Like the AlGaAs/GaAs double heterostructures described by Antypas and Edgecumbe,^[14] the universal inverted structure is applied to the t-mode GaAsP-based photocathodes. However, differing from the lattice-matched AlGaAs/GaAs heterostructure, both heterostructures like GaAsP/GaAs and AlGaAsP/GaAs are lattice mismatched systems. The misfit dislocation and stacking fault arising from the lattice mismatch between substrate and epitaxial layer would dramatically deteriorate the photoemission capability. As for GaAs_1−xP_x ternary alloy, the dependence of the band gap of GaAs_1−xP_x alloy on the P composition x is given by^[15]

In order to verify the enhancement of the blue–green response for the GaAsP-based photocathode, another type of blue-extended t-mode photocathode consisting of the AlGaAs/GaAs heterostructure was prepared as well. The inverted structures for the two t-mode photocathode samples are shown in Fig. 1, where the different feature for the GaAsP sample is that a GaAsP graded-layer with a variable P composition of 0 ∼ 0.1 is sandwiched between the GaAsP emission-layer and the AlGaAs stop-layer. In the growing of the lattice mismatched AlGaAs/GaAsP heteroepitaxial layer, the graded buffer layers introduced can smooth the transition between them and improve the crystalline quality. The two t-mode cathode samples were both grown on the same 2-inch (1 inch = 2.54 cm) diameter semi-insulation GaAs (100) substrate by using a solid source Riber C21T system equipped with a valved arsenic and phosphorus cracker cell. In the growth process, the epitaxial layers consisting of a stop-layer, an emission-layer, a window-layer, and a cap-layer, were grown according to priority, where the stop-layer and the cap-layer served as an etch-resistant stopping layer and an inside-oxidation-preventing layer, respectively. For the AlGaAs/GaAs cathode material, the growth temperature was kept to be 580 °C for the GaAs layer and 650 °C for the AlGaAs layer. While for the AlGaAsP/GaAsP cathode material, the substrate temperature was always kept at a low temperature of 480 °C for GaAsP and AlGaAsP layers. A 0.5-μm-thick p-GaAs buffer layer was grown to trap impurities that may diffuse from the substrate. To obtain the lattice match structure, a 1-μm-thick undoped GaAs_1−xP_x graded-layer was grown on the AlGaAs stop-layer. The P composition of the GaAs_1−xP_x graded-layer was adjusted by P₂ flux from zero to the desired flux value. The Ga flux and As flux were kept to be a constant value during the growth. All the samples were cooled down quickly after stopping growth. The beryllium element was incorporated as the p-type dopant for MBE growth and the doping concentration was around 5× 10¹⁸ cm⁻³. During the entire growth procedure, the growth rates for all epilayers were about 1 μm/h, and the crystalline quality of the epilayers was monitored during growth via reflection high-energy electron diffraction in situ.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. Structural diagrams of t-mode (a) GaAsP and (b) GaAs cathode samples.

After the growth, the surface photovoltage spectroscopy (SPS) data for the cleaned epitaxial materials were measured to predict the cathode emission performance, since there are many similarities between SPS and spectral response (SR), which are both related to the inherent material performance factors.^[18] Following that, as the proposed process of fabricating t-mode GaAs photocathodes by Antypas and Edgecumbe,^[14] the cap-layer was first removed with a properly selective etch to expose the window-layer to coat the Si₃N₄ antireflection film on it. Then the glass was thermocompression bonded to the full wafer structure to support the fragile epilayers. After that, the substrate, stop-layer and graded-layer were selectively etched away to expose the emission-layer to perform the activation. Finally, the AlGaAsP/GaAsP and AlGaAs/GaAs samples were made into the 18-mm-diameter cathode module with a glass/Si₃N₄/window-layer/emisison-layer structure. Based on the double optical path method, the spectrophotometer typed UV-3600 was adopted to measure the reflectivity and transmissivity curves of the module samples in a range of 400 nm–1000 nm. When measured, photons entered into the modules in the near normal incidence on the glass layer.

Following chemical etching and heat cleaning, the NEA activation for the cathode modules was performed in an ultrahigh vacuum chamber with a base pressure of ≤ 10⁻⁹ Pa by using the co-deposition activation technique,^[19] in which the cesium source was kept constant and the oxygen source was introduced periodically. For both the samples, the Cs/O flux ratio was adjusted to be the same value. After the usual “high-low temperature” two-step activation, the cathode module was transferred from the activation chamber to the sealed vacuum chamber and indium sealed into a vacuum image tube, in which the activated cathode module was equipped in association with the microchannel plate and fluorescence screen. Finally, the spectral response of the sealed image tube after being transferred into ambient air was measured by the spectral response measurement system.^[20]

