Quality Factor Enhancement of 650 MHz Superconducting Radio-Frequency Cavity for CEPC

Abstract

Medium-temperature (mid-T) furnace baking was conducted at 650 MHz superconducting radio-frequency (SRF) cavity for circular electron positron collider (CEPC), which enhanced the cavity unloaded quality factor (Q₀) significantly. In the vertical test (2.0 K), Q₀ of 650 MHz cavity reached 6.4 × 10¹⁰ at 30 MV/m, which is remarkably high at this unexplored frequency. Additionally, the cavity quenched at 31.2 MV/m finally. There was no anti-Q-slope behavior after mid-T furnace baking, which is characteristic of 1.3 GHz cavities. The microwave surface resistance (R_S) was also studied, which indicated both very low Bardeen–Cooper–Schrieffer (BCS) and residual resistance. The recipe of cavity process in this paper is simplified and easy to duplicate, which may benefit the SRF community.

1. Introduction

SRF cavities made of niobium are broadly adopted for the modern accelerators, which have demonstrated much higher Q₀ and lower R_S than cavity made of copper. Here, R_S is the electrical resistance of surface layer to current, and inversely proportional to Q₀ of SRF cavity. Recently, the technology of medium-temperature (mid-T, 250–400 °C) baking is used to increase Q₀ of 1.3 GHz cavities successfully, which dissolves the oxides in the surfaces of niobium [1,2,3]. It is more convenient and easier than nitrogen doping/infusion [4,5,6]. At the Fermi National Accelerator Laboratory (FNAL), the 1.3 GHz single-cell cavity was evacuated with pump and received mid-T baking at 300 °C in an oven. The highest Q₀ of cavities mid-T baked exceeded 5 × 10¹⁰ (at 2.0 K). The mid-T furnace baking of 1.3 GHz single-cell cavities also succeed at High Energy Accelerator Research Organization (KEK), which was conducted in a vacuum furnace. The Institute of High Energy Physics, Chinese Academy of Sciences (IHEP) also successfully developed mid-T furnace baking of 1.3 GHz single-cell and nine-cell cavities.

For low-frequency (<1 GHz) elliptical cavities used in colliders and proton/neutron sources [7,8,9], Q₀ usually decreases more quickly along with accelerating gradient than that of 1.3 GHz cavities. Nitrogen doping could increase Q₀ of 650 MHz single-cell β = 0.9 cavity for the Proton Improvement Plan II (PIP-II) project to 7 × 10¹⁰ at medium fields of approximately 15–25 MV/m [7]. However, there are very few mid-T baking attempts of low-frequency cavities, which may demonstrate differences with 1.3 GHz as well as nitrogen doping [7,10]. Thus, the mid-T furnace baking of 650 MHz cavity for CEPC was carried out at IHEP, which obtained high Q₀ and low R_S successfully.

2. The Process of Mid-T Furnace Baking

The cavity investigated in this paper is a 650 MHz single-cell β = 1 SRF cavity for CEPC, which is made of fine-grain niobium with high residual-resistivity-ratio (RRR > 300). The ratio of the peak surface magnetic field (B_peak) to accelerating gradient (E_acc) is approximately 4.2. Firstly, the 650 MHz cavity received bulk electro-polishing (EP) process with a total removal of ~315 μm, where the temperature was below 25 °C all the time. There was no high-temperature annealing for hydrogen outgassing before or after EP, because we wanted to simplify the recipe of mid-T furnace baking, which combines high-temperature annealing and mid-T furnace baking at a later stage. Therefore, the 650 MHz cavity received high-pressure rinsing (HPR), assembly, low-temperature baking (120 °C for 48 h) and vertical test directly after bulk EP [1]. In the vertical test of EP baseline (blue icon), the 650 MHz cavity quenched at a very high gradient of 37.5 MV/m (B_peak = 158 mT) with Q₀ of 1.4 × 10¹⁰ in Figure 1, which indicated smooth and defect-free surfaces of the cavity. However, the Q value was slightly lower than the CEPC vertical test (VT) spec of 4.0 × 10¹⁰ at 22 MV/m. The behavior of Q-slope may result from hydrogen diffused in the cavity during EP process. The radiation began to increase at 33 MV/m, so field emission was low during the vertical test. R_S of the 650 MHz cavity (blue icon) increased exponentially along with the accelerating gradient in Figure 2, which is typical for cavity EP processed [7,9]. Hence, the cavity was considered qualified to receive mid-T furnace baking.

