1. Introduction

Negative electron affinity (NEA) photocathodes, in which the photo-excited electrons are transported to the surface and emitted to vacuum, have been successfully used in areas such as outer space-based detection [1], high-resolution low-light-level spectral imaging [2–6], large-scale electron accelerators [7–11], low-energy electron microscopes [12–13], and electron beam lithography [14]. However, there has been an ever increasing demand for robust photocathodes with high quantum efficiency (QE), faster photoelectric response, longer operating lifetime, wider spectral response range and higher spin polarization [15–16], in particular when it comes to applications requiring high average current and/or high bunch charge typical of some large scale electron accelerators and accelerator-based light sources. Photo-electron emission from flat film-type structures, such as graded doping/band-gap layers [17–18] and distributed Bragg reflectors [19], have been explored to increase the QE of the gallium arsenide (GaAs) based photocathodes. However, the QE of the flat film-type device is always limited by the mismatch between the minority-carrier diffusion length and the optical absorption depth. In addition, a large portion of the incident light (>35%) is reflected from the surface of the flat film-like surface due to the high index of refraction of GaAs, which not only decreases QE but also generates unintended and unwanted photoemission in electron guns [15].

High-index of refraction semiconductors with sub-wavelength nanostructures have attracted increased attention due to their ability to support Mie-type geometrical resonances, and for their compatibility with standard semiconductor device fabrication techniques [20–22]. In the visible and near infrared spectral range, Si nanoparticles have been considered ideal Mie-type resonators and have attracted significant interests [23–26]. However, little attention has been paid to GaAs resonators and so far, no results have been reported relevant to photocathode applications. Compared with indirect-bandgap Si, the direct-bandgap GaAs has a much larger light absorption coefficient, which leads to strong light absorption in resonators. In the visible waveband, every photon absorbed by GaAs can excite an electron-hole pair and contribute to the QE of optoelectronic devices.

To exploit Mie-type resonance for a photocathode application, the light must be efficiently absorbed and concentrated within the fabricated nano-structure [27–28]. Due to the much smaller size of the nano-resonators compared with traditional film structures, the process of light absorption can be efficiently decoupled from that of carrier collection, and improved optoelectronic efficiency has been reported in GaAs nano-resonator devices [29–33]. For typical diode-type devices such as solar cells, photodetectors, etc., the p-n junctions and electrodes serve indispensable functions, and it is difficult to realize these complex device functions inside the nano-scale resonators [30,33]. However, because the only functional part of the NEA photocathode is the p-type GaAs [34], it is much simpler to fabricate nano-scale photocathodes compared to diodes. These features make GaAs an excellent candidate for designing NEA photocathodes with nanostructured resonator surfaces.

In this work, a new type of nano-structured NEA photocathode is presented, in which the p-type GaAs Nanopillar-array (NPA) Mie-type resonator serves to enhance the photoelectron emission of the photocathode. The photoelectron emission processes of the NPA photocathodes were analyzed based on Spicer’s three-step model [35–36], which consists of photoelectron generation, electron transport to the surface, and escape across the surface into vacuum. The photoelectron generation and transport efficiencies were modeled using Lumerical Finite Difference Time Domain (FDTD) [37] and Cogenda Technology Computer-Aided Design (TCAD) [38] tools, respectively. The surface-electron escape probability was obtained by fitting the theoretical model to measured QE of typical GaAs photocathodes [36]. Based on these methods, we show excellent light management performance of the nanophotonic resonators, which provides exceptionally low surface reflectance and significant QE enhancement across the entire 500 ∼ 850 nm waveband. Furthermore, the photoelectric response time for NPA photocathodes is expected to be ultrafast due to the much shorter photoelectron transport distance in the nanopillars compared to flat wafers. We conclude that GaAs NPA resonant structures can be designed and optimized to significantly improve photoemission performance of NEA photocathodes, which could benefit applications such as high-resolution night-vision imaging and large-scale electron accelerators.

2. Device structure and theoretical model

Figure 1(a) shows one unit-cell of the GaAs NPA. Unit cells would be repeated many times, with nano-pillars set at equal-distance and arranged on a p-type GaAs substrate. The chemicals cesium and fluorine cover the entire surface – pillars and substrate – and are used to activate the surface to the NEA state [34]. The relevant parameters that describe the geometry of the NPA are the period of the square lattice P, pillar diameter D, and pillar height H. By placing periodic boundary conditions, simulations were carried out within one unit-cell to model the properties of the periodic nanopillars on a square lattice.

