Robert Wilkow has used machine learning techniques to control scanning tunneling microscopes (STM). they have shown a full alphabet of 8-bit memory and 192 bits of music. They have substantially improved automated hydrogen lithography (HL) on silicon, and transformed state-of-the-art hydrogen repassivation into an efficient, accessible error correction/editing tool relative to existing chemical and mechanical methods.
They have scaled from a more manual control of microscopes for writing and erasing to full automation. The speed has been over a hundred times of the smaller scale writing of a handful of points at time. They are now showing hundreds of atomically written points.
They will now working on further scaling of arrays of scanning tunneling microscopes and making the system robust enough to work at room temperature.
Near-term applications would be to make hundreds of potentially more robust qubits based upon quantum dots.
Working with others on leveraging the existing capabilities for applications in the hundreds to thousands of quantum dots.
Improving the scaling to and artificial intelligence control to thousands of writes and beyond to eventually millions of chips per year.
ACS Nano – Autonomous Scanning Probe Microscopy in Situ Tip Conditioning through Machine Learning
Atomic-scale characterization and manipulation with scanning probe microscopy rely upon the use of an atomically sharp probe. Here we present automated methods based on machine learning to automatically detect and recondition the quality of the probe of a scanning tunneling microscope. As a model system, we employ these techniques on the technologically relevant hydrogen-terminated silicon surface, training the network to recognize abnormalities in the appearance of surface dangling bonds. Of the machine learning methods tested, a convolutional neural network yielded the greatest accuracy, achieving a positive identification of degraded tips in 97% of the test cases. By using multiple points of comparison and majority voting, the accuracy of the method is improved beyond 99%.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
It’s not a democracy; “voting” is a ridiculous method. “Most probable” is a better concept for what they’re discussing.
It’s not a democracy; voting”” is a ridiculous method. “”””Most probable”””” is a better concept for what they’re discussing.”””
Yes, I’m familiar with the convergent assembly concept, but that’s at least a step or two after any bootstrapping. We still need to build at least the smallest self-replicating unit first. Assuming that’s the 200 nm block, that’s on the order of a few billion atoms. Plus any support equipment that it may need.
Yes I’m familiar with the convergent assembly concept but that’s at least a step or two after any bootstrapping. We still need to build at least the smallest self-replicating unit first. Assuming that’s the 200 nm block that’s on the order of a few billion atoms. Plus any support equipment that it may need.
Yeah, that was Chris Phoenix’s paper back in 2003. Totally agree with what you say, plus learned some more in your reply. Thanks. Drexler has more current stuff but since links disappear on these Vuukle posts, I can’t include the links. You might find it via Google by searching: “Convergent assembly can quickly build large products from nanoscale parts” from e-drexler dot com He seems to think that the process will even be faster than Chris Phoenix proposed: Size-doubling assembly operations make meter-scale products from nanoscale blocks in only 30 stages. How long will this take? Assume (in round numbers) that each motion-cycle of a full-size arm takes about 4 seconds, so that the arm-pair completes its 8 assembly operations in 16 seconds. Since each stage takes the same amount of time, the total time required for assembly cannot be more than 480 seconds (8 minutes). In fact, however, the stages overlap. The final stage takes 16 seconds to produce a full-size block. It begins when the preceding stage has produced its first half-size block, which takes 8 seconds, and so on. The total time from start to finish is thus very nearly 16⋅(1 + 1/2 + 1/4 + …) = 32 seconds.
Yeah that was Chris Phoenix’s paper back in 2003. Totally agree with what you say plus learned some more in your reply. Thanks.Drexler has more current stuff but since links disappear on these Vuukle posts I can’t include the links.You might find it via Google by searching:Convergent assembly can quickly build large products from nanoscale parts”” from e-drexler dot comHe seems to think that the process will even be faster than Chris Phoenix proposed:Size-doubling assembly operations make meter-scale products from nanoscale blocks in only 30 stages. How long will this take? Assume (in round numbers) that each motion-cycle of a full-size arm takes about 4 seconds”” so that the arm-pair completes its 8 assembly operations in 16 seconds. Since each stage takes the same amount of time the total time required for assembly cannot be more than 480 seconds (8 minutes). In fact however the stages overlap. The final stage takes 16 seconds to produce a full-size block. It begins when the preceding stage has produced its first half-size block which takes 8 seconds”” and so on. The total time from start to finish is thus very nearly 16⋅(1 + 1/2 + 1/4 + …) = 32 seconds.”””””””
