Quantitative Photoemission Predictions of Semiconducting Photocathodes from Many-Body Ab Initio Theory

TL;DR

This work develops a parameter-free, ab initio extension of the three-step photoemission model by integrating $GW$ quasiparticle corrections and Bethe-Salpeter equation (BSE) excitonic effects atop density functional theory. It computes QE as a product of an ab initio excitation probability, derived from the macroscopic dielectric function, and a rigorously calculated emission probability that incorporates a vacuum barrier and exciton-aligned energy distributions, with Lorentzian broadening to reflect spectral features. Validation on binary and ternary alkali antimonides shows that the approach captures key spectral features and trends beyond what DFT can describe, and the incorporation of LayerOptics to account for thin-film interference yields quantitative agreement with absolute QE values across several datasets. The resulting framework provides a robust, parameter-free link between microscopic electronic correlations and macroscopic photoemission observables, offering a pathway to rational photocathode design; future refinements include scattering effects, more realistic barriers, and explicit surface/defect modeling for real-world materials.

Abstract

The development of high-performance electron sources requires theoretical frameworks that accurately link the microscopic electronic properties of cathode materials to their macroscopic photoemission observables. Here, we present a many-body extension of the three-step photoemission model for semiconducting photocathodes, directly integrating the $GW$ approximation and the solution of the Bethe-Salpeter equation on top of density functional theory (DFT). This approach overcomes the intrinsic limitations of standard DFT by explicitly accounting for quasiparticle and excitonic effects in the photoexcitation process. The quantum efficiency (QE) is evaluated by combining the ab initio absorption with an emission probability derived as an exciton-weighted average. We validate this model on representative alkali antimonides and demonstrate that a qualitative many-body description successfully captures complex spectral features that empirical models fail to reproduce. Furthermore, by incorporating macroscopic optical effects such as thin-film interference and polarization via Fresnel post-processing, we achieve quantitative agreement with experimental QE values without any adjustment. Minor discrepancies near the photoemission threshold are attributed to the idealized surface barrier adopted in the model and impurity effects in the samples, highlighting specific directions for future refinements. This work establishes a robust, parameter-free ab initio tool that bridges microscopic electronic correlation with macroscopic observables, providing a critical pathway for the rational design of next-generation electron sources.

Figures (5)

• Figure 1: Schematic overview of the three-step model for photoemission implemented from ab initio many-body theory. The colored blocks illustrate each computational step, including DFT calculations of the ground-state properties (grey), MBPT runs ($G_0W_0$+BSE) to access excited state properties (blue), and final post-processing to compute the photoemission yield (green).
• Figure 2: Results for K$_3$Sb: (a) Optical absorption averaged over components schi-cocc25prm, (b) probability of emission depending on the energy of the excited electron, with onset from experiment spic58pr, (c) exemplary energy distribution of an excited electron at 3.6 eV, and (d) total calculated emission probability by photon energy.
• Figure 3: Results for binary crystals composed of K/Na and Sb, matched against experimental data from Spicerspic58pr. The dotted bars indicate the QE value used for aligning the ab initio many-body prediction to the experimental data.
• Figure 4: QE of cubic Na$_2$KSb and CsK$_2$Sb crystals measured at HZB and computed from the proposed ab initio many-body photoemission model. The dotted horizontal bars indicate the QE value used to align the calculated spectrum to the experimental data.
• Figure 5: Results for cubic Cs$_3$Sb, compared against experimental data from Spicer spic58pr as well as measurements performed at HZB schm19thesis, Cornell University parz+22prl and ASU saha+22apl. The dotted bars indicate the QE value used to align the ab initio many-body prediction with the experimental data. A quantitative prediction for a 9 nm-thick thin film calculated with LayerOptics is provided without manual alignment of the QE.