Energy Bandgap Engineering of Transmission-Mode AlGaAs/GaAs Photocathode Energy Bandgap Engineering of Transmission-Mode AlGaAs/GaAs Photocathode

Aiming to enhance the photoemission capability in the waveband region of interest, a graded bandgap structure was applied to the conventional transmission-mode AlGaAs/ GaAs photocathodes based on energy bandgap engineering, wherein the composition in Al x Ga 1 (cid:1) x As window layer and the doping concentration in GaAs active layer were grad- ual. According to Spicer ’ s three-step model, a photoemission theoretical model applicable to the novel transmission-mode Al x Ga 1 (cid:1) x As/GaAs photocathodes was deduced so as to guide the cathode structural design. Then the cathode material was grown by the metalorganic chemical vapor deposition technique, and the epitaxial cathode material quality was evaluated by the means of scanning electron microscope, electrochemical capacitance-voltage, X-ray diffraction and spectrophotometry. Through a series of specific processes, the cathode material was made into the multilayered module, possessing a glass/Si 3 N 4 /Al x Ga 1 (cid:1) x As/GaAs structure. After the surface treatment including heat cleaning and Cs ▬ O activation for the cathode module, the image intensifier tube com-prising the activated cathode module, microchannel plate, and phosphor screen was fabricated by indium sealing. The spectral response test results confirm the validity of the novel structure for the enhancement of blue-green photoresponse.


Introduction
Since negative-electron-affinity (NEA) GaAs photocathode was proposed as a type of excellent photoemitter by Scheer and Laar [1], GaAs-based photocathodes have found widespread applications in photodetectors, accelerators, electron microscopes, photon-enhanced thermionic emission devices, and other fields [2][3][4][5]. In view of the high visible spectral response, good spectral extensibility to the near infrared (NIR) region and low dark current, NEA GaAs, GaAsP, and InGaAs photocathodes are important components in the vacuum photodetectors, for example, low-light-level (LLL) image intensifiers, photomultiplier tubes, and streak tubes [6]. In the modern light sources based on free electron lasers or energy recovery linacs, GaAsbased photocathodes serve as high brightness electron sources with the unique virtues of large current density driven by visible lasers, high spin polarization, low thermal emittance, and narrow energy distribution [7]. In recent years, a spin-polarized transmission electron microscope combining electron microscopy and accelerator technology using GaAs-GaAsP strained superlattice photocathodes was developed to observe dynamically a magnetic field images with high spatial and temporal resolutions [8]. Moreover, with the aid of the ultrahigh speed pulse laser, GaAs photocathodes can satisfy the requirements of fast response speed and large emission current density aiming to THz frequency vacuum devices [9].
As is well known, GaAs photocathodes can operate in the transmission-mode (t-mode) and the reflection-mode (r-mode), respectively, depending on the difference in the direction of the incident light [10,11], as shown in Figure 1. For the t-mode operation, the incident light is irradiated on the substrate surface, and the photoelectrons are extracted from the opposite surface side, whereas for the r-mode operation, the incident light and photoelectrons are located on the same emission surface side. Due to the difference in absorption length of longwave and shortwave photons, the shapes of spectral response curves for GaAs photocathodes working in the two modes are different [11]. Differing from r-mode GaAs photocathodes, t-mode ones are difficult to achieve high spectral response in a broadband region from ultraviolet to NIR spectrum. Usually in the practical applications, the researches on tmode photocathodes are more concerned. For example, the image intensifiers and related imaging systems, t-mode photocathodes conform to the optical imaging structure [12]. Besides, as polarized electron sources in photoinjector apparatus, t-mode photocathodes are more popular than r-mode ones, because the laser spot size can be reduced through the short focus lens placed on the photocathode backside, which would not hinder the path of the electron beam and is more conducive to achieve a super-high-brightness electron beam [13,14].