The experimental curves of optical properties are given in Fig. 2 for the GaAsP- and GaAs-based cathode module samples, and the absorptivity curves are easily gained according to the principle of energy conservation. As shown in Fig. 2(a), in a range of 400 nm–555 nm, the reflectivity of the GaAsP sample is evidently lower than that of the GaAs sample, while the transmissivities for both samples are equal to zero. In a region of 600 nm–950 nm, the reflectivity and transmissivity of the GaAsP sample become larger than those of the GaAs sample. On account of the multi-beam interference principle in the four-layer module, a group of alternative extreme points emerge simultaneously in the reflectivity and transmissivity curves. The maximum point in the reflectivity curve corresponds to the minimum one in the transmissivity curve when taking the energy conservation into consideration. Compared with the GaAs sample, the GaAsP sample produces a big effect on the oscillation frequency of reflectivity. Especially in a waveband of 640 nm–1000 nm, the number of peak-valley cycles in the reflectivity curve for the GaAsP sample greatly exceeds that for the GaAs sample. As shown in Fig. 2(b), in the near-infrared (NIR) band, the GaAsP photocathode gradually becomes optically transparent as the absorptivity decreases to nearly zero. In addition, it can be clearly seen from Fig. 2(b) that the absorptivity of the GaAsP sample is larger than that of the GaAs sample over the blue–green waveband. In the region larger than 620 nm, the absorptivity of the GaAsP sample decreases severely and exhibits a smaller longwave cut-off in contrast with the GaAs sample. Accordingly, it is concluded that the t-mode GaAsP and GaAs samples each manifest better absorptions of blue–green light and red-NIR light, respectively.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Experimental optical property curves of t-mode GaAsP and GaAs cathode samples.

Using the monochromator combined with a tungsten halogen lamp, the measured normalized SPS and SR curves for the two samples are shown in Fig. 3. According to the results shown in Fig. 3, we can find that there are similarities between SPS curves and SR curves, wherein SPS and SR curves possess the same variation trends. The peak wavelengths of SPS and SR curves are in the region of the blue–green spectrum for the GaAsP sample, and in the region of the NIR spectrum for the GaAs sample. It is noted that there exists a difference in peak wavelength between SPS and SR curves. For the GaAsP sample, the peak wavelength of SR is longer than that of SPS while the case for the GaAs sample is the opposite. Since SPS represents the change of potential distribution at the interface of the material due to illumination, the SPS signal of the t-mode multilayer sample is considered to be induced in the surface space charge region (SCR) and at several buried interfaces via light absorption. The reason for the difference is that the SPS signal of the t-mode cathode sample is an algebraic sum of the signals accumulated in the surface SCR and the five interface SCRs,^[18] while the SR signal only comes from the photoelectrons in the bulk emission-layer via light absorption. By comparing the SPS curve with the SR curve, it is inferred that for the t-mode GaAsP photocathode, the high SPS in the region of the blue–green spectrum can predict the better SR in the same spectrum region to some extent.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. Normalized SPS and SR curves of t-mode GaAsP and GaAs cathode samples.

The experimental quantum yield curves for the t-mode GaAsP- and GaAs-based cathode samples sealed in the vacuum image tubes are shown in Fig. 4, which are obtained through converting the spectral response curves.^[20] The measured results show that the t-mode GaAsP photocathode can obtain higher quantum efficiency in a range of 400 nm–720 nm, and the longwave cut-off wavelength is at 850 nm. For the t-mode GaAs photocathode, the quantum efficiency in the longwave region is more than 720 nm, and the longwave cut-off wavelength can extend to 920 nm. Although the thin window and high Al composition are applied to the t-mode GaAs photocathode to enhance the shortwave response,^[21] the blue–green response of interest is still higher for the t-mode GaAsP photocathode. The GaAsP sample possesses a quantum efficiency peak of 62% at 560 nm while the GaAs sample possesses a quantum efficiency peak of 42% at 690 nm. Besides, the quantum efficiencies at 532 nm reach 59% and 30% for GaAsP and GaAs samples, respectively.

The quantum efficiency curves of foreign t-mode GaAsP photocathodes from Hamamatsu and Intevac are shown in Fig. 5,^[1,22] and the spectral characteristics of the t-mode GaAsP photocathodes are shown in Table 1. From Fig. 5 and Table 1, we can see that the comparison of the quantum efficiency curve between the t-mode GaAsP photocathodes from Hamamatsu and Intevac, and the t-mode GaAsP photocathode reported in the present work exhibits that the former has a quantum efficiency peak 62% higher than the latter one at 560 nm. Owing to the high P composition exceeding 0.35, the peak wavelength positions of the t-mode GaAsP photocathodes from Hamamatsu and Intevac are less than the GaAsP photocathode reported in this work. Meanwhile, the t-mode GaAsP photocathode reported in this work, with the lower P composition, has a larger longwave cut-off wavelength. Whereas, the quantum efficiency at 532 nm of our t-mode GaAsP photocathode is higher than that of the ones from Hamamatsu and Intevac, and can reach as high as 59%, which is beneficial to the underwater range-gated imaging based on 532 nm-laser illumination.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. Experimental quantum efficiency curves of t-mode GaAsP and GaAs cathode samples.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. Quantum efficiency curves of t-mode GaAsP photocathodes from Hamamatsu and Intevac.

Table 1.

Table 1.