 

Fig. 1. Device structure and theoretical model of GaAs NPA photocathode. (a) Depiction of one unit cell of the GaAs NPA photocathode. (b) Photoelectron emission processes and electronic band structure of NEA photocathode based on Spicer’s model. (c) Cross-section of the FDTD setup for simulating the optical properties of GaAs NPA. (d) TCAD setup for analyzing the photoemission properties of GaAs NPA photocathode.

Download Full Size | PDF

Spicer’s three-step model [35–36], illustrated in Fig. 1(b), was used to assess photocathode performance. The first step is photoexcitation (process A in Fig. 1(b)), described as the probability of absorbing an incident photon which in turn generates an electron-hole pair (P_g). This process was modeled through FDTD calculations of the interaction between light and GaAs [37]. Figure 1(c) shows the cross-section of the FDTD setup. Virtual two-dimensional light power meters in the x-y plane are positioned on top and bottom of the NPA for monitoring reflection and transmission, respectively. A cuboid three-dimensional field monitor is positioned around the nanopillar to monitor the spatial distribution of the electric field intensity (E(λ, x, y, z)) for the incident plane wave light with wavelength λ. The generation rate of the electron-hole pairs (g(λ, x, y, z)) in pairs·cm⁻³·s⁻¹ can be calculated as

The third step is the emission of electrons into vacuum (process C in Fig. 1(b)) described by the surface-electron escape probability, P_e. Due to the C_S and F activation layer on the surface, the vacuum level is lowered below the bulk conduction band minima (NEA condition). Electrons reaching the surface tunnel through a small barrier formed by the activation layer and are emitted into vacuum. This tunneling process depends on the dopant concentration and the level of contamination of the photocathode surface. In Fig. 1(b), a downward band-bending occurs at GaAs surface, which affects the energy distribution of the photoelectrons in the conduction band and subsequently the surface-escape process. Two valley [36] and three valley [40] conduction band models have been used to analyze this process, which can accurately account for the dopant concentration. However, surface band-bending is difficult to model exactly because even small amounts of surface contamination can drastically impact the NEA condition. In this work, P_e was obtained by fitting the theoretical P_g and P_t to published QE spectra of flat wafer GaAs photocathode samples [34,36]. By combining these simulations, the emitted electron current, I_e(λ), from the photocathode can be calculated as

Table 1. GaAs photocathode material parameters used in simulations, taken from Refs. [39,42–44]

View Table | View all tables in this article

Photoelectron emission processes are illustrated in Figs. 2(a) and 2(b) for the GaAs NPA and flat surface (referred to as flat wafer hereafter) photocathodes, respectively. From Fig. 2, we can see the advantages of NPA over flat wafer photocathodes in terms of key photocathode performance features. Light absorption in NPA can be much larger than flat wafer because the local density of optical states (LDOS) in nanopillars can be significantly enhanced by Mie resonance. And the photoelectrons can be highly localized inside the nanopillars near the top and side surfaces where the electrons can be efficiently transported and emitted to vacuum, which helps to increase P_g and P_t, and ultimately leads to the enhancement of QE.

 

Fig. 2. Electron emission processes in GaAs photocathodes. (a) GaAs NPA photocathode, (b) GaAs flat wafer photocathode.

Download Full Size | PDF

The light scattering (σ_sca), absorption (σ_abs), and extinction cross-sections (σ_ext) can be used to analyze the Mie resonance properties which are defined as [41]

For the evenly-spaced NPA with pillars oriented vertically on the substrate and the incident light propagating along the axis of the nanopillar, f can be calculated as

Furthermore, the NPA photocathode has a larger effective electron emission area, a_NPA, compared to a flat wafer, a_flat, due to the fact that the electrons are emitted from both the top and side surfaces of nanopillars, which also permits larger QE. For the evenly-spaced NPA, the ratio of a_NPA to a_flat, defined as r_emi, provides a measure of the effective electron emission surface area enhancement and can be calculated by

It’s worth noting that, since the diameter and height of the GaAs nanopillars investigated in this paper are larger than 100 nm, the quantum confinement effect can be neglected. The optoelectronic model we employed here, combining the wave optics properties of the light-matter interaction with coupled drift-diffusion equations for the electrons and holes, has been widely used to describe the optoelectronic and carrier transportation properties of the optoelectronic devices with the nanostructures similar to this work [45–46].

