Overview of the Semiconductor Photocathode Research in China

With the growing demand from scientific projects such as the X-ray free electron laser (XFEL), ultrafast electron diffraction/microscopy (UED/UEM) and electron ion collider (EIC), the semiconductor photocathode, which is a key technique for a high brightness electron source, has been widely studied in China. Several fabrication systems have been designed and constructed in different institutes and the vacuum of most systems is in the low 10−8 Pa level to grow a high QE and long lifetime photocathode. The QE, dark lifetime/bunch lifetime, spectral response and QE map of photocathodes with different kinds of materials, such as bialkali (K2CsSb, K2NaSb, etc.), Cs2Te and GaAs, have been investigated. These photocathodes will be used to deliver electron beams in a high voltage DC gun, a normal conducting RF gun, and an SRF gun. The emission physics of the semiconductor photocathode and intrinsic emittance reduction are also studied.


Introduction
Many scientific projects such as a high repetition rate XFEL [1][2][3], electron-ion collider (EIC) [4], ultrafast electron diffraction/microscopy (UED/UEM) [5][6][7], etc., are being built or proposed in China. In these big projects, the demand for a high-performance photocathode has become more crucial and urgent, especially for a semiconductor photocathode [8]. For instance, a Cs 2 Te photocathode is the first choice of a hard XFEL application such as a Shanghai HIgh repetitioN rate XFEL and Extreme light facility (SHINE) and a Shenzhen Superconducting Soft X-Ray Free Electron Laser (S 3 FEL). The reasons for increasing interest in the photocathode as the electron source are as following: High quantum efficiency (QE) [9]. The quantum yield of the photocathode material and the available laser power should be high enough to generate the needed bunch charge required by the beam dynamics [10,11].
Long dark/charge lifetime [12]. The charge lifetime, determined by the photocathode and drive laser together, should be large enough at an acceptable length of time without interrupting the machine operation to change or rejuvenate the photocathode [13,14].
Low intrinsic emittance [15]. With the development of photocathode injectors, the intrinsic emittance of photocathode has become the main part of the final electron beam emittance and directly sets the minimum achievable emittance from the start. For an ultra-low emittance beam required application such as XFEL, it is critical to use a low intrinsic emittance photocathode or eliminate the intrinsic emittance of the photocathode with different methods [16][17][18][19][20][21][22].
Among the different kinds of photocathodes, the metallic photocathode is robust and has low intrinsic emittance, but relatively low QE. On the other hand, the metallic photocathode must be illuminated by a UV laser that cannot deliver the bunch charge for a high repetition rate/CW XFEL [23]. Therefore, a semiconductor photocathode is the only choice for a high repetition rate/CW XFEL. The first choice is Cs 2 Te as it has a high QE and long lifetime [24][25][26]. The bialkali photocathode is chosen as the backup for its high QE and low intrinsic emittance [27].
(1) The substrate is heated at about 300 • C for more than 4 h to release the adsorbed gas.
(2) About 4~6 nm Te thin film is deposited on the substrate at 120 • C.
(3) The Cs activation starts when Te deposition is finished. The photocurrent is monitored during the cesiation process and the photocathode is illuminated by UV light. When the QE reaches plateau, the Cs source is turned off and the cathode will slowly cool down to room temperature. The QE of the fresh Cs 2 Te photocathode grown on Mo and stainless steel is about 10 % (@266 nm UV laser). SHINE and S 3 FEL have chosen very high frequency (VHF) NCRF gun [43,44] and Cs 2 Te photocathode as the baseline [45][46][47]. Therefore, Shanghai Advanced Research Institute (SARI, host of SHINE project), Dalian Institute of Chemical Physics (DICP), and Tsinghua University (THU) have built their Cs 2 Te deposition systems and started the recipe research separately. The QE of Cs 2 Te photocathodes fabricated at SARI and THU have achieved around 6% (@266 nm) and no QE measurement has been reported at DICP. Up to now, there are no beam test results published by more than three institutes.
No other gun test results are published for Cs 2 Te photocathode except PKU [48]. The Cs 2 Te photocathode performed well in the DC-SRF-I photoinjector during the beam experiment [41,49]. During the beam experiments of the DC-SRF-I photoinjector [48,[50][51][52], the QE of Cs 2 Te photocathode was stabilized at 4% (@266 nm) in the photoinjector for months as shown in Figure 2. The electric field on the surface of the cathode is about 5 MV/m and the peak electric field is less than 13 MV/m. During the beam experiment, the E acc of the SRF cavity in the DC-SRF-I photoinjector reached 14.5 MV/m in CW mode and 17 MV/m in pulse mode. The loss of Cesium during the transport of the cathode from deposition chamber to the gun is essential for keeping a long lifetime of the cathode. Therefore, the Cs 2 Te photocathode is reactivated before being transferred. The bunch charge was about 6~13 pC and the repetition rate was 81.25 MHz. The injector worked in pulse mode and the pulse repetition rate was 10 Hz. The average beam current was 1 mA in the macropulses with a length of 7 ms (duty factor 7%) and can be kept at about 0.5 mA for routine operation. The normalized transverse emittance of the delivered beam is 3.0 mm.mrad [48]. No dark current was observed by measuring with a faraday cup at the exit of the injector when keeping the DC high voltage and RF field on and No other gun test results are published for Cs2Te photocathode except PK Cs2Te photocathode performed well in the DC-SRF-I photoinjector during th periment [41,49]. During the beam experiments of the DC-SRF-I photoinjector the QE of Cs2Te photocathode was stabilized at 4% (@266 nm) in the photo months as shown in Figure 2. The electric field on the surface of the cathode   [48].

