beam lithography for precision; FIB etching achieves atomic-scale accuracy at high costs. Future advances demand synergy between novel materials (e.g., 2D materials, supramolecular conductors) and atomically precise fabrication (e.g., DNA-mediated assembly) for high-performance scalable molecular electronics.

The rise of CNT research paralleled investigations into DNA’s charge transport, which exhibits insulator-to-metal transitions sensitive to environmental factors^92,93,94,95. Guo et al.⁹⁶ resolved measurement challenges by covalently linking amine-modified DNA to SWNTs via amide bonds, enabling the first direct conductivity measurements. CNT-DNA hybrids have advanced biomedical applications: Lee’s group⁹⁷ developed single-molecule field-effect transistor (smFET) biosensors in 2024 by functionalizing CNTs with DNA aptamers, achieving serotonin detection at single-molecule levels with sub-second resolution.

Graphene junction

Graphene, another critical carbon-based electrode material (Fig. 2f), consists of a single-layer two-dimensional structure formed by sp²-hybridized carbon atoms arranged in a tightly packed honeycomb lattice. It exhibits unique physical and electronic properties^98,99,100: atomic-scale stability at room temperature, ultrahigh carrier mobility with minimal sensitivity to chemical doping, natural compatibility with molecules for low-resistance contacts to reduce switching time, reduced gate screening effects compared to 3D electrodes due to its 2D nature, and superior suitability for large-scale integration into arrays compared to carbon nanotubes (CNTs).

Two primary methods have been developed to fabricate graphene nanogaps: electroburning^101,102,103 and oxygen plasma etching. In 2011, Prins et al.¹⁰³ demonstrated feedback-controlled electroburning at room temperature to create 1–2 nm gaps, with subsequent molecular assembly yielding graphene-based junctions exhibiting robust gating characteristics. The oxygen plasma etching approach, first reported by Guo’s team in 2012¹⁰⁴, introduced a “dash-line lithography” technique: photolithographically patterned graphene was selectively etched by oxygen plasma through a PMMA mask, producing carboxyl-terminated edges for stable covalent amide bonds with target molecules with an over 50% yield.

While electroburning offers higher yield, its stochastic nature limits positional control of nanogaps. Oxygen plasma etching enables wafer-scale nanogap arrays with precise positioning but suffers from lower yields. In 2014, Lau et al.¹⁰⁵ synergized both methods by transferring graphene onto lithographically patterned metal substrates for plasma etching, followed by feedback-controlled electroburning at the narrowest regions of pre-etched grooves. This hybrid strategy achieved 0.5-2.5 nm gaps with 71% yield and sub-nanometer positional accuracy, paving the way for scalable graphene device integration.

Post-Moore’s Law demands for exponential growth in data processing have driven structural optimizations of graphene devices. In 2020, Rong et al.¹⁰⁶ modulated graphene’s electronic structure via nitrogen doping, enabling reversible molecular assembly for high-sensitivity chemical switches. By 2024, Liu et al.¹⁰⁷ developed a graphene/germanium Schottky junction-based hot-carrier transistor, leveraging diverse thermal carrier generation mechanisms to populate high-energy states, achieving enhanced speed and reduced power consumption.

DNA origami positioning junction

DNA origami technology, leveraging biomolecular self-assembly, bridges molecular-scale precision with micro/nanofabrication capabilities, offering scalable solutions for highly integrated molecular devices. First proposed by Rothemund in 2006¹⁰⁸, this technique employs short staple DNA strands to fold long scaffold DNA into predefined 2D shapes. Douglas et al. extended this concept to 3D architectures in 2009¹⁰⁹, enabling spatial positioning beyond planar constraints. DNA origami serves as programmable templates for organizing nanowires, carbon nanotubes, or gold nanoparticles via specific interactions^{110,111,112,113,114,115}.

In 2010, Hung et al.¹¹⁶ selectively deposited DNA origami-gold nanocrystal hybrids onto lithographically patterned substrates, achieving large-scale ordered 2D nanoparticle arrays. However, uncontrolled deposition of origami solutions necessitated advanced positioning strategies. Kershner et al. addressed this in 2009¹¹⁷ by creating DNA origami binding sites on SiO₂ and diamond-like carbon substrates via e-beam lithography and dry oxidation, immobilizing origami through Mg²⁺-mediated electrostatic bridging. Despite progress, high Mg²⁺ concentrations risked component aggregation and material-specific limitations.

Gopinath et al. introduced covalent bonding strategies (CDAP and CTES/EDC) in 2014^118,119, replacing electrostatic interactions with stable covalent linkages for Mg²⁺-free origami arrays. By 2023, Martynenko et al.¹²⁰ advanced 3D positioning by assembling DNA origami onto substrates followed by silicification, enabling vertical alignment (Fig. 2g, h). Similarly, Teng et al. demonstrated programmable 3D origami superlattice assembly on oxide surfaces in 2025¹²¹.

DNA origami-enabled nanogaps facilitate microscopic studies of molecular optical properties and chirality^122,123,124. In 2018, Gür et al.¹²² achieved sub-2 nm gold nanoparticle spacings using DNA origami, suppressing propagation losses inherent in lithographic techniques. Ma et al.¹²³ constructed quasi-planar chiral satellite-core nanostructures in 2022, surrounding a central AuNP with six AuNRs on 2D origami sheets, revealing in-plane and out-of-plane chiral interactions under optical spin excitation.

Dynamic molecular junction

Dynamic junction techniques, highly sensitive methods for precise single-molecule conductance measurement and dynamic behavior analysis, enable real-time electrode spacing modulation through mechanical feedback or electromechanical control. Compared to static junction approaches, dynamic techniques offer distinct advantages: precise electrode gap control via real-time tip positioning to ensure stable molecular bridging; high stability and noise immunity; dynamic process monitoring to capture conductance evolution during junction formation, revealing molecular conformational changes and rupture kinetics; and statistical reliability through thousands of contact-withdrawal cycles for robust conductance histogram analysis. Currently reported dynamic junction platforms include STM-BJ (scanning tunneling microscopy break junction), AFM-BJ (atomic force microscopy break junction), MCBJ (mechanically controllable break junction), and MEMS-BJ (microelectromechanical system break junction), each of which will be elaborated in detail below.

STM-BJ

STM-BJ evolved from conventional scanning tunneling microscopy (STM) through instrumental modifications, employs a dynamic electrode pair comprising a metallic tip and gold substrate. Piezoelectric ceramics control tip retraction to create nanoscale gaps, which capture bridging molecules to form metal-molecule-metal junctions when the gap matches molecular dimensions (Fig. 3a, b)¹²⁵. Xu et al. pioneered this technique, enabling rapid acquisition and screening of electrical signals through repeated molecular capture-dissociation cycles. Conductance histograms statistically analyze thousands of conductance traces, extracting intrinsic conductance values for molecules such as 4,4′-bipyridine^126,127. Beyond direct tip-substrate contact, Haiss et al.¹²⁸ demonstrated a “soft-contact” mode where the tip hovers near the substrate without physical contact, retracting only after molecular bridging.

Fig. 3: Dynamic molecular junction.

a Schematic of a molecular junction formed via STM-BJ technique. b One-dimensional conductance histograms of L2-PF₆ under different bias voltages applied between a gold tip and substrate. a, b reproduced from ref.³²⁸, copyright 2024, with permission from Springer Nature. c Structural diagram of a single-molecule junction measured using AFM-BJ. d Two-dimensional force-displacement histogram derived from single-molecule junction rupture point measurements. c, d reproduced from ref.³²⁹, copyright 2023, with permission from Wiley. e Upper panel: Supramolecular diode constructed from Py/NDI (pyrene/naphthalene diimide); lower panel: MCBJ fabrication process. f Representative current-voltage (I–V) curves corresponding to distinct conductance states. e, f reproduced from ref.³³⁰, copyright 2025, with permission from American Chemical Society. g Schematic of the MEMS-BJ platform. h Conductance histogram for 8,653 unselected tapping curves of gold-gold contacts, with six representative junction formation curves illustrated in the inset. g, h reproduced from ref.¹⁸⁵, copyright 2020, with permission from Wiley

STM-BJ has become indispensable for probing electronic transport properties of molecular junctions^{47,129,130,131,132}, catalyzing advancements in molecular electronics. In their 2025 study, Jia’s team¹²⁹ employed sensitive STM-BJ techniques combined with SAPT theoretical calculations to analyze and further modulate complex vdW interactions within single-dimer junctions. They also revealed that applied electric fields significantly alter dimer configurations and conductance. Separately, Li’s group¹³⁰ utilized ECSTM-BJ to measure and regulate molecular energy levels, demonstrating the role of antiaromaticity in single-molecule charge transport. This work offers new insights for designing bipolar devices.

Beyond charge transport, STM-BJ has expanded into spin electronics, thermal transport, nanolithography, and optoelectronics. Li et al.¹³³ developed an electrochemically assisted jump-to-contact STM-BJ (JTCSTM-BJ) method, fabricating robust Fe-terephthalic acid (TPA)-Fe spin junctions. Applying in-plane or out-of-plane magnetic fields, they observed a 53% giant tunneling anisotropic magnetoresistance (T-AMR), advancing atomic-scale spin studies. Addressing thermal transport challenges, Lee et al.¹³⁴ integrated customized probes with nanothermocouples, revealing asymmetric heat dissipation in molecular junctions dependent on bias polarity and majority carrier type.

In 2023, Lu et al.¹³⁵ devised a coaxial focused electrohydrodynamic jet (CFEJ) printing technique using STM probes, enabling large-scale, precise fabrication of organic semiconductor nanowires on flexible substrates without lithography. For photocurrent studies, Imai-Imada et al.¹³⁶ combined STM with tunable laser-driven localized plasmonic fields in 2022, exciting single free-base phthalocyanine (FBPc) molecules to detect first-excited-state tunneling electrons, demonstrating bias-sensitive control over photocurrent directionality and spatial distribution.

AFM-BJ

AFM-BJ represents another critical approach for nanogap formation using scanning probe microscopy. Unlike STM-BJ, which relies on tunneling current feedback to adjust electrode spacing and risks molecular structural damage from excessive probe forces during repeated stretching, AFM-BJ employs a force feedback mechanism. This system maintains non-contact between the conductive probe and substrate while monitoring cantilever deflection to track mechanical responses during molecular stretching, thereby minimizing molecular compression and tunnel current-induced damage (Fig. 3c, d).

