Papers
Topics
Authors
Recent
Search
2000 character limit reached

Towards Atom-by-Atom Fabrication: Mechanosynthetic donation and abstraction

Published 11 Jun 2026 in cond-mat.mtrl-sci | (2606.13876v1)

Abstract: Enabled by inverted-mode scanning tunneling microscopy (IM-STM) and the use of functionalized molecular tools, we demonstrate positionally-controlled mechanosynthetic addition (donation) of carbon and subtraction (abstraction) of silicon atoms on a model build site: atomically clean and crystalline Si(100). The resulting structures represent the first demonstrations of an emerging ability to manipulate radical chemistry with positional control of specific atoms and moieties in 3D. Furthermore, by comparing the behavior of molecular tools designed for atomic donation versus abstraction, we highlight general principles governing molecular tool design for selective and reliable mechanosynthetic functionality.

Summary

  • The paper presents a dual mechanosynthetic method enabling atom addition (donation) and precise removal (abstraction) on Si(100) using IM-STM.
  • It employs tripodal adamantane derivatives to form Si–C bonds and create Si vacancies, achieving a 54±7% donation rate and 100% abstraction selectivity with optimized tool design.
  • The study underscores that tailored molecular tool engineering and mechanosynthetic protocols can drive deterministic atom-by-atom fabrication for advanced nanodevice engineering.

Mechanosynthetic Donation and Abstraction for Atom-by-Atom Fabrication via IM-STM

Introduction

The work "Towards Atom-by-Atom Fabrication: Mechanosynthetic donation and abstraction" (2606.13876) establishes critical advancements in atomically precise fabrication (APF) by demonstrating both addition (donation) and subtraction (abstraction) of specific atoms on Si(100) with positional control, using functionalized adamantane-based molecular tools deployed by inverted-mode scanning tunneling microscopy (IM-STM). Prior research in SPM-enabled manipulation was restricted by reliance on metallic probes and electronic bias, which limited positional selectivity and feasible structures. This study extends the capabilities through a mechanosynthetic approach, achieving bottom-up covalently bonded architectures atom-by-atom and exploring the role of molecular tool design for specific atomic manipulations.

Mechanosynthesis with Molecular Tools: Architecture and Protocol

The molecular tools are tripodal adamantane derivatives, comprising customizable flexible legs for substrate binding, an adamantane core for spatial orientation, and a reactive feedstock at the bridgehead. The prototype tool, EAOGe-C2I, employs an ethyl-legged, oxygen-terminated, germanium-bridgehead design with an acetylenic-iodo functional group, providing a protected source of ethynyl radical upon deiodination.

Mechanosynthetic interactions occur under UHV and cryogenic conditions. Tools are vapor-deposited, oriented, and deprotected (deiodinated) ex situ from the build site. Subsequently, an IM-STM protocol positions the tool with sub-ångström precision relative to a Si(100) build site on custom-fabricated Si probe chips. The targeted atom addition or abstraction is induced solely by mechanical approach/retraction, without applied bias. The process is governed by competitive bond dissociation enthalpies (BDEs) between the tool's bridgehead and the surface, with outcomes confirmed through systematic IM-STM imaging.

Atomically Precise Addition (Donation) and Subtraction (Abstraction)

Positioned ethynyl radicals on the adamantane-derived tools enable the formation of Si–C bonds via donation events, resulting in diverse arrangements such as inter-row, on-dimer, and inter-dimer C2 insertions. The approach yields unique carbon and silicon surface features consistent with prior theoretical and STM studies of acetylene/ethylene dehydrogenation on Si(100), confirming covalent C2 incorporation. A minority of trajectories yield Si atom abstraction, generating silicon vacancies (SiV) with characteristic dynamic instabilities in IM-STM images. Subsequent abstraction events form higher-order defects such as dimer vacancies (DV) and induce spontaneous dimer reconstructions.

The outcomes of a mechanosynthetic event depend sensitively on STM trajectory, the competition between Si–C and Ge–C (or C–C) BDEs, and the dynamic rearrangement of pendant groups upon retraction. Statistical analysis shows donation rates of 54±7%54 \pm 7\% with EAOGe-C2I tools, revealing the limits of selectivity with z-only trajectories and highlighting the demand for tool optimization.

Structural and Chemical Characterization via Imaging Modality

IM-STM imaging enables distinction between initial and final tool states (e.g., pre-interaction -C2I, post-donation -Ge·, post-abstraction -C2Si·) based on imaging modality signatures such as RPI contrast, lattice site symmetries, and contrast inversion. The chemical nature of terminal functional groups on the molecular tool is shown to correlate with unique imaging characteristics, facilitating identification and assignment of the mechanosynthetic outcomes.

