Ultra low emittance electron beams from multi-alkali antimonide photocathode operated with infrared light

Alkali antimonide semiconducting materials are of great interest for their use in photomultiplier devices dedicated to single photon detection¹ and as electron sources for the generation of high brightness electron beams for next generation light sources like Energy Recovery Linacs and Free Electron Lasers.² Recent studies demonstrate that these materials have the potential to generate electron beams with unprecedented, sub-room temperature, low mean transverse energies when operated near the photoemission threshold at a cryogenic temperature. Additionally, the relatively high Quantum Efficiencies (QEs) (7 × 10⁻⁵ at 690 nm) obtained highlight these materials as possible electron sources for Ultrafast Electron Diffraction (UED) experiments.³ 

The spectral response of single and bi-alkali photocathodes usually shows an emission threshold which depends on the growth conditions, and which usually does not extend beyond 750 nm at room temperature.^4,5 Of the alkali antimonide photocathodes family, the layered structure consisting of a Na₂KSb base layer with a Cs₃Sb ultrathin topmost layer (better known as the S-20 photocathode) is of particular interest for the extended response into the infrared part of the spectrum up to about 1 μm.⁶ 

Here, we report on the photoemission properties of an S-20 photocathode using a wide range of wavelengths ranging from 406 to 830 nm. We demonstrate that when operated in the IR part of the spectrum (up to a wavelength of 830 nm), this photocathode can provide bright electron beams with intrinsic emittance very close to the limits imposed by the electron thermal energy (0.22 mm-mrad/(mm rms) corresponding to a Mean Transverse Energy (MTE) of 25 meV at room temperature). The obtained lowest values for the electron beam mean transverse energy seems to indicate that at least up to the electric field intensities used to characterize the electron beam in the present work, the surface roughness contributions to the intrinsic emittance growth are negligible. In particular, a QE of 3 × 10⁻⁵ is measured at a wavelength of 830 nm, a value comparable to typical QE values produced in metals using UV photons, such as those found in RF guns.⁷ 

S-20 photocathodes were grown over the surface of stainless steel substrates that were polished by means of diamond colloidal suspension with decreasing average grain size from 30 μm down to 50 nm. Fig. 1 shows a typical surface morphology of the polished stainless steel substrate obtained using an optical interferometric profiler (ADE Phase Shift MicroXAM). The RMS roughness is determined to be about 3 nm. Substrates were inserted into a dedicated ultra high vacuum growth chamber equipped with effusion cells loaded with high purity elements (Sb, Na, K, and Cs). A detailed description of the growth chamber and of the recipe used to grow the Na₂KSb base layer has been already reported elsewhere.⁴ In order to extend the otherwise limited sensitivity (to wavelength shorter than 750 nm) of Na₂KSb photocathodes into the IR spectrum, we grew a very thin (estimated thickness is less than 10 nm) Cs₃Sb layer over the Na₂KSb surface. Our estimated thickness value is in good agreement with previous findings.⁸ Because the formation of a p-n heterojunction at the interface between the p-type Na₂KSb and the n-type Cs₃Sb (see Fig. 2), photo-excited electrons lying near the bottom of the conduction band within the Na₂KSb layer can travel through the Cs₃Sb layer and be extracted to vacuum as described in Ref. 6. The correct dosing of Cs and Sb over the surface of the Na₂KSb photocathodes is critical to extend the response of the photocathode to the infrared region of the spectrum, potentially all the way to about 1 μm wavelength. To perform the growth of Cs₃Sb topmost layers, we used alternating evaporation of Sb and Cs over the surface of several Na₂KSb photocathodes, which were first allowed to cool down to room temperature. During the growth of Na₂KSb, the QE of photocathodes at 532 nm was used as feedback to maximize the sensitivity of the photocathode. Typical QE values were measured to range between 3 × 10⁻² and 7 × 10⁻² at 532 nm. After allowing photocathodes to cool down to room temperature, the 532 nm laser diode module was replaced with another unit emitting light at 780 nm. With illumination at this wavelength, we cannot detect photoemission from the bare Na₂KSb surface. The Na₂KSb photocathodes were then exposed to very small fluxes of alternating Sb and Cs vapors (on the range of few 10¹² atoms cm⁻² s⁻¹) allowing slow growth of the Cs₃Sb layer to maximize the QE through the fine tuning of its thickness. The photocurrent extracted from the photocathode surface illuminated with light at 780 nm was used as feedback to trigger the Cs and Sb evaporation fluxes on and off.⁴ The photocurrent measured during the exposure of the Na₂KSb photocathodes to Cs and Sb vapors is reported in Fig. 3. Final QEs obtained at 780 nm with this procedure are ranging in the 3 × 10⁻⁵ to 5 × 10⁻⁴ interval.⁴ These numbers are promising and show that we have been capable of extending the sensitivity of our photocathodes to the infrared part of the spectrum. On the other hand, the efficiency of the photoemission process in the IR spectral range is still at least an order of magnitudes lower than the one obtained in commercial photomultipliers tubes for similar materials.⁹ 