As proposed by Spicer and Herreragomez [15], the photoemission process from photocathodes consists of electron excitation by incident light absorption, electron transport toward surface, and electron escape across the surface barrier into vacuum. For t-mode photocathodes, some important cathode parameters such as electron diffusion length, interface recombination velocity, and surface escape probability are crucial to the photoemission performance, especially the shortwave photosensitivity [16]. Enhancing the blue-green response of t-mode GaAs photocathodes as far as possible, would not only be beneficial to the detection in sandy or desert terrain for image intensifiers [17], but also increase the current density driven by 532 nm laser for electron sources [18]. Although the external electric field biased across the photocathodes can improve the photoemission capability, the limitations of this approach are the difficulty in making thin electrode pattern and the increased dark current with the strong field [19,20]. In view of this adverse case, internal built-in electric fields through energy band engineering design could be an alternative approach. In our research, a complex structure composed of the composition-graded structure and the doping-graded structure is proposed to prepare high efficient t-mode AlGaAs/GaAs photocathodes. Furthermore, the photoemission model, cathode structure design, cathode material epitaxy, and vacuum tube fabrication are investigated through the integrated analysis of theory and experiments. Finally, the effectivity of the designed novel structure is verified by comparison with the common photocathodes.

Graded bandgap structure
For the t-mode GaAs photocathode, the AlGaAs and GaAs materials are usually used as the window layer and the active layer, which determine the shortwave cutoff and longwave cutoff, respectively. A built-in electric field in the interior of the photocathode material can be realized by the variation of dopant or composition according to energy band engineering design [21,22]. Based on this concept, a novel structure is proposed to improve photoelectron emission capability, wherein a composition-graded structure and a doping-graded structure are employed to the Al x Ga 1Àx As window layer and GaAs active layer, respectively [23,24], as shown in Figure 2.
To form a built-in constant electrical field in the GaAs active layer of the photocathode, the p-type dopant concentration can follow the exponential variation, and the doping formula is expressed by [22] where A denotes the exponential-doping (e-doping) factor, N 0 is the doping concentration at the surface of GaAs active layer, y is the distance from the coordinate origin (i.e., the surface of GaAs active layer), and N(y) is the p-type doping concentration in the GaAs active layer. As a result of the variation of dopant concentration, the initial Fermi level is different. In thermal equilibrium, the Fermi level at different positions in the active layer is unified, and the electric potential energy qV(y) is varied as follows: where q is the electron charge, T is the cathode temperature, and k B is the Boltzmann constant. The diagram of band structure with the downward shape in the GaAs active layer is shown in Figure 2, wherein the built-in electric field E 0 in a certain thick (T e ) active layer is given by In the Al x Ga 1Àx As window layer, the bandgap is decreased from the substrate interface to the GaAs interface due to the composition-graded structure. Because of the high p-type doping concentration, the valence bands of the Al x Ga 1Àx As/GaAs heterojunction are aligned, as shown in Figure 2. The graded Al composition in the window layer results in a built-in electric field E 1 , which is treated to be uniform as follows [25]: where ΔE g is the energy bandgap difference of Al x Ga 1Àx As material, and Δd is the overall thickness of Al x Ga 1Àx As window layer. Figure 2 illustrates the transport process of photoexcited electrons in the t-mode graded bandgap Al x Ga 1Àx As/GaAs photocathodes. As for the t-mode photocathodes, the photoelectrons generated by shortwave light excitation are Figure 2. Energy band structure diagram of the t-mode Al x Ga 1Àx As/GaAs photocathode with the graded bandgap structure. E g is the bandgap, E 0 is the vacuum level, E F is the Fermi level, E C is the conduction band minimum, and E V is the valence band maximum.
distributed in the Al x Ga 1Àx As window layer. Under the first-stage built-in electric field, the thermalized photoelectrons in the Al x Ga 1Àx As layer move toward the GaAs interface. After that, the second-stage built-in electric field in the active layer can promote these photoelectrons toward the emission surface. On the other hand, the GaAs active layer can absorb the longwave light, and the excitated photoelectrons are promoted to move toward the surface with the help of the built-in electric field in the active layer. Consequently, there are reasons to believe that, by virtue of this unique graded bandgap structure, the quantum efficiency over the broadband spectrum, especially in the shortwave response region would be enhanced to some extent.

