Broadband Ultrafast Electron Microscopy Using Electrically Driven Pulse Generation

Le résumé fourni par la source

Advancements in electron microscopy technology have expanded the power of a TEM such that sub-angstrom spatial resolution and microsecond temporal resolution are common with modern detectors. Ultrafast electron microscopy (UEM) is a recent development that expands TEM capabilities to a broader temporal regime. At Brookhaven National Laboratory, we have been developing and optimizing a laser-free UEM (manufactured by Euclid Techlabs) with wider experimental parameterization and easy-to-use interface [1,2]. Two advantages of an electrically driven UEM are the large frequency range (up to tens of GHz) and ease of synchronization to DC or microwave excitation sources. As a result, our system can synchronize sample excitation to the pulsed electron beam for pump-probe microscopy using frequencies ranging from 1 Hz to 12 GHz and pulse widths as small as a few picoseconds. Here, we describe the modifications to our UEM system that expand the frequency tunability down to 1 Hz (previous minimum was 1 GHz). At high frequencies, the electron beam is swept between two RF strip-lines across a chopping aperture, then corrected with 180° out-of-phase RF to undistort the sweeping motion [1–3]. The pulses can reach temporal widths as small as a few picoseconds and be controlled by a user-friendly interface without the need for comprehensive knowledge of laser optics. At low frequencies, a constant DC potential deflects the electron beam off the optic axis of the microscope. Pulses are created by using a pulse generator that momentarily corrects the electron beam back to the optic axis. Consequently, the pulsing mechanism is limited to the quality of the pulse generator. Fig. 1 contains an example of the electron beam and corresponding pulse durations using the high and low frequency setups. At high frequencies, more efficient RF transmission corresponds to shorter pulse durations because the electron beam is swept a larger magnitude beyond the aperture (see inset of Fig. 1b). For low frequencies, the pulser reciprocates the input signal without significant losses (Fig. 1c). Examples of high frequency operation are well established [1–3], so here explore the efficacy of the low frequency mode. Specifically, we excited a 3000-mesh copper grid with a 1 MHz oscillating DC current alternating between +5 V and -5 V for 500 ns each with a 5 ns edge time. Fig. 2 shows the resulting dynamics when the imaged time points are near the 5 ns transition. This technique images the probe by sampling the dynamics that are contained within the temporal window of the electron pulses. During the transition, the finite probe width coupled to the rise time of the input excitation causes a gradient of states to be super imposed onto one another in the micrograph. Outside of this temporal domain, the images are clearer because they contain one static state. Overall, we will discuss the innovations and developments that enable exciting new possibilities in UEM when the tunable frequency range is broadened by nine orders of magnitude and coupled to a wider variety of excitation sources. Pulse durations of the electron beam for both operation modes of the laser-free electron pulser. (a) Image of the pulsed electron beam with the indicated potentials applied to the two pulser cavities using a 3 GHz frequency. (b) Measured pulse durations using the intensity ratio between the pulsed and continuous beams for various frequencies. The inset is the measured power that transmits through the pulser with a 100-mV input potential. Error bars are the fluctuating variance of the instrument. (c) Measured potential of 1 MHz pulses (2.2 ns) after transmitted through the pulser. Case study using the low frequency pulsing method to observe dynamics in a 3000-mesh copper grid. (a) Comparison of a hole in the copper mesh during the transition from -5 to +5 V (0 ns, left) versus constant +5 V (12 ns, right). Notice that Fresnel fringes lose visibility when the potential is dynamic instead of static. (b) Space-time contour plot of the hole edge with respect to time. When the transition (0 ns) occurs, the contrast from the Fresnel fringes level out, indicating a loss of coherence in time. (c) Intensity along the black dashed line in (b). When a smaller potential is used, the intensity reduction of the Fresnel fringes is reduced. Error bars represent the standard deviation of both edges for three different holes.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.