One of these cathodes was moved from the growth chamber to one of the high voltage DC guns hosted by Cornell University,¹⁰ and here, QEs and the intrinsic emittances of the electron beam were measured for different laser wavelengths ranging from 406 to 830 nm. For each wavelength, a corresponding low power laser diode was used to illuminate circular apertures of different diameters (0.3, 0.5, 0.7, and 1.0 mm). The illuminated pinhole was then imaged over the photocathode surface by a suitable optical system to generate electron beams with different initial beam sizes. Preliminary measurements of beam emittance have been performed with two methods: the solenoid scan technique and the Emittance Measurement System (EMS).^11,12 The main advantage of the solenoid scan technique is that it does not rely on the knowledge of the initial size of the electron beam which is retrieved along with the initial beam emittance by solving the system of linear equations derived from applying linear optics matrices to the beam propagation as detailed in Ref. 11. The propagation of the measurement errors yields on average a 5% error for the beam sizes and emittances. Much larger systematic errors can be produced from the best fit procedure due to magnetic hysteresis of the focusing solenoid if a proper degaussing procedure is not followed before each measurement.

The EMS system allows a direct mapping of the transverse phase space of the electron beam through direct beam sampling performed by with a pair of slits and a pair of fast scanner magnets. Because this method relies on a direct mapping of the electron beam in phase space to retrieve the beam emittance, hysteresis of the solenoid is irrelevant when the beam dynamics is not affected by space charge. An accurate calibration of the magnification obtained from the optical imaging system allows retrieving the RMS value of the laser spot size over the cathode surface with negligible error so that the errors in measuring the intrinsic emittance are dominated by the uncertainty associated with the beam phase space area measurements. The emittance measurements here reported have been collected using the EMS system while keeping the electron beam currents intensity lower than 1 μA DC so that space charge effects can be neglected.

QE and intrinsic emittance measurements in the high voltage DC gun have been carried out only at a fixed voltage of 150 kV (which correspond to a photocathode surface field of 1.65 MV/m). This was due to the presence of a single field emitter which we believe originated from a defect on the stainless steel substrate surface. At voltages above 150 kV, current from this emitter was generating non negligible levels on the radiation monitors installed around the gun. A linear relationship is used to fit the experimental data of beam emittance as a function of laser beam size. Linear fit examples of experimental datasets are reported for three selected wavelengths in Fig. 4. The intrinsic emittances values $ϵx,i$ defined as $ϵx,i=ϵx/σx$⁠, where ϵ_x is the RMS normalized emittance of the beam and σ_x the RMS value of laser spot size are obtained from the linear fit shown in Fig. 4. The MTE of the electron beam can be calculated using the following equation under the assumption of isotropic emission with no correlation between position and momentum

