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.
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%.