3. Results and discussion

3.1 Optical design of Mie-type GaAs NPA resonators for NEA photocathodes

To achieve the desired QE enhancement, the optical structure of NPA must be optimized for maximum light absorption at a specific wavelength. Systematic studies on the Mie resonant properties of Si nanopillars have been reported, however, similar work has not been performed with GaAs. As shown in Fig. 3(a), at 400 ∼ 900 nm wavelength (λ) range, GaAs and Si have similar refractive index n, but the light extinction coefficient k of GaAs is significantly higher. Figure 3(b) shows the cross-sectional spectra calculated by FDTD for GaAs and Si nanopillar arrays on respective substrates, with pillar diameter D = 100 nm and pillar height H = 100 nm, with resonant peaks observed at λ ∼ 518 nm for both materials. It can be seen that σ_abs is more than 2 times larger than σ_phy for GaAs nanopillar at the resonance peak (e.g. C > 2), and is much larger than that of Si due to much larger k of GaAs. This means photons are more readily absorbed in GaAs within the 500 ∼ 800 nm waveband, i.e. every absorbed photon can generate an electron-hole pair and contribute to the generation of photoelectron current from NEA photocathodes. Figures 3(c) and 3(d) present the normalized electric/magnetic field intensity (|E|²/|H|²) profiles and field lines in vertical crosscuts through the center of a GaAs nanopillar around λ ∼ 518 nm, which clearly show the loops induced by the driving electric/magnetic field also drive the Mie-type magnetic/electric dipole resonance modes (MD/ED). The MD enhanced |H|² and ED enhanced |E|² are also shown in Figs. 3(e) and 3(f), respectively, which clearly indicate the enhanced LDOS in nanopillars.

 

Fig. 3. Mie resonance characteristics of GaAs and Si nanopillars on substrate. (a) Refractive index, n, and extinction coefficient, k, of GaAs and Si vs. wavelength (λ) from 400 to 900 nm, the data is adopted from Ref. [42]. (b) Cross-section (σ_scat, σ_abs, σ_ext) spectra of GaAs and Si nanopillars on respective substrates, the parameters of incident light and geometry are shown as insets. (c)–(d) Normalized electric and magnetic field intensity (|E|² and |H|², colored) and field lines (white) in vertical crosscuts through the center of the GaAs nanopillar at the resonance wavelength λ ∼ 518 nm. (e)–(f) The same crosscuts as in c and d, but showing the resonance enhanced |H|² and |E|² (colored), respectively. The magnetic and electric dipole (MD and ED) resonance modes are labeled in (e) and (f), respectively.

Download Full Size | PDF

Mie resonances depend on the NPA geometry parameters. An FDTD analysis was used to model the light absorption efficiency, η_a, and light concentration, C, of GaAs NPAs as a function of the geometric parameters and the wavelength λ of the plane-wave radiation propagating along the nanopillar axes. For lattice period P = 600 nm, Figs. 4(a) and 4(b) give the dependence of η_a spectra on pillar diameter D for pillar heights H = 350 nm and 750 nm, respectively. The NPA filling factors, f, are also shown in the figures and were used to calculate the light concentration coefficient C. Some dominant η_a spectral branches are observed in Figs. 4(a) and 4(b), and agree with Mie’s resonant theory [20], namely, the lowest-order magnetic/electric dipole (MD/ED) mode is excited when (Dn)/λ=1, whereas the quadrupole (MQ/EQ) and higher-order multipole modes are excited for larger values of (Dn)/λ which are labeled as the corresponding η_a peaks. Figures 4(c)–4(f) give the |E|²/|H|² profiles and field lines of dipole (black dot M1 in Fig. 4(a)) and quadrupole (black dot M2 in Fig. 4(a)) modes, which confirm the Mie-type resonance modes for the corresponding η_a peaks.

