A detector for electron ptychography

Anyone who writes about physics research must have a part of their headspace currently taken up by assessing a new and potentially groundbreaking claim out of the IISc: the discovery of superconductivity at ambient pressure and temperature in a silver nanostructure embedded in a matrix of gold. Although The Hindu has already reported it, I suspect there’s more to be said about the study than is visible at first glance. I hope peer review will help the dust settle a little, but we all know post-publication peer-review is where the real action is. Until then, other physics news beckons…

Unlike room-temperature superconductivity, odds are you haven’t heard of ptychography. In the field of microscopy, ptychography is a solution to the so-called phase problem. When you take a selfie, the photographic sensor in your phone captures the intensity of light waves scattering off your face to produce a picture. In more sophisticated experiments, however, information about the intensity of light alone doesn’t suffice.

This is because light waves have another property called phase. When light scatters off your face, the phase change doesn’t embody any useful information about the selfie you’re taking. But if physicists are studying, say, atoms, then the phase change can tell them about the distribution of electrons around the nucleus. The phase problem comes to life when microscopes can’t capture phase information, leaving scientists with only a part of the picture.

Sadly, this constraint only exacerbates electron microscopy’s woes. Scientists in various fields use electron microscopy to elucidate structures of matter that are much smaller than the distances across which photons can act as probes. Thanks to their shorter wavelength, electrons are used to study the structure of proteins, the arrangement of atoms in solids and even aid in the construction of complex nanostructure materials.

However, the technique’s usefulness in studying individual atoms is limited by how well scientists are able to focus the electron beams onto their samples. To achieve atomic-scale resolution, scientists use a technique called high-angle annular dark-field imaging (ADF), wherein the electrons are scattered at high angles off the sample to produce an incoherent image.

For ADF to work better, the electrons need to possess more momentum, so scientists typically use sophisticated lenses to adjust the electron beam while they boost the signal strength to take stronger readings. This is not desirable. If the object of their study is fragile, the stronger beam can partially or fully disintegrate it. Thus, the high-angle ADF resolution for scanning transmission electron microscopy has been chained to the 0.05 nm mark, and going up to 0.1 nm for more fragile structures.

Ptychography solved the phase problem for X-ray crystallography in 1969. The underlying technique is simple. When X-rays interact with a sample under study and return to a detector, the detector produces a diffraction pattern that contains information about the sample’s shape.

In ptychography, scientists iteratively record the diffraction pattern obtained from different angles by changing the position of the illuminating beam, allowing them to compute the phase of returning X-rays relative to each other. By repeating this process multiple times from various directions, scientists will have data about the sample that they can reverse-process to extract the phase information.

Ptychography couldn’t be brought to electron microscopy straightaway, however, because of a limitation inherent to the method. For it to work, the microscope has to measure the diffraction intensity values with equal precision in all the required directions. “However, as electron scattering form factors have a very strong angular dependence, the signal falls rapidly with scattering angle, requiring a detector with high dynamic range and sensitivity to exploit this information” (source).

In short, electron microscopy couldn’t work with ptychography because these detectors didn’t exist. As an interim solution, in 2004, researchers from the University of Sheffield developed an algorithm to fill in the gaps in the data.

Then, on July 18, researchers from the US reported that they had built just such a detector (preprint), which they called an “electron microscope pixel array detector” (EMPAD), and claimed that they had used it to retrieve images of a layer of molybdenum disulphide with a resolution of 0.4 Å. One image from their paper is particularly stunning: it shows the level of improvement ptychography brings to the table, leaving the previous “state of the art” resolution of 1 Å achieved by ADF in the dust.

Source: https://arxiv.org/pdf/1801.04630.pdf
Source: https://arxiv.org/pdf/1801.04630.pdf

The novelty here isn’t that the detector is finally among us. The same research group (+ some others) had announced that it had built the EMPAD in 2015, and claimed then that it could be used for better electron ptychography. What’s new now is that the group has demonstrated it.

a) Schematic of STEM imaging using the EMPAD. b) Schematic of the EMPAD physical structure. The pixelated sensor (blue) is bump-bonded pixel-by-pixel to the underlying signal processing chip (pink). Source: https://arxiv.org/pdf/1511.03539.pdf
a) Schematic of STEM imaging using the EMPAD. b) Schematic of the EMPAD physical structure. The pixelated sensor (blue) is bump-bonded pixel-by-pixel to the underlying signal processing chip (pink). Source: https://arxiv.org/pdf/1511.03539.pdf

According to their 2015 paper, the device

consists of a 500 µm thick silicon diode array bump-bonded pixel-by-pixel to an application-specific integrated circuit. The in-pixel circuitry provides a 1,000,000:1 dynamic range within a single frame, allowing the direct electron beam to be imaged while still maintaining single electron sensitivity. A 1.1 kHz framing rate enables rapid data collection and minimizes sample drift distortions while scanning.

For the molybdenum disulphide imaging test, the EMPAD had 128 x 128 pixels, operated in the 20-300 keV energy range, possessed a dynamic range of 1,000,000-to-1 and with a readout speed of 0.86 ms/frame. The scientists also modified the ptychographic reconstruction algorithm to work better with the detector.