Optimizing Mechanosynthetic Selectivity

To enhance selectivity for silicon atom abstraction, a new tool (MAOC-C2I) was synthesized by substituting the germanium bridgehead with carbon and employing shorter methyl-alcohol legs for fewer configuration isomers. This single-atom modification led to perfect selectivity for abstraction (100%100\% abstraction, n=163n=163, across both 4 K and 77 K), with no observed donation events. Extended silicon vacancy structures, including dimer vacancy arrays and arbitrary patterns such as "L" shapes of up to 10 sequentially abstracted atoms, were realized even at elevated cryogenic temperatures.

This result constitutes strong evidence that molecular tool structure alone can deterministically bias mechanosynthetic trajectories in additive or subtractive atomic construction, ensuring precise outcome control for targeted device architectures.

Implications, Limitations, and Future Prospects

This research advances the field of bottom-up APF by providing protocols for both atom addition and subtraction, verified for both carbon and silicon at technologically relevant Si(100) interfaces. The robust mechanosynthetic control, achievable by rational tool design combined with IM-STM, suggests feasible pathways toward constructing more complex and defect-corrected 3D atomically precise nanostructures, moving beyond planar surface chemistry.

The demonstration of single-atom resolution subtraction with 100%100\% selectivity, device-quality patterning, and in situ error correction opens new practical paradigms in nanofabrication, quantum device prototyping, and ultimately scalable 3D atom-by-atom manufacturing. However, stochastic elements still limit the deterministic assembly of intricate multi-atom structures, particularly under simple z-only STM manipulation protocols and for tool variants with lower selectivity. Further understanding of substrate dynamics and bond energetics, as well as the design of new mechanosynthetic trajectories and tool chemistries, will be critical to extending these results to more complex functional building blocks and temperature regimes.

The demonstration here of subtractive and additive APF at the single-atom level on semiconductor surfaces also underpins theoretical models of mechanosynthesis, reinforcing the practicality of positional control in radical surface chemistry and providing a foundation for developing precise, programmable manufacturing techniques at the atomic scale.

Conclusion

The paper presents a significant step in atomically precise fabrication by establishing both additive and subtractive mechanosynthetic methods using rationally designed molecular tools and IM-STM on Si(100). The robust selectivity achieved through molecular tool engineering underlines the critical importance of tool chemistry for deterministic atom manipulation. These findings enable new technical routes for single-atom device engineering, error correction, and bottom-up nanofabrication, bringing atom-by-atom manufacturing closer to practical realization.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Explain it Like I'm 14

What is this paper about?

This paper shows a new way to build and reshape matter one atom at a time. The team uses a special microscope and tiny custom-made “molecular tools” to either add single carbon atoms to a surface or remove single silicon atoms from it. Think of it like using ultra-precise tweezers to place or pluck LEGO pieces, except the “pieces” are atoms and the “board” is a super-clean silicon surface.

What questions were they trying to answer?

The researchers wanted to find out:

  • Can we control chemical reactions so precisely that we add or remove individual atoms exactly where we want, without relying on big electrical pushes that can blur or disturb things?
  • What kinds of tiny “tools” (molecules) work best for adding atoms versus removing them?
  • Can we string together several single-atom steps to make more complex patterns on a surface?

How did they do it?

They combined two big ideas: a flipped (inverted) scanning tunneling microscope and specially designed molecular tools.

The flipped microscope (inverted-mode STM)

A scanning tunneling microscope (STM) “feels” surfaces with a sharp tip and lets scientists see and move atoms. Normally, the metal tip does the touching and reacting. Here, they flipped the role: the “tip side” is a small, flat patch of silicon that acts as the build site, and the “sample side” holds the molecular tools. Because the tools stand up like tiny tripods, each one acts like a local probe that can reach and react with chosen spots on the silicon build site.

They do everything in ultra-high vacuum (no air) and at very low temperatures (as cold as 4 K, near absolute zero, and also at 77 K). That keeps atoms still and surfaces clean, like working in a dust-free, motionless room.

The tiny tools

Each tool is a molecule shaped like a tripod with:

  • “Feet” that stick it to the surface.
  • A sturdy core (adamantane) that holds its shape.
  • A “payload” at the top (the part that will react), protected by an iodine “cap.”

They first pop off the iodine cap with a small electrical nudge. That reveals a “radical,” which you can imagine as a tiny sticky hook eager to grab onto something.

The basic steps

Here is the simple loop they follow to make or erase atoms on the build site:

  • Find a well-oriented tool (by its image signature) and choose a spot on the silicon build patch.
  • Remove the iodine cap to uncover the sticky end.
  • Gently bring the build site and tool together without applying voltage, so a bond forms exactly where they meet.
  • Pull back. One of two things usually happens:
    • Donation: the tool leaves behind a carbon pair (C2) on the silicon surface (you “added” atoms).
    • Abstraction: the tool pulls a silicon atom off the surface (you “removed” an atom), leaving a tiny vacancy.
  • Take images to confirm what happened, and repeat to build more complex patterns.