 

Fig. 4. Mie resonance enhanced light absorption in GaAs NPA. (a), (b) dependence of η_a (color) absorption spectra on diameter-D for GaAs NPA with H = 350 nm and 750 nm, respectively. The incident light and geometry parameters are shown as insets. The black dots M1 (D = 120 nm), M2 (D = 300 nm), M3 (D = 120 nm), M4 (D = 300 nm) correspond to wavelength λ and geometry combinations used for (c) - (j). (c)/(d) and (e)/(f) are the normalized |E|²/|H|² (color) and field lines (white) in vertical crosscuts through the center of the nanopillars M1 (MD/ED resonance at 630 nm) and M2 (MQ/EQ resonance at 620 nm), respectively. (g) - (j) are the resonance enhanced |H|² (color) for M1 (MD at 630 nm), M2 (MQ at 620 nm), M3 (MD at 650 nm) and M4 (MQ at 640 nm).

Download Full Size | PDF

By comparing Figs. 4(a) with 4(b), the same resonance modes are excited in both photocathodes at similar values of D. And for both cases, the dipole modes are excited by the smallest D, and therefore can provide the largest C. The pillar height H has the most profound impact, with stronger absorption resonance peaks obtained by increasing pillar height H. This can be understood from Figs. 4(g)–4(j), which show the MD and MQ enhanced |H|² profiles in vertical crosscuts parallel to the driving electric field through the center of GaAs nanopillars M1, M2, M3, and M4 (indicated by black dots in Figs. 4(a) and 4(b)). As can be seen, a much larger fraction of |H|² can be coupled into taller nanopillars, giving rise to increased η_a in nanopillars. A similar argument holds for the ED and EQ modes.

From Fig. 4, the dipole and quadrupole resonance-dominated η_a absorption spectra branches cover the wavelength range 500 ∼ 850 nm, consistent with photoelectron emission requirements of GaAs photocathodes. The largest C, and potentially the strongest QE enhancement, can be achieved in MD/ED resonance mode. For a practical GaAs NEA photocathode electron gun, a typical operating wavelength is 532 nm for un-polarized electron beam [7,9–11,34] and 780 nm for spin-polarized beam [8,47–48]. Figure 4 shows that GaAs NPAs with D of 100 nm and 160 nm can excite dipole resonance at ∼532 nm and ∼780 nm, respectively. Figure 5 explores the dependence of dipole resonance enhanced absorption, η_a-res, and resonance wavelength, λ_res, on P and H for these two types of GaAs NPAs.

 

Fig. 5. Dependence of the dipole resonance enhanced light absorption and resonance wavelength upon lattice period, P, and pillar height, H, of evenly-spaced GaAs NPA. (a) D = 100 nm. (b) D = 160 nm.

Download Full Size | PDF

It can be seen from Fig. 5 that λ_res barely changes with P, whereas η_a-res appears quite sensitive to P. Increasing P decreases f and light coupling in nanopillars and significantly weakens η_a-res. However, stronger light coupling can be achieved with taller nanopillars which allow larger P for the highest η_a-res and also larger light concentration C due to smaller f. With respect to NPA in Fig. 5(a) with D = 100 nm, H = 700 nm and P = 300 ∼ 400 nm, nearly 100% η_a-res is obtained at λ_res ∼ 532 nm, and the corresponding light concentration values, C, are between 11 and 20. However, for the NPA of Fig. 5(b) with D = 160 nm, a maximum η_a-res of only about 95% is obtained at λ_res ∼ 780 nm with P = 500 nm and much larger H = 1200 nm, and the corresponding C is about 12. This means much taller pillars are needed to fully concentrate and absorb the light at longer wavelengths in nanopillars due to the decreased light extinction coefficient, k, of GaAs (see in Fig. 3(a)), which increases the difficulty of NPA fabrication. However, light concentration and absorption enhancement at longer wavelengths has special meaning for certain practical electron source applications as discussed in the following section.

The analysis above serves to inform the optical design philosophy to achieve the largest QE enhancement in NPA photocathodes. Namely, for a practical NPA photocathode operating at a specified wavelength, the nanopillar diameter, D, was selected to excite the corresponding resonance mode, with the nanopillar height, H, chosen to be tall enough to fully absorb the incident light and the lattice period, P, selected to balance the light concentration and the electron emission area of the nanopillars (see in Eq. (10)).

