pn-type substrate dependence of CsK2Sb photocathode performance

The CsK2Sb photocathode is capable of generating a high-intensity and low-emittance electron beam with visible laser light. In this study, we examined CsK2Sb photocathode evaporation on nand p-type Si(100), GaAs(100), and Si(111) substrates, and compared their cathode performance. We found that the quantum efficiency of the p-type substrate was superior to that of the n-type substrate for the same substrate material and surface orientation. We show that this can be qualitatively analyzed by an energy-band model for the semiconductor(metal)-semiconductor junction between CsK2Sb and the substrate. We also evaporated the Cs3Sb cathode on nand p-type Si(100) and GaAs(100) substrates, and confirmed that the cathode performance of all substrates was consistent with the model results. This finding indicates the possibility of improving the thin-film cathode performance by revisiting the semiconductor(metal)semiconductor junction between the cathode and the substrate, which improves the understanding of high-performance photocathodes.


I. INTRODUCTION
Future accelerator projects based on linear accelerators, such as the Energy Recovery Linac (ERL) [1], Free Electron Laser (FEL) [2], and Linear Collider (LC) [3] will extend the knowledge base in scientific applications. Photocathodes are one of the most important subsystems in advanced linear accelerators because the accelerator performance strongly depends on the initial beam quality. For example, in the baseline design of the LCLS II project [4], which is the first continuous wave x-ray FEL project in the United States, a large current up to 0.3 mA is required with 0.4 mm · mrad emittance.
Since the 1980s, photocathodes have been operated in many accelerators [5]. The most popular photo-cathode materials are metals such as Cu [6] or Ag [7]. To generate an intense electron beam against the low quantum efficiency (QE) of photoelectron emissions, typically in the order of 0.01%, and the large work function, a high-power UV laser is required, which may be produced by harmonic generations [8] but with low efficiency.
The negative electron affinity-GaAs cathode has excellent properties such that the QE can be higher than 20% with the infrared (IR) laser. However, the negative electron affinity surface is fragile and easily damaged by residual gas adsorption or beam extraction.
Alkali (Multi) antimonide photocathodes composed of one or more alkali metals, such as Cs 3 Sb and CsK 2 Sb, are widely used in photomultiplier tubes [9]. The Cs 3 Sb is normally a p-type material [10] and has high QE(1%-5%) with visible lasers [11].
CsK 2 Sb is a nearly intrinsic semiconductor [12] fabricated as a thin film using the vacuum evaporation technique. The material can be driven by visible green light (532 nm) for photoemissions with QE values higher than 10% [13,14]. The material is robust against a large emission current [15]. Owing to these characteristics, CsK 2 Sb is one of the best candidates for a high-brightness electron source at a large current density [16,17].
Although it is well known that the CsK 2 Sb cathode performance depends on the substrate material and surface conditions [13,[18][19][20][21], the underlying physics has not been extensively studied to date because of the lack of standardization of the cathode evaporation process. For example, the dependence of the cathode performance on the surface orientation of the CsK 2 Sb or the substrate crystal was studied but not significantly observed [22,23] because the cathode evaporation condition was not fully optimized in these studies. Indeed, the QE in these studies was relatively low (3%).
Recently, it was shown that about 5% QE and nearly atomically smooth films of K-Cs-Sb could be obtained with good reproducibility by a triple-element co-deposition in which all three elements of K, Cs and Sb were evaporated simultaneously onto the substrate [24,25]. One the other hand, in our previous study, we established a cathode evaporation process, which produced high QE with good reproducibility by the sequential evaporation of Sb, K and Cs on a substrate [26]. Based on this process, we proved that the cathode performance depends not only on cleanness and crystallinity, but also on the surface orientation of the crystalline substrate.
To improve our understanding for the underlying physics of CsK 2 Sb photo-cathode, we studied the effect of the semiconductor type (p or n) of the substrates on the cathode performance. For this purpose, CsK 2 Sb cathode was fabricated on Si(100), Si(111), and GaAs(100) of the pand n-type with the process established in our previous study [26].

