Directly detecting defect domains

Figure: Scanning electron diffraction images of UiO-66. Top: a single crystal of UiO-66 visualised in (purple). Bottom: the defect nanodomains (green).

Defects are key to the properties of all kinds of materials, from silicon chips to turbine blades. Metal-organic frameworks (MOFs) are no exception. Previously, we have shown that nanodomains consisting of a more open structure can be produced within crystals of the most important MOFs, so-called UiO-66, by controlling the synthesis. In this paper, we collaborated with teams based at the Universities of Cambridge and Leeds who are experts in an emerging technique, scanning electron diffraction (SED), which allowed us to image directly entire defect nanodomains in a MOF for the first time.

The shape and distribution of domains within a material makes a big difference to how it behaves: in metals the domain structure determines how strong or ductile it is and in porous materials like zeolites or MOFs, the domain structure will determine how easily guests or reactants can diffuse through. Domains are typically too small to see with ordinary light microscopes, but too large to see easily with the X-ray or neutron diffraction techniques used by crystallographers. Materials scientists therefore rely on electron microscopes to visualise domains because electron microscopes have the right resolution and field of view to see these nanoscale features. One serious limitation of electron microscopes is that the electron beam in a microscope will destroy soft materials like MOFs very rapidly, meaning a reliable image cannot be recorded before the crystal is destroyed.

In this paper, we used the advanced microscopes situated at the UK’s Diamond Light Source equipped with ultrafast detectors together with the new technique of SED o capture images of domains without requiring long exposures to the electron beam. SED uses a very small electron beam to measure a diffraction pattern at each point in a finely spaced grid across the sample. This can be done automatically and it minimises the amount of time the sample is exposed to the electron beam. Once this ‘four dimensional’ dataset (the two dimensions of the image and the two dimensions of the diffraction pattern at each point) has been recorded, the defect domains can be visualised by finding in every grid point whether the features in the diffraction pattern which indicate defects are present or not, and then converting this grid back into an image.

These ‘virtual’ dark field images allowed us not only to directly see defect domains in UiO-66 for the first time, but also gave us the first glimpse of other kinds of defects, including low-angle domain walls, which are well known in dense materials but had not yet been seen in MOFs.