Photoemission model derivation
As is well known, the one-dimensional continuity equation can afford a useful avenue to establish the photoemission model of t-mode or r-mode III-V group photocathodes, which takes account of the spatial photon adsorption, spatial carrier distribution, and interface electron recombination [10,11]. As shown in Figure 2, the photoelectrons generated in the Al x Ga 1Àx As layer are able to move into the GaAs layer and contribute to the total emitted electrons. For the composition-graded Al x Ga 1Àx As layer, some physical properties, for example, electron mobility (μ), electron diffusion coefficient (D n ), and electron recombination lifetime (τ) are the functions of the Al composition x, which are expressed as follows [26,27]: Because of the aforesaid variable physical properties regarding to Al composition, the continuity equation of electron transport in the Al x Ga 1Àx As window layer is quite complex. For simplicity, the Al x Ga 1Àx As layer is treated to be of a series of sublayers with different Al compositions. As shown in Figure 2, the Al x Ga 1Àx As window layer can be considered to be of n sublayers, wherein T wn denotes the thickness of nth sublayer, and T dn denotes the coordinate point along the y-axis. In this case, the transport of photoelectrons in the Al x Ga 1Àx As window layer follows the one-dimensional continuity equation through diffusion and drift under the built-in electric field, which is as follows: where g i (y) represents the photoelectron generation function in each Al x Ga 1Àx As sublayer and is expressed as [28,29]: In Eqs. (8) and (9), i represents the Al x Ga 1Àx As sublayer along the y axis direction, n i (y) and α hvi denote the excess electron concentration and the absorption coefficient in each part of Al x Ga 1Àx As sublayer, I 0 is the incident light intensity, R hv is the reflectivity at the light incident surface, and T e is the active layer thickness. Besides, the three physical properties, that is, D ni , μ i , and τ i in each sublayer are expressed by aforesaid Eqs. (5)- (7).
The excess electron concentration in the former sublayer should contribute to the latter sublayer, accordingly, the boundary conditions adequate for each sublayer are expressed as [28,29]: where S vi is the electron recombination velocity at each interface. By recursively solving the above continuity equations, the excess electron concentration n 1 (T e ) reaching the Al x Ga 1Àx As/ GaAs interface can be calculated.
As for the GaAs active layer, the excess electrons consist of electrons contributed from the Al x Ga 1Àx As window layer and electrons generated in the GaAs active layer. Under the secondstage built-in electric field, the photoelectron transport process in GaAs active layer follows the one-dimensional continuity equation as described by where n 0 (y) is the excess electron concentration in the GaAs active layer, α hv0 is the absorption coefficient of the GaAs active layer, and D n0 , μ 0, and τ 0 denote the electron diffusion coefficient, the electron mobility, and the electron recombination lifetime in the GaAs active layer, respectively. Considering that the electrons from the Al x Ga 1Àx As window layer can contribute to the GaAs active layer, the boundary conditions adequate for Eq. (13) are given by [28]: By solving Eq. (13) via the boundary conditions Eq. (14) and the electron concentration n 1 (T e ) from the Al x Ga 1Àx As window layer, the concentration of electrons n 0 (y) in the active layer can be figured out. Finally, the quantum efficiency Y(hv), defined as the emitted electron number per incident photon, for the complex Al x Ga 1Àx As/GaAs photocathode is calculated as follows: where P is the surface electron escape probability. If E 0 = 0, the quantum efficiency model of Al x Ga 1Àx As/GaAs photocathode with the graded-composition (g-composition) and uniformdoping (u-doping) structure can be obtained. In the same way, when E 1 = 0 and E 0 = 0, we can also deduce the quantum efficiency model of common t-mode AlGaAs/GaAs photocathodes with the uniform-composition (u-composition) and u-doping structure. In a word, the aforementioned derivation method of photoemission model is applicable to those t-mode photocathodes with a common or complex structure.