I’ve read this before. Thermal noise is a larger issue then expected at the time, but not unsolvable. Drexler talks about how they had hopes for all of this to occur so much faster, but honestly being able to manipulate atoms , even if only on a 2d surface, is a massive first step. We are developing the tools to develop the tools still. And AI systems that can simulate the results before running experiments may show us some rapid advances. I feel like this is entering the stage where it may be worth it for folks to drop a few billion in developing it, because honestly once we CAN produce a nanofactory…..everything changes.
I’ve read this before. Thermal noise is a larger issue then expected at the time but not unsolvable. Drexler talks about how they had hopes for all of this to occur so much faster but honestly being able to manipulate atoms even if only on a 2d surface is a massive first step. We are developing the tools to develop the tools still. And AI systems that can simulate the results before running experiments may show us some rapid advances. I feel like this is entering the stage where it may be worth it for folks to drop a few billion in developing it because honestly once we CAN produce a nanofactory…..everything changes.
Yes, I provided a snippet from Chris Pheonix’s ‘Design of a Primitive Nanofactory’ (2003) in response (above) that gets into that. In that design, we only need ~200nm nanoblocks to be used in convergent assembly.
Yes I provided a snippet from Chris Pheonix’s ‘Design of a Primitive Nanofactory’ (2003) in response (above) that gets into that. In that design we only need ~200nm nanoblocks to be used in convergent assembly.
You only need to make nanoblocks ~200nm in size, and then join them together: 2.4. Nanofactory overview The nanofactory system described here incorporates a large number of fabricators under computer control. In a single product cycle, each fabricator produces one nanoblock, approximately the same size as the fabricator. The blocks are then joined together, eight sub-blocks making one block twice as big. This process is repeated until eight large blocks are produced, and finally joined in an arrangement that is not necessarily cubical. The output of multiple product cycles may be combined to produce large products. The production system is arranged in a three-dimensional hierarchical branching structure (see Section 4.3) which allows the sub-block assembly to be done by machinery of appropriate size. Eight factories of a given size can be combined to form one larger factory; the 64 blocks produced are joined into eight blocks twice as big. The design is easily scalable to tabletop size, with a ~1 meter factory producing eight ~5 cm blocks per product cycle. As discussed in Section 8.4, depending on the capabilities of the mechanochemical fabricator, the time required for a product cycle will be conveniently measured in hours. The blocks need not be solid cubes, and their interior may be quite complex. As discussed in Section 5.1.5, products can be unfolded after manufacture, greatly increasing the range of possible product structures and allowing products to be much larger than the nanofactory that produced them. For this design, a 200-nm cube is convenient: it is large enough to contain a simple 8086-equivalent CPU, a microwatt worth of electrostatic motors/generators, a shaft carrying 0.4 watts (Freitas, 1999, sec. 6.4.3.4), or the Merkle assembler (1999), but small enough to be fabricated quickly and to survive background radiation for a useful period of time. As discussed in Section 6, the partitioning of the product into nanoblocks, and the use of
You only need to make nanoblocks ~200nm in size and then join them together:2.4. Nanofactory overviewThe nanofactory system described here incorporates a large number of fabricators under computer control. In a single product cycle each fabricator produces one nanoblock approximately the same size as the fabricator. The blocks are then joined together eight sub-blocks making one block twice as big. This process is repeated until eight large blocks are produced and finally joined in an arrangement that is not necessarily cubical. The output of multiple product cycles may be combined to produce large products. The production system is arranged in a three-dimensional hierarchical branching structure (see Section 4.3) which allows the sub-block assembly to be done by machinery of appropriate size. Eight factories of a given size can be combined to form one larger factory; the 64 blocks produced are joined into eight blocks twice as big. The design is easily scalable to tabletop size with a ~1 meter factory producing eight ~5 cm blocks per product cycle. As discussed in Section 8.4 depending on the capabilities of the mechanochemical fabricator the time required for a product cycle will be conveniently measured in hours. The blocks need not be solid cubes and their interior may be quite complex. As discussed in Section 5.1.5 products can be unfolded after manufacture greatly increasing the range of possible product structures and allowing products to be much larger than the nanofactory that produced them.For this design a 200-nm cube is convenient: it is large enough to contain a simple 8086-equivalent CPU a microwatt worth of electrostatic motors/generators a shaft carrying 0.4 watts (Freitas 1999 sec. 6.4.3.4) or the Merkle assembler (1999) but small enough to be fabricated quickly and to survive background radiation for a useful period of time. As discussed in Section 6 the partitioning of the product into nanoblocks and the use of relatively lar
