Submission + - Mechanosynthesis Just Got Demonstrated. Smalley's Fat Fingers Were Wrong. (arxiv.org)
tylorsama writes: There's a preprint up on arXiv that claims the result this field has been arguing about for the better part of two decades, and unless the supplementary data quietly falls apart, it looks like the real thing: atomically precise mechanosynthesis, on an actual surface, with an actual tip. [arxiv.org/abs/2605.27250]
The work uses an inverted-mode STM on hydrogen-passivated Si(100) to donate Câ units off surface-deposited molecules onto reactive sites that were also patterned in-situ. They show it at a single site, then across multiple patterned sites, and then they go and assemble polyyne chains by forming the C-C bonds one after another. That last part is the part that matters, because it is positional covalent chemistry rather than self-assembly or something falling out of solution statistically. You pick where the carbon goes, you put it there, and then you do it again somewhere else on purpose.
It is worth remembering that Nobel winning chemist Richard Smalley spent the early 2000s explaining to Drexler, in print, that this could never happen. Fat fingers, sticky fingers, no room at the bottom for a tool that sets individual atoms down on command. The author list on this paper includes Freitas and Merkle, who have been publishing the positional toolset papers since back when that debate was still live, and who appear to have spent the intervening twenty years being right and waiting for the instrumentation to catch up. Ironically, it looks like they leverage the 'sticky fingers' argument as part of their engineering, ensuring the place they donate is stickier than the finger letting go.
Which brings us to the argument actually worth having. A few dozen bonds placed inside a multi-million-dollar UHV rig is a very long way from a gram of anything, and the regulars who are already drafting replies about throughput and serial deposition rate are not wrong to do it. The honest open question is whether any of this parallelizes past hero-experiment status, and nobody has shown that yet. But the wall that a Nobel laureate told the whole field was made of physics rather than engineering now looks like it was engineering the entire time.
For the reproducibility hawks in the back: the supplementary information is "available upon request," which in 2026 is its own quiet little editorial. Have at it.
The work uses an inverted-mode STM on hydrogen-passivated Si(100) to donate Câ units off surface-deposited molecules onto reactive sites that were also patterned in-situ. They show it at a single site, then across multiple patterned sites, and then they go and assemble polyyne chains by forming the C-C bonds one after another. That last part is the part that matters, because it is positional covalent chemistry rather than self-assembly or something falling out of solution statistically. You pick where the carbon goes, you put it there, and then you do it again somewhere else on purpose.
It is worth remembering that Nobel winning chemist Richard Smalley spent the early 2000s explaining to Drexler, in print, that this could never happen. Fat fingers, sticky fingers, no room at the bottom for a tool that sets individual atoms down on command. The author list on this paper includes Freitas and Merkle, who have been publishing the positional toolset papers since back when that debate was still live, and who appear to have spent the intervening twenty years being right and waiting for the instrumentation to catch up. Ironically, it looks like they leverage the 'sticky fingers' argument as part of their engineering, ensuring the place they donate is stickier than the finger letting go.
Which brings us to the argument actually worth having. A few dozen bonds placed inside a multi-million-dollar UHV rig is a very long way from a gram of anything, and the regulars who are already drafting replies about throughput and serial deposition rate are not wrong to do it. The honest open question is whether any of this parallelizes past hero-experiment status, and nobody has shown that yet. But the wall that a Nobel laureate told the whole field was made of physics rather than engineering now looks like it was engineering the entire time.
For the reproducibility hawks in the back: the supplementary information is "available upon request," which in 2026 is its own quiet little editorial. Have at it.