Meanwhile, it is noted that the quantum efficiency has a close relation with the reflectivity R (hv) of photocathode, as shown in Eqs. (9) and (13), thus the optical properties of t-mode graded bandgap Al x Ga 1Àx As/GaAs photocathodes need to be investigated. In fact, the usual t-mode photocathode can be treated as a multilayer module, which comprises the glass faceplate, the antireflection layer, the window layer, and the GaAs active layer. The typical structure of t-mode Al x Ga 1Àx As/GaAs photocathodes is shown in Figure 3. The glass substrate with a thickness of several millimeters is much thicker than other thin layers in the order of nanometers or micrometers, so the glass is treated as the incident medium rather than the thin film. The reflectivity of incident light permeating the glass substrate is greatly declined by the silicon nitride (Si 3 N 4 ) antireflection film, and then the light in the wave range of interest is absorbed by the Al x Ga 1Àx As window layer and GaAs active layer in succession. The optical  properties of multilayer module can be calculated based on the transfer matrix of thin-film optics, and the characteristic matrix of the multilayered cathode module is given by [30]: In Eqs. (16)- (18), δ j and η j are the optical phase difference and complex refractive index of the jth film layer, η K + 1 is the optical constant of emergent medium, n j and k j constituting the complex refractive index are the refractivity and the extinction coefficient, d j is the thickness of the jth film layer, and θ j is the refraction angle of incident light. When light is perpendicularly incident on the surface of the glass substrate, the refraction angle is equal to zero. The reflectivity R hv and transmittivity T hv of the multilayered photocathode module can be calculated by the following expressions [30] where η g denotes the optical constant of the glass substrate. For the Al composition-varied window layer, the optical parameters, for example, the refractivity and extinction coefficient are different in each AlGaAs sublayer [31]. When the Al x Ga 1Àx As window layer is composed of n sublayers, the t-mode photocathode module can be treated as the thin film system of n + 2 layers to calculate the optical properties changing with incident photon wavelength, which are used as the necessary supplement to the quantum efficiency model.

Quantum efficiency simulation
As to the t-mode AlGaAs/GaAs photocathodes, the optical properties between the gcomposition and u-composition structures should be different. For simplified calculation, the composition-graded Al x Ga 1Àx As window layer is assumed to be of five sublayers with the fixed Al composition in each sublayer. The five Al composition values are assumed to be 0.9, 0.675, 0.45, 0.225, and 0, respectively, distributed from the AlGaAs/Si 3 N 4 interface to AlGaAs/ GaAs interface. For the u-composition AlGaAs/GaAs photocathode, the Al composition in the AlGaAs window layer is assumed to be 0.7. The optical properties including the reflectivity R hv and transmittivity T hv can be simulated by utilizing Eqs. (16)-(20) by referring to the structure of Figure 3. In the simulations, the refractivity and extinction coefficients of Al x Ga 1Àx As with different Al compositions are referred to [31], the refractivity coefficients of glass and Si 3 N 4 are 1.49 and 2.06, respectively, the extinction coefficients of glass and Si 3 N 4 are zero due to no absorption, and the thicknesses of Si 3 N 4 , AlGaAs, and GaAs layer are assumed to be 100, 500 nm and 1.0 μm, respectively. Besides, each sublayer in the g-composition Al x Ga 1Àx As window layer is supposed to have the equal thickness of 0.1 μm. The simulated optical property curves between the two t-mode AlGaAs/GaAs photocathode modules with different window layer structures are shown in Figure 4. It is clear to see that the oscillation number in the entire 400-1100 nm region for the g-composition structure is less than those for the ucomposition structure. In other words, compared with the u-composition Al 0.7 Ga 0.3 As/GaAs photocathode, the g-composition Al x Ga 1Àx As/GaAs photocathode exhibits the much smoother reflectivity curve in the spectrum region of 400-900 nm, which is the concerned photon absorption waveband for the AlGaAs/GaAs material. Besides, in the 900-1100 nm region, the locations of peaks and valleys of the reflectivity curves for the g-composition Al x Ga 1Àx As/ GaAs photocathode move toward the shortwave direction, nevertheless, this has little effect on the photoemission performance of GaAs photocathodes since these photons with the wavelength greater than 900 nm are hardly absorbed by the GaAs material.