Bialkali (K2CsSb, K2NaSb) Photocathode
The first K2CsSb photocathode system [53,54] in China was designed by the Shanghai Institute of Applied Physics (SINAP, now SARI) in 2012, where the first K2CsSb photocathode was grown with this system. The vacuum is as low as 10 −9 Pa, which is similar to that of bialkali deposition systems at other laboratories around the world [10,12,18,36,55,56].
The deposition recipe of K2CsSb photocathode has been investigated at SARI since 2014. The recipe is as follows [57]: (1) The Mo substrates is degassed at 370 °C for six hours to release the absorbed gas such as H2, O2, N2 and H2O. The metal oxide and old photocathode thin film can also be removed during the heating process. (2) The substrate temperature is stabilized at about 75 °C, and Sb is evaporated to about 10 nm. The deposition rate is about 0.01 nm/s. (3) Co-deposition method is adopted at SARI. The temperature of the substrate is kept at 75 °C, and the K and Cs is evaporated together at a ratio of about 1:1. The heating power of the sources should be controlled in time. The K and Cs sources are turned off when the photocurrent reached a plateau. After preparation, the photocathode is cooled down to room temperature.
As shown in Figure 3, the activation process cost about 10 h. The reason for such a long activation process may be caused by the long distance (80 mm) between sources and substrate. During the activation process, the reflectivity of the thin film is monitored constantly. The reflectivity decreases with the increasing thickness of Sb. With the deposition of K and Cs, the reflectivity increased at the beginning and then decreases. When the reflectivity decreases to the lowest, the photocurrent starts to increase. After the QE reaches the plateau, the reflectivity stabilizes too. If K and Cs continue to deposit and the reflectivity will decrease, the QE will decrease too, which means that the reaction stopped and no new photocathode is formed. The reflectivity is a useful method to monitor the growth process of the bialkali photocathode.  [48]. Reproduced with the permission from reference [48].

Bialkali (K 2 CsSb, K 2 NaSb) Photocathode
The first K 2 CsSb photocathode system [53,54] in China was designed by the Shanghai Institute of Applied Physics (SINAP, now SARI) in 2012, where the first K 2 CsSb photocathode was grown with this system. The vacuum is as low as 10 −9 Pa, which is similar to that of bialkali deposition systems at other laboratories around the world [10,12,18,36,55,56].
The deposition recipe of K 2 CsSb photocathode has been investigated at SARI since 2014. The recipe is as follows [57]: (1) The Mo substrates is degassed at 370 • C for six hours to release the absorbed gas such as H 2 , O 2 , N 2 and H 2 O. The metal oxide and old photocathode thin film can also be removed during the heating process. (2) The substrate temperature is stabilized at about 75 • C, and Sb is evaporated to about 10 nm. The deposition rate is about 0.01 nm/s. (3) Co-deposition method is adopted at SARI. The temperature of the substrate is kept at 75 • C, and the K and Cs is evaporated together at a ratio of about 1:1. The heating power of the sources should be controlled in time. The K and Cs sources are turned off when the photocurrent reached a plateau. After preparation, the photocathode is cooled down to room temperature.
As shown in Figure 3, the activation process cost about 10 h. The reason for such a long activation process may be caused by the long distance (80 mm) between sources and substrate. During the activation process, the reflectivity of the thin film is monitored constantly. The reflectivity decreases with the increasing thickness of Sb. With the deposition of K and Cs, the reflectivity increased at the beginning and then decreases. When the reflectivity decreases to the lowest, the photocurrent starts to increase. After the QE reaches the plateau, the reflectivity stabilizes too. If K and Cs continue to deposit and the reflectivity will decrease, the QE will decrease too, which means that the reaction stopped and no new photocathode is formed. The reflectivity is a useful method to monitor the growth process of the bialkali photocathode.  Figure 3. The activation process of K2CsSb photocathode at SARI. "R" is the reflectivity of the thin film during deposition, and the drive laser wavelength is 532 nm [57].
The temperature dependence of QE of the K2CsSb photocathode during the activation process has been studied by SARI researchers [57]. The decayed photocathode could be recovered by heating without Cs atmosphere. Heating at 88 °C for 2 h can recover 50% of its original QE. But the photocathode QE after thermal annealing decreases faster. During the measurement of the bialkali photocathode lifetime, the QE at different time annealing are measured. Figure 4 shows the photocathode QE map before and after thermal annealing. The QE decreased and the uniformity became worse after thermal annealing, which means that some elements were missed during the deposition process.  . The activation process of K 2 CsSb photocathode at SARI. "R" is the reflectivity of the thin film during deposition, and the drive laser wavelength is 532 nm [57].
