X-rays allow you to see small nanometer-sized edges buried deep in the CPU

Advanced technology is applied to detect errors before putting a new CPU on the market.

The semiconductor industry is famous for its complexity and requires precise processes. A modern integrated circuit is not only a single circuit but also contains many layers stacked on top of each other. This is done by photolithography, which uses light radiation to transform the photosensitive coating on the surface to create the image to be created on a silicon chip.

Structure diagram of Apple A8 chip.

Each class requires a separate image and all images are meticulously calibrated. The mask layer will adjust the light on the part of the circuit that needs to be created. At the 14 nm line, for example, the alignment operations require very high accuracy. So how can you clearly see such small overlapping architectures from which to shape your will?

The answer is due to X-rays.

Stacking architectural 3D images

We are used to the method of using X-rays to capture objects that light cannot illuminate. Bone absorbs more X-ray radiation than the surrounding meat, so doctors can see the difference as well as detect cracks or other lesions.

But surely, you don't know much about X-ray crystallography, the science determines the arrangement of atoms inside a crystal based on X-rays. Specifically, X-rays are scattered by the atom of the crystal. . The crystals are composed of fixed ordered atoms, so X-ray images taken with X-rays will be made into 3D models of electron density.

X-rays are used for bone imaging.

An integrated circuit is a place of different density materials: some powerful X-ray scattering, others are very weak. There is a crystal structure, others are not. However, X-rays go so deep that scattered radiation contains all the information about the entire structure.

In principle, scattering model photography will allow reconstructing 3D images.

However, it is not so. The reason is that we only record the intensity of the X-ray, not its amplitude. While the infinite recording of the 3D model gives the same intensity distribution, we cannot imagine the crystal structure.

To solve this problem, people put chips into extremely powerful X-ray beams to create noise patterns. The chip is rotated to get individual samples. We will create many different X-ray scattering models, repeated until a single structure is identified.

This is a common technique in many 3D imaging programs. However, it takes a lot of time to get each model and longer with the rest.

Free electron lasers become useful

For this reason, Paul Scherrer Institute has focused on finding an alternative solution. The researchers approached the X-ray laser method using ultra-strong and ultra-short X-rays to capture the image of the object before it destroyed the specimen.

Core diagram of free electron laser technique.

They rely on free-electron laser technology, a type of laser capable of changing color and homogeneous luminescence, which helps to obtain a single diffraction pattern of a structured object at the nanoscale. However, the new method is not yet ready for widespread use in industry.

The report by Paul Scherrer Institute said that they have obtained quality 3D images, even seeing rough spots next to the wires of the old IC circuit generation ASIC and Intel Pentium G3260. It is clear that new technology works very well.

Another good signal is that the shooting speed is much faster, allowing the structure of complex chips to be recorded in a short time. Free electron laser techniques will contribute to minimize the error of the microprocessor before selling to the market.

With such great progress, users will benefit because they can buy quality products, high performance and very few errors.