Spectral DQE of the Volta phase plate

Abstract

The Volta Phase Plate (VPP) consists of a heated, thin film that is placed in the same plane as the focused diffraction pattern of an electron microscope. A change in surface potential develops at the point irradiated by the intense, unscattered electron beam, and this altered surface potential produces, in turn, a phase shift between the unscattered and scattered parts of the electron wave. While the VPP thus increases the image contrast for weak-phase objects at low spatial frequencies, we report here that it also leads to the loss of an increasing fraction of the signal at higher resolution. The approximately linear dependence (with increasing resolution) of this loss has been quantified at 200 kV and 300 kV, using evaporated-carbon films of different thicknesses as Volta phase plates. In all cases, the loss of signal remains almost independent of variation of the conditions and parameters that were tested. In spite of having done a number of additional, discovery-based experiments, the cause of this loss of signal remains unexplained at this point.

Figure 1.

Power spectral density (PSD) of 300 kV images of thin amorphous carbon obtained with a VPP (blue) and without a VPP (red, dash-dotted line), using an image pixel size of 0.67Å (referred to the specimen) and 2k x 2k pixels. The flat curves are taken without a specimen (“background” curves indicated by an arrow). The inset zooms in on the high-resolution region. The VPP clearly introduces a reduction of modulation depth of the PSD. This VPP had an estimated thickness of 15–20 nm. Microscope optical parameters: gun lens 4, nanoprobe mode, 50 μm C2 aperture, spot size 2, 1 μm under-focus.

Figure 2.

The ratio of the power spectra of background curves as shown in Figure 1 (VPP divided by no VPP). This ratio is constant and has an average value of 0.84.

Figure 3.

Example of a spline-fit through minima and maxima of a PSD graph.

Figure 4.

Spectral DQE of the VPP, defined as the ratio of modulation depth of the PSD with and without a VPP. The PSD graphs have been scaled to the same noise spectral density as defined by the background curves. For comparison, the ratio of the upper-envelope spline fits is also shown, and this ratio demonstrates the same behavior as does the ratio of modulation depths. Beam voltage was 300 kV and the VPP thickness was 15–20 nm. The asterisk shows the DQE value at zero Nyquist, which is the overall signal loss due to scattering by the phase plate. Oscillations in the falloff of the DQE are believed to be an artifact of the spline-fitting to the local maxima and minima of the CTF curves.

Figure 5.

Similar DQE graph as in Figure 4 for a VPP with an estimated thickness of 5 nm, again at 300 kV. The DQE curve fall-off is the same as for a VPP with standard thickness value of 15–20 nm.

Figure 6.

Comparison of EELS spectra obtained either with or without a VPP, inserted as a sample into a Titan column (FEG, 300 kV) fitted with an energy filter (Gatan), and heated to 250 ºC. A Volta potential was created and the zero-loss peak width was recorded, both with (lower, more narrow, dark blue spectrum) and without using a gun monochromator (upper, broader, light blue spectrum). The spectrum extending out to a loss of 40 eV, in the latter case, is inserted in the upper right-hand corner of the figure. As is shown by the red curves, no significant change in the full width at half maximum of the zero-loss peak was observed when moving the sample in and out. All spectra are normalized to a value of 1.0 at zero energy loss. The dashed grey line shows the 50% intensity level from which the full width at half maximum can be judged.

Figure 7.

Spectral DQE curve obtained with a ~3 nm thin amorphous carbon film serving as a hole-based phase plate (300 kV). The hole in the phase plate was 15 μm wide, which corresponds to 0.11/Å cut-on frequency. The unscattered beam was centered in the hole. The DQE curve fall-off is the same as for a VPP without a hole. The DQE at zero frequency is one, showing that the unscattered beam is not attenuated by the VPP.

Figure 8.

Rotationally averaged Thon rings of thin amorphous carbon obtained with a ~3nm thin phase plate (300 kV). The inserted figure shows the phase plate layout with two holes that have a diameter of 15 μm. The unscattered beam was focused on the carbon film between these two holes at the position of the white arrow. The rotational average of the PSD of the image of amorphous carbon is taken in an azimuthal section that covers the hole (dashed blue curve) and a similar-sized azimuthal section outside the hole (solid orange curve). The modulation depth is the same, within experimental error. The blue curve has somewhat higher overall intensity because there is less scattering by the phase plate (as the electron wave is passing the two holes).