X-ray imaging of microchips fills the gap

Three-dimensional interconnections in electronic devices are now so small and intricate that they cannot be imaged without destroying them in the process. This leaves a ‘metrology’ gap between the design of devices and the actual output of manufacturing.

But a technique known as ptychographic X-ray computed tomography (PXCT) that uses X-rays instead of light or electrons to examine samples non-destructively could hold the answer, suggests a team of researchers. Mirko Holler and colleagues from the Paul Scherrer Institute used the synchrotron beam at the Swiss Light Source to examine integrated circuits (ICs) in a new way [Holler et al., Nature (2017), doi: 10.1038/nature21698].

“ICs are typically investigated with focused ion beam scanning electron microscopy (FIB SEM) methods, which are destructive,” explains Holler. “Such imaging causes artifacts, so-called curtaining, because the cutting does not produce a flat surface.”

Instead of taking physical slices through a sample, PXCT takes of many (hundreds of thousands) of individual X-ray diffraction patterns, where a coherent beam of X-rays is scattered by the sample. The multiple images are then added together – or reconstructed – into three-dimensional, real-space renditions of density variations in the sample. In contrast to conventional scanning imaging techniques, the resolution is not determined by the diameter of the beam or the optics of the microscope but by noise levels in the signal and the accuracy with which the sample is positioned.

“In our case, imaging artifacts do not occur, and the resolution and measurement time is comparable [to FIB SEM],” says Holler.

Holler and his colleagues put the technique through its paces using a microchip of known design and then examined a commercial chip of unknown design. The reconstructed three-dimensional images allow the identification of the key parts of the chip, as well as manufacturing artifacts. The technique is sensitive enough to detect regions of silicon doped with n- or p-type impurities.

With its resolution of less than 15 nm, PXCT has sufficient contrast and sensitivity to allow imaging of even the smallest circuitry but the time required to acquire images is long. With the advent of more advanced synchrotron systems, Holler believes that the resolution and/or imaging speed of PXCT can be improved dramatically.

“X-ray imaging can now produce images of high quality and resolution of ICs that allow seeing the individual transistors,” says Holler. “I believe that the state-of-the-art FIB SEM methods used in chip inspection will be replaced by PXCT imaging.”

The researchers believe the technique could support the optimization of production processes, identification of failure mechanisms, and validations of microchips. While the technique currently needs a small sample to be removed from an IC, the team is working on an approach whereby a whole two-dimensional chip could be imaged in three-dimensions.

Jianwei (John) Miao of the University of California, Los Angeles believes that the real novelty of the method lies in its ability to non-destructively image heterogeneous features buried in a three-dimensional material at very high spatial resolution. However, he points out that the current work was performed using a third generation synchrotron radiation source.

“As a coherent diffractive imaging technique, PXCT requires a coherent X-ray source (i.e. the beam is parallel and the bandwidth is narrow). To make this technique widely accessible to both academia and industry, one needs to implement it using tabletop coherent X-ray sources. Presently, several groups around the world are developing such X-ray sources," he says.