Teaching Atoms to Bond: A Molecular Hand for Mechanosynthesis
15 Jun 2026, Yanjiang
A molecular tool on an STM tip donates carbon atoms to or abstracts silicon atoms from a surface, enabling atom-by-atom mechanosynthesis.
The dream has been alive for decades: to build materials, devices, perhaps entire machines, by placing atoms one at a time exactly where we want them. It sounds like a kind of nanoscale alchemy—turning the stochastic dance of chemical reactions into a deliberate, mechanically controlled choreography. But the gap between dream and reality is vast. Atoms don’t come with handles; thermal jostling and the stubbornly probabilistic nature of chemical bonding make atom-by-atom assembly feel, most days, like trying to stack grains of sand using tweezers in a hurricane.
Now, a preprint (arXiv:2606.13876) from a team led by Mathieu Durand at CBN Nano Technologies in Ottawa shows that the hurricane can be quieted, and that the tweezers can have a delicate enough grip. Using a specially designed molecular tool mounted on a scanning tunneling microscope, the researchers have demonstrated the positionally controlled addition (donation) of carbon atoms to, and removal (abstraction) of silicon atoms from, an atomically clean silicon surface—a proof of concept for a chemical fabrication technology that treats atoms as building blocks, not just unpredictable reactants.
The Molecular Crane
Think of the STM tip as a microscopic crane, its boom so precise that it can be positioned with sub‑atomic accuracy. At its tip, instead of a hook, sits a single organic molecule—the molecular gripper. In this work, the gripper is a molecule called EAOGe–C₂I, terminated by a carbon–iodine bond. Under electronic bias in ultra-high vacuum, the iodine atom is cleaved away, leaving behind a long-lived ethynyl radical: a dangling bond ready to form a new connection.
When the tip retracts after making contact, the weakest bond in the system breaks. With the original EAOGe–C₂I tool, the germanium–carbon bond at the bridgehead is most likely to cleave, donating the carbon to the silicon surface—mechanosynthetic donation. But in a fraction of cases, the silicon–carbon bond breaks instead, and the tool pulls a silicon atom from the crystal lattice—mechanosynthetic abstraction. The outcome depends on the tool’s molecular design, not on operator choice. By replacing the germanium bridgehead atom with carbon in a second tool called MAOC–C₂I, the team flipped this selectivity entirely: silicon abstraction became the only observed outcome, turning the tip into a reliable atomic-scale chisel for subtractive patterning.
Adding and Subtracting Atoms
The researchers imaged the resulting structures and confirmed that they had indeed placed carbon atoms or created vacancies at predetermined locations. They observed inter‑row, on‑dimer, and inter‑dimer carbon deposits—three distinct adsorption geometries—as well as single silicon vacancies. By repeating the abstraction, they carved out divacancies and even reconstructed dimers where the surface re‑organized its bonds.
The really striking demonstration came when the team used the subtraction capability to carve a deliberate pattern. By removing silicon atoms along a path, they etched a tiny “L” shape into the surface. When an accidental extra abstraction left a mobile silicon adatom on the surface, they were able to remove it with the same subtractive process—a rudimentary form of error correction, though the original damage to the lattice could not be reversed. This is not just chemical manipulation; it is digital editing of matter.
The Grammar of Molecular Tools
Perhaps the deepest contribution of the paper is a systematic comparison of molecular tools optimized for donation versus abstraction. The difference hinges on the strength of the bond at the bridgehead atom: replacing germanium with carbon increases the bond dissociation enthalpy, making it more energetically favourable for the silicon–carbon bond to break instead, abstracting the silicon. By tuning these factors, the team extracted general principles of molecular tool design: a budding grammar for mechanosynthetic operations. Just as a machinist selects different cutting bits for different materials, the molecular engineer will need a library of tool molecules, each tailored for a specific atom‑placing task.
This is the kind of insight that moves the field from serendipity toward engineering. For decades, chemists have designed reactions by controlling temperature, concentration, and catalysts. Positional control—using a tip to dictate where a reaction occurs and what exactly happens—adds an entirely new dimension. It turns the messy, statistical business of radical chemistry into something resembling a programmable synthesis.
The demonstration takes place at cryogenic temperatures—mostly at 4 K, with subtractive patterning also demonstrated at 77 K (about –196 °C)—on an atomically clean silicon surface in ultra-high vacuum. Scaling up to practical, room‑temperature manufacturing will require new generations of tools that work on arbitrary substrates and with higher throughput. What the CBN Nano Technologies team has done, however, is confirm that the underlying physics works: a single molecule, properly designed, can serve as a controllable delivery system for individual atoms.
For anyone who has ever dreamed of building the world from the bottom up—of assembling a material layer by layer, atom by atom—that is a very good start. The road ahead is long, but the first stones have been laid. We are no longer just watching atoms dance; we are beginning to teach them the steps.
— Yanjiang
Yanjiang is the founding editor of LoomSci.com, specializing in physics and science communication.
References
- Brandon Blue et al., Towards Atom-by-Atom Fabrication: Mechanosynthetic donation and abstraction, arXiv:2606.13876