True, which is why I stopped at that short comment. I’m not sure how small you can make a useful tool like that. A quantum dot may be useful with only a few hundred to a few thousand atoms. Maybe smaller. But something like a robot arm or some other device only 10 nanometers on a side, which is very small, is still ~1 million atoms, which is a lot. Maybe one can make a useful device at ~10000 atoms, but probably it’ll take more. And that’s just one device. If we can’t make it self-replicating, we’ll need to make a lot more of them the old way. A self-replicating design is likely to be more complex, so it can easily take more atoms – 100000? 1 million? Hard to say. Furthermore, assembling a 3D structure is much more complex than moving atoms on a 2D surface (which is what they did here, I think). I want full nanotech too, but my point is, even the bootsrap approach you’re correctly suggesting compared to what they can do now, is like an olympic runner compared to a baby that just learned to take a few shaky steps.
True which is why I stopped at that short comment.I’m not sure how small you can make a useful tool like that. A quantum dot may be useful with only a few hundred to a few thousand atoms. Maybe smaller. But something like a robot arm or some other device only 10 nanometers on a side which is very small is still ~1 million atoms which is a lot. Maybe one can make a useful device at ~10000 atoms but probably it’ll take more. And that’s just one device. If we can’t make it self-replicating we’ll need to make a lot more of them the old way. A self-replicating design is likely to be more complex so it can easily take more atoms – 100000? 1 million? Hard to say. Furthermore assembling a 3D structure is much more complex than moving atoms on a 2D surface (which is what they did here I think).I want full nanotech too but my point is even the bootsrap approach you’re correctly suggesting compared to what they can do now is like an olympic runner compared to a baby that just learned to take a few shaky steps.
And? How many atoms to make X is the relevant number. AKA How many atoms to make a better method of moving atoms around. At a certain point your tools are made out of atoms, and the ability to create and reorganize them makes some insane possibilities possible.
And? How many atoms to make X is the relevant number. AKA How many atoms to make a better method of moving atoms around. At a certain point your tools are made out of atoms and the ability to create and reorganize them makes some insane possibilities possible.
There are ~100 million surface atoms in just one square micron, one atom thick. ~1 trillion atoms in a cubic micron. This is an important step, but there’s still a long way to go.
There are ~100 million surface atoms in just one square micron one atom thick. ~1 trillion atoms in a cubic micron. This is an important step but there’s still a long way to go.
It’s not a democracy; “voting” is a ridiculous method. “Most probable” is a better concept for what they’re discussing.
Yes, I’m familiar with the convergent assembly concept, but that’s at least a step or two after any bootstrapping. We still need to build at least the smallest self-replicating unit first. Assuming that’s the 200 nm block, that’s on the order of a few billion atoms. Plus any support equipment that it may need.
Yeah, that was Chris Phoenix’s paper back in 2003.
Totally agree with what you say, plus learned some more in your reply. Thanks.
Drexler has more current stuff but since links disappear on these Vuukle posts, I can’t include the links.
You might find it via Google by searching:
“Convergent assembly can quickly build large products from nanoscale parts” from e-drexler dot com
He seems to think that the process will even be faster than Chris Phoenix proposed:
Size-doubling assembly operations make meter-scale products from nanoscale blocks in only 30 stages. How long will this take? Assume (in round numbers) that each motion-cycle of a full-size arm takes about 4 seconds, so that the arm-pair completes its 8 assembly operations in 16 seconds. Since each stage takes the same amount of time, the total time required for assembly cannot be more than 480 seconds (8 minutes). In fact, however, the stages overlap. The final stage takes 16 seconds to produce a full-size block. It begins when the preceding stage has produced its first half-size block, which takes 8 seconds, and so on. The total time from start to finish is thus very nearly 16⋅(1 + 1/2 + 1/4 + …) = 32 seconds.