In 2000, Wold et al.¹³⁷ pioneered conductive AFM (CP-AFM) by coating AFM tips with metal to form metal-molecule-metal junctions with alkanethiol monolayers on gold substrates, enabling AFM-based electronic transport measurements^{138,139,140,141,142}. Beyond electronics, AFM’s unique mechanical sensitivity facilitates studies of molecular mechanical properties—chemical bond strength, intermolecular forces, and force-electric coupling effects¹⁴³. Frei et al.¹⁴⁴ employed modified CP-AFM to investigate rupture forces in alkanes terminated with four distinct chemical groups, revealing that thiol-terminated molecules induce electrode restructuring via Au-S bond formation, yielding unique conductance and mechanical signatures.

AFM also advances nanolithography¹⁴⁵ and molecular imaging¹⁴⁶, though conventional metal-coated tips suffer from limited lateral resolution due to conical geometry and tip degradation^147,148. Carbon nanotube (CNT)-functionalized AFM probes offer superior wear resistance, nanoscale diameter, and high aspect ratios, yet CNT growth methods face complexity and poor controllability^{149,150,151,152}. In 2021, Cheng et al.¹⁵³ achieved a 93% yield in synthesizing single high-quality CNTs on AFM tips by adjusting trigger thresholds for solution-based growth, enabling precise measurements of intricate nanostructures with enhanced aspect ratios and resolution.

MCBJ

MCBJ generate molecular junctions through lateral electrode stretching, distinct from the longitudinal configurations formed by SPM-BJ techniques. The conceptual precursor to MCBJ emerged in 1985 when Moreland et al.¹⁵⁴ fractured a Nb–Sn filament on a flexible glass substrate to create an electronic tunneling junction. Müller et al.^155,156 formally introduced the MCBJ concept during investigations of quantum conductance in noble metals. In 1997, Reed et al.¹⁵⁷ first applied MCBJ to molecular electronics, successfully measuring the conductance of benzene-1,4-dithiol.

The core functional unit of a classic MCBJ system typically consists of a pre-notched metallic microbridge fabricated via lithography then electron-beam deposition on a flexible polyimide substrate, anchored by symmetrically distributed support structures. During operation, a piezoelectric actuator applies nanoscale vertical displacement (Z-axis), transmitting substrate deformation to the microbridge via a three-point bending mechanism. Stress concentration at the notch induces plastic deformation, and continued bending progressively stretches the metallic bridge until fracture, producing a pair of atomic-scale electrodes with precisely controlled spacing (Fig. 3e, f). The system achieves nanoscale precision through macroscopic mechanical control, offering unique advantages in single-molecule studies: (a) the elastic substrate and closed-loop piezoelectric control provide superior mechanical stability compared to STM-BJ^158,159,160; (b) pristine molecular junctions form without additional chemical treatments^161,162; (c) sub-angstrom reversible gap adjustment enables repeated junction formation^163,164,165.

Reported MCBJ fabrication methods fall into three categories: mechanical cutting, electrochemical deposition, and nanolithography. Mechanical cutting, an early low-cost approach, involves manually scoring epoxy-fixed metal wires on flexible substrates^155,161. While simple and equipment-minimal, this method suffers from contamination risks, poor mechanical stability, and insufficient precision for molecular-scale studies. Electrochemical deposition significantly improved these limitations. Yi et al.¹⁶⁶ pre-patterned wide electrode gaps on silicon substrates in 2010 and narrowed them via electrochemical gold deposition under PMMA encapsulation, producing stable nanoelectrodes with reduced contact areas. To enhance vibration resistance, nanolithography was integrated into MCBJ fabrication. van Ruitenbeek et al.¹⁶⁷ employed electron-beam lithography with PMMA resist in 1996 to fabricate Al and Cu electrodes, achieving high reproducibility and scalability^168,169.

A critical advancement was the integration of gate electrodes, transforming two-terminal into three-terminal devices. Two-terminal configurations only probe molecular states relative to one electrode’s Fermi level, whereas three-terminal systems enable independent tuning of molecular states relative to both electrodes. Existing gate designs include back-gate and side-gate architectures. In 2005, Champagne et al.¹⁷⁰ pioneered back-gate structures using degenerately doped Si substrates coated with SiO₂, followed by e-beam lithography and HF etching to suspend metallic nanowire with a sub-40 nm molecule-gate distance. By 2013, Xiang et al.¹⁶⁹ developed side-gate architectures, positioning non-contact electrodes nanometers from molecular junctions to directly modulate charge transport via applied electric fields without mechanical compromise.

MCBJ’s structural versatility facilitates broad applications in molecular conductance measurement^{171,172,173,174,175}, spin-state control^176,177,178, and vibrational studies^179,180. Its compatibility with complementary techniques further enhances functionality. Combined MCBJ-Raman spectroscopy¹⁸¹ demonstrated accelerated nanoscale Menshutkin reactions via rational system design, exceeding macroscopic benchmarks by 10⁴-fold and substantiating electrostatic catalysis mechanisms at confined interfaces. Concurrently, Zhao et al.¹⁸² developed a fiber-based break junction (F-BJ) technique, enabling in situ probing of single-molecule photoconductance through optical waveguides and establishing a novel paradigm for light-molecule interaction studies at the single-molecule level.

MEMS-BJ

Current techniques for constructing molecular junctions face inherent limitations such as operational complexity and low yield, driving researchers to develop more reliable methodologies. Microelectromechanical Systems (MEMS), a novel technology that integrates microscale mechanical structures, sensors, actuators, and electronic circuits onto a single chip via micro/nanofabrication¹⁸³, offers a promising solution. The core principle involves using semiconductor processes, such as photolithography, thin-film deposition, and ion etching, to fabricate micron-scale (1–100 μm) mechanical components on diverse substrates, enabling synergistic operation with signal-processing circuits. MEMS excels in miniaturization, low power consumption, cost-effective batch production, and multifunctional integration. Its fabrication workflow is compatible with tranditional semiconductor manufactering processes, allowing mass production of tens of thousands to millions of devices on a single wafer, significantly reducing per-unit costs¹⁸⁴.

In 2020, Jeong et al.¹⁸⁵ pioneered the application of MEMS to molecular junction fabrication, termed MEMS-BJ. They employed this technique to study the electronic properties of 1,6-hexanedithiol and biphenyldithiol, as well as junction stability under ambient and cryogenic conditions (Fig. 3g, h). However, MEMS development faces challenges in multi-physics (mechanical-electrical-thermal) coupling design complexity, encapsulation reliability (e.g., dust/humidity resistance), and high-precision manufacturing consistency, necessitating further exploration and optimization.

Similarly, Scheer et al.¹⁸⁶ developed two methods for in situ tuning of in-plane electrode gaps with Ångström-scale resolution: leveraging laterally expanding piezoelectric sheets or deformable polymeric substrates. This technology evolved into planar molecular break junctions capable of >10,000 reversible opening/closing cycles, enabling reproducible electrical characterization of benzenedithiol (BDA) molecules. This approach provides critical insights for bridging single-molecule science and nano-optical field modulation technologies.

In summary

STM-BJ plays a pivotal role in single-molecule electron transport studies due to its operational simplicity and ease of implementation, making it the most widely adopted method. AFM-BJ enables quantitative bond energy analysis via synchronized mechano-electrical detection, yet operational complexity restricts its broad adoption. MCBJ excels in sub-picometer control and wide-temperature operation, ideal for quantum transport under extreme conditions, but requires complex and expensive micro/nanofabrication processes for device preparation, which are time-consuming. MEMS-BJ enables high-throughput automation and chip integration, yet the suspended MEMS structures constrain applications in solution phase environment for reliable BJ measurements, which introduces capillary forces that can potentially damage the MEMS cantilevers. Technique selection should thus align with specific research objectives.

Atomic manufacturing methods

Synergistic interface and intermolecular interactions

The construction of molecule–electrode interfaces plays a decisive role in forming stable and reliable molecular junctions, as interfacial interactions not only determine coupling strength but also govern charge transport properties and mediate environmental modulation of electron transfer^187,188,189. These interfaces typically involve molecules functionalized with anchoring groups that bind to electrodes via covalent or non-covalent interactions. The concept of covalent bonds, first proposed by Langmuir in 1919¹⁹⁰, describes strong intermolecular interactions formed through shared electron pairs. In contrast, non-covalent bonds, including coordination bonds, hydrogen bonds, van der Waals forces, and other secondary interactions, exhibit weaker strength but critically influence interfacial behavior. The following sections elaborate on these two interaction classes within molecular junctions.

Covalent interactions

Covalent bonding interactions primarily utilize diverse terminal functional groups as anchoring moieties to bridge molecular junctions between electrodes. Thiol groups (–SH), the earliest and most extensively studied anchoring groups^45,191, dominate due to their straightforward synthesis and strong Au–S bonding affinity, which enhances junction formation probability. However, Au–S covalent bonds in single-molecule applications face limitations: (a) Ideal bond formation occurs exclusively in solution due to hydrogen’s critical role in thiol functionality, restricting environmental adaptability; (b) Thiol oxidation alters molecular structures, destabilizing junctions; (c) The higher Au-S bond strength relative to Au–Au induces electrode surface restructuring, complicating interfacial properties⁴⁶.

To address these challenges, researchers have explored alternative anchoring groups. In 2010, Xing et al.¹⁹² substituted thiols with dithiocarboxylate groups (–CSSH) in phenylethynylene molecular wires, observing enhanced electrode coupling and reduced charge transport barriers. Zeng et al.¹⁹³ demonstrated in 2020 that thioacetyl groups (R-SAc) enable selective junction formation in mixed solvents, with applied bias modulating molecular dipole orientation to improve yield (Fig. 4a, b). Similarly, methylthiol (R-SMe)¹⁹⁴ and thiophenyl groups¹⁹⁵ exhibit analogous selectivity.