They also run computer simulations and use the way the images look (“imaging modality”) to tell which tool ending they have and which structure they made, a bit like recognizing a pencil by the kind of line it draws.

What did they find?

The main results are:

  • They achieved atom-by-atom control in both directions:
    • Addition (donation): They placed carbon pairs (C2) on clean silicon in several distinct positions that match known chemistry.
    • Removal (abstraction): They plucked out single silicon atoms to create vacancies, and even pairs of vacancies that reconstruct in predictable ways.
  • They could chain steps together to make more complex features:
    • Two removals at the same spot made a stable “divacancy.”
    • Patterns of vacancies rearranged nearby silicon into “reconstructed dimers,” which they could see and match with simulations.
    • They combined add-and-remove steps to sculpt small, designed features.
  • They learned to “steer” the outcome by redesigning the tool:
    • With one tool (EAOGe-C2I), addition happened a bit over half the time under simple straight-in approaches.
    • By swapping one atom in the tool (replacing germanium with carbon and tweaking the legs to make MAOC-C2I), they got perfect selectivity for removal: 63 out of 63 tries at 4 K and 100 out of 100 at 77 K pulled a silicon atom with no unwanted carbon additions.
  • They demonstrated subtractive patterning at 77 K, carving sequences of up to ~10 removed silicon atoms into shapes and even correcting mistakes (removing stray silicon “adatoms” that appeared).

Why this matters: This is the first time radical chemistry has been guided in 3D space with such positional control to add or remove specific atoms on a clean, technologically important surface (Si(100)), all without applying voltage during the critical bond-forming step.

Why is this important?

  • It brings the long-standing dream of “atom-by-atom manufacturing” closer to reality. Being able to add and remove individual atoms where you choose is like having a nanoscale 3D printer and eraser.
  • It avoids relying on large electrical biases to drive reactions, which can cause side effects. That means cleaner, more predictable results and potentially finer details.
  • It reveals design rules for molecular tools: by tuning bond strengths and geometry, you can strongly favor adding or removing atoms as needed.

What could this lead to?

  • New ways to build ultra-small devices: precisely patterned silicon is the foundation of modern electronics. Better atomic control could help with next-generation chips and quantum devices.
  • Designer materials made from the bottom up: placing specific atoms in exact spots could create properties not possible with traditional methods.
  • A toolkit of molecular “end effectors”: expanding beyond carbon pairs to other building blocks and more complex motion paths should enable richer structures and faster, more reliable fabrication.

In short, this work shows both the “pen” (adding atoms) and the “eraser” (removing atoms) can be guided with sub-atomic precision. With improved tools and strategies, atom-by-atom construction of practical, complex objects looks increasingly achievable.

Knowledge Gaps

Knowledge gaps, limitations, and open questions

Below is a focused list of what remains missing, uncertain, or unexplored in the study, phrased to guide concrete follow‑up research.

Mechanism, energetics, and modeling

  • Quantitative reaction pathways and barriers are not established for approach/retraction events; no ab initio dynamics or force–distance measurements validate whether bond formation/cleavage is quasi-adiabatic, thermally activated, or force-activated.
  • The exact chemical state of the silylated tool after abstraction is explicitly “uncertain”; its structure, charge state, and stability window need spectroscopic confirmation (e.g., STS/IETS/XPS in situ).
  • Bond order and detailed bonding geometry of the transferred C2 on Si(100) are unresolved; STM alone cannot distinguish σ/π character or possible rehybridization and subsurface rearrangements.
  • The reliance on tabulated BDEs and simplified models is acknowledged; there is no explicit, surface-inclusive, trajectory-dependent potential energy surface mapping to rationalize selectivity (donation vs abstraction) or site dependence (on-dimer, inter-row, inter-dimer).
  • The mechanism of sequential Si–Si bond cleavage during abstraction (order of bond rupture, role of local strain/electric field, partial bond formation to C2 during pull-off) is not elucidated.

Selectivity and trajectory control

  • Donation fidelity remains limited (~54±7% for z-only, on-dimer with EAOGe‑C2I); no demonstrated method yet achieves high-fidelity (>90%) donation analogous to the perfect selectivity achieved for abstraction with MAOC‑C2I.
  • The origin of donation failures is not deconvolved (e.g., pendent C2 relaxation pathways, Ge–C bridgehead weakness, local site geometry, approach speed, or lateral misalignment).
  • Only z-only trajectories were tested; there is no systematic exploration of lateral approach paths, approach/retraction velocities, hold times, or multi-step force/position waveforms that might improve donation selectivity.
  • The generality of MAOC‑C2I’s “perfect” abstraction selectivity was shown only for z‑only, on‑dimer interactions; robustness across other sites (inter-row, step edges), biases, speeds, and SPC geometries is untested.
  • No closed-loop, force- or signal-based feedback scheme was used to detect bond formation in real time; lack of concurrent nc-AFM or force sensing precludes deterministic “stop-on-bond” control.