By using the deduced quantum efficiency models which take into account the reflectivity varying with the wavelength, the quantum efficiency curves of the t-mode Al x Ga 1Àx As/GaAs photocathode with those unique graded bandgap structures are simulated, wherein the window layer is of the g-or u-composition structure, and the active layer is of the e-or u-doping structure, respectively. Figure 5 exhibits the superiority of the Al x Ga 1Àx As/GaAs photocathode with g-composition window layer and e-doping active layer. In Figure 5, some structural parameters such as the Al composition in the u-composition window layer, the Al composition distribution in each sublayer of Al x Ga 1Àx As window layer, and the thicknesses of Si 3 N 4 , AlGaAs and GaAs layers are identical to those in Figure 4. In the GaAs active layer, the doping concentration for e-doping structure is exponentially varied from 1 Â 10 19 to 1 Â 10 18 cm À3 , and that for the u-doping structure is 1 Â 10 19 cm À3 . In addition, the surface electron escape probability P is assumed to be 0.5. As a result of the reduced lattice mismatch by the seamless Al x Ga 1Àx As/GaAs heterojunction, the interface recombination velocity S vi for g-composition structure cannot exceed 10 4 cm/s, while S v for the common u-composition structure is usually 10 6 cm/s [5].
It is seen clearly from Figure 5 that the t-mode g-composition and e-doping photocathode can obtain the highest quantum efficiency in the spectrum region from 400 to 900 nm in contrast to other photocathodes. The quantum efficiency in the shortwave region, that is, blue-green region are enhanced greatly for the two former photocathodes with the g-composition structure. In the g-composition Al x Ga 1Àx As window layer, the photoelectrons excitated by shortwave light would be promoted toward the GaAs active layer under the g-composition induced electric field. Then, these shortwave photoelectrons are successively boosted toward the emission surface under the built-in electric field formed by the e-doping structure. As shown in Figure 5, the e-doping structure for the g-composition Al x Ga 1Àx As/GaAs photocathodes can slightly enhance the quantum efficiency, which is not like the case for the u-composition AlGaAs/GaAs photocathodes. The possible reason is that the g-composition Al x Ga 1Àx As layer can also absorb some extra longwave photons, which are originally absorbed by the GaAs active layer. In other words, more enough absorption space for longwave photons can be provided by the g-composition structure. While for the u-composition AlGaAs/GaAs photocathodes, the case is different. The GaAs active layer just absorbs the longwave photons, and the transport efficiency for these generated photoelectrons can just be improved by the doping-induced electric field.
To guide the structural design of t-mode graded bandgap Al x Ga 1Àx As/GaAs photocathode, the changes of quantum efficiency with the active layer thickness and the window layer thickness are analyzed, as shown in Figure 6. Figure 6(a) shows the changes of quantum efficiency curves with the active layer thickness T e , assuming Al composition distribution, S vi and T wi are the same as those in Figure 5. As T e increases, more space in the bulk for absorption of the longwave photons in the region of 650-900 nm is provided to generate more electrons to increase the quantum efficiency. If the GaAs active layer is thin, photoelectrons generated by shortwave light in the Al x Ga 1Àx As window layer would easily transport toward the GaAs active layer through diffusion and drift under the two-stage built-in electric field and finally escape into vacuum. In such a case, the quantum efficiency in the shortwave region would remain unchanged. Nevertheless, the thickness of the GaAs active layer must be controlled within a certain range, and the sufficiently thick active layer would decrease the quantum efficiency in the shortwave region, as shown in Figure 6(a). Therefore, the thickness of the active layer should be designed to balance the longwave response and shortwave response. When the Al x Ga 1Àx As window layer is 500 nm in total thickness, the appropriate thickness is thought to be in the range of 1.0-1.5 μm.
Considering that the built-in electric field in the window layer is inversely proportional to window layer thickness, the effect of the window layer thickness on quantum efficiency in the shortwave region, especially in the blue-green waveband for g-composition photocathodes, is more pronounced than that for the u-composition ones. Figure 6(b) shows the quantum efficiency changing with the window layer thickness T w , assuming T e = 1.0 μm. As T w decreases, the quantum efficiency in the waveband region from 400 to 720 nm is greatly enhanced arising from the enhanced g-composition induced electric field. When T w is thin, Figure 6. Quantum efficiency simulations with the changes of (a) active layer thickness and (b) window layer thickness for the t-mode graded bandgap Al x Ga 1Àx As/GaAs photocathodes. there is not enough space to absorb shortwave photons, and these shortwave photons can be absorbed by GaAs active layer. The quantum efficiency in the shortwave region would get increased as T w decreases since that the transport capacity of photoelectrons in GaAs layer is better than that in AlGaAs layer. Nevertheless, a passivation layer is necessary to prevent impurities from the substrate into the active layer for the t-mode photocathodes, and thus the Al x Ga 1Àx As window layer also utilized as the passivation layer should not be extremely thin.