Fig. 4: Molecular interactions and self-assembly.

a Schematic illustration of the STM-BJ device and the construction of molecular junction under an external electric field and the anchoring groups corresponding to M1-M5, respectively. M1 and M2 (-SMe), M3 (-SAc), M4 (-NH₂), and M5 (-PY). b Plots of the junction formation probability of M1 (green), M2 (red), M3 (purple), M4 (orange), and M5 (blue) versus applied biases. The error bar is determined from the chip-to-chip variation of three independent experiments. a, b reproduced from ref.¹⁹³, copyright 2020, with permission from Wiley. c Schematic of the single-molecule junction constructed by WTe₂ electrodes and the molecular structure of M-TPP. d 1D conductance histograms for M-TPP (M = Fe, Co, Ni, Cu, Zn) and TPP single-molecule junctions. c, d reproduced from ref.²¹², copyright 2023, with permission from American Chemical Society. e Schematic illustration of the preparation of photosensitizer-assembled nanocluster. f Representative TEM photographs of C3TH nanocluster. e, f reproduced from ref.²²⁰, copyright 2025, with permission from Wiley

Nitrogen-based groups, including amines (R-NH₂)¹⁹⁶, pyridines (R-Py)¹⁹⁷, and imidazoles¹⁹⁸, represent a second major category. In 2006, Venkataraman et al.¹⁹⁹ showed that diamine-terminated molecules form isotropic Au-N bonds via lone-pair interactions, yielding consistent conductance across junction types, resolving variability issues in dithiol-Au systems.

Carbon-based anchors, such as alkynes²⁰⁰, methylene groups²⁰¹, and fullerenes¹⁷³, offer another pathway. Hong et al.²⁰² pioneered in 2012 the use of trimethylsilyl (TMS)-protected alkynes, which generate Au-C σ-bonds via in situ cleavage. These junctions exhibit higher conductance than thiol-based analogs while eliminating toxic precursors.

Oxygen-containing groups like carboxyl (–COOH)²⁰³, nitro (–NO₂)²⁰⁴, and ester/ether derivatives²⁰⁵ complete the fourth category. Building on earlier carboxyl-Au studies, Peng et al.²⁰⁶ employed ECSTM-BJ to systematically analyze alkanedicarboxylic acids on Cu/Ag electrodes, revealing conductance dependencies on molecular length and electrode-specific electronic coupling.

Non-metallic electrodes expand interfacial design. Carbon-based interfaces often utilize carboxyl-amine amide bonds, as detailed in prior CNT/graphene junction discussions. Vezzoli et al.²⁰⁷ demonstrated GaAs–molecule–Au junctions via As–S covalent bonds, where GaAs doping modulates electronic properties, while direct bandgap characteristics enable optoelectronic applications. Silicon electrodes, explored by Peiris et al.²⁰⁸, form durable Si–C bonds with diazonium radicals, achieving room-temperature junction lifetimes (1.1 s) exceeding Au–S bonds. These mechanically stable junctions enable precise I–V characterization and rectification control, circumventing Au’s intrinsic limitations.

Non-covalent interactions

Non-covalent interactions, though weaker than covalent bonds, enable spontaneous formation of ordered structures under mild conditions (e.g., solution-phase self-assembly) and permit reversible structural modulation of molecular junctions in response to external stimuli such as light, heat, electric fields, or chemical environments. This dynamic tunability addresses the rigidity of covalent bonds, opening new dimensions for functional molecular device design. This section briefly introduces common non-covalent interactions in molecular junctions: coordination bonds, van der Waals forces, hydrogen bonds, and π-π interactions.

Coordination bonds in molecular devices typically involve terminal groups acting as electron donors, pairing lone electron pairs with vacant orbitals of metal electrodes¹⁸⁸. In 2019, Vladyka et al.²⁰⁹ demonstrated in situ formation of 1D coordination polymers within junctions using MCBJ. The high affinity and anisotropic binding of isocyanide-terminated groups to gold atoms significantly improved junction formation efficiency.

Hydrogen bonds, intermediate in strength between covalent and van der Waals interactions, balance structural stability with responsiveness to temperature, solvent polarity, or pH. This allows dynamic control over junction properties²¹⁰. In 2016, Wang et al.²¹¹ investigated quadruple hydrogen bonds in UPy derivatives functionalized with thiol (–SH), pyridyl (–Py), and amino (–NH₂) groups. Thiol-terminated molecules exhibited the highest conductance, while solvent polarity modulated hydrogen bond formation. Directional specificity of multiple hydrogen bonds enhanced binding strength and preserved molecular conjugation.

Van der Waals forces, the weakest yet most ubiquitous interactions, govern junction assembly and dynamics via transient dipole-induced attractions. In systems lacking strong covalent/hydrogen bonds, their nondirectional and short-range nature enables weak coupling for self-assembly and synergizes with other forces to enhance mechanical stability. In 2023, Lu et al.²¹² constructed transition metal dichalcogenide (TMDC) molecular devices solely via van der Waals interactions, avoiding interfacial chemical bonds (Fig. 4c, d). This approach boosted conductivity (~736%) and enabled broad conductance modulation by varying metal centers.

π–π interactions, driven by delocalized π-electron clouds in aromatic systems, exhibit strength surpassing van der Waals forces but below covalent bonds. Their directional and distance-sensitive nature promotes ordered self-assembly of π-conjugated molecules (e.g., porphyrins, acenes, fullerene derivatives) through π-orbital overlap, reducing interfacial energy barriers and enhancing charge delocalization for improved conductivity and carrier mobility²¹³. In 2015, Yoshida et al.²¹⁴ used conductive scanning AFM (CSAFM) to track conductance and rupture forces in aromatic diphenylacetylene derivatives. Pyridyl-terminated molecules formed high-conductivity coplanar connections with gold electrodes via π–π interactions, exhibiting lower conductance and rupture forces compared to other terminal groups.

Self-assembly

Non-covalent interactions between molecules drive the self-assembly of molecular devices, enabling spontaneous formation of ordered structures with specific spatial arrangements and functionalities without external intervention (Fig. 4e, f). The essence of self-assembly lies in leveraging weak intermolecular forces to achieve macroscopic-scale order. This bottom-up strategy transcends the resolution limits of traditional lithography, offering energy-efficient pathways to fabricate electronic devices with molecular-scale precision.

Biological macromolecules, such as DNA and proteins, play pivotal roles in self-assembled molecular junctions due to their inherent molecular recognition capabilities and structural programmability²¹⁵. DNA’s utility, particularly through DNA origami, has been extensively discussed in prior sections on nanogap fabrication. Proteins, with their precise and complex architectures, can guide directional alignment of biomolecular components via specific interactions, forming stable junctions. In 2005, Alessandrini et al.²¹⁶ demonstrated the feasibility of constructing single-molecule bio-transistors using disulfide-bonded Azurin proteins under physiological humidity. They further showed that altering the protein’s oxidation state could modulate rectification. By 2013, Raccosta et al.²¹⁷ enhanced this design by attaching 5 nm gold nanoparticles (GNPs) to Azurin proteins, assembling them onto gold electrodes. Conductive probe atomic force microscopy (CP-AFM) revealed that this optimized top-contact configuration improved current stability and reduced the mechanical force required for reliable electrical characterization.

Beyond natural biomolecules, zwitterionic materials have emerged in recent studies as promising candidates for self-assembled molecular junctions. These materials, featuring spatially separated positive and negative charges, often serve as interfacial mediators. Their unique charge distribution enhances junction stability through synergistic non-covalent interactions with electrodes or adjacent molecules. Zwitterions’ high hydrophilicity and anti-fouling properties mitigate environmental interference from moisture or contaminants, while their bipolar charge distribution suppresses electrostatic repulsion-induced aggregation, promoting densely packed monolayers for homogeneous junctions²¹⁸. In 2014, Wu et al.²¹⁹ explored the role of hydrophilic counteranions in directing the self-assembly of cationic bola-amphiphiles. They found that zwitterionic interactions facilitated 2D planar aggregates, with aggregate morphology tunable via counteranion substituents and organic chain lengths, offering strategies for controlled molecular device assembly. A study by He et al.²²⁰ highlighted zwitterionic self-assembly in biomedical applications. They designed nanoclusters from self-assembling zwitterionic small molecules, where cross-aligned molecular arrangements minimized intermolecular electron-vibration coupling. This configuration enhanced light-induced electron transfer, triggering self-ionization to generate cationic and anionic PS radicals with potent antimicrobial efficacy.

Functional molecular electronic devices

In the field of molecular electronics research, its central scientific objective focuses on constructing single-molecule-scale devices with tailored electrical functionalities through atomic precision manufacturing techniques, combined with the precise design of molecular architectures and synergistic modulation of interfacial states, ultimately aiming to realize integrated circuit systems based on single-molecule units. Although current technologies have yet to fully overcome critical bottlenecks in device integration, significant progress has been achieved at the fundamental component level. This advancement is attributed to continuous breakthroughs in the foundational theories of molecular electronics and innovations in characterization techniques. Core functional units such as molecular switches with reversible switching characteristics, molecular diodes demonstrating rectification effects, and gate-voltage-modulated molecular transistors have been successfully implemented. This section will systematically review recent research advances in major functional molecular devices, emphasizing their operational mechanisms and technological implications.

Molecular switch

The electronic switch stands as one of the most pivotal components in circuits, operating through the reversible switching of materials between two stable states. Molecular switches achieve conductance modulation via transitions in molecular conformation or electronic states. As detailed in recent reviews²²¹, this necessitates functional molecules to exhibit at least two stable isomers that can reversibly interconvert under external stimuli²²². For example, optical stimulation induces photoisomerization, charge manipulation alters molecular charge states through redox reactions, and mechanical stimuli provoke conformational changes via stretching or compression. Notably, molecular switches can be simultaneously triggered by multiple stimuli, enabling synergistic effects²²³. These mechanisms furnish diverse switching strategies for molecular electronics, propelling the advancement of highly integrated nanodevices^224,225.

Photoischromic molecular switches enable conductance modulation through photoresponsive conformational changes of molecules, whose operational mechanisms hold significant research value in the field of molecular electronics. Taking the diarylethene system as an example, research teams have substantially reduced the effective coupling strength between the molecule and electrodes by introducing three methylene (CH₂) units on both sides of the molecular backbone. Experimental results demonstrate that under a gate voltage of 0 V and source-drain bias of ±1 V, the current switching ratio of this molecular switch can be enhanced to approximately 100-fold. As illustrated in Fig. 5a, when single-molecule switch devices are exposed to UV and visible light irradiation, the diarylethene molecules exhibit reversible interconversion between closed-ring and open-ring configurations, with current response curves displaying pronounced real-time switching behavior²²⁶. Furthermore, in photo-controlled molecular junction systems, azobenzene derivatives have emerged as critical research targets due to their synthetic accessibility and chemical stability²²⁷. These molecules consist of two benzene rings connected by an azo (–N = N–)-bond, whose trans-cis configurational isomerization can be reversibly regulated via light irradiation. Studies confirm that under low-bias conditions, this isomerization process triggers asymmetric stochastic conductance switching phenomena²²⁸. Additionally, spiropyran systems featuring dual photo/chemical responsiveness exhibit unique isomerization modulation mechanisms. Experimental investigations reveal that bifunctional spiropyran derivatives achieve rapid reversible isomerization transitions in bulk solutions through photoexcitation or chemical stimuli, accompanied by significant conductance switching between high and low conductance states²²⁹.