Tool chemistry and design space

  • Design rules are only partially validated: a single-atom change (Ge→C bridgehead) improved abstraction, but there is no systematic scan of bridgehead identity, feedstock types, leg chemistries, and protecting groups to tune both donation and abstraction.
  • Only C2 feedstock is demonstrated; higher-complexity moieties (e.g., CH, CH2, C3/C4, heteroatom-containing groups like BN, NO, SiC, or dopant species) are not explored.
  • Reusability and lifetime of tools (number of cycles to failure, deiodination side reactions, radical quenching pathways) are not quantified.
  • Post-donation tool state (EAOGe•) utility is not assessed; it is unknown whether the residual tool can be refunctionalized in situ or repurposed for subsequent steps.

Imaging, characterization, and classification

  • Imaging modalities are used as chemical proxies, but they lack independent spectroscopic ground truth and quantified classification accuracy (false positives/negatives); the suggested FFT/ML approaches are not implemented or validated.
  • Electric-field/current-induced switching and reconstructions occur during imaging; their influence on apparent structure and on the stability of freshly fabricated features is not quantified.
  • Local band-bending and doping-related effects on contrast and imaging windows are acknowledged but not systematically measured or modeled across different SPC anneals and dopant densities.

Substrate, build-site, and environmental scope

  • Demonstrations are restricted to atomically clean Si(100) SPC build sites; performance on technologically common H:Si(100) (referenced to a companion paper), other semiconductors (e.g., SiC, GaAs), or insulating/metallic substrates is not shown here.
  • Temperature scope is limited to 4 K and 77 K; feasibility, selectivity, and retention of fabricated structures at higher (e.g., 300 K) or intermediate temperatures are unknown.
  • The stability of fabricated features (C2 insertions, vacancies, reconstructed dimers) under thermal cycling, applied fields, or gentle anneals is untested.
  • Build-site evolution after multiple transfers (accumulated defects, adatoms, changes in electronic structure) and its impact on subsequent reaction yields and imaging are not quantified.
  • Residual iodine or other deiodination byproducts near the build site are not characterized; potential contamination effects on yields and electronic structure are unknown.

Structure formation and post-reaction dynamics

  • Spontaneous reconstructions (RD, MRD) are observed but not controlled; rules governing when/where reconstructions occur and how to intentionally invoke or suppress them remain undefined.
  • The suggestion of a newly opened subsurface trough in extended patterning is based on STM contrast/simulation; direct experimental validation (e.g., bias-dependent STS, controlled adatom/dimer manipulation) is lacking.
  • Mobile adatoms are intermittently observed and used for “preliminary error correction”; comprehensive error taxonomies and correction playbooks for both donation and abstraction are not developed.

Throughput, scaling, and automation

  • The method is inherently serial; no strategy is provided for parallelization (e.g., arrays of tools, multiple RPIs per field) or for automated, high-throughput mechanosynthesis.
  • Deposition density/orientation control of surface-bound tools is only heuristically managed via RPI isolation; statistical control over orientation distributions and binding geometries remains unquantified.
  • There is no demonstration of multi-layer (out-of-plane) growth or truly 3D architectures beyond surface-local modifications; pathways to build vertical or cross-layer covalent structures are not specified.

Generality, reproducibility, and data access

  • Site dependence is mentioned (Fig. S2) but not comprehensively mapped (yield heatmaps vs. lattice site, local defects, step edges); reproducible protocols for target site selection are not provided.
  • Device-integration relevance is stated but not demonstrated (e.g., compatibility with subsequent passivation, lithography, or dopant activation workflows).
  • Underlying datasets and raw images are not publicly available due to IP; independent validation and meta-analysis by the community are hindered.

Concrete questions for future work

  • What trajectory, speed, and force profiles maximize donation fidelity while suppressing abstraction for a given tool and site?
  • Which bridgehead/feedstock/leg combinations systematically tune bond strengths and sterics to achieve high (>90%) selectivity for both donation and abstraction across multiple sites?
  • Can in situ spectroscopy (STS/IETS/XPS) or combined nc‑AFM/STM be integrated to deliver bond-order identification and real-time bond-formation feedback?
  • How do temperature, substrate doping/type, and local electric fields jointly modulate reaction barriers, selectivity, and post-reaction reconstructions?
  • Can deterministic strategies be developed for multi-layer growth and vertical covalent linkages to realize true 3D APF?
  • What automation and parallelization architectures (e.g., patterned arrays of bound tools, multi-probe systems) can raise throughput without compromising positional accuracy?