Epitaxial growth and quality characterization 4.1. Epitaxial growth of photocathode materials
In modern epitaxial growth techniques, the metalorganic chemical vapor deposition (MOCVD) technique is suitable for growing the complex ultrathin multilayer materials with the composition-graded or doping-graded structures. To confirm the actual effect of the gcomposition and e-doping structure on the quantum efficiency of t-mode AlGaAs/GaAs photocathodes, the 2-inch-diameter Al x Ga 1Àx As/GaAs epilayers with two different structures were grown on the low-defect n-type GaAs (100) substrates in the horizontal low-pressure MOCVD reactor from AIXTRON. As shown in Figure 7(a), the multiple epitaxial layers consist of four AlGaAs/GaAs heterostructures, which follow the "inverted structure" technology [32,33]. In Figure 7(a), the AlGaAs stop layer serves as an etching-resistance layer, and the GaAs cap layer serves as an oxidation-blocking layer. The detailed structures of the two types of cathode materials are shown in Figure 7(b) and (c). The difference between the two samples is the structure of window layer, wherein one is of g-composition Al x Ga 1Àx As layer, and the other is of u-composition Al 0.7 Ga 0.3 As layer. Note that, as a result of the current epitaxial limitation, the GaAs active layer exhibits a quasi-exponential doping structure with the p-type dopant concentration varying from 1 Â 10 19 to 1 Â 10 18 cm À3 .
During the epitaxial growth process of the multiple layers, the group III sources are the trimethylgallium (TMGa) and trimethylaluminum (TMAl), the group V source was the AsH 3 , the dopant source was the diethylzinc (DEZn), and the carrier gas was the H 2 gas. Additionally, the growth process was monitored in situ using the LayTech EpiRAS-200 spectrometer. The parameters of the epitaxial growth process are as follows: the growth rate was about 2.5 μm/h, the V/III flux ratio was adjusted at 10-15, the Al composition was controlled by the flow ratio of TMGa to TMAl, and the growth temperature was set as 680 C and 710 C for GaAs and Al x Ga 1Àx As, respectively.

Quality characterization of photocathode materials
To understand the profile structure of the multilayered photocathode samples, the crosssectional photographs of the multilayered structure for the two cathode material samples were measured by the scanning electron microscope (SEM) from Hitachi. It is clearly seen from Figure 8 that differing from the case for u-composition sample, no sharp borderline exists at the interface of the Al x Ga 1Àx As window layer and GaAs active layer for the g-composition sample. This seamless interface would greatly reduce the interface electron recombination. It is noted that many cracks in Figure 8(b) are caused by the inappropriate cleavage, which cannot reflect the true quality of the epitaxy. From the SEM photographs, it is judged that the vertically multilayered constructions of the epitaxial cathode materials agree well with the structural design.  The depth distribution of carrier concentration in the multilayered p-type AlGaAs/GaAs materials was measured by the electrochemical capacitance-voltage (ECV) system from Bio-Rad. As shown in Figure 9, a series of sublayers forming the graded doping structure can be realized by the MOCVD technique. The carrier concentration of no more than 8 Â 10 18 cm À3 in the GaAs active layer shows a gradient distribution. For the Al x Ga 1Àx As window layer in Figure 9 (a), the carrier concentration decreases with the increase in Al composition, which exactly reflects the composition-graded structure.
To investigate the crystalline quality of the epitaxial cathode materials, the X-ray diffraction (XRD) curves were measured by the X'Pert Pro MRD system. As shown in Figure 10, the  rightmost peak represents the GaAs material, which is the superposition of the diffraction peaks of the GaAs cap layer, active layer, and substrate. The only one diffraction peak indicates that the crystalline perfection of the GaAs epilayers is consistent with the GaAs substrate. The left two diffraction peaks for the u-composition sample represent the AlGaAs window layer and stop layer, respectively. In the g-composition sample, there is no diffraction peak denoting the window layer, and a series of diffraction peaks exist nearby the peak of the GaAs layer, which are caused by the g-composition Al x Ga 1Àx As epilayer. The slightly narrower full width at half maximum of the GaAs diffraction peak indicates that the GaAs active layer in the g-composition sample has a better crystalline quality.

