Technologies for verifying and validating electronic devices using electroluminescence

A method of inspecting integrated circuits, comprising, with a controller:

• obtaining a first electroluminescent (EL) test data of a device under test (DUT) at least in part by measuring EL emissions from the DUT using at least one first detection system;
• obtaining a second EL test data of the DUT at least in part by measuring the EL emissions from regions of the DUT using at least one second detection system;
• comparing the first EL test data and the second EL test data to a reference model of a reference device, the reference model developed based at least in part on measured EL reference data, synthetic EL reference data;
• or a combination thereof obtained from the reference device or a reference design of the reference device;
• converting the reference model to one or more reference images in an image format, wherein the image format is chosen from the group consisting of photon data analysis results mapped to a single channel image of multiple single channel images of frequency results, rise time results, spectral data mapped to the single channel image, and multiple signal channel images converted to a single multichannel reference image of n channels or a stack of images with the n channels;
• reducing a dimensionality of the reference images using machine learning techniques to create one or more reduced reference images, wherein the machine learning techniques include at least one of image to image conversion or principal component analysis;
• applying a same machine learning techniques to one or more measured images of the DUT to create one or more reduced measured images;
• comparing the one or more reduced reference images to the one or more reduced measured images;
• and
• determining whether the DUT is in accordance with the reference device or a reference specification of the reference device based at least in part on the comparison.

A method for verifying and validating integrated circuits, comprising, with a controller:

• obtaining synthetic or measured EL data from a plurality of test vectors based on a known good reference;
• producing a reference model from the synthetic or measured EL data;
• converting the reference model to one or more reference images in an image format;
• reducing a dimensionality of the reference images using machine learning techniques to create one or more reduced reference images, wherein the machine learning techniques include at least one of image to image conversion or principal component analysis;
• applying a same machine learning techniques to one or more measured images of the DUT to create one or more reduced measured images;
• comparing the one or more reduced reference images to the one or more reduced measured images;
• measure EL data using multiple test vectors from a device under test (DUT);
• aggregate the measured EL data;
• comparing the reference model to the measured EL data;
• determining whether the DUT is within specification of the reference model based on the comparison;
• and
• reporting whether the DUT is within specification to a user.