3.2 Enhanced QE from GaAs NPA resonator photocathodes

Photoelectron emission processes in GaAs NPA photocathodes were analyzed based on the theoretical models and material parameters shown in section 2. From Eq. (6), QE was calculated as the product of P_g, P_t and P_e. The quantities P_g and P_t were simulated using FDTD and TCAD tools, respectively while P_e was obtained by fitting the theoretical P_g and P_t to published QE spectra of a GaAs “epi-ready” flat wafer photocathode over the wavelength range 500 ∼ 800 nm [34]. The results are shown in Fig. 6 which indicates QE is primarily limited by P_e (i.e., the smallest of the three quantities). But note that P_e is still approximately twice as large as the measured QE across the entire wavelength range, indicating there is considerable opportunity for improving QE via P_g and P_t, which can be done by NPA resonators.

 

Fig. 6. Fitted P_e from GaAs “epi-ready” flat wafer photocathode: QE (black line) was measured in a previous report [34], P_g (red line) and P_t (blue line) were simulated using the FDTD and TCAD tools, respectively, P_e (green line) was then obtained by fitting P_g and P_t to measured QE data.

Download Full Size | PDF

It is possible for electrons to be emitted from the nanopillar surfaces and the substrate of the GaAs NPA photocathode, although for this application, the intent is for electrons to emit only from the nanopillars. Electrons emitted from nanopillars and the substrate were simulated and analyzed separately as illustrated in Fig. 7(a). Some photoelectrons might be transported between pillar and substrate, but this contribution is considered to be small and as such, it was ignored in the model. Since no experimental QE data of nano-structured photocathode is available, the fitted P_e from Fig. 6 was used to calculate QE of the investigated GaAs NPA and flat wafer photocathodes. Based on the optical design consideration described in section 3.1 and the targeted operation at specific wavelengths, GaAs NPA photocathodes with MD/ED resonance wavelengths at ∼532 nm (D = 100 nm, H = 700 nm, P = 300 nm, labeled as N1) and ∼780 nm (D = 160 nm, H = 1200 nm, P = 500 nm, labeled as N2) were analyzed.

 

Fig. 7. (a) Cross-section illustration of electron emission from nanopillar and substrate of GaAs NPA photocathode. (b) Simulated surface reflectance spectra for GaAs NPA (labeled as N1, N2) and flat wafer (labeled as W) in the 400 ∼ 900 nm waveband.

Download Full Size | PDF

Figure 7(b) presents the simulated surface reflectance spectra for these two NPAs together with flat wafer results (labeled as W) within the 400 ∼ 900 nm waveband. The reflectance of flat wafer was greater than 30% across the entire waveband, whereas the reflectance of N1 and N2 were only ∼1% at the MD/ED resonance wavelengths of ∼532 nm and ∼780 nm, respectively. With such ultralow reflectance, the NPA photocathode can be considered an optically “dark” photocathode. Reduced reflectance is a very useful feature for DC high voltage photoguns. Reflection of the incident laser light from the photocathode surface leads to unwanted/unintended photoemission from the edge of the photocathode producing electron beam that is not properly transported from the photogun. This unintended photoemission strikes the vacuum chamber walls, degrading vacuum and hastening photocathode QE decay and leading to accelerator downtime [15]. GaAs NPA photocathodes and NPA coatings applied to the photogun vacuum chamber walls could significantly improve the operating lifetime of the photogun.

Simulated photocathode QE spectra of GaAs NPAs N1 and N2 together with flat wafer in 500 ∼ 800 nm wavelength range are presented in Figs. 8(a) and 8(b), respectively. It clearly shows that QE is enhanced by NPA resonators. Since the same P_e was used for NPA and flat wafer, this enhancement is mainly due to P_g and P_t presented in Figs. 8(c) and 8(d), respectively.

 

Fig. 8. Simulated QE and related performances of GaAs photocathodes. Simulated total QE (solid blue line) and QE from nanopillars (dashed blue line) and substrate (dotted blue line) for (a) GaAs NPA photocathodes N1 (D = 100 nm, H = 700 nm, P = 300 nm), (b) N2 (D = 160 nm, H = 1200 nm, P = 500 nm). The QE and fitted P_e of GaAs flat wafer (W) photocathode are also presented as solid red and black lines in (a) and (b) for comparison. (c-d) Simulated P_g and P_t of N1 (blue lines), N2 (red lines) and GaAs flat wafer (black line) photocathodes.

