Photocathode Quantum Efficiency of Ultra-Thin Cs2Te Layers On Nb Substrates

The quantum efficiencies (QE) of photocathodes consisting of bulk Nb substrates coated with thin films of Cs2Te are reported. Using the standard recipe for Cs2Te deposition developed for Mo substrates (220 {\AA} Te thickness), a QE ~11% - 13% at light wavelength of 248 nm is achieved for the Nb substrates, consistent with that found on Mo. Systematic reduction of the Te thickness for both Mo and Nb substrates reveals a surprisingly high residual QE ~ 6% for a Te layer as thin as 15 {\AA}. A phenomenological model based on the Spicer 3-Step model along with a solution of the Fresnel equations for reflectance, R, leads to a reasonable fit of the thickness dependence of QE and suggests that layers thinner than 15 {\AA} may still have a relatively high QE. Preliminary investigation suggests an increased operational lifetime as well. Such an ultra-thin, semiconducting Cs2Te layer may be expected to produce minimal ohmic losses for RF frequencies ~ 1 GHz. The result thus opens the door to the potential development of a Nb (or Nb3Sn) superconducting photocathode with relatively high QE and minimal RF impedance to be used in a superconducting radiofrequency (SRF) photoinjector.


I. Introduction
Future free-electron-laser (FEL)-based light sources will require low emittance, high brightness and high average-current electron beams, necessitating high duty cycle (> 1 MHz) or effectively CW operation [1].
Superconducting radiofrequency (SRF) photoinjectors made of pure Nb are currently a favored choice for producing such beams as they dissipate significantly less power than normal RF guns. [2] The photocathode is an integral component of the photoinjector, contributing to the surface RF impedance, and therefore ideally it should be superconducting as well. A replaceable, superconducting plug cathode would be particularly attractive for a compact SRF linac as it has a simple design [2,3]. Recent advances in Nb-based SRF cavities, including record high Q values at 15-20 MV/m via a nitrogen doping process [4], as well as the successful in-situ growth of higher TC Nb3Sn on the inside surface [5,6 ] suggests that a compact electron linac operating at 4.2K is feasible in the future. A limiting factor is that the present choices of superconducting photocathode have relatively low quantum efficiencies (QE), e.g., for Nb QE < 0.01% and for Pb QE < 0.1% at 248 nm wavelength [3].  [7,8]. Such an approach seems particularly attractive for thin films of Mg on Nb where earlier proximity effect studies have shown an induced gap essentially the same as that of Nb [9], similar to results using Al overlayers. [7] Also, Mg has one of the highest QE values of any metallic element, ~ 0.1% at 262 nm. [10] Preliminary results on such Nb/Mg hybrid structures are reported elsewhere. [11] High peak current is more easily obtained with semiconductor cathodes such as cesium telluride (Cs2Te) [12,13]. It has a QE as high as 20% and has consistently produced a QE > 1% during normal accelerator operations over a period of at least a year, providing a relatively large bunch charge per laser pulse, and has been shown to be robust in a photoinjector environment. It has been used as an electron source in SRF photoinjectors, but only as a normal-state photocathode [14]. Here we consider the use of Cs2Te for a hybrid superconducting photocathode.
Given that it is a semiconductor, the proximity effect might be weak or nonexistent. [7,8,9]

II. Experiment
Cs2Te was fabricated at the Argonne Wakefield Accelerator (AWA) photocathode growth chamber. Details of the photocathode facility and the fabrication have been described in Ref. [15]. Briefly, the polished and cleaned substrate plug Here the Mo plug has been substituted by a Nb plug as the substrate. It was first investigated if Cs2Te deposited on Nb using the standard recipe (Te thickness = 210 A and will be labelled as Te210) had the similar QE to that deposited on Mo. Figure 1 shows the profile of the QE during Cs between 5%-7% is reproducible. Fig. 2. Dependence of the quantum efficiency (QE) of Cs2Te films on Nb (blue circles) and Mo (red squares) substrates vs. estimated Cs2Te thickness. Data points correspond to the average values while error bars represent the maximum and minimum measured QE for multiple samples of any given thickness. Solid curves are the phenomenological model fits described in the text using n and k, the real and imaginary parts of the index of refraction, respectively, for the film and substrate.

III. Data Analysis
Analysis of the data shown in Fig. 2 (red and blue fit curves) is achieved by using the Spicer 3-Step model for photoemission. [16] Briefly, this model breaks down the overall process into three separate parts: i.  First, for a given thickness of Cs2Te the QE for Nb substrates is generally higher than for Mo and appears to originate in the higher transmission term 1-R for Nb seen in Fig. 3.
More importantly, Eq. 1 shows that in the limit of L << 1, the QE is proportional to
As another example, we note that in suppressing the presence of multipactor at RF windows, a thin layer of TiN is often deposited to coat onto these windows. The thickness of the TiN can be ~7-15 nm or more [25]. It is important that the TiN layer be relatively transparent to the RF to ensure that the RF is not reflected at the window.