I’ve read this before. Thermal noise is a larger issue then expected at the time, but not unsolvable. Drexler talks about how they had hopes for all of this to occur so much faster, but honestly being able to manipulate atoms , even if only on a 2d surface, is a massive first step. We are developing the tools to develop the tools still. And AI systems that can simulate the results before running experiments may show us some rapid advances.
I feel like this is entering the stage where it may be worth it for folks to drop a few billion in developing it, because honestly once we CAN produce a nanofactory…..everything changes.
Yes, I provided a snippet from Chris Pheonix’s ‘Design of a Primitive Nanofactory’ (2003) in response (above) that gets into that. In that design, we only need ~200nm nanoblocks to be used in convergent assembly.
You only need to make nanoblocks ~200nm in size, and then join them together:
2.4. Nanofactory overview
The nanofactory system described here incorporates a large number of fabricators under computer control. In a single product cycle, each fabricator produces one nanoblock, approximately the same size as the fabricator. The blocks are then joined together, eight sub-blocks making one block twice as big. This process is repeated until eight large blocks are produced, and finally joined in an arrangement that is not necessarily cubical. The output of multiple product cycles may be combined to produce large products. The production system is arranged in a three-dimensional hierarchical branching structure (see Section 4.3) which allows the sub-block assembly to be done by machinery of appropriate size. Eight factories of a given size can be combined to form one larger factory; the 64 blocks produced are joined into eight blocks twice as big. The design is easily scalable to tabletop size, with a ~1 meter factory producing eight ~5 cm blocks per product cycle. As discussed in Section 8.4, depending on the capabilities of the mechanochemical fabricator, the time required for a product cycle will be conveniently measured in hours. The blocks need not be solid cubes, and their interior may be quite complex. As discussed in Section 5.1.5, products can be unfolded after manufacture, greatly increasing the range of possible product structures and allowing products to be much larger than the nanofactory that produced them.
For this design, a 200-nm cube is convenient: it is large enough to contain a simple 8086-equivalent CPU, a microwatt worth of electrostatic motors/generators, a shaft carrying 0.4 watts (Freitas, 1999, sec. 6.4.3.4), or the Merkle assembler (1999), but small enough to be fabricated quickly and to survive background radiation for a useful period of time. As discussed in Section 6, the partitioning of the product into nanoblocks, and the use of relatively large sub-blocks at each step, allows the use of relatively simple robotics and control algorithms in the nanofactory. As discussed in Section 5, such division also simplifies product design without imposing many practical limits on product complexity.
At the smallest scale, the organization of the factory changes to allow simpler distribution of feedstock, cooling, power, and control, and simpler error handling. A production module consists of one computer and a few thousand fabricators. It produces a few blocks, a few microns in size, by combining a few thousand nanoblocks. These rectilinear production modules incorporate a few block assembly stages. They are combined into the smallest factories, which are also rectilinear—and so on to any size.
Design of a Primitive Nanofactory
December 4, 2003 by Chris Phoenix
True, which is why I stopped at that short comment.
I’m not sure how small you can make a useful tool like that. A quantum dot may be useful with only a few hundred to a few thousand atoms. Maybe smaller. But something like a robot arm or some other device only 10 nanometers on a side, which is very small, is still ~1 million atoms, which is a lot. Maybe one can make a useful device at ~10000 atoms, but probably it’ll take more. And that’s just one device. If we can’t make it self-replicating, we’ll need to make a lot more of them the old way. A self-replicating design is likely to be more complex, so it can easily take more atoms – 100000? 1 million? Hard to say. Furthermore, assembling a 3D structure is much more complex than moving atoms on a 2D surface (which is what they did here, I think).
I want full nanotech too, but my point is, even the bootsrap approach you’re correctly suggesting compared to what they can do now, is like an olympic runner compared to a baby that just learned to take a few shaky steps.
And? How many atoms to make X is the relevant number. AKA How many atoms to make a better method of moving atoms around. At a certain point your tools are made out of atoms, and the ability to create and reorganize them makes some insane possibilities possible.
There are ~100 million surface atoms in just one square micron, one atom thick. ~1 trillion atoms in a cubic micron. This is an important step, but there’s still a long way to go.