Transmission-mode cathode module fabrication
Following the recipe of fabricating glass-sealed t-mode AlGaAs/GaAs photocathodes [32,33], the epitaxial cathode materials cutted from the 2-inch-diameter epitaxial wafer were fabricated into the multilayered t-mode cathode module. The schematic process flow for fabricating t-mode AlGaAs/GaAs photocathode modules is shown in Figure 11. First, the GaAs cap layer was removed by chemical etching to expose the AlGaAs window layer, and by plasma enhanced chemical vapor deposition (PECVD), a thin antireflective layer of 100 nm-thick Si 3 N 4 was deposited on the exposed window layer surface. Then, the 7056 glass, serving as the incident window and support layer, was bonded on the Si 3 N 4 antireflection layer by thermocompression. Following that, through selective etching process, the GaAs substrate and AlGaAs stop layer were etched away to expose the GaAs active layer to prepare the NEA surface [32]. Finally, the Cr-Ni ring electrode applied to bias on the cathode was prepared by the physical vapor deposition (PVD), such as magnetron sputtering method. After these processing steps, the multilayered cathode module with a glass/Si 3 N 4 /AlGaAs/GaAs structure was finished. In addition, to eliminate etching-induced damage at the active layer surface, the polishing treatment was implemented, which slightly decreased the thickness of the GaAs active layer.
The optical property curves of the t-mode cathode modules with two different structures were measured by utilizing the Shimadzu UV-3600 spectrophotometer, which possesses three detectors working from ultraviolet to NIR waveband. The optical properties were measured based on the double optical path method, and light was incident on the surface of glass faceplate in a normal direction. Figure 12 shows the experimental reflectivity and transmissivity curves of the two different multilayered module samples. It is found that, just as the simulated results in Figure 4, the reflectivity curve in the region of 400-800 nm for g-composition structure is relatively smoother than that for u-composition structure. In other words, the smooth reflectivity curve verifies the composition-graded structure in the Al x Ga 1Àx As window layer from another aspect. Thereby, the characterization results regarding the cross-sectional photographs, carrier concentration distributions, X-ray diffraction peaks, and optical properties all reflect the special design structure.

Activation of photocathode surface
Prior to activation, the 18-mm-diameter cathode modules experienced the chemical cleaning and vacuum annealing to obtain an atomic level clean surface. The heat treatment with a suitable temperature under ultrahigh vacuum (UHV) condition is particularly important for the activation, and the quadrupole mass spectrometer (QMS) was adopted to monitor the change of residual gas components during the programmed temperature rose and fell. Figure 13 shows the changes of mainly concerned residual gas components for the two t-mode cathode module Figure 11. Schematic of the process flow for fabricating t-mode AlGaAs/GaAs photocathode modules following the "inverted structure" technology. samples. Through detecting the gas presence of the QMS traces at m/e = 18 (H 2 O), 75 (As), 91 (AsO), 150 (As 2 ), and 156 (Ga 2 O), it can be judged that whether the oxides on the GaAs surface, such as As 2 O 3 and Ga 2 O 3 , are cleared away with the increased temperature or not [34]. It can be inferred from Figure 13 that both the cathode module samples obtained an oxide-free clean surface after the heat treatment procedure in terms of these obvious QMS trace peaks.
After the sample cooled to room temperature, the Cs▬O activation to form the NEA state at the cathode surface was performed in the UHV chamber with a base pressure of 10 À9 Pa. The Cs and O sources used in the activation are solid dispensers easily controlled by direct current, and the flux is proportional to the operating current [35]. During the activation, the Cs source was on all the time, and the O source was switched on and off [35]. The operating current of Cs and O dispensers was regulated by program control current supply, and the photocurrent induced by a white light source was monitored in real time by the computer-controlled test system [35]. The initial Cs supply caused the gradual increase of the photocurrent. With the continuous Cs flux, when the photocurrent dropped to 80% of its peak, the O source was open. In subsequent alternate activation cycles, the O source was closed when the photocurrent reached its peak and was open again when the photocurrent dropped to 80% of the peak. The operating current ratio of Cs source to O source for both samples was regulated as the same 1.65/1.8. Until the photocurrent peak no longer increased, the O source and Cs source were closed successively, and the activation process was finished. To further improve the photoemission performance, the second heat treatment with a lower temperature was employed to the samples [36]. After that, the samples were activated again using the same co-deposition activation. As seen from Figure 14, the second activation can dramatically enhance the final cathode performance. Meanwhile, the final photocurrent peaks of the two samples are approximately the same.