Practical Applications

Immediate Applications

The following immediate applications can be deployed in advanced research and instrumentation settings that already operate ultra-high vacuum (UHV) scanning probe platforms. They leverage the paper’s demonstrated IM-STM workflows, molecular tool designs, and defect-engineering results at 4–77 K.

  • Atomically precise subtractive patterning on Si(100)
    • What: Deterministic abstraction of individual silicon atoms and creation of divacancies/reconstructed dimers using MAOC-C2I with near-perfect selectivity, including basic error correction by adatom removal.
    • Sector: Semiconductors, surface science, instrumentation.
    • Potential tools/products/workflows: “Atomic subtractive lithography kit” (MAOC-C2I molecular tools, silicon probe chips with flat apex mesas, control recipes for z-only trajectories at 77 K), reference process recipes for DVs/RDs.
    • Assumptions/dependencies: UHV, cryogenic operation (4–77 K), IM-STM access, stable tool supply, operator training; patterns’ stability at room temperature not established.
  • Reference defect libraries and calibration standards for STM/STS
    • What: On-demand creation of Si vacancies, divacancies (along-row and cross-trough), and reconstructed dimers that mimic native defects for instrument benchmarking and method validation.
    • Sector: Metrology, academic research, STM vendors.
    • Potential tools/products/workflows: Certified reference samples for tip-state diagnostics and tip-conditioning protocols; standard operating procedures for generating known features on Si(100).
    • Assumptions/dependencies: Low-temperature stability of features; reproducible SPC fabrication; accessible comparison simulations.
  • Mechanosynthetic tool design and screening platform
    • What: Use of demonstrated selectivity shift (EAOGe-C2I vs MAOC-C2I) to iterate molecular tool chemistries for specific additive/subtractive tasks.
    • Sector: Synthetic chemistry, materials design, nanofabrication R&D.
    • Potential tools/products/workflows: Contract synthesis and screening services; a growing “tool library” with characterized imaging modalities for rapid end-state identification.
    • Assumptions/dependencies: Reliable synthesis pipelines, IP/licensing access, UHV stability of radicals and tool legs, access to computational support.
  • Imaging modality-informed classification and operator aids
    • What: Use “imaging modalities” of the tool as a chemical signature to label tool state and interaction outcomes (donation vs abstraction) during/after operations.
    • Sector: Software for SPM, AI/ML in scientific instrumentation.
    • Potential tools/products/workflows: On-instrument ML classifiers (FFT-based feature extraction; modality recognition) to reduce operator ambiguity, suggest next actions, and log outcomes for QA.
    • Assumptions/dependencies: Labeled datasets (subject to IP), integration with existing SPM control software, validation across instruments and dopant levels.
  • Rapid prototyping of novel on-silicon surface motifs
    • What: Opportunistic formation of C2 additions (IR-C2, OD-C2, ID-C2) and combined patterns (e.g., IR-C2 flanked by Si vacancies), enabling exploratory surface chemistry and surface-state studies.
    • Sector: Materials discovery, semiconductor surface physics.
    • Potential tools/products/workflows: Protocols for bias-free bond formation on Si(100), IM-STM workflows for constructing and probing local electronic structures.
    • Assumptions/dependencies: Low-temperature imaging, limited deterministic control across >5 sequential steps using z-only trajectories, unknown thermal robustness.
  • Vendor offerings for IM-STM-compatible hardware and consumables
    • What: Commercialization of SPCs (atomically flat Si(100) mesas), sample preparation holders, and pre-functionalized molecular tools for IM-STM.
    • Sector: Scientific instrumentation and consumables.
    • Potential tools/products/workflows: Turnkey IM-STM add-on packages; pre-characterized tool batches with traceable modality fingerprints; in situ deiodination procedures.
    • Assumptions/dependencies: Manufacturing consistency of SPCs, supply-chain for tool synthesis, training and support.
  • Education, workforce development, and STEM outreach
    • What: Use well-visualized atom-by-atom operations (donation/abstraction) and error-correction demos as teaching modules in advanced nanotechnology courses and public outreach.
    • Sector: Education and training.
    • Potential tools/products/workflows: Courseware on APF (atomically precise fabrication) workflows; virtual labs using simulated STM images aligned with experimental modalities.
    • Assumptions/dependencies: Access to datasets/simulations; limited institutions have the required UHV/cold SPM infrastructure.
  • IP licensing and consulting around APF workflows
    • What: Transfer of know-how on IM-STM, molecular tool design, and process selectivity, including safety and QA practices.
    • Sector: Industrial R&D, instrumentation.
    • Potential tools/products/workflows: Licenses for patented methods; joint development agreements; playbooks for APF lab setup and validation.
    • Assumptions/dependencies: Patent scope/terms; client access to UHV-SPM; trained personnel.