Table 1. Spectral characteristics of this work and t-mode GaAsP photocathodes from Hamamatsu and Intevac. . Type Peak wavelength/nm Quantum efficiency Peak/% Quantum efficiency at 532 nm/% This work 560 62 59 Hamamatsu 520 52 51 Intevac 510 48 47

Table 1. Spectral characteristics of this work and t-mode GaAsP photocathodes from Hamamatsu and Intevac. .

In order to evaluate the important intrinsic cathode performance parameters such as electron diffusion length, back interface recombination velocity and surface electron escape probability of the two different t-mode photocathodes, we compare the measured quantum efficiency curves with the theoretical curves respectively according to the t-mode uniform-doping quantum efficiency equation, which is given by^[23]

When simulating the theoretical quantum yield curves, about the absorption coefficient values varying with wavelength in the layers like GaAs, Al_0.7Ga_0.3As and GaAs_0.9P_0.1 we referred to Refs. [24] and [25]. Nevertheless, in the absence of any reference about the absorption coefficient of the Al_0.7Ga_0.3As_0.9P_0.1 window-layer in the t-mode GaAsP photocathode, about the absorption coefficient of Al_0.7Ga_0.3As_0.9P_0.1 alloy with the low P composition we just referred to that of Al_0.7Ga_0.3As alloy. By using Eq. (3), we can simulate the theoretical quantum efficiency curves, and the simulated curves are consistent with experimental curves as shown in Fig. 6, and the three cathode performance parameters obtained by a best fit of the experimental quantum efficiency curves are listed in Table 2.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. Experimental and theoretical quantum efficiency curves of t-mode GaAsP and GaAs cathode samples.

From Fig. 6, it is seen that the inconsistency between experimental and theoretical quantum efficiency in the shortwave region for the GaAsP sample is ascribed to the improper absorption coefficient of the window-layer. According to Table 2, we find that the electron diffusion length and back interface recombination velocity of the GaAsP sample are nearly the same as those of the GaAs sample, which indicates that the crystalline quality of the grown cathode material, whether GaAsP- or GaAs-based sample, can meet the requirements. From Table 2, we can also find that the surface electron escape probability of the GaAsP sample is larger than that of the GaAs sample, which is similar to the experimental results observed by Escher and Antypas.^[16]

Table 2.

Table 2.

Table 2. Fitted performance parameters of t-mode GaAsP and GaAs cathode samples. . Type Electron diffusion length/μm Interface recombination velocity/cm·s−1 Surface electron escape probability GaAsP 3.0 105 0.62 GaAs 3.2 105 0.52

Table 2. Fitted performance parameters of t-mode GaAsP and GaAs cathode samples. .

With an attempt to clarify the difference in surface electron escape probability between GaAsP and GaAs photocathodes, we should carry on an investigation into the inner band structure and surface barrier, which is shown in Fig. 7. The surface barrier profile based on the double dipole model comprises two approximately straight lines with different slopes, namely barriers I and II,^[26] which are related to the Cs–O activation layer. As shown in Fig. 7, the high-energy photoelectrons excited by the shortwave photons near the back interface would relax into the Γ conduction-band minimum and preferentially traverse barrier II, while the low-energy photoelectrons excited by the longwave photons near the surface would consequently traverse barrier I without experiencing a thorough relaxation. As is well known, the profiles of barriers I and II would play a crucial role in surface electron escape.^[20] On the one hand, for the small-bandgap GaAs sample, the longwave photoelectrons would become easier to escape to a vacuum because of the narrower barrier I. Besides, the larger red-NIR absorption in the thicker GaAs photocathode can also contribute to the number of longwave photoelectrons. On the other hand, for the large-bandgap GaAsP sample with a small electron affinity, the shortwave photoelectrons would escape to a vacuum more easily because of the narrower barrier II, while part of shortwave photoelectrons would possibly penetrate barrier I for the GaAs sample. Compared with barrier II, barrier I is more difficult to pass through for photoelectrons. Therefore, there are higher quantum efficiencies respectively in the blue–green waveband for the GaAsP photocathode and in the red-NIR waveband for the GaAs photocathode.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. Schematic band structures and surface barriers of the GaAsP and GaAs photocathode. EV is the valence band maximum, EF is the Fermi level, EC1 and EC2 are the conduction-band minima of GaAsP and GaAs respectively, and E01 and E02 are the vacuum levels of GaAsP and GaAs respectively.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. Theoretical changes of (a) peak wavelength and quantum efficiency at 532 nm with emission-layer thickness.

In order to obtain the optimum thickness of the GaAs_0.9P_0.1 emission-layer for better blue–green response, we investigate the effects of emission-layer thickness on peak wavelength and quantum efficiency at 532 nm, respectively, and the simulation results are shown in Fig. 8. From Fig. 8(a), we can see that when the emission-layer thickness is within 0.6 μm, the peak wavelength can be in the blue–green waveband. Furthermore, it can be found from Fig. 8(b) that the quantum efficiency at 532 nm can reach its maximum when the emission-layer thickness ranges from 0.4 μm to 0.6 μmm. To sum up, the optimum thickness of the emission-layer is in a range of 0.4 μm–0.6 μm which would result in a better blue–green response, especially for the underwater range-gated imaging based on 532-nm-laser illumination.