The temperature dependence of QE of the K 2 CsSb photocathode during the activation process has been studied by SARI researchers [57]. The decayed photocathode could be recovered by heating without Cs atmosphere. Heating at 88 • C for 2 h can recover 50% of its original QE. But the photocathode QE after thermal annealing decreases faster. During the measurement of the bialkali photocathode lifetime, the QE at different time annealing are measured. Figure 4 shows the photocathode QE map before and after thermal annealing. The QE decreased and the uniformity became worse after thermal annealing, which means that some elements were missed during the deposition process.
Micromachines 2021, 12, x FOR PEER REVIEW 5 of 19 Figure 3. The activation process of K2CsSb photocathode at SARI. "R" is the reflectivity of the thin film during deposition, and the drive laser wavelength is 532 nm [57].
The temperature dependence of QE of the K2CsSb photocathode during the activation process has been studied by SARI researchers [57]. The decayed photocathode could be recovered by heating without Cs atmosphere. Heating at 88 °C for 2 h can recover 50% of its original QE. But the photocathode QE after thermal annealing decreases faster. During the measurement of the bialkali photocathode lifetime, the QE at different time annealing are measured. Figure 4 shows the photocathode QE map before and after thermal annealing. The QE decreased and the uniformity became worse after thermal annealing, which means that some elements were missed during the deposition process. The QE (@532 nm) map of photocathode before and after thermal annealing at SARI [57]. Figure 4. The QE (@532 nm) map of photocathode before and after thermal annealing at SARI [57].
By heating the substrate at 300 • C [57], the low QE K-Cs-Sb photocathode can be removed from the substrate. The reflectivity of the degraded photocathode is monitored during the removal process. During the heating process (as shown in Figure 5), the reflectivity increases first and then decrease, finally increase to a stable value. The different reflectivity values corresponds to different compounds in the substrate and will be analyzed using in-situ X-ray in the future. When the reflectivity does not change any more, the used photocathode layer is completely removed. This is a useful method to reuse the substrate plug without breaking the vacuum. The used substrates have to be mechanically polished at many laboratories around the world, which cost a lot of time and the surface roughness may be influenced during the polishing process. Complete removal can be characterized with the reflectivity method. According to the results at SARI, no big difference on QE is observed from the heating removal and mechanically polished substrates.
icromachines 2021, 12, x FOR PEER REVIEW 6 By heating the substrate at 300 °C [57], the low QE K-Cs-Sb photocathode can b moved from the substrate. The reflectivity of the degraded photocathode is monit during the removal process. During the heating process (as shown in Figure 5), the re tivity increases first and then decrease, finally increase to a stable value. The differen flectivity values corresponds to different compounds in the substrate and will be anal using in-situ X-ray in the future. When the reflectivity does not change any more, the photocathode layer is completely removed. This is a useful method to reuse the subs plug without breaking the vacuum. The used substrates have to be mechanically poli at many laboratories around the world, which cost a lot of time and the surface rough may be influenced during the polishing process. Complete removal can be character with the reflectivity method. According to the results at SARI, no big difference on Q observed from the heating removal and mechanically polished substrates. INFN (National Institute for Nuclear Physics, Italy) type substrates and suitcase used at SARI [28,58]. Figure 6 shows the suitcase, load lock and transferring cham which will be used in the SHINE project at SARI. The suitcase comprised of a sto chamber, a 20 L/s sputtering ion pump and a 400 L/s SAES NEG pump, an anode to m ure the photocurrent from a viewport in one side of the storage chamber. Up to four p can be stored in the photocathode carrier in the suitcase. After the photocathodes is f cated, the suitcase is demounted from the deposition chamber and connected to the t fer chamber through the load lock. The load lock comprised of an ion pump and an metal angle valve. The load lock is heated up to 200 °C to derive UHV before the two valves can be opened. Two UHV manipulators is mounted on the transfer chamber. manipulator is to transfer the photocathode carrier between the suitcase and tra chamber. The other manipulator is to transfer the photocathode in and out of the gun vacuum level is kept at low 10 −9 Pa level in these chambers. INFN (National Institute for Nuclear Physics, Italy) type substrates and suitcases are used at SARI [28,58]. Figure 6 shows the suitcase, load lock and transferring chamber, which will be used in the SHINE project at SARI. The suitcase comprised of a storage chamber, a 20 L/s sputtering ion pump and a 400 L/s SAES NEG pump, an anode to measure the photocurrent from a viewport in one side of the storage chamber. Up to four plugs can be stored in the photocathode carrier in the suitcase. After the photocathodes is fabricated, the suitcase is demounted from the deposition chamber and connected to the transfer chamber through the load lock. The load lock comprised of an ion pump and an all-metal angle valve. The load lock is heated up to 200 • C to derive UHV before the two gate valves can be opened. Two UHV manipulators is mounted on the transfer chamber. One manipulator is to transfer the photocathode carrier between the suitcase and transfer chamber. The other manipulator is to transfer the photocathode in and out of the gun. The vacuum level is kept at low 10 −9 Pa level in these chambers. The Institute of High Energy Physics [59] (IHEP), China Academy of Enginee Physics (CAEP), and PKU started to design and build their K2CsSb photocathode syst from 2014. THU, DICP are also planning to build their deposition systems for bia (K2CsSb or K2NaSb) photocathode.
