Implementing electron energy‐gain spectroscopy in scanning transmission electron microscope

Abstract

Spectro‐microscopy techniques like electron energy‐loss spectroscopy (EELS) and cathodoluminescence (CL) are routinely used nowadays for optical characterization of nanostructures with wonderful spatial resolution. With advanced monochromators, the energy resolution in EELS can be as good as 10 meV [1]. Nevertheless, with monochromators, there is a trade‐off between the signal intensity and the energy resolution. Conversely, the spatial resolution in common optical microscopes is diffraction limited, though the spectral resolution is very often down to the sub meV regime. Hence, it is a high time to look for a technique, which can combine the spectacular spectral resolution of optical probes and spatial resolution of the electron probe. The possibility of integrating these two probes was first anticipated by Prof. Archie Howie in 1999 [2], and a few years later, a compact theoretical formalism of the technique named electron energy‐gain spectroscopy (EEGS) was proposed [3]. In EEGS, electrons pick up energy from external illumination, which is very often the coherent output of a finely tuned laser. Although the electron and photon do not couple linearly in free space due to energy‐momentum mismatch, the presence of a nanostructure can break the mismatch by providing the extra momentum through induced light fields, like the evanescent plasmonic field of the nanostructure. Employing synchronized femtosecond pulses of electrons and very intense pulse of photons, electron energy gain (EEG) has been demonstrated, a technique commonly called as photon‐induced near‐field electron microscopy (PINEM) [4]. It has shown that electrons can absorb or emit photons and the resulting spectra consist of peaks positioned at energies integral multiples of the laser photon energy on both sides of the zero loss peak. Multiphoton absorption/emissionis a non‐linear phenomenon, occurring at very high laser peak intensity (~ GWcm ‐2 ), but this can also be triggered at comparatively lower laser intensity, if the laser wavelength is tuned at particle plasmon wavelength [5]. The multiphoton emission and absorption probabilities depend only on the temporal ratio of the electron and photon pulse and the delay between them. The laser can be described as a coherent photon state, which excites a coherent plasmon state and the electron can absorb photon resulting in gain. And the reverse process is a stimulated photon emission by the electron (SEELS) in which the electron loses one photon energy. For large number of incident photons, these two probabilities are the same [5]. For a moderate laser intensity the EEGS or SEELS probability is few ten times larger than the EELS probability [3,5]. In our lab, we are developing an EEGS setup in one of the existing STEM (VG HB 501). The goal is to do spectroscopy by varying the laser energy and detecting the energy gain of electrons. We use a paraboloid mirror (used in CL) to focus the laser output on the sample surface, which can be used over a wide wavelength range without any change of optics. In a normal EEG set up, a pulsed laser is used as excitation source and a pulsed gun is used to detect only those electrons which goes through energy gain or loss in interaction with the laser pulse. In this work we are not using a pulsed gun. A pulsed gun is expensive to make, pulse duration is not easily tunable. A high brightness and thus high spatial resolution is not achieved easily with pulsed gun. We will use a Q‐switched pulsed dye laser with tunable wavelength (570‐900 nm) and typical pulse duration of approximately 30 ns. The instrumental details and some preliminary results will be discussed. We expect that geometries like nanostars (Figure 1(b)) which can confine a huge amount of field might be a promising candidate that can yield much stronger EEGS signals at moderate laser intensity.