Afterwards, the 650 MHz cavity was subjected to high-temperature annealing at 900 °C for 3 h (hydrogen degassing) [11,12], and the following light EP was cancelled. Then, the cavity was exposed to air for 3 days in cleanroom (oxidation). Finally, it received mid-T furnace baking at 300 °C for 3 h [13] as well as 1.3 GHz nine-cell cavity [3], which is shown in Figure 3. During the entire process, both cavity flanges were covered with niobium foils, which had been chemically polished, to keep the cavity inner surface from dust. The process of mid-T furnace baking was key to improve Q₀ of 650 MHz cavity. Prior to mid-T furnace baking, the furnace was pumped adequately to obtain ultimate vacuum (<5 × 10⁻⁶ Pa). The highest pressure of furnace vacuum was about 3.5 × 10⁻⁵ Pa during the process of mid-T furnace baking (Figure 4), which was maintained with two cryopumps. After cooling down, the vacuum pressure was about 2.5 × 10⁻⁶ Pa, and the cavity was taken out.

3. Vertical Test after Mid-T Furnace Baking

Following mid-T furnace baking, the 650 MHz cavity was subjected to HPR and assembled with flanges. Then, it received a vertical test immediately, while the usual low-temperature baking was cancelled. The cavity demonstrated significantly high Q₀ at all fields during the vertical test, which is shown in Figure 1. Extremely high Q₀ was achieved at low fields, which exceeded 1.0 × 10¹¹. The cavity also exceeded the CEPC vertical test spec of 4.0 × 10¹⁰ at 22 MV/m. The Q₀ reached 6.3 × 10¹⁰ at 31 MV/m, which was much higher than 2.1 × 10¹⁰ of EP baseline. There was no field emission during the test, and the cavity quenched at 31.2 MV/m (B_peak = 131 mT) finally. Furthermore, there was no anti-Q-slope behavior, which is usual for 1.3 GHz cavities after mid-T baking [1,2,3].

The surface resistance after mid-T furnace baking is depicted in Figure 2, which was remarkably lower than EP baseline. It was about 2.5 nΩ at low fields and 4.5 nΩ at high fields, which was significantly low and increased slowly as a function of gradient.

The 650 MHz cavity also received vertical tests at lower temperatures (1.5–2.0 K) to investigate the composition of R_S, which consists of BCS resistance (R_BCS) and residual resistance (R_res). The value of R_BCS increases exponentially as a function of temperature. However, R_res remains consistent when the temperature increase from 1.5 to 2.0 K [14].

$$R_{\mathrm{S}}=R_{\mathrm{BCS}}+R_{\mathrm{res}}$$

(1)

$$R_{\mathrm{BCS}}=A\frac{f^2}{T}\exp \left(-\frac{\Delta \left(T\right)}{kT}\right)$$

(2)

Here, k is the Boltzmann constant, T is the temperature, f is the frequency of SRF cavity, 2Δ is the energy gap of niobium. A is the coefficient that depends on the superconducting parameters of niobium, such as mean free path, coherence length and London penetration.

R_BCS at 2.0 K (Figure 5) and R_res (Figure 6) of 650 MHz cavity were calculated according to Equations (1) and (2), and compared with 1.3 GHz cavity, which adopted the same recipe of mid-T furnace baking. R_res remained <2 nΩ at both low and high fields in Figure 6, which is rather low at this frequency. It was less than 1 nΩ at low fields. R_BCS was 2 nΩ constantly at low fields, and slightly increased as a function of accelerating gradient above 20 MV/m in Figure 5. The decrease in R_BCS as a function of accelerating gradient for 1.3 GHz cavity [1,2,3] did not appear for 650 MHz cavity. It confirms that mid-T furnace baking could not reverse R_BCS at the frequency of 650 MHz. Therefore, the anti-Q-slope behavior cannot be observed at 650 MHz cavity mid-T furnace baked. Similarly, nitrogen doping could not reverse R_BCS of 650 MHz cavity, either [7,15].

4. Conclusions

Mid-T furnace baking was attempted at 650 MHz single-cell cavity, which achieved a state-of-the-art high Q₀ at the accelerating gradient of 30 MV/m. Extremely low R_S of 3.4 nΩ at 25 MV/m was demonstrated, while the typical value of R_S is 5~10 nΩ.

Nitrogen doping is another kind of processing for SRF cavities, which can also improve Q₀. It is also carried out in a vacuum furnace along with the injection of nitrogen gas by thermal diffusion, which is followed by light EP to remove the niobium nitrides inside cavity surface [16]. Compared with nitrogen doping, mid-T furnace baking is much easier and more convenient. These studies can help to enhance the performance and comprehension of large elliptical (<1 GHz) superconducting cavities, which are broadly used by synchrotron light sources, proton/neutron sources and high-energy colliders.