Tube package and spectral response test
After the two-step Cs▬O activation process, the cathode module in the UHV activation chamber was transferred to the UHV seal vacuum chamber and indium sealed into an image  intensifier tube, wherein the t-mode AlGaAs/GaAs cathode module was equipped in association with the filmed microchannel plate (MCP), phosphor screen, output window, ceramics, and Kovar sealing parts [37]. The schematic structure and the photograph of the LLL proximity focused image intensifier are shown in Figure 15. As shown in Figure 15(a), the proximity focused image intensifier is capable of enhancing a LLL image from several thousands to tens of thousands of times. The input LLL image is converted into photoelectrons by the AlGaAs/ GaAs photocathode, and then the number of photoelectrons is multiplied several thousands of times by the MCP coated with a thin ion barrier film which can prevent ion feedback. Lastly, the multiplied photoelectrons bombard the phosphor screen and are converted into photons.  Thus, the input LLL image is intensified and appears as the output image on the phosphor screen. In addition to the function of direct eye observation, the LLL image intensifier can be coupled with CCD/CMOS array by the fiber optic taper to realize video output and remote monitoring [38,39].
The sealed image intensifiers were extracted from the seal vacuum chamber into ambient air, and the spectral response curves were measured by the spectral response testing instrument [22]. Through the spectral response values corresponding to the wavelength, the quantum efficiency values corresponding to the wavelength for the two different cathode samples were obtained [40]. In the spectral region of 600-750 nm, the quantum efficiency exceeds 40%. As shown in Figure 16, it is found that in contrast to the u-composition structure, the gcomposition structure is especially useful to the enhancement of shortwave quantum efficiency, which conforms to the original intention of our design concept. By fitting the experimental optical property and quantum efficiency data based on the theoretical photoemission model, the internal cathode parameters difficult to be measured directly can be obtained. The thickness values of each layer calculated by fitting the experimental reflectivity and transmittivity curves are listed in Table 1. It is seen that the Al composition in the gcomposition window layer is not distributed uniformly, and the sublayers with low Al composition are relatively thinner compared to those with high Al composition. For the two samples, the thicknesses of the GaAs active layer are smaller than the design values, which are caused by the polishing treatment after the fabrication of cathode modules.
By means of fitting the experimental quantum efficiency curves, we can obtain some performance parameters, for example, interface recombination velocity S v and surface escape probability P. It is seen from Table 1 that the two samples have the same P, which means that the  surface barrier shapes changed by the activation process are the same. Whereas, S v for the gcomposition sample, it is considerably reduced in contrast to that for the u-composition sample. It is known that the S v is mainly determined by the crystal quality of photocathode itself, such as misfit dislocations and stacking faults at the AlGaAs/GaAs interface. The gcomposition structure in the window layer can not only mitigate the interface discontinuity caused by the interface lattice mismatch, but also form an internal electric field to facilitate the transport of shortwave photons excitated electrons toward the emitting surface.

Conclusions
In this chapter, we have carried out systematically theoretical and experimental researches on t-mode AlGaAs/GaAs photocathodes, with regard to bandgap structure design, photoemission model derivation, epitaxial growth, surface activation, device fabrication, and performance evaluation. Compared with the common t-mode AlGaAs/GaAs photocathode, the graded bandgap t-mode Al x Ga 1Àx As/GaAs photocathode with a g-composition and e-doping structure can achieve higher quantum efficiency in the shortwave response region, particularly the blue-green spectral region of interest. In addition, this g-composition structure is helpful to mitigate the interface recombination and enhance the absorption of the longwave light, which leads to the enhanced photoemission capability. This work has reference significance for the design of other graded bandgap III-V group photocathodes.