The deposition chamber and the recipe for K2CsSb photocathode at IHEP (as sh in Figure 7) and CAEP are similar to that of SARI. During preliminary experiment, th of K2CsSb photocathode was about 1%. The deposition chambers at CAEP are mainly u for Cs2Te preparation at present after a test of K2CsSb photocathode fabrication. The Institute of High Energy Physics [59] (IHEP), China Academy of Engineering Physics (CAEP), and PKU started to design and build their K 2 CsSb photocathode systems from 2014. THU, DICP are also planning to build their deposition systems for bialkali (K 2 CsSb or K 2 NaSb) photocathode.
The deposition chamber and the recipe for K 2 CsSb photocathode at IHEP (as shown in Figure 7) and CAEP are similar to that of SARI. During preliminary experiment, the QE of K 2 CsSb photocathode was about 1%. The deposition chambers at CAEP are mainly used for Cs 2 Te preparation at present after a test of K 2 CsSb photocathode fabrication. At PKU, a deposition chamber with four arms has been designed and manufactured since 2017. A different recipe with very fast cesiation process was adopted and optimized. The advantage of the verified recipe is that the K2CsSb photocathodes can have a longer lifetime. Recent experiments show that the QE of K2CsSb photocathodes are 4.1-7.4% and no degradation has been observed in DC-SRF gun for 2 weeks. The photocathodes were transported from deposition chamber to DC-SRF gun by a suitcase at a vacuum level of 9 × 10 −9 Pa as shown in Figure 8.

GaAs Photocathode
There are two GaAs photocathode deposition systems at IHEP and CAEP in China. The vacuum of these two systems is about 2~4 × 10 −10 Pa. Sputtering ion pumps (200~400 L/s, homemade) and NEG pumps (2000~4000 L/s, SAES) are used in these systems with Leybold IE514 as the vacuum monitor.
The fabrication of a GaAs photocathode has been accomplished by IHEP and CAEP. 10% QE (@532 nm) and weeks of lifetime have been derived during past years, which is At PKU, a deposition chamber with four arms has been designed and manufactured since 2017. A different recipe with very fast cesiation process was adopted and optimized. The advantage of the verified recipe is that the K 2 CsSb photocathodes can have a longer lifetime. Recent experiments show that the QE of K 2 CsSb photocathodes are 4.1-7.4% and no degradation has been observed in DC-SRF gun for 2 weeks. The photocathodes were transported from deposition chamber to DC-SRF gun by a suitcase at a vacuum level of 9 × 10 −9 Pa as shown in Figure 8. At PKU, a deposition chamber with four arms has been designed and manufactured since 2017. A different recipe with very fast cesiation process was adopted and optimized. The advantage of the verified recipe is that the K2CsSb photocathodes can have a longer lifetime. Recent experiments show that the QE of K2CsSb photocathodes are 4.1-7.4% and no degradation has been observed in DC-SRF gun for 2 weeks. The photocathodes were transported from deposition chamber to DC-SRF gun by a suitcase at a vacuum level of 9 × 10 −9 Pa as shown in Figure 8.

GaAs Photocathode
There are two GaAs photocathode deposition systems at IHEP and CAEP in China. The vacuum of these two systems is about 2~4 × 10 −10 Pa. Sputtering ion pumps (200~400 L/s, homemade) and NEG pumps (2000~4000 L/s, SAES) are used in these systems with Leybold IE514 as the vacuum monitor.
The fabrication of a GaAs photocathode has been accomplished by IHEP and CAEP. 10% QE (@532 nm) and weeks of lifetime have been derived during past years, which is

GaAs Photocathode
There are two GaAs photocathode deposition systems at IHEP and CAEP in China. The vacuum of these two systems is about 2~4 × 10 −10 Pa. Sputtering ion pumps (200~400 L/s, homemade) and NEG pumps (2000~4000 L/s, SAES) are used in these systems with Leybold IE514 as the vacuum monitor.
The fabrication of a GaAs photocathode has been accomplished by IHEP and CAEP. 10% QE (@532 nm) and weeks of lifetime have been derived during past years, which is reliable and repeatable. At present, no polarized electron source based on superlattice GaAs has been investigated for an accelerator-based project in China [60][61][62]. The typical recipe for high QE GaAs photocathode at CAEP is as follows: (1) The GaAs substrate is cleaned by high temperature heating [63] to remove the residual oxygen and other impurity on the surface of the GaAs sample. The time for the high temperature cleaning is about 45 min. (2) The Cs source is heated to release Cs atom to activate the GaAs. The photocurrent will increase after 20 min. The vacuum of the deposition chamber will decrease to 1 × 10 −7 Pa from 4 × 10 −8 Pa. (3) When the photocurrent reaches the plateau, the Cs source is shut off and oxygen is brought into the chamber to increase the photocurrent. When the photocurrent decreased, shut off oxygen and open the Cs source again. Then repeated the above process for several cycles to the highest photocurrent.
