Adding and Removing Atoms and Molecules Shown

A new paper provides clear, reproducible proof-of-concept for subtractive mechanosynthesis on silicon. It is not yet a complete minimal toolset”demonstration on diamond surfaces, but it validates the core principles of selective, mechanically driven atom abstraction and shows how molecular tool design can control reaction outcome. Together with the additive C₂ donation results, it represents the strongest experimental progress to date toward the early phases of the Freitas/Merkle mechanosynthesis roadmap.

This paper (companion to the May 2026 additive C₂ donation paper) demonstrates positionally controlled subtractive mechanosynthesis — specifically the selective removal (abstraction) of silicon atoms from a Si(100) surface — using inverted-mode STM and custom molecular tools. It provides strong experimental validation of key subtractive capabilities required by the molecular mechanosynthesis roadmap (rep058/rep059).

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

Key Points of the Subtractive Demonstrations

1. Core Achievement: Selective Silicon Abstraction

Using functionalized molecular tools (primarily MAOC–CI and EAOGe–CI), silicon atoms are abstracted from the Si(100) build site during controlled approach-retraction cycles.

The tool forms a temporary bond with a surface silicon, and upon retraction the Si atom is removed (creating single vacancies SiV or divacancies DV), while the tool is released.

High selectivity is achieved through molecular design: MAOC–CI (carbon bridgehead) shows 100% selectivity for abstraction (0 donation observed in 63 trials at 4 K and 100 at 77 K).

EAOGe–CI shows partial selectivity (~46% abstraction rate).

2. Mechanistic Control (Bond-Strength Engineering)

Selectivity comes from relative bond dissociation energies (BDEs): Stronger tool–Si bond vs. weaker surface Si–Si bonds favors abstraction on retraction.

Tool design (choice of bridgehead atom: C vs. Ge) tunes this preference.

Reactions are positionally controlled and occur with minimal or zero bias, emphasizing mechanical force over electrochemical driving.

3. Reproducibility and Patterning

Reliable creation of single Si vacancies, divacancies (along-row and cross-trough), and reconstructed dimers.

Extended subtractive patterning demonstrated (e.g., an “L”-shaped structure from sequential abstraction of ~10 silicon atoms).

Mobile silicon adatoms observed, enabling potential error-correction pathways.
Works at both 4 K and 77 K, showing robustness across temperatures.

4. Validation Methods

Before/after STM imaging with multiple modalities.

Density functional theory (DFT) simulations of reaction pathways and bond strengths.
Correlation between imaging signatures and specific outcomes (vacancies vs. other defects).

5. Relation to the Molecular Mechanosynthesis Roadmap

Directly addresses the subtractive half of the minimal toolset (complementing the H-abstraction shown in the companion paper and the additive C₂ donation).

Proves that designed molecular tools can perform selective atom removal with positional control — a foundational requirement for error correction, surface modification, and complex 3D structure building in the roadmap.
Shows that tool chemistry (bridgehead atom, legs for orientation, capping groups) can be engineered for desired additive vs. subtractive behavior.
Advances the experimental platform (inverted-mode STM + custom molecular tools) toward the Molecular Workstation concept in rep059.