High-extinction electron pulses by laser-triggered emission from a Schottky emitter

Electron sources are important tools in many fields of science, from imaging of biological samples to the study of fundamental physical phenomena.^1–3 Sharp cold field and Schottky emitters are common electron sources—conventionally chosen for their high brightness, transverse coherence, and small energy spread—required for many techniques in electron microscopy and imaging.^4–6 In contrast to cold field emitters, Schottky tips benefit from high probe current stability and extended lifetime, exceeding 10 000 h of continuous operation, making them a very robust and efficient source choice for many electron microscopes.^7,8

Laser-triggered electron sources have been developed, first for flat cathodes¹ and then for field emitters triggered by ultrafast lasers,^9–14 extending electron imaging techniques to the femtosecond time scale. Only recently, Schottky emitter laser triggering has been demonstrated in the femtosecond^15–18 and the microsecond¹⁹ regimes.

One inherent drawback of pulsing electrons in femtosecond time scales is that they are typically limited to the one electron per pulse regime, in order to avoid space charge. This sets a limit on the maximum possible electron flux, which is often low compared to continuous sources. We note that, as a rule of thumb, electron pulses that are 1 ns or longer can achieve the brightness required for single-shot imaging, while not being limited by space charge at energies and fluxes that are common in electron microscopy.^20,21 Furthermore, for some applications of ultrafast electron microscopy techniques such as pump-probe, stroboscopic, and single-shot imaging^1,20 or for techniques that rely on fast gating of electrons, e.g., multi-pass microscopy,^22–24 a source having temporal resolution in the few nanoseconds to sub-nanosecond range is sufficient.

Here, we experimentally characterize the emission properties of a Schottky emitter triggered by nanosecond laser pulses. We study the temporal profile of the generated pulses by electron counting in the single-electron-per-pulse regime, as well as their dependence on dc bias voltage, laser intensity, and polarization.

A diagram of the experimental apparatus is depicted in Fig. 1. A pulsed Nd:YAG, actively Q-switched second harmonic laser (Standa Q10-SH) with a 1.6 ns FWHM pulse width and a 532 nm wavelength is triggered by a pulse generator at a 5 kHz repetition rate. Attenuation of the beam is varied by a rotation controlled half-wave plate (HWP) and a polarizer. The laser is focused to a 5 μm waist (⁠$1/e2$ intensity radius), centered on the tip apex, and where the laser polarization is aligned along the direction of the Schottky tip.

The Schottky tip used in this work is a DENKA TFE 174 ZrO/W emitter with a curvature radius of r = 540 nm, protruding by 250 μm through a complementary suppressor cap of 400 μm opening diameter. The tip (100)-oriented tungsten, is heated to a temperature of 1600 K for about an hour, to allow diffusion of zirconium oxide from a reservoir located at the base of the tip to reach and coat the tip apex. Using high fields at the barrier lowers the work function of the emitting surface to 2.5 eV through the Schottky effect. A voltage bias is applied to tip, $Vtip$⁠, against a grounded extractor anode (not shown in Fig. 1) placed 1.78 mm away from the suppressor cap. In the experiment, this tip bias value is set so that dc electron emission is fully suppressed, and photoelectrons are only generated by the focused laser. These photoelectrons are projected on a dual chevron configuration micro-channel plate (MCP) and imaged on a phosphor screen (not shown in the diagram). An anode placed at the MCP back is used to count and measure electron arrival times, in the single electron per pulse regime. A 1 GHz preamplifier and constant fraction discriminator (ORTEC 9327) with a multi-channel analyzer (Swabian Instruments, TimeTagger 20) time electron arrivals with 50 ps precision.

We start by measuring the temporal profile of the electron pulses by counting photoelectron arrival times integrated over many pulses. Figure 2(a) shows such a measurement, counting the arrival time of $9×105$ photoelectrons in 56 ps time bins, from $4.2×106$ pulses having 0.21 counted electrons per pulse. We observe the electron emission probability for 1 and 10 ns after the laser pulse to be suppressed by a factor of 40 dB and 92 dB, respectively. Timing histograms have been examined for a range of tip voltages; pulses remained short, without any appreciable change. While electron pulses can exhibit high extinction factors (suppression factors of 40 dB or higher, as discussed above), this is no longer the case when the temperature of the tip is increased beyond 1400 K; longer tails become increasingly prominent after the laser is off for 1500 K and 1600 K, as shown in Fig. 2(b). In these measurements, the bias voltage is adjusted to fully suppress the dc current, $Vtip=$ −1524 V, −1493 V, and −1442 V, while the laser pulse energy is adjusted to 72 nJ, 144 nJ, and 144 nJ, for 1400 K, 1500 K, and 1600 K, respectively, to maintain a constant electron count rate of about 0.2 e/pulse.