Long-Term Applications

The following applications require further research to increase selectivity, expand tool chemistry, automate trajectories, raise temperature windows, and scale throughput beyond single-head SPM.

  • Atomically precise manufacturing (APM) platform for covalent nanostructures
    • What: Generalized, CAD-to-atom mechanosynthesis of 2D/3D covalent architectures via libraries of additive (e.g., Cx, Nx, Bx, Six) and subtractive tools with deterministic toolpaths.
    • Sector: Advanced manufacturing, materials.
    • Potential tools/products/workflows: APF toolchain (design, simulation, verification, process-control); standardized building-block kits; qualification protocols.
    • Assumptions/dependencies: Expanded tool libraries, complex multi-axis trajectories, robust in situ verification, higher-temperature operation or effective passivation.
  • Atomic-scale semiconductor device fabrication
    • What: Fabrication of single-atom transistors, deterministic donor/acceptor arrays, atomically thin interconnects, and Si–C heterostructure patches directly on Si(100).
    • Sector: Semiconductors, quantum computing.
    • Potential tools/products/workflows: APF-defined qubit arrays and couplers; atomic interconnect routing; hybrid Si/SiC features for band-engineering.
    • Assumptions/dependencies: Integration with CMOS process flows, room-temperature stability, dopant incorporation beyond C2/Si, yield and error-correction at scale.
  • Wafer-scale atomic lithography via parallel IM-STM arrays
    • What: Massive parallelization of IM-STM heads to raise throughput from single sites to millimeter–wafer scales.
    • Sector: Semiconductor manufacturing equipment.
    • Potential tools/products/workflows: Multi-head cryo-UHV APF tools; synchronization and alignment systems; vibration/isolation engineering.
    • Assumptions/dependencies: Precision parallel mechatronics, inter-head calibration, cost/reliability, clean handoff to downstream process steps.
  • On-surface covalent synthesis and carbon growth
    • What: Extension from C2 donation to controlled formation of larger carbon constructs (e.g., diamondoid fragments, graphene nanoribbons, SiC phases) with positional control.
    • Sector: Materials, energy devices, sensors.
    • Potential tools/products/workflows: Mechanosynthetic polymerization workflows; site-specific SiC interfaces; designer bandgaps via atom-by-atom construction.
    • Assumptions/dependencies: Diverse feedstocks and reaction sequences, trajectory control beyond z-only, thermal/chemical stability of products.
  • Catalysts and sensors engineered at the single-atom level
    • What: Design of catalytic active sites or sensing hot-spots by placing/removing atoms and moieties with angstrom precision.
    • Sector: Energy, chemicals, environmental sensing.
    • Potential tools/products/workflows: Catalyst testbeds on Si or other substrates; single-atom sensor arrays; on-instrument screening of structure–function relationships.
    • Assumptions/dependencies: Extension to catalytically relevant surfaces (metals/oxides), operation in reactive/ambient environments, durability.
  • Closed-loop, AI-driven autonomous mechanosynthesis
    • What: Real-time decision-making using imaging modality recognition, outcome prediction, and adaptive toolpath planning for high-yield APF.
    • Sector: Software, robotics for nanofabrication.
    • Potential tools/products/workflows: RL-based controllers; digital twins linking STM simulations to process control; automated error detection/correction via silylated/adatom signatures.
    • Assumptions/dependencies: Large labeled datasets, robust physics-informed models, fast and reliable state estimation under varying conditions.
  • Hybrid top-down/bottom-up process integration
    • What: Use APF-defined atomic features as nucleation sites, masks, or electrical elements that integrate with e-beam/DUV/EUV lithography and deposition/etch processes.
    • Sector: Semiconductors, advanced packaging.
    • Potential tools/products/workflows: Process design kits (PDKs) that include APF steps; selective ALD/CVD enabled by APF-defined chemical cues; local strain/band tuning.
    • Assumptions/dependencies: Cross-compatibility of chemistries and temperatures, contamination control, throughput alignment.
  • Standards, certification, and policy frameworks for atomic fabrication
    • What: Metrology standards for verifying atomically precise features; safety, IP, and export-control policies specific to APF equipment and molecular tools.
    • Sector: Policy, metrology, compliance.
    • Potential tools/products/workflows: Certification protocols for “atomically precise” claims; traceable artifacts; best-practice documents for safe tool handling.
    • Assumptions/dependencies: Consensus across standards bodies; transparent measurement chains; risk assessments for dual-use concerns.
  • High-density atomic data storage and reconfigurable surfaces
    • What: Encoding bits by presence/absence of atoms or specific moieties and re-writing via donation/abstraction.
    • Sector: Data storage, security.
    • Potential tools/products/workflows: Prototype atomic memories on Si(100); nanoscale patterning workflows for physically unclonable functions (PUFs).
    • Assumptions/dependencies: Room-temperature stability and read/write speed, error rates, endurance and retention.
  • Nanomedicine and molecular machines (very long horizon)
    • What: Translation of positional mechanosynthesis to biocompatible environments for molecular machines or targeted therapeutics.
    • Sector: Healthcare, biotech.
    • Potential tools/products/workflows: Atomically designed drug carriers or surface-functionalized biosensors; mechanosynthetic assembly of biocompatible nanostructures.
    • Assumptions/dependencies: Operation in liquids at physiological temperatures, biocompatible substrates and tool chemistries, extensive safety and regulatory validation.