Fig. 5: Photoischromic molecular switches.

a Schematic of a graphene-diarylethene-graphene reversible photoswitching molecular device (left) and I–V curve of open/closed state and photocurrent response curve(right), reproduced from ref.²²⁶, copyright 2016, with permission from American Association for the Advancement of Science. b Schematic illustration of the electrochemically gated break junction experiment (left) and molecular structure of NDI-BT in the neutral state, radical-anion state and dianion state (right), reproduced from ref.²³¹, copyright 2015, with permission from Wiley. c Schematic illustrating the reversible transformation of a neutral or negatively charged Sn-up molecule via a planar molecule geometry to a Sn-down molecule (left) and time series of tunneling current measured on a Sn-up molecule with open feedback loop (right), reproduced from ref.²³², copyright 2009, with permission from American Chemical Society. d Schematic of the hydrogen tautomerization reaction responsible for the switching (left) and current trajectory was obtained when the tip of STM was located at one end of the molecule (right), reproduced from ref.²³³, copyright 2020, with permission from American Chemical Society. e As the molecule unfolds/refolds, the average conductance reversibly changes over three orders of magnitude, reproduced from ref.²³⁵, copyright 2011, with permission from American Chemical Society. f Schematic of a single-molecule pulling/molecular electronics setup (left) and the change of conductance corresponding to the molecular conformation change (right), reproduced from ref.²³⁶, copyright 2011, with permission from American Chemical Society

Charge-regulated molecular switches can be categorized into three distinct classes based on their redox mechanism: electrochemical gating systems, charge-transfer-driven systems, and electron-induced switching systems²³⁰. The core mechanism of electrochemical gating lies in the precise modulation of the electrode Fermi level to alter the redox state of molecular functional units, thereby achieving high-precision switching functionality. Taking the NDI-B molecular system as an example (Fig. 5b), this molecule enables reversible transitions between neutral (NDI-N) and dianionic (NDI-D) states through electrochemical gating techniques. Experimental studies confirm that this system exhibits tri-state charge storage characteristics, with marked conductivity differences between charge states, and allows reversible switching between any two states via fine-tuned molecular potential control²³¹. In the domain of charge-transfer-driven switches, research demonstrates that phthalocyanine molecular layers adsorbed on metal electrode surfaces can establish bistable switching systems. These systems induce reconstruction of intramolecular charge distribution through resonant electron or hole injection into molecular orbitals, subsequently triggering vertical displacement modulation of the central tin ion. Ordered lamellar array architectures based on this charge-transfer mechanism have achieved reversible conductance switching at the single-molecule scale for the first time²³² (Fig. 5c). Electron-induced switching operates through current-driven energy excitation to activate molecular conformational transitions. As shown in Fig. 5d, combined density functional theory (DFT) calculations and spectroscopic analyses reveal that the switching effect originates from the tautomerization process of the imino hydrogen within the molecular central cavity. Notably, the observed switching behavior in confined environments, such as step edges and molecular arrays, excludes the possibility of whole-molecule rotation, confirming the dominant role of hydrogen positional rearrangement²³³.

Mechanically controlled switches, as critical modulation mechanisms in molecular/atomic devices, enable conductance switching through mechanical force-induced molecular conformational transitions, interfacial contact evolution, and charge state manipulation. Their operational mechanisms primarily involve three dimensions: molecular conformational reorganization altering electron transport pathways, dynamic evolution of molecule–electrode interfaces modulating electron coupling efficiency, and mechanical stress-induced charge state shifts in molecules. Studies demonstrate that factors including molecular isomer configuration differences, spatial distribution of redox-active sites, and structural deformations under external forces (e.g., molecular folding/twisting, dihedral angle variations from stretching/compression) significantly regulate device conductance characteristics^233,234. Experimentally, research teams have achieved precise manipulation of folding-extension conformational transitions at the single-molecule scale by integrating conductive atomic force microscopy (C-AFM) with molecular dynamics (MD) simulations. Upon transitioning from folded to extended states, the molecular system exhibits reversible conductance modulation spanning three orders of magnitude, demonstrating exceptional mechano-electrical responsiveness²³⁵ (Fig. 5e). Theoretical simulations further reveal counterintuitive quantum transport behavior in a model molecule comprising complementary aromatic moieties (1,4-naphthoquinone) interconnected via ether linkages. During mechanical stretching, the tunneling current displays a non-monotonic increase with applied force (Fig. 5f), challenging conventional frameworks of molecular force-electronic coupling²³⁶.

Molecular diode/rectifier

Molecular diodes, as core functional units in molecular electronics, exhibit a rectification mechanism sharing physical homology with semiconductor p-n junctions. By constructing molecular-scale heterojunctions of p-type and n-type doped materials to achieve unidirectional charge carrier transport characteristics—forming highly efficient conductive channels under forward bias while demonstrating high impedance under reverse bias—they display rectification effects comparable to macroscopic diodes^237,238. Importantly, while realizing diode functionality requires novel molecular designs and diverse operational conditions, gating the diode behavior remains equally critical²³⁹. Current molecular rectifier architectures have expanded across material systems including quantum-confined structures in single-molecule systems, interface engineering strategies utilizing SAMs, and band-alignment regulation designs in heterojunction composites. These architectures successfully realize controllable rectification behavior spanning nanoscale to mesoscale through precise modulation of molecular orbital energy level alignment, interfacial dipole orientation, and spatial charge distribution^240,241.

The electrode-molecule-electrode architecture of single-molecule rectifiers, representing the ultimate manifestation of molecular electronics, confronts dual scientific challenges in quantum confinement effects and interface coupling modulation. These nanoscale systems exhibit unique quantum tunneling transport characteristics, with rectification performance metrics surpassing the physical limitations of conventional semiconductor devices. Although precise control of molecular-electrode interfaces remains a significant technical bottleneck, breakthrough advancements in nanomanipulation techniques have provided critical support for high-reproducibility device fabrication. Employing STM-BJ technology, the research team achieved stable assembly of 1,8-nonadiyne molecules on low-doped silicon electrode surfaces. Experimental results demonstrate that this single-molecule rectification system attains a rectification ratio exceeding 4 × 10³ under ambient conditions, comparable to traditional silicon-based diodes. This breakthrough enables performance compatibility between single-molecule devices and conventional semiconductor circuits, establishing fundamental technical frameworks for molecular-semiconductor hetero-integration (Fig. 6a)²⁴². Notably, subsequent studies utilizing CP-AFM revealed that environmental humidity (5–60%) can dynamically modulate the rectification ratio by three orders of magnitude (10³) in ruthenium polypyridyl complex monolayers fabricated on indium tin oxide substrates. This tunable characteristic provides novel insights for developing environment-responsive molecular devices (Fig. 6b)²⁴³.

Fig. 6: Molecular diode/rectifier.

a 1,8-Nonodiyne: Chemical structure and self-assembled monolayer formation on H-Si via UV-assisted hydrosilylation (left) and current-voltage properties of single-nonadiyne junctions (right), reproduced from ref.²⁴², copyright 2017, with permission from Springer Nature. b Schematic representation of the 2-Ru-N ITO-molecule-ITO junction (left) and logarithmically binned RR for the humid case (right), reproduced from ref.²⁴³, copyright 2018, with permission from Springer Nature. c Schematic of alkanethiolate SAMs: SC₁₁Fc (ferrocene headgroup) vs. Fc-free controls (SC₁₀CH₃, SC₁₄CH₃) for charge transport, reproduced from ref.⁷⁴, copyright 2009, with permission from American Chemical Society. d Schematic of Ag-SAM//GaOx/EGaIn junction (left) and J (V) curve with R1/R2 conductance states and W/R/E voltages (right), reproduced from ref.²⁴⁷, copyright 2020, with permission from Springer Nature. e Schematic of MoS₂|MPc heterojunction via electrostatic adsorption (left) and J(V) rectification curves for MoS₂|MPc vs. MPc|MoS₂ with single molecule magnets alignment-induced orbital modulation (right), reproduced from ref.²⁵¹, copyright 2017, with permission from Royal Society of Chemistry. f Representative I-V characteristics (left) and statistical histograms of RR for OPT2/1_L-MoS₂ and OPT2/1_L-WSe₂ junctions (right), reproduced from ref.²⁵², copyright 2020, with permission from Springer Nature. g Schematic representation of the gMCBJ principle (left) and Scanning electron micrographs of a gMCBJ sample (right). h I–V characteristics (left) and corresponding gate-induced rectification ratio modulation (right) at different gate voltages. g, h reproduced from ref.²⁵³, copyright 2016, with permission from Royal Society of Chemistry

SAMs rectifiers, as archetypal two-dimensional molecular devices, demonstrate significant advantages in functional integration and engineering applications. Compared with single-molecule systems, this technical approach offers superior characteristics including simplified fabrication processes, enhanced structural stability, scalable production compatibility, and cost-effectiveness²⁴⁴. Through interface engineering strategies, high-density ordered molecular layers can achieve controlled assembly on diverse substrates spanning metals, oxides, and oxygen-free silicon surfaces²⁴⁵. Among these, thiol/dithiol-based molecular systems have emerged as preferred materials for large-area molecular architectures due to their strong affinity with noble metal electrodes, excellent fabrication reproducibility, and structural design flexibility²⁴⁶. Although thiol-based SAMs face limitations such as prolonged assembly time, structural defects, and interfacial coupling degradation caused by chemical oxidation, precise rectification behavior can still be engineered through molecular structure optimization and electrode property modulation⁴². The rectification mechanism fundamentally correlates with molecular backbone length. The hydrocarbon chain not only determines tunneling barrier thickness but also regulates charge transport pathways through conformational uniformity. A pivotal breakthrough was achieved in 2009 by the Nijhuis group, who constructed alkylthiol monolayers terminated with ferrocene groups between silver and eutectic GaIn electrodes, realizing a rectification ratio of ~10² under ±1 V bias⁷⁴ (Fig. 6c). Mechanistic studies revealed that this performance originates from asymmetric alignment of the ferrocene unit’s highest occupied molecular orbital (HOMO) within the junction, enabling resonant tunneling channels under negative bias⁷⁰. Recent advancements transcend the single-energy-level tunneling paradigm by integrating resistive switching with rectification functionalities. As shown in (Fig. 6d), the Ag/GaIn heterojunction incorporating S(CH₂)₁₁MV²⁺X⁻₂ molecules (MV²⁺: methyl viologen dication, X⁻: counterion) achieves an exceptional rectification ratio of 2.5 × 10⁴ while maintaining a resistive switching ratio of 6.7 × 10³, demonstrating synergistic diode-variable resistor characteristics²⁴⁷.

