Method for testing the mounting of semiconductor chips and a computing device for doing so.

Abstract

This invention provides a mounting inspection method, a computing device, and a computer program that calculate the height information of a semiconductor chip using only a planar image of the semiconductor chip after the mounting process. [Solution] A method for inspecting the mounting of a semiconductor chip by a computing device including a processor and a memory for storing one or more programs executed by the processor, includes: acquiring a semiconductor chip image for mounting inspection S110; calculating height information of the semiconductor chip image based on a height prediction model learned to predict height information of a two-dimensional image S120; acquiring an inspection image from the semiconductor chip image based on the height information S130; and inspecting the semiconductor chip mounting based on the inspection image S140.

Claims

1. Processor and A computing device including a memory for storing one or more programs executed by the processor, The aforementioned processor, The operation of acquiring a semiconductor chip image for implementation testing, The operation of calculating the height information of the semiconductor chip image based on a height prediction model trained to predict the height information of a two-dimensional image, Based on the height information, the operation of acquiring an inspection image from the semiconductor chip image, A computing device that performs the operation of mounting and testing the semiconductor chip based on the aforementioned inspection image.

2. The aforementioned processor, In the image of the semiconductor chip, the operation of recognizing objects that are not subject to the implementation inspection as noise based on first pixel information, which is information about the color of each pixel, The computing device according to claim 1, further performing an operation to eliminate noise by changing the color of the noise based on the color of the background layer of the semiconductor chip.

3. The operation of recognizing objects that are not subject to the aforementioned implementation inspection as noise is, If the pixel information of each adjacent pixel changes by more than a predetermined standard, an action is taken to separate the boundary between the two pixels. In the aforementioned semiconductor chip image, the operation involves connecting multiple consecutive boundaries, The computing device according to claim 2, further comprising the operation of recognizing the region of the closed curve as noise when the connected boundaries form a closed curve.

4. The operation of recognizing the region of the closed curve as noise is, The operation of calculating the representative color of the closed curve region, The operation of determining the shape of the closed curve, The computing device according to claim 3, which includes the operation of recognizing the closed curve region as noise when it is determined that the representative color of the closed curve region is darker than the color of the background layer and the shape of the closed curve is a predetermined shape.

5. The operation for calculating the height information is as follows: The operation of inputting the semiconductor chip image to which the height information is labeled into the height prediction model, The computing device according to claim 1, comprising the operation of training the height prediction model so that the height information predicted by the height prediction model is similar to the labeled height information for the semiconductor chip image.

6. The operation of training the aforementioned height prediction model is as follows: The operation involves calculating a loss value that represents the difference between the predicted height information and the labeled height information, The computing device according to claim 5, further comprising the operation of adjusting the neural network constituting the height prediction model so that the loss value is reduced.

7. The operation to acquire the aforementioned inspection image is as follows: The operation of expanding the dimensions of the semiconductor chip image based on the height information to generate a three-dimensional image, In the aforementioned three-dimensional image, the operation involves determining a set of multiple pixels exhibiting the same characteristics as an inspection object based on second pixel information, which is information about each pixel. The computing device according to claim 1, comprising the operation of distinguishing and displaying a plurality of inspection objects in the three-dimensional image and generating the inspection image.

8. The same characteristics mentioned above are, The computing device according to claim 7, which is determined based on one or more of the color information and predicted height information of the pixel.

9. The aforementioned second pixel information is This includes the color information of the pixel and the predicted height information of the pixel. The aforementioned processor, The operation of calculating the average height of the inspection object based on the second pixel information of the inspection object, The operation further involves determining the type of the inspection object according to the average height, The computing device according to claim 7, wherein the aforementioned type is one of a plurality of objects classified in the implementation inspection.

10. The aforementioned types fall under either package or solder. The aforementioned processor, An operation to designate the area including the solder adjacent to the package as the area to be inspected, The computing device according to claim 9, further performing the operation of displaying the area to be inspected on the three-dimensional image and acquiring the inspection image.

11. The aforementioned processor, The operation of comparing the average height of the package and solder in the area to be inspected with the design data of the semiconductor chip, The computing device according to claim 10, further performing an operation to determine whether the package implementation is normal based on the comparison.

12. Processor and A method executed on a computing device including a memory for storing one or more programs executed by the processor, Steps include acquiring a semiconductor chip image for implementation testing, A step of calculating the height information of the semiconductor chip image based on a height prediction model that has been trained to predict the height information of a two-dimensional image, The steps include: acquiring an inspection image from the semiconductor chip image based on the height information; A semiconductor chip mounting inspection method comprising the step of mounting and inspecting the semiconductor chip based on the inspection image.

13. A computer program stored in a non-transitory computer-readable storage medium, The computer program includes one or more instructions, and when the instructions are executed by a computing device having one or more processors, the computing device causes the computer device to: Steps include acquiring a semiconductor chip image for implementation testing, A step of calculating the height information of the semiconductor chip image based on a height prediction model that has been trained to predict the height information of a two-dimensional image, The steps include: acquiring an inspection image from the semiconductor chip image based on the height information; A computer program that performs the steps of mounting and testing the semiconductor chip based on the aforementioned inspection image.

Images