To further experimentally characterize the emission mechanism, we studied the emission characteristics as a function of laser intensity, dc voltage bias, and laser polarization. In Fig. 3, electron pulse narrowing is shown with increasing laser intensity. We have measured pulse widths at the FWHM from 0.35 ns to 0.7 ns, shorter than the optical pulse by a factor ranging from 2 to 4. Figure 4(a) shows photocurrent measurements (squares) as a function of pulse energy of the laser for the tip voltage bias of −1580 V, −1590 V, −1600 V, and −1610 V. We note that in these measurements, dc current from the tip is suppressed to < 100 e per second on average or $10−7$ e per 1 ns window. Furthermore, in the data of Fig. 4(a), we plot the electron count in a 100 ns window following the laser pulse so that the measured current shows only counts resulting from the laser. Solid lines represent power law fits to the measurements, resulting with powers of 13.5, 9.3, 4.5, and 4.1, for the voltage bias of −1580 V, −1590 V, −1600 V, and −1610, respectively, showing that emission is highly nonlinear.

In order to characterize the emission dependence on laser polarization, we remove the polarizer (see Fig. 1) and rotate the HWP to rotate the angle of linear polarization with respect to the direction of the Schottky tip θ. In Fig. 4(b), the photocurrent is measured as a function of the polarization direction, resulting in maximal emission when the laser polarization is along the tip direction and minimal when the polarization is perpendicular to the tip. The measured data are fitted well with $cos2n(θ)$ over a flat background, for n = 1.4.

While the energy of a single photon from the laser is lower than the potential energy of the Schottky tip, it is interesting that the polarization dependence measured appears to be sharper than single-photon field emission. However, given the highly nonlinear dependence of photocurrent shown in Fig. 4(a) matching a 13.5 photon scaling, for the same tip parameters at fixed laser intensity as in the polarization dependence data [Fig. 4(a)], namely, measured with a laser pulse energy of 58 nJ, a tip temperature of T = 1400 K, and a dc bias voltage of $Vtip=$ −1590 V, it could be that some of the nonlinearity of the polarization dependence results from laser-induced heating of the tip. It was also previously noted in simulations that laser heating has some polarization dependence, where tip heating is increased when the laser polarization is along the tip direction,²⁵ which results in photoemission, which appears nonlinear with polarization rotation. Furthermore, we note that the peak intensity at the focus of our laser is lower by 1–2 orders of magnitude than typical in photofield emission in the femtosecond regime, therefore making two-photon over the barrier emission highly unlikely.

Given the prompt emission, which exhibits narrowing and highly nonlinear photoemission dependence on laser intensity and long tail distribution for high temperatures, it seems that a photoassisted thermally enhanced field emission (PA-TFE) model, assisted by single-photon field emission, is possible. Figure 5 represents a schematic description of PA-TFE, where a laser heats an electron distribution in the metal, an electron then is excited by a laser photon to an intermediate state, followed by tunneling through the barrier, by voltage applied to the tip. We note that a similar model has been described for HfC tips in the femtosecond regime, where the laser-induced temperature rise of several hundreds Kelvin was observed.²⁵ The regime of our work differs from that associated with ultra-fast excitation, where a linear scaling with intensity has been previously observed;¹⁶ in the ultra-fast regime, excitation and emission occur on a timescale faster than the electron-phonon thermalization time,²⁵ thus avoiding thermally induced non-linearities. The long tail seen in our measurements for higher temperatures of the Schottky tip [Fig. 2(b)] is explained by slow nanosecond scale cooling of the tip shank. Similar cooling behavior has been shown in simulations for PA-TFE heating by a laser.²⁵ In the PA-TFE model, heating or temperature change is linear with laser intensity;^25,26 as a result, emission probabilities are strongly nonlinear with laser intensity, as observed in our measurements [Fig. 4(a)]. The nonlinearity of photocurrent as a function of laser intensity changes with the tip voltage bias, as previously been observed.²⁵ It is expected that PA-TFE could yield a reduced energy spread electron source compared to thermally enhanced field emission (TFE), since the photo field emission allows populating a narrower band of energies.²⁷ 

In summary, we have performed an experimental characterization of laser-triggered emission from a Schottky emitter. Our measurements showed the generation of prompt sub-ns and ns pulses of electrons in the single-electron-per-pulse regime, having a high ratio and suppressing long emission below tip temperatures of 1400 K. We have shown a highly nonlinear response to laser intensity, as well as a nearly linear polarization response. These characteristics could be explained by a photo-assisted thermally enhanced emission mechanism. Such a prompt sub-ns response and high ratio of suppression, combined with the high stability of the Schottky emitter, could prove highly desirable in developing technologies for electron optics and electron microscopy, particularly in applications that require fast gating of electrons.

We thank Alexander Stibor, Brian Schaap, Thomas Juffmann, and Marian Mankos for fruitful discussions and assistance with the initial stages of the experiment. This work was done as part of the Quantum Electron Microscope collaboration funded by the Gordon and Betty Moore foundation. Adam Bowman and Brannon Klopfer acknowledge support from the Stanford Graduate Fellowship. Adam Bowman acknowledges support from the National Science Foundation Graduate Research Fellowship Program under Grant No. 1656518.