In the study of heterostructures for molecular rectifier devices, non-functionalized molecules have traditionally been excluded from the molecular diode paradigm due to their symmetric tunneling charge transport characteristics²⁴⁸. However, heterojunction interface band engineering can overcome this limitation by regulating band alignment to achieve asymmetric current-voltage (I–V) characteristics. Recent research has developed hybrid molecular junction and heterojunction design strategies that integrate inorganic materials (e.g., nanoparticles, two-dimensional semiconductors) with molecular films for functional synergy²⁴⁹. A representative case involves constructing n-ZnO/p-ZnO-organic molecule (Rose Bengal B or copper phthalocyanine)-mercury multilayer heterojunctions on n-type silicon substrates via electrostatic assembly²⁵⁰. This design leverages the donor properties of nanoparticles and the acceptor characteristics of organic molecules to achieve unidirectional charge transport and rectification effects through synergistic interplay. Another breakthrough design employs layer-by-layer electrostatic adsorption to fabricate metal phthalocyanine/MoS₂ heterojunctions (Fig. 6e)²⁵¹. Here, the 3d orbital characteristics of central metal atoms in phthalocyanines determine magnetization vector intensity. Following external magnetic field-induced planar orientation alignment of molecular layers, polyallylamine cationic layers are utilized for orientation fixation. This alignment regulation significantly modifies molecular orbital energy level distributions and optimizes charge transfer pathways at electrode-phthalocyanine interfaces, enabling precise control over interfacial energy level alignment and rectification performance. Organic-inorganic heterointerface studies have further revealed critical principles: Thiol-based SAMs on gold substrates coupled with transition metal dichalcogenides (MoS₂/WSe₂) exhibit pronounced rectification characteristics (Fig. 6f)²⁵². While biphenyl-4-thiol (OPT2) SAMs alone demonstrate near-unity rectification ratios (≈1), their integration with MoS₂ and WSe₂ enhances rectification ratios to 1.79 × 10³ and 2.3, respectively, under ±1 V bias. These results conclusively demonstrate the decisive role of heterointerface band engineering in establishing asymmetric transport properties in molecular devices.

Compared to traditional fixed molecular rectifying structures, three-terminal gated devices constructed using the MCBJ technique provide molecular diodes with in situ electrical tunability. As demonstrated by Perrin et al. using the gateable mechanically controllable break junction (gMCBJ) technique, the DPE-2F single-molecule rectifier (Fig. 6g) utilized a gate electrode to precisely regulate the relative alignment between molecular orbitals and the electrodes’ Fermi levels. This enabled the first experimental verification of dynamically controllable rectification based on the dual-site quantum interference mechanism. At a cryogenic temperature of 6 K, the device exhibited a high rectification ratio of 600, with the gate voltage enabling continuous tuning of this ratio (Fig. 6h). The observed modulation behavior aligns remarkably well with the dual-site level alignment model predicted by DFT calculations²⁵³. This work not only transforms the Aviram-Ratner rectifier proposal into a dynamically operable molecular component but also reveals the decisive role of the synergistic effect between gate voltage, energy level alignment, and quantum interference on rectification performance. It establishes a key technological paradigm for developing next-generation programmable molecular integrated circuits.

Molecular transistor

As the core functional component of modern integrated circuits, transistors operate based on the principle of regulating current transport characteristics between source-drain electrodes through gate voltage modulation, where the gate channel length directly determines the ultimate integration density limit of devices. In the field of molecular electronics, the molecular system constructed within molecular junctions can be analogously regarded as the conductive channel in transistor gate architectures, sharing fundamental physical mechanisms with silicon-based transistors in terms of gate-modulated charge transport environments. Molecular transistors employing bottom-up assembly strategies can compress conductive channel dimensions to the sub-nanometer scale, overcoming the physical constraints of conventional photolithography and providing innovative solutions for ultrahigh-density integrated circuits. By systematically investigating gate-controlled rectification behavior, carrier mobility modulation mechanisms, and the influence of interfacial state density on switching ratios, research on molecular transistors will drive the paradigm shift in nanoelectronics from the microelectronic era to the molecular electronic era. These performance enhancements, including tunable rectification, high switching ratios, and sub-nanometer channel control, collectively provide crucial scientific support for expanding integrated circuit technology pathways in the post-Moore’s Law era.

Electrostatic-gated molecular transistors achieve silicon-like current modulation functions through gate electric field regulation of molecular orbital energy levels, whose core mechanism lies in the synergistic interaction between the molecular junction system and the gate dielectric layer. In 2000, Park et al.²⁵⁴ pioneered the construction of a three-terminal device using C₆₀ molecules as conductive channels through electromigration technology, employing doped silicon wafers as gate electrodes. This groundbreaking work experimentally demonstrated for the first time that single-molecule systems could achieve significant alterations in current-voltage characteristics via gate voltage modulation. Notably, their single-C₆₀ transistor exhibited quantized nanomechanical oscillations at ~1.2 THz (5 meV) with an ON/OFF conductance ratio of ~300. This work thereby established the experimental foundation for molecular transistor research (Fig. 7a). With advancements in nanofabrication techniques, Song et al.²⁵⁵ constructed molecular junctions of 1,4-benzenedithiol and 1,8-octanedithiol between gold electrodes in 2009, innovatively implementing a 3-nm-thick aluminum oxide gate dielectric layer. Their work demonstrated continuous modulation of molecular conductance through gate voltage application, with experimental data revealing a 0.25 eV shift in molecular orbital energy levels under 1 V gate bias, corresponding to a shift rate of ~0.25 eV/V via electrostatic gating. These data established a linear response model between gate voltage and orbital energy-level displacement (Fig. 7b), while also enhancing inelastic electron tunneling spectroscopy (IETS) intensity by over 30 times and yielding Γ ≈ 0.12 eV for HOMO resonance. Recent breakthroughs in theoretical simulation technologies have provided new insights for molecular transistor design. In 2020, Weckbecker et al.²⁵⁶ employed first-principles calculations to reveal the bistable conductive characteristics of molecular-graphene heterojunctions during photoinduced isomerization processes, demonstrating that external gate electric fields could precisely regulate molecular configurations and interfacial coupling states to selectively stabilize distinct conductance states, as evidenced by an achieved ON/OFF conductance ratio of ~20 at 0.05 V via proton-transfer gating, with HOMO energy at −2.0 eV and coupling strength Γ = 0.12 eV (Fig. 7c). These research advancements collectively validate the effectiveness of electrostatic gate control mechanisms at the molecular scale.

Fig. 7: Molecular transistor.

a Single-molecule C₆₀ device structure with electromigration-induced nanogap (~1 nm) and corresponding I–V curves at varying gate voltages, reproduced from ref.²⁵⁴, copyright 2000, with permission from Springer Nature. b Device structure and schematic diagram (left) and representative I–V curves measured at 4.2 K for different values of V_G (right), reproduced from ref.²⁵⁵, copyright 2009, with permission from Springer Nature. c Schematic diagram of a proton-controlled molecule-graphene nanojunction (up) and gate-modulated current-voltage switching characteristics (down), reproduced from ref.²⁵⁶, copyright 2020, with permission from American Chemical Society. d Schematic diagram of the Au STM tip and molecular wire setup (up) and current-distance decay curves (down), reproduced from ref.¹²⁸, copyright 2003, with permission from American Chemical Society. e Schematic diagram of a gated molecular junction setup (left) and linear sweep voltammetry curve (right), reproduced from ref.²⁵⁸, copyright 2014, with permission from American Chemical Society. f Schematic diagram of an ionic-liquid-gated graphene-molecule-graphene junction (up) and molecular-dependent current-voltage/transmission spectroscopy curves (down), reproduced from ref.²⁵⁹, copyright 2018, with permission from Wiley

Electrochemical gated transistor technology leverages the electrical double layer potential at metal-electrolyte interfaces as a gate modulation mechanism, effectively circumventing the intricate three-terminal molecular device fabrication processes inherent in conventional electrostatic gating while enabling precise molecular conductance regulation at room temperature²⁵⁷. This approach establishes a novel paradigm for single-molecule device research through its capability to reversibly modulate molecular redox states via electrochemical reactions. In 2003, Haiss et al. pioneered the first single-molecule electrochemical transistor using a four-electrode STM-BJ configuration¹²⁸ (Fig. 7d), demonstrating a ~100-fold conductance increase at −0.6 V (vs. Ag/AgCl) for single 6V6 viologen molecules due to redox switching, with plateau currents quantized at ~98 pA. This foundation was further advanced by the Venkataraman research group in 2014 through the development of an optimized three-electrode system²⁵⁸. The device architecture (Fig. 7e) employs encapsulated tips and substrates as working electrodes, with platinum serving dual roles as counter and gate electrodes. Experimental investigations demonstrated that modulating the gate voltage (Vg) alters the double-layer structure at the working electrode surface, thereby enabling continuous regulation of molecular conductance—modulating conductance by about 3 times for LUMO-conducting molecules and about 2 times for HOMO-conducting molecules using electrolytic gating, with gate efficiency α ≈ 0.22–0.25 eV/V. A significant breakthrough emerged in 2018 when Xin et al.²⁵⁹ introduced ionic liquids as gate dielectrics, achieving synergistic control of graphene electrode Fermi levels and molecular orbital energy levels through Vg adjustments. All three electrochemically inert aromatic molecular junctions studied exhibited pronounced gate voltage-dependent conductance characteristics—enabling ambipolar transport in electrochemically inert aromatic molecules (E_G 3.5 eV) at |V_G | ≤ 1.5 V, with a conductance attenuation coefficient of β = 0.23 Å⁻¹. Notably, the structural stability of ionic liquid double layers below their freezing points renders this single-molecule transistor an ideal platform for investigating gate-modulated quantum transport phenomena and novel single-molecule quantum states under cryogenic conditions (Fig. 7f).