Cross-cutting assumptions and dependencies that affect feasibility

  • Environmental requirements: Current demonstrations require UHV and cryogenic temperatures (4–77 K); raising operation temperature and moving beyond UHV are major milestones.
  • Chemistry/toolkit breadth: Expansion beyond C2/Si to include dopants (e.g., B, P, N), heteroatoms, and larger building blocks is necessary for many applications.
  • Control and selectivity: Improved trajectory control (beyond z-only), deterministic multi-step sequences, and robust error correction are needed to scale complexity and yield.
  • Throughput and scale: Parallelization (multi-head arrays) and automation are required for industrial relevance; mechatronic stability and synchronization are non-trivial.
  • Verification and metrology: In situ, fast, and unambiguous characterization (modality-aware imaging, spectroscopy) must close the loop for autonomous operation.
  • IP and access: Patents and proprietary datasets may affect adoption; collaboration and licensing models will shape ecosystem growth.
  • Process integration: Compatibility with standard semiconductor and materials processing (thermal budgets, contamination control) will determine manufacturability.
  • Stability and durability: Many applications depend on features remaining intact at or near room temperature and under ambient or reactive environments.

Glossary

  • Abstraction: Removal of an atom from a surface by forming a new bond and retracting the tool. "Silicon abstraction is another possible outcome (Fig. 1D/F)"
  • Adamantane: A rigid, diamondoid hydrocarbon framework used as a scaffold in molecular tools. "Adamantane molecular tools are generally described in three parts"
  • Adatom: An atom adsorbed on a surface that is mobile relative to the lattice. "with the second silicon remaining on the SPC surface as a mobile silicon adatom (Fig. 5C)"
  • Ad-dimer: A dimer of atoms adsorbed on a surface, distinct from lattice dimers. "distinct from ad-dimers simulated by others (29)"
  • Annealing: Thermal treatment used to modify surface or probe properties. "which vary based on annealing time in the case of arsenic-doped Si(100) (20)"
  • Apex atom: The atom or moiety at the very tip of a probe that dominates interactions. "the mesoscopic shape of the probe supporting the apex atom or moiety"
  • APF (Atomically Precise Fabrication): Fabrication method where individual atoms are added/removed with positional control. "which we refer to as atomically precise fabrication (APF) via 'positionally-controlled mechanosynthesis' (8)"
  • Arsenic-doped Si(100): Silicon (100) surface intentionally doped with arsenic to modify electronic properties. "arsenic-doped Si(100) (20)"
  • Band-gap: Energy difference between valence and conduction bands governing tunneling bias windows. "The available biases are restricted by the relative band-gaps of the probe and substrate materials chosen"
  • Bond dissociation enthalpy (BDE): Energy required to homolytically cleave a bond. "bond dissociation enthalpy, BDE = 5.213 eV"
  • Bridgehead atom: An atom at a bridgehead position in a rigid framework, anchoring a functional group. "a germanium bridgehead atom"
  • Chemisorption: Strong, chemical (covalent) binding of a molecule to a surface. "surface binding by chemisorption of the 'feet'"
  • Charge transfer: Movement of charge between tip and molecule/surface affecting stability during imaging. "especially to charge transfer"
  • Cryogenic: Very low temperature conditions used to stabilize structures and control reactions. "mechanosynthesis under cryogenic (4 K) ultra-high vacuum conditions"
  • Dangling bond: An unsatisfied valence on a surface atom, often highly reactive. "the adjacent silicon dangling bonds appeared similarly bright"
  • Dehydrogenation: Removal of hydrogen atoms from a molecule, often electron-induced on surfaces. "electron-induced dehydrogenation of acetylene (C2H2) and ethylene (C2H4) on Si(100) (17)"
  • Deiodination: Removal of an iodine atom to unmask a reactive radical center. "The iodine atom protects the reactive feedstock until it is deiodinated (9, 12)"
  • Dimer (silicon dimer): Pair of surface silicon atoms bonded on Si(100) forming the 2×1 reconstruction. "the splitting of the silicon dimer beneath the C2 subunit"
  • Dimer buckling: Dynamic tilting of silicon dimers on Si(100) that can switch under imaging conditions. "perhaps analogous to other instances of dynamic dimer buckling (17)"
  • Dimer vacancy (DV): A missing silicon dimer defect on the Si(100) surface. "silicon dimer vacancy ('DV', Fig. 4A)"
