Influence of cathode geometry on electron dynamics in an ultrafast electron microscope

Abstract

Efforts to understand matter at ever-increasing spatial and temporal resolutions have led to the development of instruments such as the ultrafast transmission electron microscope (UEM) that can capture transient processes with combined nanometer and picosecond resolutions. However, analysis by UEM is often associated with extended acquisition times, mainly due to the limitations of the electron gun. Improvements are hampered by tradeoffs in realizing combinations of the conflicting objectives for source size, emittance, and energy and temporal dispersion. Fundamentally, the performance of the gun is a function of the cathode material, the gun and cathode geometry, and the local fields. Especially shank emission from a truncated tip cathode results in severe broadening effects and therefore such electrons must be filtered by applying a Wehnelt bias. Here we study the influence of the cathode geometry and the Wehnelt bias on the performance of a photoelectron gun in a thermionic configuration. We combine experimental analysis with finite element simulations tracing the paths of individual photoelectrons in the relevant 3D geometry. Specifically, we compare the performance of guard ring cathodes with no shank emission to conventional truncated tip geometries. We find that a guard ring cathode allows operation at minimum Wehnelt bias and improve the temporal resolution under realistic operation conditions in an UEM. At low bias, the Wehnelt exhibits stronger focus for guard ring than truncated tip cathodes. The increase in temporal spread with bias is mainly a result from a decrease in the accelerating field near the cathode surface. Furthermore, simulations reveal that the temporal dispersion is also influenced by the intrinsic angular distribution in the photoemission process and the initial energy spread. However, a smaller emission spot on the cathode is not a dominant driver for enhancing time resolution. Space charge induced temporal broadening shows a close to linear relation with the number of electrons up to at least 10 000 electrons per pulse. The Wehnelt bias will affect the energy distribution by changing the Rayleigh length, and thus the interaction time, at the crossover.

FIG. 1.

Schematic illustration of the principal elements of the ultrafast electron microscope. Conventional truncated tip and guard ring cathode geometries are shown as insets. The green and purple pulse trains indicate laser pulses for pump and probe (directed to the sample and photocathode, respectively).

FIG. 2.

Time series PINEM spectra of silver nanowires using 2.4 eV (515 nm) pump photon energy. (a) Probed by a few electrons per pulse pulses and (c) multi-electron pulses (approximately 1000 electrons per pulse at the detector). (b) The temporal change in intensity of the zero loss peak as indicated by the red dashed line in figure (a). The blue line α and the red line β in (d) are the temporal intensity changes at ±10 eV electron energy loss corresponding to the indicated blue α and red β lines in image (c).

FIG. 3.

Temporal cross-correlation of the electron bunch and the pump laser as a function of Wehnelt bias and increasing UV pulse energies (blue 7.3 μJ/cm², red 62 μJ/cm², yellow 740 μJ/cm²). (a) Guard ring cathode at 700 μm gap and (b) truncated tip cathode at 600 μm gap.

FIG. 4.

Energy dispersion of electron bunches photoemitted from (a) a guard ring cathode and (b) a truncated tip cathode at increasing UV fluence (blue 7.3 μJ/cm², red 62 μJ/cm², yellow 740 μJ/cm²). The corresponding average numbers of electrons per pulse measured at the retractable CCD camera are shown in panels (c) for the guard ring cathode and (d) for the truncated tip cathode.

FIG. 5.

Effective source distributions of emission from (a) a guard ring cathode and (b) a truncated tip cathode at increasing bias voltage on Wehnelt. The scale bars represent 200 pixels. (c) and (d) Line profiles (200 pixels) and fitted Gaussians around the center spot showing the effective source distribution for the guard ring cathode and truncated tip cathode, respectively. The curves with shoulders in (c) are a result from a four terms Gaussian fitting. The UV laser fluence was 740 μJ/cm².

FIG. 6.

Electron positions after 1300 ps flight time for a 2000 electron bunch photoemitted from a truncated tip cathode. The simulated Wehnelt bias was set to: (a) 150 V, (b) 350 V, and (c) 550 V. The inset in (c) shows the initial lateral position of the electrons on the surface of a tip cathode (color-coded).

FIG. 7.

Electron positions after 1300 ps flight time for a 2000 electron bunch photo-emitted from (a) a guard ring cathode (750 μm gap) and (b) a truncated tip cathode (650 μm gap). For the truncated tip cathode, only electron emission from the flat central truncated area was considered (shank-emitted electrons were omitted). The figure shows the influence of the Wehnelt bias on the dispersion of the electron bunch. The inset show the initial lateral distribution of the photoelectrons on the central flat surface of the cathode (in the relevant color-code).

FIG. 8.

(a) Changes in temporal dispersion as a function of flight distance for electron bunches emitted from guard ring and truncated tip cathodes (only electrons emitted from the truncated surface is considered). The gray dashed line indicates the position of the limiting aperture. (b) The simulated FWHM temporal width of an electron pulse as a function of bias voltage for a guard ring cathode at 750 μm and 500 μm gap, and a truncated tip cathode at 650 μm gap. (c) The influence of the Wehnelt bias on the extraction field in the longitudinal direction around the cathode surfaces. (d) and (e) Equipotential lines and beam trajectories around the guard ring cathode and truncated tip cathode, respectively. The range of the color scale bar is from −200.0 kV to −199.5 kV for both kinds of cathode. In all panels except (b), the guard ring cathode is placed at a 750 μm gap and the truncated tip cathode at a 650 μm gap.

FIG. 9.

(a) Simulated temporal dispersion for electron bunches emitted from a guard ring cathode containing increasing number of photoelectrons at 50 V Wehnelt bias. (b) Simulated electron energy distribution for a 10 000 electron bunch in a relative time domain (at 50 V bias) immediately after the limiting aperture. The electrons are color coded according to their initial normal velocity in m/s. (c) Energy spectrum obtained projecting the electrons in (b) onto the energy dimension.

FIG. 10.

Normalized effective transverse emittance of emission from guard ring (gap 750 μm) and truncated tip (Gap 650 μm) cathodes under increasing Wehnelt bias voltage. Each simulated electron bunch contains 2000 electrons.