Large-scale integration of molecular devices

3D integration technology

In the ongoing miniaturization process of semiconductor manufacturing technology, although the exponential reduction in critical linewidth dimensions of transistors and damascene interconnect layers has significantly enhanced device integration density, current strategies relying solely on physical dimensional scaling now face dual physical constraints: quantum tunneling-induced leakage current surges and extreme ultraviolet (EUV) lithography precision limits²⁶⁰. Consequently, 3D integrated circuits (3D ICs) have emerged as a revolutionary integration paradigm that breaks through traditional 2D planar limitations, achieving multidimensional heterogeneous integration through architectural innovation. This approach establishes vertical interconnection channels via Through-Silicon Vias (TSVs), synergistically integrating the micro-bump array arrangement of redistribution layers (RDLs) with high-precision bonding processes to realize Z-axis heterogeneous chip stacking. By constructing 3D interconnect architectures, this technology accomplishes effective chip area reduction while preserving core device feature dimensions and interconnect spacing, not only circumventing the physical limitations inherent in traditional lithography’s continuous lateral dimension compression but also sustaining performance enhancement through optimized vertical integration^261,262,263. This section first elaborates on the fundamental physical architecture of transistor interconnect layers and the engineering implementation principles of damascene electroplating processes, followed by a systematic analysis of 3D integration methodologies based on TSV vertical interconnects, RDL routing optimization, and high-density bump bonding coordination mechanisms.

Transistor interconnect layers and Damascene electroplating processes

In the transistor interconnect layer process, the M₀ layer, as a core component of middle-end-of-line (MEOL) technology, plays a pivotal role in bridging front-end-of-line (FEOL) transistor devices and back-end-of-line (BEOL) metal interconnect layers. This layer establishes low-resistance electrical connections to FEOL devices through direct contact with silicide materials such as NiSi or TiSi₂ formed on transistor source/drain surfaces, while vertically overlaying shallow trench isolation structures and gate architectures, achieving 3D interconnection with the M₁ layer via Via₀ contacts (Fig. 8a, b). Its fundamental functions extend beyond optimizing device performance through significant contact interface resistance reduction, effectively mitigating stress mismatch issues arising from thermal expansion coefficient differences between FEOL devices and BEOL interconnects via unique structural designs^264,265. From a process flow perspective, M₀ layer fabrication encompasses critical steps including silicide contact interface formation, high-precision lithographic contact hole etching, TaN/TiN composite barrier layer construction using atomic layer deposition, tungsten metal filling via chemical vapor deposition, and advanced integration of novel materials like ruthenium for enhanced conductivity in cutting-edge nodes²⁶⁶ (Fig. 8c), culminating in surface planarization through chemical mechanical polishing²⁶⁷. As technology nodes advance, the M₀ layer confronts key technical challenges such as filling uniformity control in high aspect ratio structures, interface defect suppression, and interfacial stability between emerging metal materials and silicides, driving innovations in atomic-scale deposition processes, selective metal deposition technologies, and interface engineering methodologies.

Fig. 8: Cross-sectional microstructure of a chip.

a Schematic diagram of the cross-sectional structure of a chip, reproduced from ref.²⁶⁴, copyright 2025, with permission from Elsevier. b Cross-sectional electron micrograph, reproduced from ref.²⁶⁵, copyright 2017, with permission from Springer. c Cross-sectional transmission electron micrographs along a Ru interconnect, indicating void-free filling, reproduced from ref.²⁶⁶, copyright 2016, with permission from American Chemical Society

Above the M₀ layer, the M₁ layer, serving as the first metal interconnect in the BEOL process, employs the single Damascene electroplating technique to construct vertical interconnect structures, while the M₂ to M_x layers, requiring simultaneous formation of vias and horizontal interconnects, predominantly utilize the dual Damascene electroplating process²⁶⁸ (Fig. 9a, b). Dual Damascene technology encompasses three typical modes based on the sequence of via and trench formation: the via-first scheme prioritizes vertical interconnect via creation before horizontal metal trench patterning, the trench-first approach reverses the sequence by defining horizontal trenches prior to via etching, and the self-aligned mode achieves concurrent via-trench formation through synergistic processes. The single Damascene process for the M₁ layer integrates critical steps such as dielectric film deposition via plasma-enhanced chemical vapor deposition (PECVD) or atomic layer deposition (ALD), photolithographic pattern transfer through exposure and development, reactive ion etching for trench formation, physical vapor deposition of Ta/TaN composite barrier layers, sputter deposition of copper seed layers, electrochemical copper plating for metal filling, and surface planarization via chemical mechanical polishing. Taking the self-aligned dual Damascene process as an example, its workflow initiates with precise multilayer dielectric stack construction involving sequential deposition of nitride-doped carbide (NDC), barrier dielectric (BD), and silicon nitride (SiN), followed by photolithographic patterning and reactive ion etching of SiN to define via regions. Subsequent deposition cycles of BD, TiN, and silicon dioxide (SiO₂) establish composite structures, after which secondary photolithography defines horizontal trench patterns. A single-step etching process then synchronously forms 3D trench-via profiles, concluding with metalization steps analogous to single Damascene processing—barrier layer deposition, seed layer formation, copper filling, and CMP. The coordinated application of Damascene technology not only enables vertical integration and horizontal extension of nanoscale interconnects but also significantly enhances interconnect density and signal transmission efficiency through via-trench co-formation. Its multilayer dielectric stacking and selective etching strategies effectively mitigate parasitic capacitance and crosstalk, while the integration of copper metallization with ultrathin barrier layers ensures electromigration resilience under high current densities. Furthermore, the self-aligned dual Damascene process reduces lithographic mask iterations and simplifies manufacturing steps, substantially shortening cycle times and lowering defect rates, thereby establishing a robust process foundation for interconnect reliability and performance scaling in advanced-node ultra-large-scale integrated circuits.

Fig. 9: Schematic diagram of the damascene electroplating filling process.

a Single Damascene electroplating filling process. b Dual Damascene electroplating filling process

While Damascene technology excels at creating high-density intra-die interconnects (within a single chip), the relentless pursuit of higher performance and integration density necessitates extending connectivity beyond the planar confines of a single die. This demand drives the adoption of 3D integration schemes, where multiple dies are stacked vertically. The critical component for such vertical stacking is the TSV technology, which provides efficient electrical pathways through the silicon substrate itself, connecting different strata in the 3D stack.

TSV-based vertical interconnects

TSV technology has emerged as the cornerstone for realizing 3D integrated circuits, addressing the fundamental challenge of establishing low-resistance, high-bandwidth vertical connections between stacked dies or between a die and its package substrate. This technique stands in contrast to the planar Damascene interconnects discussed earlier, which operate within a single chip layer. Over the past decades, TSV technology has undergone significant technological evolution and achieved industrial-scale implementation^263,269,270. This technique employs deep reactive ion etching to fabricate high-aspect-ratio vertical channels within silicon substrates, followed by filling with high-conductivity metallic materials such as copper, tungsten, or polysilicon, thereby establishing low-resistance vertical interconnection architectures that effectively overcome the physical limitations of conventional two-dimensional planar interconnects in terms of density and signal transmission bandwidth^271,272. The Bosch etching process²⁷³ (Fig. 10a) has become the predominant technique for TSV fabrication due to its exceptional etching rate (5-10 μm/min), photoresist selectivity (50-100), and oxide mask selectivity (200:1). This cyclic etching-passivation mechanism comprises three sequential phases: initial silicon substrate etching using SF₆ plasma, subsequent C₄F₈ gas deposition to form lateral passivation layers preventing sidewall etching, and directional ion bombardment to simultaneously remove bottom passivation films and deepen the silicon trench, ultimately achieving vertical sidewall profiles. Prior to metallization, a dielectric isolation layer (SiO₂/Si₃N₄) with superior step coverage, minimal leakage current, high breakdown voltage, and low thermal stress must be deposited²⁷⁴ (Fig. 10b), with process parameters rigorously controlled to ensure compatibility with subsequent integration requirements, particularly thermal stability during 400°C annealing processes. Furthermore, to simultaneously prevent copper diffusion and enhance interfacial adhesion, sequential deposition of barrier materials (Ti, Ta, and their nitrides TiN/TaN) followed by the deposition of a copper seed layer is implemented on the dielectric surface²⁷⁵ (Fig. 10c), with deposition techniques (PVD, CVD, or ALD) selected based on TSV dimensional specifications. Final TSV metallization employs three principal electroplating strategies: conformal plating, sealed bump bottom-up plating, and super-conformal plating²⁷⁶ (Fig. 10d), with methodology selection dictated by TSV geometric parameters (10–200 μm depth ranges, aspect ratios determined by combined dielectric/barrier/seed/filling processes) and specific demands of 2.5D/3D integration architectures.

Fig. 10: Schematic diagram of the TSV electroplating filling process.

a SEM image of blind vias etched via Bosch process, reproduced from ref.²⁷³, copyright 2011, with permission from Elsevier. b SEM image of dielectric isolation layer. c SEM image of copper seed layer. b, c reproduced from ref.²⁷¹, copyright 2017, with permission from Springer Nature. d SEM image of copper electroplating filling in TSV process, reproduced from ref.²⁷⁶, copyright 2012, with permission from Electrochemical Society

Synergistic integration of RDL and bump bonding for 3D chip stacking

In 3D IC integration, RDLs and bump bonding technologies, through functional complementarity and process synergy, collaborate with TSV to construct high-density heterogeneous integration architectures^277,278. Although early packaging technologies achieved basic electrical connectivity, traditional pad layouts constrained by planar interconnection density led to deteriorated high-frequency signal integrity²⁷⁹. Therefore, the semiconductor packaging field has introduced RDL technology, which utilizes photolithography, electroplating, and etching processes to reconfigure micron-scale copper/aluminum interconnects on dielectric layers (e.g., polyimide/SiO₂). This redistributes I/O ports to chip edges or functional regions while optimizing wiring topology through finite element analysis (FEA). Simultaneously, it enhances high-frequency performance by refining conductor parameters (width/spacing/dielectric thickness) to reduce parasitic capacitance and signal delay²⁸⁰. However, RDLs primarily address intra-layer horizontal routing challenges²⁸¹ (Fig. 11a), whereas vertical interconnections in 3D stacking necessitate bump bonding technologies. These employ electroplated or placed micro-bump (Fig. 11b) combined with thermocompression bonding or reflow soldering to achieve interlayer alignment^282,283.