  • Ethynyl radical: A carbon–carbon triple-bond fragment with an unpaired electron (–C2•). "a long-lived ethynyl radical (-C2°, where . represents any moiety's radical character)"
  • Fast-Fourier Transform (FFT): Frequency-domain image analysis used to categorize imaging modalities. "Fast-Fourier Transforms (FFT (24), see Supporting Information Fig. S6)"
  • Feedstock (chemical): The transferable atomic/molecular unit on a tool that participates in bond formation. "the 'feedstock' at the bridgehead position which will take part in the reaction"
  • Functionalization: Modification of a probe or molecule by attaching a specific chemical group to control behavior. "requiring CO or other functionalization methods (4-6)"
  • Germyl radical: A radical centered on germanium, here on the tool after donation. "resulting in a germyl radical- terminated molecular tool (EAOGe®)"
  • H:Si(100): Hydrogen-terminated Si(100) surface used as a model system. "H:Si(100) surfaces (1)"
  • Homolytic cleavage: Bond breaking where each fragment retains one electron, forming radicals. "Homolytic cleavage of C-Ge during retraction yields a pendent ethynyl radical"
  • IM-STM (Inverted-Mode Scanning Tunneling Microscopy): STM mode where a surface-bound molecule acts as a local probe of a flat apex. "inverted-mode scanning tunneling microscopy (IM-STM)"
  • Imaging modality: Distinctive imaging signature tied to the tool’s terminal chemistry and orientation. "collectively referred to as the tool's imaging modality"
  • Inter-dimer C2 (ID-C2): A donated C2 configuration bridging between dimers. "inter- dimer C2 (ID-C2) outcome"
  • Inter-row C2 (IR-C2): A donated C2 configuration positioned between dimer rows. "inter-row C2 (IR-C2, Fig. 2A)"
  • In situ: Performed within the experimental environment without exposure to ambient conditions. "and all reactions are done in situ, without exposure to gaseous precursors"
  • Isocurrent value: Constant-current contour used in STM simulations to render images. "with 50 nA isocurrent value"
  • Mesoscopic: Size regime between microscopic and macroscopic, affecting local fields and imaging. "the mesoscopic shape of the probe"
  • Moiety: A specific functional part of a molecule. "specific atoms and moieties in 3D"
  • OD-C2 (On-dimer C2): A donated C2 configuration centered on a silicon dimer. "on-dimer C2 (OD-C2, Fig. 2B)"
  • qPlus: A type of AFM/STM sensor whose tip state can affect imaging. "qPlus probe shape effects (22, 23)"
  • Reconstructed dimer (RD): A reformed silicon dimer structure created after nearby defects. "reconstructed dimer (RD)"
  • Reflected probe image (RPI): The image of the broad probe apex reflected by a tall molecule acting as a local probe. "Each tool thus generates its own RPI"
  • Sample bias: The applied voltage between tip and sample controlling tunneling conditions. "applied sample bias"
  • Scanning probe microscopy (SPM): Family of techniques manipulating/imaging surfaces with a nanoscale probe. "scanning probe microscopy (SPM) based manipulation of atoms"
  • Scanning tunneling microscopy (STM): A technique imaging surfaces via tunneling current controlled by tip-sample distance. "pausing STM z- axis feedback"
  • Si(100): The (100) crystallographic surface of silicon with characteristic dimer reconstruction. "atomically clean and crystalline Si(100)"
  • Silicon vacancy (SiV): A missing single silicon atom defect on the surface. "a silicon vacancy was formed (SiV, Fig. 2D)"
  • Silylated: Bearing a silicon-containing group attached to the tool. "the exact chemical state of the silylated tool (Fig. 1F) is uncertain"
  • Step edge: A monatomic step on a crystal surface that defines terrace boundaries. "akin to Si(100) step edges"
  • Tip-induced band-bending: Local bending of semiconductor bands due to the electric field from the probe. "tip-induced band-bending (7, 8, 21)"
  • Tripodal geometry: Three-legged arrangement of a surface-bound tool that fixes orientation. "the tripodal geometry ensures that the feedstock is available (Fig. 1C)"
  • Ultra-high vacuum (UHV): Extremely low-pressure environment minimizing contamination. "under ultra-high vacuum (UHV) conditions (13)"
  • z-only trajectory: An approach–retract path varying primarily along the surface normal (z-axis). "z-only trajectories under these conditions"

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Collections

Sign up for free to add this paper to one or more collections.

Tweets

Sign up for free to view the 4 tweets with 42 likes about this paper.