During the heating removal process, the temperature distribution on the GaAs photocathode is derived with the ANSYS program. The QE distribution of the GaAs photocathode fabricated is relevant with the temperature distribution in the reference [63]. The HV DC gun (500 kV) is the key element of the THz-FEL facility at CAEP. With the QE of about 5% in DC gun, an average CW beam current of 5 mA was accelerated to 8 MeV in the SRF cryomodule [64]. The normalized emittance was about 6 mm.mrad at the CW 4.8 mA average current. The intrinsic emittance of GaAs photocathode was measured to be 0.6 mm.mrad/mm from the beam line at CAEP as shown in Figure 9a,b. The factors that affect the GaAs photocathode operation lifetime in the high voltage DC gun are investigated. The vacuum and ion backbombardment are the principle factors affecting the lifetime. The effects and limitations of the beam off-axis emission to suppress the ion back bombardment is simulated at CAEP [65]. The simulation suggested that the ions mainly bombard the electrostatic center of the photocathode and 95% can be reduced if the electron is excited off-axis. The QE maps are measured before and after the DC gun (shown in Figure 9b) CW operation in both the on and off-axis emission, which agrees well with the simulation results.
reliable and repeatable. At present, no polarized electron source based on superlattice GaAs has been investigated for an accelerator-based project in China [60][61][62]. The typical recipe for high QE GaAs photocathode at CAEP is as follows: (1) The GaAs substrate is cleaned by high temperature heating [63] to remove the residual oxygen and other impurity on the surface of the GaAs sample. The time for the high temperature cleaning is about 45 min. (2) The Cs source is heated to release Cs atom to activate the GaAs. The photocurrent will increase after 20 min. The vacuum of the deposition chamber will decrease to 1 × 10 −7 Pa from 4 × 10 −8 Pa. (3) When the photocurrent reaches the plateau, the Cs source is shut off and oxygen is brought into the chamber to increase the photocurrent. When the photocurrent decreased, shut off oxygen and open the Cs source again. Then repeated the above process for several cycles to the highest photocurrent. During the heating removal process, the temperature distribution on the GaAs photocathode is derived with the ANSYS program. The QE distribution of the GaAs photocathode fabricated is relevant with the temperature distribution in the reference [63]. The HV DC gun (500 kV) is the key element of the THz-FEL facility at CAEP. With the QE of about 5% in DC gun, an average CW beam current of 5 mA was accelerated to 8 MeV in the SRF cryomodule [64]. The normalized emittance was about 6 mm.mrad at the CW 4.8 mA average current. The intrinsic emittance of GaAs photocathode was measured to be 0.6 mm.mrad/mm from the beam line at CAEP as shown in Figure 9a,b. The factors that affect the GaAs photocathode operation lifetime in the high voltage DC gun are investigated. The vacuum and ion backbombardment are the principle factors affecting the lifetime. The effects and limitations of the beam off-axis emission to suppress the ion back bombardment is simulated at CAEP [65]. The simulation suggested that the ions mainly bombard the electrostatic center of the photocathode and 95% can be reduced if the electron is excited off-axis. The QE maps are measured before and after the DC gun (shown in Figure 9b) CW operation in both the on and off-axis emission, which agrees well with the simulation results.
At IHEP, a GaAs photocathode was used as the electron source in the 500 kV DC gun. During the beam experiment, the QE of GaAs photocathode was about 5% (@532 nm) and a 5 mA average current CW beam was derived. Now the DC gun at IHEP is used as the high energy photon source (HEPS) test bench, which is used to test the SRF cavities and cryo-modules for XFEL and the proposing circular electron-positron collider (CEPC).

Physics of the Emission Process
The researchers at PKU [55], SARI [53], and THU [66] have simulated the emission process of the alkali antimony photocathodes.
The Monte Carlo simulation [67] is based on the Spicer's three step model. The initial electron distribution is derived with the convolution of the density of states (DoS) in the valence band and in the conduction band. The initial electron distribution is calculated with the localized spherical-wave method. In Figure 10, the percentage of electrons with different energy and the energy distribution in the thickness direction is calculated. The number of the photons/electrons used in the simulation is 100,000. The spectral response, intrinsic emittance, response time, and cryogenic performance can be derived from this code as shown in Figure 11a, which agrees well the experimental results [55]. The spectral response is simulated with and without electron-hole scattering, respectively, which will greatly influence the results by changing the energy loss during the electron transport step in the conduction band, especially in the high photon energy range [6]. The difference is that in the high energy part, the QE will decrease if the electron-hole scattering is included. The emission time of the electrons, which is an important factor for the bunch length, is less than 200 fs from the simulation, which agree well with the experimental results. At IHEP, a GaAs photocathode was used as the electron source in the 500 kV DC gun. During the beam experiment, the QE of GaAs photocathode was about 5% (@532 nm) and a 5 mA average current CW beam was derived. Now the DC gun at IHEP is used as the high energy photon source (HEPS) test bench, which is used to test the SRF cavities and cryo-modules for XFEL and the proposing circular electron-positron collider (CEPC).

Physics of the Emission Process
The researchers at PKU [55], SARI [53], and THU [66] have simulated the emission process of the alkali antimony photocathodes.