In the estimate of intrinsic emittances and derived mean transverse energies, the surface roughness induced contribution is included. The derived MTEs of electron beam with respect to the photon energies are reported in Fig. 5 along with a theoretical curve estimating the MTEs of electron beams emitted from a polycrystallyne surface having a work function of 1.75 ± 0.15 eV.¹³ Fig. 5 also includes MTE curves obtained for work function values of 1.6 and 1.9 eV. The work function range has been chosen to better match our experimental data for photon energies lower than 2 eV. It should be again stressed that this photoemission model does not include electron-phonon or electron-electron scattering. Accounting for these relaxation processes will result in lower predicted MTEs and possibly a more accurate work function estimate. For photon energies above 2 eV, much stronger deviations from the simple analytical model are expected as more channels become available for electrons to lose their energy through inelastic scattering.

The QE of the sample was also measured as a function of the different photon energy and is reported in Fig. 6. QE measurements indicate that efficiencies comparable with the ones of commonly used metals like Cu,¹⁴ Pb,¹⁵ and Mg¹⁶ can be obtained from the present multi-alkali photocathodes using longer laser wavelengths than the ones needed to operate metal photocathodes. In addition to that as reported in Fig. 5, the MTEs of the electron beams generated with photon energies in the IR part of the spectrum are closely approaching the thermal limit at room temperature and providing electron beams with extremely small intrinsic emittances (0.247 mm mrad/mm RMS of the laser spot size has been measured at 808 nm along with a corresponding QE of 6 × 10⁻⁵). The response time of these photocathode materials is expected to be shorter than 1 ps as reported for similar materials when operated in streak camera devices¹⁷ and the upper limit of that being determined by the thickness of the photocathode layer that in our case is estimated to be less than 200 nm. Further lowering of the MTEs can be achieved by reducing the photocathode temperature as demonstrated for Cs₃Sb photocathode.³ Under the assumption that the electronic temperature T is locally in equilibrium with the temperature of the lattice within the irradiated area and neglecting any surface induced emittance growth, then the mean transverse energy of the electron beam is, indeed, expected to tend asymptotically to k_bT, with k_b being the Boltzmann's constant, as the photon energy gets lower than the material work function as indicated in Ref. 13.

The availability of photocathode materials with the above shown properties opens new possibilities for developing a compact UED system capable of delivering the extremely low electron beam emittances required for single shot diffraction imaging of large biomolecules. The extended sensitivity in the infrared part of the spectrum enabled by the growth of a thin Cs₃Sb layer over the surface of a high QE Na₂KSb photocathode makes possible operating these fast photocathodes with a laser based on Ti:Sapphire oscillators operated in their fundamental wavelength at about 810 nm.¹⁸ The weak dependence of electron beam MTEs on photon energies lower than 1.6 eV (Fig. 5) coupled with a QE similar to that of metallic photocathodes make these cathode materials attractive to develop “table top” experimental setups capable of producing ultrashort bright electron beams.

An experimental study of the influence of the laser power density on the mean transverse energies of the electron beam generated with ultrashort laser pulses falls beyond the scope of the present work because it requires an ultrafast light source which is not presently available to us. Nevertheless, it has to be noted that if the electronic excitation due the incoming laser beam occurs in a very short period of time (much less then 1 ps) which is the case for a photocathode operated in a photoinjector with an ultrafast laser pulse generated from a Ti:Sapphire laser system, then the electronic and lattice temperatures are expected not to be in equilibrium any longer.¹⁹ An equilibrium condition will be eventually reached after the initial excitation depending on the electron-lattice coupling constant characteristics of the material and that is usually on a few ps time scale. The degree of departure from the thermal equilibrium of the electron energy distribution depends essentially on the laser power density to which the photocathode surface is exposed. The power density needed to operate the photocathode, on the other hand, is defined by the QE of the material at the operating laser wavelength and by the required electron charge per bunch. Due to the small charge per bunch needed to perform single shot UED experiment (less than 0.1 pC), emittance growth due to photoemission from a non thermalized electron distribution can still be found insignificant.

This work has been funded by the National Science Foundation (Grant No. PHY-1416318) and Department of Energy (Grant Nos. DE-SC0014338 and DE-SC0011643).