Fig. 11: SEM image of RDL and bumps.

a SEM image of RDL, reproduced from ref.³³¹, copyright 2021, with permission from Springer. b SEM image of bumps, reproduced from ref.³³², copyright 2022, with permission from Multidisciplinary Digital Publishing Institute

From a synergistic perspective, RDLs and bump bonding collectively establish a “horizontal-vertical” integrated interconnection framework. The precision wiring of RDLs reduces signal crosstalk while their dielectric layers provide mechanical support for bumps. Conformal electroplating enables seamless integration between metal traces and bumps, effectively mitigating interfacial defects. Conversely, the low-impedance connections of bumps minimize transmission losses, forming cascading enhancement effects with RDL-based signal optimization to jointly improve system bandwidth and energy efficiency^284,285,286. Therefore, their process compatibility and functional coupling have emerged as critical drivers for breakthroughs in interconnection reliability and performance within 3D integration technologies.

3D integration of molecular devices

Molecular electronics focuses on designing novel electronic components through precise manipulation at atomic and molecular scales, aiming to complement or even surpass the physical limitations of conventional silicon-based devices and construct next-generation molecular electronic systems with multidimensional functional characteristics. While current research has achieved preliminary progress in two-dimensional molecular array integration based on self-assembly techniques, it remains constrained by planar topological restrictions. The strong lateral coupling effects between molecules hinder independent regulation of carrier transport properties, while random distribution of heterogeneous interfacial contact resistance, functional homogeneity of devices, and thermodynamic metastability collectively severely hinder practical applications^{287,288,289,290}. Therefore, it is imperative to incorporate the top-down 3D integration methodology from traditional semiconductor integrated circuits while leveraging the bottom-up assembly characteristics of molecular electronic devices to achieve high-density 3D integration of molecular systems. This section first reviews current attempts at 2D array integration, then analyzes the distinctive features of molecular electronic devices, and subsequently discusses how molecular electronics might adapt and utilize 3D integration technologies for 3D high-density integration.

2D array integration of molecular devices

Despite the multiple technical bottlenecks in the integration of molecular electronic devices, critical breakthroughs have been achieved through innovations in self-assembled monolayer (SAM)-based interface engineering^291,292,293. Since Green et al. first demonstrated a 160-kilobit molecular memory circuit with a crossbar array architecture in 2007, the field has progressively revealed the core challenges and potential solutions for molecular-scale integration. Green et al. employed mechanically interlocked rotaxane molecules as memory units and constructed high-density crossbar arrays using the superlattice nanowire pattern transfer (SNAP) technique. By incorporating fault-tolerant principles inspired by the Teramac supercomputer, they preliminarily validated the functional integration of molecular circuits²⁹⁴ (Fig. 12a).

Fig. 12: 2D integrated molecular device.

a SEM image of 400 Si and Ti/Pt nanowire array structures, reproduced from ref.²⁹⁴, copyright 2007, with permission from Springer Nature. b Schematic of a flexible molecular array device, reproduced from ref.²⁹⁵, copyright 2012, with permission from Springer Nature. c Schematic of a molecular layer array node, reproduced from ref.²⁹⁶, copyright 2016, with permission from American Chemical Society. d Schematic diagram of SAM device arrays, reproduced from ref.²⁹⁷, copyright 2018, with permission from Springer Nature. e Top and bottom electrodes arranged to form 300 individual cross-bar devices per sample, reproduced from ref.²⁹⁸, copyright 2021, with permission from American Chemical Society. f Schematic illustrations (not drawn to scale) of the reversible placement of the top-electrode on the SAM, reproduced from ref.²⁹⁹, copyright 2015, with permission from Royal Society of Chemistry

However, this prototype exhibited issues such as ambiguous switching states, molecular layer degradation, and short-circuit defects. To improve the yield of molecular-layer devices, the introduction of protective layers to mitigate short circuits caused by assembly defects has emerged as an effective strategy, leading to diverse protective layer paradigms. The first paradigm focuses on soft interface engineering, pioneered by Park et al. with their poly(3,4-ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) polymer buffer layer. This flexible conductive matrix not only suppresses metal penetration but also provides mechanical compliance, which is critical for molecular devices on deformable polyimide substrates²⁹⁵ (Fig. 12b). Others, such as Najarian et al., developed a carbon-based integration strategy utilizing electron-beam-deposited carbon electrodes. By forming covalent bonds with SAMs and achieving atomic-level planarization, this approach created electromigration-resistant all-carbon junctions, particularly suitable for transparent electronics²⁹⁶ (Fig. 12c). Beyond these material-centric solutions, geometric confinement techniques have reached new heights with the Puebla-Hellmann group’s gold nanoparticle barriers. Their microcavity assembly strategy leverages steric exclusion effects to construct fault-tolerant molecular junctions, demonstrating that spatial control can synergize with chemical protection to enhance production yield²⁹⁷ (Fig. 12d). While protective layers enhance reliability, achieving higher integration density necessitates complementary strategies. Kos et al. designed a nanoparticle-on-mirror (NPoM) plasmonic architecture that uniquely integrates optical modulation with molecular junctions, expanding the functionality of optoelectronic devices²⁹⁸ (Fig. 12e). Additionally, soft interface engineering strategies have gained attention, exemplified by the Wan group’s approach using polydimethylsiloxane (PDMS) microcavity arrays to immobilize gallium oxide/gallium-indium eutectic (GaOx/EGaIn) liquid metal electrodes. This soft-contact paradigm circumvents edge defects induced by conventional fabrication processes, enabling unprecedented reproducibility in large-area self-assembled monolayer tunnel junctions²⁹⁹ (Fig. 12f).

Although these advances propelling theoretical integration densities from 10⁹ to 10¹² devices/cm², fundamental challenges persist. On one hand, heterogeneous interfaces introduced by protective layers exacerbate electrical performance fluctuations. On the other hand, a significant gap remains between the current maximum crossbar array density and molecular-scale precision requirements, alongside the lack of universal low-variability interface protocols. Future progress demands deep integration of top-down micro-nano fabrication (e.g., 3D integration, ultrahigh-precision lithography) with bottom-up molecular self-assembly techniques to develop cross-scale collaborative manufacturing strategies.

Molecular devices integrated with 3D integration technology

The current research on the integration of molecular electronic devices primarily focuses on two-dimensional planar architectures. While this manufacturing approach offers relative simplicity, it faces inherent limitations in achieving high-density functional integration due to constrained planar space and insufficient system-level integration capabilities when relying solely on bottom-up molecular self-assembly. Consequently, the development of molecular electronic device integration urgently requires the convergence of advanced 3D integration technologies with molecular-scale manipulation techniques to establish novel hybrid manufacturing paradigms featuring cross-scale characteristics. Achieving this objective necessitates a profound understanding of the compatibility mechanisms between the intrinsic properties of molecular devices and 3D integration processes, particularly the constraints imposed by critical parameters like interface sensitivity and thermal stability on manufacturing workflows^{21,300,301,302}.

The fundamental characteristics of molecular devices essentially redefine the design logic inherited from conventional semiconductor technologies. The chemical bonding states at electrode-molecule interfaces directly govern carrier transport efficiency, where energy level alignment determines the controllability of quantum tunneling effects³⁰³. Such interface sensitivity mandates atomic-level precision in contact control during manufacturing, as rough interfaces generated by traditional processes like photolithography and metal deposition significantly degrade device performance³⁰⁴. Meanwhile, organic molecular materials generally exhibit thermal stability thresholds below 200 °C^305,306, fundamentally conflicting with key semiconductor manufacturing processes such as high-temperature annealing (>400°C)^307,308 and chemical vapor deposition (300–600 °C)³⁰⁹. These intrinsic properties necessitate disruptive manufacturing strategies for 3D integration of molecular devices that preserve structural integrity while enabling cross-scale system integration.

To address these challenges, back-end-of-line (BEOL)-compatible integration strategies offer a viable pathway for constructing 3D architectures. By incorporating molecular devices into the BEOL process, the high-temperature environments generated during front-end operations can be effectively circumvented. Specifically, TSVs can be pre-fabricated to establish vertical interconnects, while redistribution layers (RDLs) form 3D wiring networks. Bump bonding technology enables multi-layer stacking and external circuit connections, with small-scale BEOL interconnect layers inversely fabricated on TSVs. After completing Mx-M₁ metal layers, the smallest-dimension M₀ layer is prepared to reserve specific integration sites for molecular devices. This architectural design requires deep collaboration with low-temperature assembly processes, such as spatially directed positioning of functional molecules through molecular self-assembly technology that constructs DNA-origami-based gold nanorod electrode gaps at predefined integration sites³¹⁰ (Fig. 13). Furthermore, aluminum oxide (Al₂O₃) passivation layers deposited via atomic layer deposition (ALD) reduce moisture/oxygen sensitivity in molecular devices by over 80%, significantly enhancing long-term stability in 3D stacked structures³¹¹. The implementation of this methodology demands precise multi-physical field coupling control. During 3D interconnect framework construction, optimizing the aspect ratio and stress distribution of TSVs proves critical to prevent mechanical deformation from damaging pre-designed structures during subsequent molecular assembly³¹². Additionally, molecular self-assembly processes involve complex solid-liquid interface dynamics regulation. Given the inherent directionality and selectivity of non-covalent intermolecular interactions, meticulous management of molecular design, assembly condition optimization, and real-time characterization/monitoring becomes imperative^313,314,315.

Fig. 13

Schematic diagram of molecular device integration

Looking forward, the deep integration of molecular devices with 3D technologies may catalyze a new generation of information device systems. By combining molecular-scale quantum effects with system-level 3D architectures, this approach promises to break through the energy efficiency limitations of von Neumann architectures and realize in-memory computing intelligent chips. However, achieving this vision requires breakthroughs in interface state precision control, cross-scale manufacturing equipment, and device-system co-design methodologies. Particularly crucial is the development of in-situ characterization techniques to monitor real-time structure-property evolution during 3D integration, providing dynamic feedback for process optimization. With continuous advancements in materials science and micro/nano-fabrication technologies, molecular electronic devices are poised to transition from laboratories to practical applications, heralding a new era of integrated circuit technology in the post-Moore era.