The Monte Carlo simulation [67] is based on the Spicer's three step model. The initial electron distribution is derived with the convolution of the density of states (DoS) in the valence band and in the conduction band. The initial electron distribution is calculated with the localized spherical-wave method. In Figure 10, the percentage of electrons with different energy and the energy distribution in the thickness direction is calculated. The number of the photons/electrons used in the simulation is 100,000. The spectral response, intrinsic emittance, response time, and cryogenic performance can be derived from this code as shown in Figure 11a, which agrees well the experimental results [55]. The spectral response is simulated with and without electron-hole scattering, respectively, which will greatly influence the results by changing the energy loss during the electron transport step in the conduction band, especially in the high photon energy range [6]. The difference is that in the high energy part, the QE will decrease if the electron-hole scattering is included. The emission time of the electrons, which is an important factor for the bunch length, is less than 200 fs from the simulation, which agree well with the experimental results. The Schottky effect [68] on cryo-cooled K2CsSb photocathode at h SRF gun is also simulated, which has a nice correlation with the exp shown in Figure 11b [68]. When the RF phase changes, the electric fi surface of the cryo-cooled K2CsSb photocathode will change, and the s be lower according to the Schottky effect. Therefore, the effective QE RF phase. At the beginning of the RF phase scan, the photocurrent is li charge limit. The author explained this phenomenon with an analyti Monte Carlo simulation [68], which agreed well with the experimen helpful to understand the mechanism of the cryo-effect on the intrinsi tion of the K2CsSb photocathode and is essential to derive an ultralow photocathode. The heterojunction photocathodes (K2NaSb/Cs3Sb, GaAs/Cs3Sb) are also simulated at PKU with this code [69]. These het cathodes aim to find the physics of the emitted electrons and to achi high QE, and a long lifetime photocathode, especially for a long lifetim cathode [70].  The Schottky effect [68] on cryo-cooled K2CsSb photocathode at high gradient SRF gun is also simulated, which has a nice correlation with the experimental d shown in Figure 11b [68]. When the RF phase changes, the electric field applied surface of the cryo-cooled K2CsSb photocathode will change, and the surface barri be lower according to the Schottky effect. Therefore, the effective QE will grow w RF phase. At the beginning of the RF phase scan, the photocurrent is limited by the charge limit. The author explained this phenomenon with an analytical model a Monte Carlo simulation [68], which agreed well with the experimental results. T helpful to understand the mechanism of the cryo-effect on the intrinsic emittance tion of the K2CsSb photocathode and is essential to derive an ultralow intrinsic em photocathode. The heterojunction photocathodes (K2NaSb/Cs3Sb, GaAs/Cs2Te GaAs/Cs3Sb) are also simulated at PKU with this code [69]. These heterojunction cathodes aim to find the physics of the emitted electrons and to achieve low emi high QE, and a long lifetime photocathode, especially for a long lifetime polarized cathode [70].
(a) At SARI [53], the researcher gave a special view on the emission process of K2CsS photocathode as shown in Figure 12: the electron bunch is divided by "electron slice" in side the cathode material, which separately contributes to the intrinsic emittance, just a the slice emittance contributes to the final emittance [71]. The scattering probability in creases with the transport distance and more electron energy would lose during the scat tering process. The emission probability and transverse momenta of the electrons would decrease with the photocathode depth. Therefore, the distribution of the photoelectron to final QE and intrinsic emittance decreases with increasing depth. This is very helpfu to understand the mechanism of the intrinsic emittance, which in turn can help to deriv a high brightness electron beam by adjusting the initial distribution of the photoelectro along the depth direction. Figure 11. (a) The simulation results of spectral response with and without electron-hole (e-h) scattering, and intrinsic emittance at room temperature. The dots are the experimental results of the spectral response of the K 2 CsSb photocathode at room temperature [55]. (b) The Schottky effect on the cryo-cooled K 2 CsSb photocathode in the 704 MHz SRF gun at Brookhaven National Lab (BNL) [68]. Reproduced with the permission from reference [68].
The Schottky effect [68] on cryo-cooled K 2 CsSb photocathode at high gradient in the SRF gun is also simulated, which has a nice correlation with the experimental data as shown in Figure 11b [68]. When the RF phase changes, the electric field applied on the surface of the cryo-cooled K 2 CsSb photocathode will change, and the surface barrier will be lower according to the Schottky effect. Therefore, the effective QE will grow with the RF phase. At the beginning of the RF phase scan, the photocurrent is limited by the space charge limit. The author explained this phenomenon with an analytical model and the Monte Carlo simulation [68], which agreed well with the experimental results. This is helpful to understand the mechanism of the cryo-effect on the intrinsic emittance reduction of the K 2 CsSb photocathode and is essential to derive an ultralow intrinsic emittance photocathode. The heterojunction photocathodes (K 2 NaSb/Cs 3 Sb, GaAs/Cs 2 Te, and GaAs/Cs 3 Sb) are also simulated at PKU with this code [69]. These heterojunction photocathodes aim to find the physics of the emitted electrons and to achieve low emittance, high QE, and a long lifetime photocathode, especially for a long lifetime polarized photocathode [70].