II. EXPERIMENT
The experimental setup was described in Ref. [26]. The cathode was fabricated in a vacuum chamber at a typical pressure of 1.0 × 10 −8 Pa. The cathode substrate is fixed on a molybdenum puck. The puck is mounted on the cathode holder during evaporation and electron emissions. In this study, Si(100) and Si(111) p-type wafers with resistivity of ≤ 0.002 Ω cm, Si(100) and Si(111) n-type wafers with resistivity of ≤ 0.002 Ω cm, GaAs(100) p-type wafers with resistivity of ≤ 0.007 Ω cm and GaAs(100) n-type wafers with resistivity of ≤ 0.002 Ω cm were employed as the substrates.
The CsK 2 Sb cathode was formed by the sequential evaporation of Sb, K, and Cs on a substrate, which is summarized as follows. (1) The substrate was heated to 500-600°C and held at that temperature for one hour to remove the surface oxidized layer (heat cleaning process).
(2) The temperature was lowered to 100°C. During evaporation, this temperature was maintained. (3) A 10 nm-thick layer of Sb was evaporated. (4) Then, K was evaporated. Simultaneously, QE was observed by illuminating the cathode with the 532 nm laser light. The evaporation was stopped when QE was saturated, which usually occurs between 0.3 and 1.0%. 5. The Cs layer was evaporated. The Cs evaporation was controlled in the same manner as that for the K evaporation. The QE was typically between 1 and 10%. 6. The substrate was cooled to room temperature.
An example of the QE evolution during the cathode evaporation is shown in Fig. 1 for the p-Si(100) substrates. Typically, QE on p-Si(100) was 9.8% at 532 nm laser. The laser spot size is 0.5 mm 2 and the cathode was biased at −100 V. The laser power was adjusted to 0.6 mW, and the typical photocurrent was 25 μA. The photocurrent was measured as the current of the bias supplier and the limit of photocurrent is 10 nA. The results (maximum QE) of all of substrates are summarized in Table I including the results of the previous studies. The evaporation was done on eight samples for each substrate type. The samples were heated to initialize the surface before each evaporation. We observed no photocurrent above the detection limit (10 nA) for the 400 nm laser after heating. The error was calculated as the standard deviation for the eight samples, indicating the high reproducibility of the evaporation process.
The QE of the Si(111) substrate was 1.6-2.8% at 532 nm. The p-Si(100) and p-GaAs(100) substrates showed a satisfactory QE as high as 10%. These results are similar to those of the Mo(100) [13] and Si(100) [14] substrates. The cathode performance of the p-type substrate was higher than that of n-type substrate with the same substrate surface orientation and material. This is the experimental evidence that the CsK 2 Sb cathode performance depends on the pn-type of the substrate. The difference of CsK 2 Sb cathode performance between Si  (100) and Si(111) substrates is caused by the dependence on the substrate surface orientation [26]. After evaporating CsK 2 Sb, we measured the spectral response of the photocathode as a function of the wavelength. A lamp (Asahi, High Power xenon lamp, MAX-303) and a monochromator (Asahi, Czerny-Turner type single monochromator, CMS-100) provided a variable wavelength light in the range of 400-900 nm with a typical bandwidth of 1%. Figures 2 and 3 show the spectral response of the CsK 2 Sb photo-cathode on various substrates. QEs larger than 20% are obtained on p-Si(100) and p-GaAs(100) at 400 nm. These results are similar to those on the p-type Si(100) substrate obtained by the preceding studies [29].
We also evaporated Cs 3 Sb on p-type Si(100), n-type Si (100), p-type GaAs(100), and n-type GaAs(100) substrates using the same process as CsK 2 Sb but only 5 nm Sb. The results (maximum QE) of all substrates are summarized in Table II including the results of the preceding studies. The evaporation was done on eight samples for each substrate type. The error was calculated as the standard deviation for the eight samples. The QE of all the substrates was approximately 5.0% at 532 nm. These results are consistent with those of the preceding studies [30]. We did not observe the dependence of cathode performance on the pn-types of the substrates for Cs 3 Sb.