Summary and perspective

The 3D integration of molecular electronic devices through atomic manufacturing technology bridges the gap between bottom-up molecular assembly and top-down functional architectures, unlocking unprecedented opportunities for advancing molecular electronics. As the field transitions from proof-of-concept demonstrations to scalable applications, future developments are anticipated to focus on two key directions: the potential of integrated molecular electronic components such as molecular memristors in chip-based memory and logic computing, and molecular-level sensing and characterization aimed at real-time, high-resolution monitoring of chemical and physical processes at the molecular scale. This necessitates the development of integrated architectures, such as a “Memory-Logic-Sensing” tri-functional chip schematic, where the exceptional density of molecular memristors (~10¹⁴ devices/cm²) enables orders-of-magnitude higher integration compared to conventional static random-access memory (~10¹⁰ devices/cm²), unlocking new paradigms (Fig. 14). This section outlines challenges and opportunities in these emerging domains, emphasizing advancements in molecular memristor-based memory-logic computing systems and the elucidation of chemical reaction kinetics coupled with physical feedback mechanisms.

Fig. 14

Molecular-Dominant Memory-Logic-Sensing Tri-functional Chip Architecture, density: 10¹⁴ devices/cm² (Left) and Schematic of a conventional 6-Transistor Static Random-Access Memory (6T SRAM) cell and its integration, density: 10¹⁰ devices/cm² (Right)

Molecular memristors for memory and logic computation

The atomic-scale precision of molecular manufacturing enables the fabrication of devices with tailored electronic properties. However, maintaining functional accuracy while scaling up these systems remains a formidable challenge. Future advancements will rely on the integration of molecular engineering with 3D integration strategies to overcome limitations in device density, signal integrity, and thermal management. This synergistic optimization approach has demonstrated significant advantages in the design of novel molecular memory devices and molecular logic computing architectures.

The atomic-level precision of molecular manufacturing enables the fabrication of devices with tailored electronic properties, yet maintaining functional accuracy while scaling up systems remains a formidable challenge. Future advancements necessitate the deep integration of molecular engineering with 3D integration strategies to synergistically address issues such as device density, signal integrity, and thermal management. This optimization paradigm has demonstrated significant advantages in the design of novel molecular memory devices and molecular logic computing, particularly in constructing brain-inspired computing systems under non-von Neumann architectures. Single-molecule memristors, leveraging their intrinsic structural flexibility, nanoscale superiority, and programmable characteristics, have emerged as effective platforms for realizing highly nonlinear and ergodic neuromorphic computing³¹⁶. Their compatibility with self-assembly strategies offers a unique technological pathway toward developing memory-computing integrated chips that combine high density, superior computing power, and low energy consumption³¹⁷.

Molecular memristors, as novel electronic devices emulating biological synaptic plasticity, derive their core operational mechanism from the dynamic response characteristics of redox-active molecules or charge-transfer complexes. These devices achieve non-volatile conductance modulation analogous to synaptic weight adaptation through external electric field-regulated intramolecular charge redistribution or conformational changes³¹⁸. Compared to conventional silicon-based memory, molecular memristors exhibit revolutionary advantages in memory-computing integrated architectures: their nanoscale dimensions and solution-processable nature enable ultrahigh-density 3D integration, while their single-molecule conductance-switching storage mechanism fundamentally overcomes the energy-efficiency and bandwidth limitations inherent to von Neumann architectures³¹⁹. Experimental studies demonstrate that both the BPDN-DT molecular system constructed via mechanically controllable break junction techniques¹⁶⁸ and the electrically controlled localized thermal activation single-molecule memory based on STM-BJ technology have achieved stable bistable memory functionality. Furthermore, their charge-transfer-regulated devices exhibit non-stochastic switching characteristics, establishing the physical foundation for constructing molecular-scale memory with brain-inspired computing capabilities³²⁰.

In the integration of molecular logic computing with neuromorphic computing, a breakthrough was achieved in 2022 by Li et al. through a two-terminal memristive device based on the Sc₂C₂@Cs(hept)-C₈₈ fullerene system³²¹. This device utilizes bias-regulated dipole-state transitions within single-molecule junctions, where the energy barrier height surpassing thermal fluctuation energy ensures room-temperature state stability. By employing set voltages to induce conductive-state switching and read voltages to detect resistive states, the integration of memory and logic operations at the single-molecule scale was successfully demonstrated. Innovatively defining electrode voltages as input logic values and molecular resistive states as output logic values, the device realized fundamental Boolean logic operations such as AND and OR. Although nonlinear classification operations like XOR and XNOR (requiring multi-device cascading) remain unrealized, this work first validated the feasibility of molecular devices executing logic operations at room temperature. Future advancements through synergistic innovations in 3D crossbar architecture design and molecular material-micro/nanofabrication technologies may enable molecular electronics to realize neuromorphic chips with in situ machine learning acceleration capabilities. Such chips would transcend conventional computing paradigms by directly performing logic operations and neuromorphic computations within memory-centric architectures, thereby pioneering new pathways for constructing high-efficiency brain-inspired computing systems.

Molecular sensing and characterization

In the field of molecular sensing and characterization research, existing technological systems have successfully achieved dynamic response mechanisms based on molecular devices for exogenous physical stimuli and intrinsic chemical reactions, thereby establishing diverse sensing and characterization methodologies. However, constrained by the integration bottleneck of molecular devices, traditional approaches predominantly rely on single or limited-component operational modes, which may not only lead to the accumulation of systematic errors or random deviations but also exhibit inherent limitations such as restricted response efficiency and relatively simplistic functional architectures. Consequently, breakthroughs in large-scale integration technologies for molecular devices have opened new pathways for constructing parallel monitoring systems with high spatiotemporal resolution, particularly creating essential prerequisites for real-time correlation analysis between molecular-scale chemical reaction kinetic processes and multivariate physical stimulus responses. This integrated strategy not only significantly enhances the reproducibility and statistical significance of detection outcomes but also enables multidimensional in situ analysis of complex molecular systems through synergistic effects of multimodal sensing units, thereby driving a paradigm shift in molecular sensing technology from discrete detection toward systematic monitoring.

In the study of response mechanisms of molecular devices to external physical stimuli, light and magnetic field-regulated changes in electrical conductance characteristics have demonstrated clear structure-property relationships. The core mechanism of photo-induced molecular conductance modulation originates from the reconstruction of electron transport channels caused by conformational changes in photosensitive groups, with a seminal example traced to the first photoresponsive single-molecule conductance switch device reported by Dulić et al.²²² in 2003. This study employed diarylethene derivatives as functional units, achieving reversible open and closed ring conformational transitions through alternating visible and ultraviolet light irradiation in solution environments. However, in electrode interface systems, only the irreversible open-ring process induced by visible light could be observed due to excited-state quenching caused by the proximity between the open-ring state excitation energy level and the gold electrode’s Fermi level, while the ultraviolet-driven closed-ring process was significantly suppressed. Compared with optical regulation, research on magnetic field responses focuses on the miniaturization demands of spintronic devices and the controllable modulation potential of molecular spin states. Schmaus et al.³²² achieved spin transport regulation in non-magnetic phthalocyanine molecular junctions by integrating magnetic materials into scanning probe systems, observing giant magnetoresistance effects at the single-molecule scale with a specific tunneling magnetoresistance (TMR) value reaching approximately 60%. Beyond engineered magnetic electrodes, strategies utilizing external fields to directly regulate molecular intrinsic spin states provide novel approaches for constructing high-sensitivity molecular spin sensors. With the synergistic development of molecular assembly techniques and 3D integration, the multidimensional response characteristics of molecular devices to external stimuli will advance sensing technology toward single-molecule precision. By establishing light-magnetic-electrical multiphysical field coupling regulation systems, real-time correlation analysis between molecular conformation, spin states, and charge transport properties becomes achievable, providing an in situ detection platform for dynamic characterization of complex molecular systems. Particularly for quantum effect-dominated molecular junction devices, their ultrahigh sensitivity to external stimuli may surpass the detection limits of existing sensing technologies.

The correlation effects between chemically stimulated single-molecule-scale reaction kinetics and electrical conductance have opened new research dimensions in molecular electronics. Aragones et al.³²³ first revealed the unique modulatory role of strong electric fields in electrode nanogaps on molecular reactions through the STM-BJ technique in 2016. When the nanogap electric field intensity reached magnitudes of 10⁹ V/m, the Diels-Alder reaction rate increased fivefold. This discovery challenged the conventional understanding that external electric fields could only catalyze redox reactions, demonstrating that molecular electronics approaches can overcome the quasi-equilibrium limitations of ensemble physical chemistry to precisely resolve microscopic differences in reaction pathways. Subsequently, the synergistic study of single-molecule-scale electric field effects and chemical reaction characterization rapidly emerged as a cutting-edge research focus. Zhang et al.³²⁴ extended this strategy in 2023 to enzyme-catalyzed reactions central to life sciences, employing STM-BJ to distinguish distinct catalytic cycle states of Candida boidinii formate dehydrogenase via characteristic conductance values and reconstructing its revolutionary catalytic pathway using clustering analysis. Multiscale simulations further revealed that during the catalytic cycle, bound reduced nicotinamide adenine dinucleotide (NADH) undergoes direct conversion to oxidized NAD⁺ through in situ hydrogen transfer, a mechanism that overturns the classical enzyme-cofactor dissociation paradigm in the Theorell-Chance theory, signifying that single-molecule electrical characterization techniques now possess the capability to resolve the operational mechanisms of complex biomolecular machines. Chemically responsive molecular devices provide unprecedented spatiotemporal resolution for in situ, real-time monitoring of chemical reactions. By developing electric field-chemical dual-modal regulation systems, cross-scale correlation models bridging electronic transport properties to molecular conformational evolution, and single-molecule reaction kinetics to ensemble macroscopic behaviors can be established. In biomedical applications, this technology may enable real-time tracking of aberrant catalytic behaviors in pathology-related enzymes, offering novel biomarkers with molecular-level sensitivity for early disease diagnosis. However, key technical challenges such as dynamic stability at molecule–electrode interfaces, weak signal denoising, and standardized data processing protocols must be addressed to ensure comparability across research systems. With the deep integration of single-molecule manipulation techniques and in situ characterization methods, chemically stimulated molecular devices are poised to become revolutionary tools for elucidating molecular mechanisms in life processes and optimizing industrial catalytic systems.