Download Full Size | PDF

The photoemission performance represented by P_g, P_t and QE for the nanopillar and substrate of NPA photocathodes in Fig. 8 make it possible to analyze the contributions from nanopillar and substrate separately. For GaAs NPA photocathodes N1 and N2, a P_t of nearly 100% nanopillars can be obtained across the entire 500 ∼ 800 nm waveband (see Fig. 8(d)), indicating excellent photoelectron transport properties of nanopillars due to their much smaller transport distances compared to flat wafers as mentioned in section 2 (see Fig. 2). However, photoelectrons in the substrate need to be transported to the surface free of nanopillars, or through a nanopillar, to be emitted into vacuum. This increases the effective transport distance and therefore reduces P_t compared to the flat wafer, as seen in Fig. 8(d). So, it is necessary to decrease the surface reflectance and fully absorb the incident light inside the nanopillars to enhance QE as much as possible. Theoretical results show P_g of nearly 99% for NPA photocathode sample N1 and 95% for sample N2, exciting peak MD/ED resonances at wavelengths 532 nm and 780 nm (see Fig. 8(c)), respectively. In addition, nearly 100% P_t is obtained from pillars at these two wavelengths (see Fig. 8(d)), which leads to QE enhancement of factors 1.8 for N1 and 2.2 for N2 compared to the flat wafer photocathode (see Fig. 8(a)).

These data indicate larger resonance QE enhancement is achieved for N2 even with a smaller P_g at λ ∼ 780 nm compared to sample N1 at ∼ 532 nm. This can be explained by excellent electron transport properties of nanopillars. Similar to other optoelectronic devices such as photovoltaic solar cells [27–33], photoelectrons in photocathodes are transported to the surface through the diffusion process. The electron diffusion length, L_e, which is ∼ 1500 nm for GaAs with p-type doping concentration of 1×10¹⁹ cm⁻³ as shown in Table 1, can be used to evaluate the transport properties of the photoelectrons. Based on the optical constants in Fig. 3(a), the GaAs thickness needed to absorb ∼95% of the incident light (referred to hereafter as the absorption depth, D_a) was calculated to be ∼ 450 nm at 532 nm, and ∼ 2500 nm at 780 nm. When the absorption depth is much longer than the diffusion length, as is the case for 780 nm light, the photoelectrons are not transported as efficiently to the surface due to carrier recombination, which causes decreased P_t for the GaAs flat wafer photocathode as observed in Fig. 8(d). But this is not such a problem for NPAs, where light can be concentrated near the surface. A pillar height of 1200 nm is needed for N2 to absorb ∼95% of the incident light (λ ∼ 780 nm) and the P_t of nanopillar does not decrease with wavelength due to the fact that the photoelectrons can be emitted from the pillar side surfaces. These different wavelength dependent properties of P_t for nanopillars and flat wafers lead to the intensified enhancement of P_t and QE by NPAs as wavelength becomes longer as observed in Fig. 8. As mentioned in section 3.1, stronger enhancement on photocathode QE at long wavelengths bears special implication for certain practical electron source applications. Namely, the highest electron spin polarization can be obtained at λ ∼ 780 nm for GaAs photocathode [47–48]. In addition, the mean transverse energy of the electron emission can be effectively suppressed at long wavelengths near band edge (λ ∼ 850 nm), which may lead to lower thermal emittance of the electron beams [9,16,40].

To better understand the impact of enhanced light concentration, we introduce QE_nor, which stands for the nanopillar QE normalized by the projected physical area, as reported in Refs. [29–33]. Specifically, QE_nor was calculated by considering only the incident photons projected onto the physical area of nanopillars. Figure 9(a) presents the calculated QE_nor spectra, it can be seen that the QE_nor of two nanopillars (N₁ and N₂) extends to 420% and 340% at the MD/ED resonance wavelengths of 532 nm and 780 nm, respectively, which further confirms the strong light concentration ability of NPA resonators. Note here, QE_nor can exceed 100% because the actual absorbed light power is greater than the projected light power within the nanopillars, which should not be confused with the classic definition of QE that is the ratio of number of electrons to the number of absorbed photons. The impact of the MD/ED resonance enhanced light concentration upon the distribution of the photoelectron generation rate in NPAs is shown in Fig. 9(b), clearly the photo generation rates are concentrated near the sides of the nanopillars for both N1 and N2. This means most of the photoelectrons can be transported efficiently to side surfaces of the nanopillars, which not only improves P_t and enhances QE as discussed before, but also decreases the electron transport time due to the shortened transport distance, which can be potentially very useful in applications such as ultrafast photoemission.