At SARI [53], the researcher gave a special view on the emission process of K 2 CsSb photocathode as shown in Figure 12: the electron bunch is divided by "electron slice" inside the cathode material, which separately contributes to the intrinsic emittance, just as the slice emittance contributes to the final emittance [71]. The scattering probability increases with the transport distance and more electron energy would lose during the scattering process. The emission probability and transverse momenta of the electrons would decrease with the photocathode depth. Therefore, the distribution of the photoelectrons to final QE and intrinsic emittance decreases with increasing depth. This is very helpful to understand the mechanism of the intrinsic emittance, which in turn can help to derive a high brightness electron beam by adjusting the initial distribution of the photoelectron along the depth direction. To enhance the electron emission, the surface polarized plasmons (SPPs) are also investigated at SARI [72][73][74][75][76]. The silver substrate with a nanopattern is employed to achieve wave vector matching and the SPPs is excited between the substrate and photocathode [73]. The nanopattern on the silver substrate (the width, depth, and the thickness of the photocathode) is optimized by CST studio to achieve the needed SPPs excited by specific wavelength. When the absorption rate of the incident laser to photocathode with nanopattern rises to 95%, the SPPs are successfully excited. The electromagnetic field near the surface between the substrate and the photocathode is redistributed, periodic and aligned with the nanopattern structure. Figure 13 shows the PNF of two unit cells with a phase of 90 degree caused by the SPPs. The electromagnetic field is enhanced by the SPPs in the interface between the substrate and the photocathode film. Hence the increased absorption and the designed PNF can help to improve the photocathode's performance, such the QE and intrinsic emittance. ure 13. Electric field distribution of PNF over cross section. Left and right images show electric field distributions of X d Y components of PNF, respectively [73]. To enhance the electron emission, the surface polarized plasmons (SPPs) are also investigated at SARI [72][73][74][75][76]. The silver substrate with a nanopattern is employed to achieve wave vector matching and the SPPs is excited between the substrate and photocathode [73]. The nanopattern on the silver substrate (the width, depth, and the thickness of the photocathode) is optimized by CST studio to achieve the needed SPPs excited by specific wavelength. When the absorption rate of the incident laser to photocathode with nanopattern rises to 95%, the SPPs are successfully excited. The electromagnetic field near the surface between the substrate and the photocathode is redistributed, periodic and aligned with the nanopattern structure. Figure 13 shows the PNF of two unit cells with a phase of 90 degree caused by the SPPs. The electromagnetic field is enhanced by the SPPs in the interface between the substrate and the photocathode film. Hence the increased absorption and the designed PNF can help to improve the photocathode's performance, such the QE and intrinsic emittance. To enhance the electron emission, the surface polarized plasmons (SPPs) are also investigated at SARI [72][73][74][75][76]. The silver substrate with a nanopattern is employed to achieve wave vector matching and the SPPs is excited between the substrate and photocathode [73]. The nanopattern on the silver substrate (the width, depth, and the thickness of the photocathode) is optimized by CST studio to achieve the needed SPPs excited by specific wavelength. When the absorption rate of the incident laser to photocathode with nanopattern rises to 95%, the SPPs are successfully excited. The electromagnetic field near the surface between the substrate and the photocathode is redistributed, periodic and aligned with the nanopattern structure. Figure 13 shows the PNF of two unit cells with a phase of 90 degree caused by the SPPs. The electromagnetic field is enhanced by the SPPs in the interface between the substrate and the photocathode film. Hence the increased absorption and the designed PNF can help to improve the photocathode's performance, such the QE and intrinsic emittance. The results of the Monte Carlo simulations in Figure 14 showed that with SPPs, the absorption efficiency for 532 nm laser would rise two times as before. This will help to raise the QE of the K2CsSb photocathode, and the emittance can be kept at 0.5~0.8 Figure 13. Electric field distribution of PNF over cross section. Left and right images show electric field distributions of X and Y components of PNF, respectively [73].
The results of the Monte Carlo simulations in Figure 14 showed that with SPPs, the absorption efficiency for 532 nm laser would rise two times as before. This will help to raise the QE of the K 2 CsSb photocathode, and the emittance can be kept at 0.5~0.8 mm.mrad/mm. These results show that part of the emittance growth is caused by plasmonic near field (PNF). Therefore, this can be suppressed and new possibilities for tailoring the nature of materials with improved performance can be achieved from the simulation results. The simulated results show that the QE of a K 2 CsSb photocathode with SPPs is 2-3 times that of the photocathode without SPPs [73]. Furthermore, the intrinsic emittance of the photocathode remains substantially constant, which means it is possible to obtain a photocathode with high QE and low emittance at the same time. mm.mrad/mm. These results show that part of the emittance growth is caused by plasmonic near field (PNF). Therefore, this can be suppressed and new possibilities for tailoring the nature of materials with improved performance can be achieved from the simulation results. The simulated results show that the QE of a K2CsSb photocathode with SPPs is 2-3 times that of the photocathode without SPPs [73]. Furthermore, the intrinsic emittance of the photocathode remains substantially constant, which means it is possible to obtain a photocathode with high QE and low emittance at the same time. Figure 14. Simulated data showing the relationship between QE of K2CsSb photocathode and incident photon energy with/without surface polarized plasmons (SPPs). Black and red curves are QEs of K2CsSb photocathodes with and without SPPs, respectively; the blue curve represents the ratio of QE of K2CsSb photocathodes with SPPs to that of K2CsSb photocathodes without SPPs [73].
