Artificial intelligence driven automated atomic manipulation
Typical atomic manipulation techniques require long fabrication times and are not readily scalable due to necessary input from an operator. A method and system for performing automated atomic manipulation is described. The method includes obtaining an image of an atomic manipulation surface for fabricating an atomic structure thereon. A processor performs image processing and determines a movable elements, and defects, in the image of the atomic manipulation surface. The processor further determines respective positions of each of the movable elements, and forbidden regions of the atomic manipulation surface, each forbidden region determined from the determined defects. A fabrication design plan is then determined from the positions of the movable elements, and forbidden regions.
1. A method for performing automated atomic manipulation, the method comprising:
obtaining, by an imaging sensor, a surface image of an atomic manipulation surface;
determining, by a processor and from the surface image, a plurality of candidate features of the atomic manipulation surface, each candidate feature being one of a movable atom or a defect, and each candidate feature being in a respective region of the image of the surface;
identifying, by the processor, which of the plurality of candidate features are movable nanoelements;
identifying, by the processor, respective positions of each of the movable nanoelements;
identifying, by the processor, which of the plurality of candidate features are defects;
identifying, by the processor, one or more forbidden regions, wherein each of the one or more forbidden regions contains an identified defect, and wherein each forbidden region is indicative of a region of the surface that is not to be traversed by a probe tip of a scanning probe microscope.
2. The method of claim 1 , wherein identifying which of the candidate features are movable nanoelements comprises implementing a machine learning technique.
3. The method of claim 1 , wherein identifying which of the candidate features are movable atoms comprises:
retrieving, by the processor and from a memory, characteristic properties of images of surfaces having movable nanoelements thereon;
performing, by the processor, image processing on the surface image and determining characteristic properties of regions of the surface image; and
comparing, by the processor, the characteristic properties of the regions of the surface with characteristic properties of images of surfaces having movable nanoelements thereon.
4. The method of claim 3 , wherein determining characteristic properties of regions of the surface image comprises determining at least one of an image sharpness, a contrast value, an edge detection value, or a geometry of a candidate feature in the surface image.
5. The method of claim 1 , wherein identifying which of the candidate features are defects comprises implementing a machine learning technique.
6. The method of claim 1 , wherein identifying which of the candidate features are defects comprises:
retrieving, by the processor and from a memory, characteristic properties of images of surfaces having defects;
performing, by the processor, image processing on the surface image and determining characteristic properties of regions of the surface image; and
comparing, by the processor, the characteristic properties of the regions of the surface with characteristic properties of images of surfaces having defects.
7. The method of claim 6 , wherein determining characteristic properties of regions of the surface image comprises determining at least one of an image sharpness, a contrast value, an edge detection value, or a geometry of a candidate feature in the surface image.
8. The method of claim 1 , wherein each forbidden region is bounded by one of a polygonal geometric boundary, a circular geometric boundary, or an elliptical geometric boundary.
9. The method of claim 1 , wherein each forbidden region is bounded by a boundary determined from the size and shape of a respective defect contained in the respective forbidden region.
10. The method of claim 1 , further comprising:
accessing, by the processor, an atomic structure design indicative of a desired atomic structure on the surface;
determining, by the processor, a design plan for fabricating the atomic structure on the surface, the design plan determined from (i) the positions of the movable nanoelements and (ii) the one or more forbidden regions; and
performing, by a scanning microscope, the design plan and fabricating the atomic structure, wherein the design plan prevents a probe tip of the scanning microscope from traversing any forbidden region.
11. The method of claim 10 , wherein determining the design plan further includes performing a global path minimization for moving the movable nanoelements from their identified positions, to respective positions required for fabricating the atomic structure.
12. A system for performing atomic scale manipulation, the system comprising:
a scanning microscope, the scanning microscope having a scanning probe that provides electrostatic manipulation of positions of atoms on a surface;
a processor configured to execute machine readable instructions; and
a non-transitory computer-readable memory having machine readable instructions stored thereon, that when executed by the processor, cause the system to:
obtain, by the scanning microscope, a surface image of a surface having movable naoelements and defects;
identify, by the processor, a plurality of movable nanoelements and positions of each movable nanoelement on the surface;
identify, by the processor, a plurality of defects and a position of each defect on the surface;
identify, by the processor and from the plurality of defects, a plurality of forbidden regions, each forbidden region containing an identified defect, and wherein each forbidden region is indicative of a region of the surface that is not to be traversed by the scanning probe of the scanning microscope.
13. The system of claim 12 , wherein to identify a plurality of movable nanostructures, the machine readable instructions further cause the system to implement a machine learning technique.
14. The system of claim 12 , wherein to identify a plurality of movable nanostructures, the machine readable instructions further cause the system to:
retrieve, by the processor and from a memory, characteristic properties of images of surfaces having movable atoms thereon;
perform, by the processor, image processing on the surface image and determining characteristic properties of regions of the surface image; and
compare, by the processor, the characteristic properties of the regions of the surface with characteristic properties of images of surfaces having movable atoms thereon.
15. The system of claim 12 , wherein to identify a plurality of defects, the machine readable instructions further cause the system to implement a machine learning technique.
16. The system of claim 12 , wherein to identify a plurality of defects, the machine readable instructions further cause the system to:
retrieve, by the processor and from a memory, characteristic properties of images of surfaces having defects;
perform, by the processor, image processing on the surface image and determine characteristic properties of regions of the surface image; and
compare, by the processor, the characteristic properties of the regions of the surface with characteristic properties of images of surfaces having defects.