 

Fig. 9. Impact of resonance enhanced light concentration on the photoelectric properties of the NPA photocathodes. (a) QE_no (see definition in the text) for NPA photocathodes N1 and N2. (b) Distribution of photoelectron generation rate in pairs.cm⁻³s⁻¹ within the vertical cross section of the samples N1 and N2 in the plane parallel to the E-field of the linearly polarized incident light at the resonance wavelengths. Light intensity is 0.1Wcm⁻².

Download Full Size | PDF

As a final assessment of the NPA photocathode described above, the performance at the resonance wavelengths are summarized in Table 2. We can see P_g and P_t of the nanopillars can be enhanced to nearly 100%, resulting in the QE enhancement by a factor of 1.8 for NPA sample N1 at the resonance wavelength of 532 nm, and by a factor of 2.2 at 780 nm for NPA sample N2, compared to the flat wafer photocathodes. Some accelerator applications benefit from electron beams with small thermal emittance. For these applications, the GaAs NPA provides a way to operate at the bandgap, to obtain small emittance, with a high QE that is typically only obtained by operating at shorter wavelengths. It is also worth considering that P_e can also be improved by increased thermal electron emissions [36,40,49] at shorter wavelengths due to the increased electron emission area of the NPA, which represents future study using experimental data to be obtained from practical NPA photocathodes. In this work, the r_emi, which provides a measure of the effective electron emission area enhancement of NPA, was calculated by Eq. (11) and listed in Table 2 for references.

Table 2. Photoemission performance parameters of GaAs NPA and flat wafer photocathodes at the resonance wavelengths (λ_r)

View Table | View all tables in this article

4. Conclusions

We presented studies on a new type of NEA photocathode with enhanced photoemission performances by patterning the photocathode surface with GaAs NPA Mie-type resonant structures. Large resonance-enhanced absorption can be achieved in GaAs nano-pillars due to the large absorption coefficient of direct-bandgap GaAs, which allows strong localization of the photoelectrons near the top and sides of the nanopillars surfaces where electrons can be efficiently transported and emitted into vacuum, resulting in significant photocathode QE enhancement. Mie-type resonance modeling of GaAs NPAs shows that resonance conditions can be obtained over the entire 500 ∼ 850 nm waveband by adjusting pillar diameter, height and spacing, thereby satisfying the requirements of most photocathode applications. This paper described NPA designs that provide strong dipole resonances at wavelengths 532 nm and 780 nm, to support un-polarized and polarized photoemission applications. The spectrally-resolved photoemission predictions based on Spicer’s three-step model show that these structures, when properly optimized, provide QE limited only by the surface-electron escape probability and significantly outperform traditional flat wafer photocathodes.

Besides the QE enhancement, the ultralow surface reflectance (∼1%) of these NPA photocathodes clearly indicates the use of NPA can be a very effective measure for suppressing unwanted photoemission to improve the operating lifetime of DC high voltage photoguns. An ultrashort photoelectric response can be also expected due to the much shorter photoelectron transportation distance in nanopillar than in flat wafer. The only functional part of the semiconductor NPA photocathode is the p-doped GaAs NPA, which can be directly fabricated on substrates using inexpensive industrial fabrication processes such as nanoimprint lithography. We believe these nanostructured GaAs resonators are very promising materials that could significantly improve the photoemission efficiency and photoelectric response of NEA photocathodes to meet the stringent requirements by applications such as high-resolution spectral low-light-level imaging and large scale electron accelerator applications. Due to the limitation of the present simulation method, some practical issues related to the cathode activation such as surface roughness and shading effect from Cs deposition, lifetime due to the backward ion bombardment can’t be addressed in this paper, but will be interesting subjects to pursue in future research, in particular experimental studies.