Reciprocity based phase contrast in a conventional TEM setup

Abstract text (incl. figure legends and references)

The advent of fast pixelated detectors has greatly increased the capabilities for phase retrieval in scanning transmission electron microscopes (STEM) using center of mass (COM) imaging or various ptychographic techniques. In a conventional TEM (CTEM) setup phase contrast either requires modification of the instrument by using phase plates or applying a defocus, thereby limiting the interpretability and resolution.

Here a method is presented by which reciprocity is exploited to create STEM datasets in a CTEM setup from a series of beam tilts. The reciprocity theorem equates a single CTEM beam tilt with a STEM detector pixel, furthermore the radius of the objective aperture and the convergence semi angle of the STEM probe are related [1].

In a STEM setup the whole diffraction pattern is created for each scan point but only parts are sampled by the detector. However, in most cases the crucial information is carried by a small portion of the Ronchigram. Hence a method would be desirable to dedicatedly record the required reciprocal information, especially for low-dose applications. By exploiting the reciprocity relation only the wanted parts of the diffraction space are created. This can be used to reduce measurement times and increase dose efficiency, for example by not sampling the center of the primary beam which does not contain much information [2].

In this work, two specimen were investigated. The application of the method to semiconductor crystals allowed the visualization of the weak built-in electric fields of a pn junction, a feature formerly restricted to STEM setups employing pixelated, DPC detectors or electron holography (Figure 1).

Application to DNA Origamis showed greatly increased contrast when compared to normal CTEM images (Figure 2). Even the minimum of four different tilts around the edge of an aperture greatly increased contrast. The measurement time is reduced drastically compared to similar STEM acquisitions (1 s per 2k x 2k image, four images total).

The recording time can be reduced further by precessing the beam around the edge of the aperture while simultaneously recording images using a fast camera. In this approach the measurement time is not delayed by the speed of the beam tilt coils or delays resulting from the used software interface. Using a Merlin detector 100 images were recorded in 0.1 seconds which reduced the influence of specimen drift.

Of particular interest is the application to biological or other dose-sensitive specimen, due to the promise of increased contrast while keeping measurement times low. Additionally the method allows the introduction of phase contrast into labs which do not have scanning units or segmented detectors. The relation between the chosen beam tilts and the observed contrast transfer is shown and problems like specimen drift will be discussed.

[1] Cowley, J. M. (1969). Applied Physics Letters, 15(2), 58-59.

[2] Lazić, I., Bosch, E. G., & Lazar, S. (2016). Ultramicroscopy, 160, 265-280.

[3] K. M.-C., B. M. and T.L. acknowledge funding from the DFG, contract EXC 2089/1 – 390776260 (e-conversion).

Figure 1: CTEM image of the GaAs specimen (a), the pn-junction is contained between the AlAs marker layers. While invisble in the CTEM image, the pn-junction is visible in the x-component of the COM (b), the average COM-shift perpendicular to the pn-junction is shown.

Figure 2: CTEM image of L-shaped DNA Origamis (a) and calculated integrated COM of the same area (b).