The cryogenic performance of the Cs3Sb cathode and its degradation mechanism [66], caused by residual gas in the chamber, is also simulated at THU. The acceptor level of the Cs3Sb photocathode is considered in the Monte Carlo model. The change of the acceptor level is considered the reason for the decay of the Cs3Sb cathode at a cryogenic temperature. The model agrees well with the experimental results achieved by Cornell [17]. The impact of residual gas on intrinsic emittance is also investigated by the dynamic model [77]. The main mechanism in the dynamical model is that the electron affinity and electron transmission probability will change with the electron affinity affected by the Cs atoms reacting with residual gas on the surface of the cathode. The simulation results are shown Figure 15a. The analytical results agree well with the experiment of QE degradation in reference, where the QE degrades to about 50% of its original value. The cathode intrinsic emittance decreases a little and stabilizes at about 0.762 mm.mrad/mm. The author also discussed when the incident photon energy is lower than the threshold under typical DC gun and RF gun conditions (4 and 50 MV/m, respectively) [78,79]. The QE degrades in both cases while the intrinsic emittance differs a lot, and the main reason is the different contribution to the acceptor level under these two electric fields. At 4 MV/m, the photoelectrons mainly come from the acceptor level and at 50 MV/m from the valence band. In addition, the simulated results suggest that the intrinsic emittance of the near-threshold emission (driven by a 690 nm laser) of the Cs3Sb photocathode is better than UV laserdrived metal cathode when the QE remains above an acceptable minimum level. This is Figure 14. Simulated data showing the relationship between QE of K 2 CsSb photocathode and incident photon energy with/without surface polarized plasmons (SPPs). Black and red curves are QEs of K 2 CsSb photocathodes with and without SPPs, respectively; the blue curve represents the ratio of QE of K 2 CsSb photocathodes with SPPs to that of K 2 CsSb photocathodes without SPPs [73].
The cryogenic performance of the Cs 3 Sb cathode and its degradation mechanism [66], caused by residual gas in the chamber, is also simulated at THU. The acceptor level of the Cs 3 Sb photocathode is considered in the Monte Carlo model. The change of the acceptor level is considered the reason for the decay of the Cs 3 Sb cathode at a cryogenic temperature. The model agrees well with the experimental results achieved by Cornell [17]. The impact of residual gas on intrinsic emittance is also investigated by the dynamic model [77]. The main mechanism in the dynamical model is that the electron affinity and electron transmission probability will change with the electron affinity affected by the Cs atoms reacting with residual gas on the surface of the cathode. The simulation results are shown Figure 15a. The analytical results agree well with the experiment of QE degradation in reference, where the QE degrades to about 50% of its original value. The cathode intrinsic emittance decreases a little and stabilizes at about 0.762 mm.mrad/mm. The author also discussed when the incident photon energy is lower than the threshold under typical DC gun and RF gun conditions (4 and 50 MV/m, respectively) [78,79]. The QE degrades in both cases while the intrinsic emittance differs a lot, and the main reason is the different contribution to the acceptor level under these two electric fields. At 4 MV/m, the photoelectrons mainly come from the acceptor level and at 50 MV/m from the valence band. In addition, the simulated results suggest that the intrinsic emittance of the near-threshold emission (driven by a 690 nm laser) of the Cs 3 Sb photocathode is better than UV laser-drived metal cathode when the QE remains above an acceptable minimum level. This is helpful in high brightness electron sources which low intrinsic emittance and high QE are needed at the same time.

Conclusions
In China, research on semiconductor photocathodes has been attracting growing interest. Different kinds of deposition systems with imported UHV devices for a semiconductor photocathode such as Cs2Te, bialkali and GaAs have been built. The vacuum level of deposition systems is in the 10 −8~1 0 −10 Pa level. The fabrication recipe, transport, gun test and emission physics have been widely investigated by local institutes and universities. The reliability and consistency of deposition systems need to be improved in the future.

Conclusions
In China, research on semiconductor photocathodes has been attracting growing interest. Different kinds of deposition systems with imported UHV devices for a semiconductor photocathode such as Cs 2 Te, bialkali and GaAs have been built. The vacuum level of deposition systems is in the 10 −8~1 0 −10 Pa level. The fabrication recipe, transport, gun test and emission physics have been widely investigated by local institutes and universities. The reliability and consistency of deposition systems need to be improved in the future.

Forward Look
In the following years, high QE and long lifetime semiconductor photocathodes will be used for scientific projects by institutes in China, especially the Cs 2 Te photocathode for the XFEL application. Research on reducing the intrinsic emittance will be studied, such as SPPs, cryogenic photocathodes, heterojunction photocathodes, etc. The semiconductor photocathode will be tested in different type of guns and injectors (DC, RF, and SRF) to derive the required electron beam. The emission physics will also be studied to understand and improve the performance of the semiconductor photocathode.
Funding: This research received no external funding.