III. DISCUSSION
The experimental results clearly show that the CsK 2 Sb cathode performance depends on the pn-types of the substrates but this is not the case for Cs 3 Sb. This is the first experimental evidence that photocathode performance can depend on the pn-types of the substrates.
CsK 2 Sb is a direct transition type with 1.2 eV bandgap and 0.7 eV electron affinity [31]. CsK 2 Sb is also a nearly intrinsic semiconductor [12]. CsK 2 Sb and the substrate make the semiconductor(metal)-semiconductor junction. Fig. 4 shows the energy-band models for the junctions between CsK 2 Sb and substrates, p-Si, n-Si, p-GaAs, n-GaAs and Mo. The vertical dashed line indicates the position of the substrate-cathode interface, and the area of the stripes shows the depletion region. By semiconductor (metal)-semiconductor junction, the energy band inside the cathode bends and the depletion region is formed. The conduction-band minimum of p-Si and p-GaAs are higher than that of CsK 2 Sb and the conduction-band minimum of n-Si and n-GaAs are lower than that of CsK 2 Sb.
In a previous x-ray reflectivity study [32], it was indicated that the CsK 2 Sb photo-cathode is thicker by about a factor of 5 to 6 than that of evaporated Sb as in the case of the Cs 2 Te cathode [33]. Therefore, the thickness of the photocathodes in this study may be around several tens FIG. 2. Spectral response for a CsK 2 Sb photo-cathode, with red circles, blue triangles, green squares and yellow rhombuses for the p-Si(100), n-S(100), p-Si(111) and n-Si(111) substrates, respectively. The measurement limit of photocurrent is 10 nA.
FIG. 3. Spectral response for a CsK 2 Sb photo-cathode, with red circles and blue triangles for the p-GaAs(100) and n-GaAs(100) substrates, respectively. The measurement limit of photo-current is 10 nA. 4-5 [30] of nanometers, which is larger than the photon penetration depth (about 20 nm). It is well known that the width of depletion region is typically several tens of nanometers even in heavily doped semiconductors and it is much wider in lightly doped or nearly intrinsic semiconductors [34]. We think that the CsK 2 Sb photocathode is in the depletion region almost up to the surface, because CsK 2 Sb is a nearly intrinsic semiconductor [12]. Due to the energy-band bending, the excited electrons in the depletion region cannot diffuse toward the substrates, p-Si and p-GaAs.
On the other hand, they can easily diffuse toward the substrates, n-Si and n-GaAs. In this way, we can qualitatively understand why the quantum efficiency of the CsK 2 Sb cathode on the n-type substrate is less than that on p-type substrate. Figure 4(c) shows that the metal-semiconductor junction is similar to that between CsK 2 Sb and p-type semiconductor substrates. The band bends upward on the CsK 2 Sb cathode side and the excited electrons cannot diffuse toward the Mo side. For this reason, the CsK 2 Sb cathode on the Mo substrate shows QE values as high as those on the p-type semiconductor substrates [13].
The cathode performance on n-Si(111) is smaller than that on other samples in this study because of the effect of the substrate surface orientation [26], in addition to the junction effect described above.
We found that the Cs 3 Sb cathode performance of the substrates does not show a significant difference. This result also can be explained by the semiconductorsemiconductor junction between Cs 3 Sb and the substrates. Cs 3 Sb is a direct transition type semiconductor with 1.6 eV bandgap and 0.45 eV electron affinity [35]. It is well known that Cs 3 Sb is a p-type semiconductor [10]. Figure 5 shows the energy-band models for the semiconductorsemiconductor junction between Cs 3 Sb and the substrates, p-Si, n-Si, p-GaAs and n-GaAs. The vertical dashed line indicates the position of the substrate-cathode interface, and the area of the stripes shows the depletion region. The conduction-band minimum of all substrates are lower than that of Cs 3 Sb. The excited electrons closing to the depletion region (or in the depletion region) diffuse toward the substrate side in all cases; thus, the quantum efficiency of the Cs 3 Sb cathode on all substrates is the same. The difference in the conduction-band minimum between Cs 3 Sb and the substrates do not make affect the QE.
These results can be explained in a unified manner by the energy-band models for semiconductor(metal)semiconductor junction between the cathode and the substrates. It indicates that, in order to obtain a high QE, the conduction-band minimum of the substrate should be higher than that of the cathode.

IV. SUMMARY
We studied the substrate dependence of the performance of the CsK 2 Sb photocathode by employing an evaporation process which gives high QE with high reproducibility. We found that the cathodes on p-GaAs(100), p-Si(100), and Mo(100) had significantly higher QE values than those on n-GaAs(100) and n-Si(100). We also found that Cs 3 Sb did not depend on the pn-type. These results clearly show that the cathode performance depends not only on the substrate material, surface state, and surface orientation, but also the pn-type. When the conduction-band minimum of the substrate is higher than that of the cathode, we can obtain high QE. This finding indicates a possibility of improving the thin-film cathode performance by choosing a suitable combination between the cathode and the substrate.