Introduction

High Q₀ (quality factor) cryomodules equipped with superconducting radio-frequency (SRF) cavities are key components of modern accelerators, such as high-repetition X-ray FEL facilities [1], high-power extreme ultraviolet lithography light sources [2], and other future high-duty factor colliders [3]. At the Shanghai High Repetition Rate XFEL and Extreme Light Facility (SHINE) [4], more than 50 high Q₀ cryomodules operating in 1.3 GHz continuous wave (CW) mode will be installed to generate an 8 GeV electron beam. In contrast, extreme ultraviolet lithography light sources require approximately 10 high-Q₀ cryomodules operating in 1.3 GHz CW mode for a 1 GeV energy recovery linac. Generally, a high Q₀ cryomodule includes eight 9-cell TESLA cavities [5], eight fundamental power couplers (FPC), one superconducting quadrupole magnet package, and one cold beam-position monitor. The design and advancements of 1.3 GHz cryomodules are primarily attributed to the R&D efforts undertaken in large-scale facilities like the TESLA Test Facility [6], the European XFEL [7], the Linac Coherent Light Source 2 (LCLS-2) [8], LCLS-2 HE [9] and SHINE.

Currently, nitrogen doping (N-doping) [10-14] and medium-temperature (mid-T) baking [15-17] are the two main methods used to enhance the Q₀ values of SRF cavities made of high-purity niobium. N-doping incorporates nitrogen atoms as interstitial impurities into the niobium lattice, lowering the mean free path of the RF penetration layer of niobium, hence the BCS resistance, and reducing the residual resistance [18]. Fermilab and Jefferson Lab developed cryomodules equipped with SRF cavities treated by 2/6 N-doping recipes for LCLS-2 [19, 20]. In 2023, 35 cryomodules with 2/6 N-doped cavities were commissioned for the LCLS-2, demonstrating Q₀ of 2.8×10¹⁰ at an average accelerating gradient of 16 MV/m in the CW operation of a superconducting linac [3]. The LCLS-2-HE cryomodule with a 2/0 N-doping recipe achieved a maximum acceleration voltage of 208 MV in CW mode, corresponding to an average accelerating gradient of 25.1 MV/m, and Q₀ of 3.0×10¹⁰ at a gradient of 21 MV/m [21]. Mid-T baking is a novel and simplified high-Q₀ recipe that yields results similar to those of N-doping, while preventing the formation of NbN precipitates, which act as defects, and reducing the risk of contamination. The Institute of High Energy Physics (IHEP) developed the first high-Q₀ cryomodule equipped with eight mid-T-baked cavities, achieving Q₀ of 3.8×10¹⁰ at 16 MV/m and Q₀ of 3.6×10¹⁰ at 21 MV/m in horizontal testing [22].

At the Shanghai Advanced Research Institute (SARI), we conduct experimental cavity treatments using facilities for SRF cavity surface treatments on a platform located in Wuxi, China [23]. Both N-doped and mid-T baking recipes have been studied, achieving high accelerating gradients exceeding 25 MV/m on 1.3 GHz 9-cell cavities [24]. Since 2020, several sets of cryomodules with high-Q₀ cavities have been assembled and tested [25]. In this letter, we report the first 1.3 GHz high Q₀ cryomodule dedicated to CW operation up to 1 mA, developed at SARI. It is equipped with eight mid-T-baked cavities and eight 30 kW FPCs [26, 27], demonstrating world-leading, ultrahigh Q₀ and ultrahigh accelerating gradient performance in CW horizontal testing.

Mid-T baked cavities

The eight cavities were mechanically fabricated by the HE-Racing Technology Company in Beijing, treated by the SHINE cavity surface treatment facilities in Wuxi [23], and tested at the SARI and IHEP vertical test (VT) stands. These cavities underwent 200 μm electropolishing, 3 h of 900 ℃ high-temperature baking, exposure to air, and 3 h of furnace baking at 300 ℃ [24].

Figure 1 shows the vertical test results of the eight dressed cavities treated with the mid-T-baked recipe, where Q₀ is corrected by the 0.8 nΩ loss of the stainless-steel flanges. The sudden drop of Q₀ at around 16 MV/m for FV007-J, 23 MV/m for FV006-J, and 25 MV/m for GY002-J is attributed to flux trapping after soft quenches caused by multipacting [28], which can be recovered after warming up and performing a fast cooldown. For a direct comparison between the vertical and horizontal tests, the Q₀ drop was compensated for by the gap. The average Q₀ was 4.0×10¹⁰ at 20 MV/m, and the average maximum gradient was 29.4 MV/m. Three cavities exhibiting field emission during the vertical test were cleaned by long, high-pressure rinsing with two rounds of six turns before delivery to the cryomodule assembly. Although no further vertical tests were performed on these three cavities due to the tight schedule, as shown in the horizontal test results, all field emissions were eliminated.

Fig. 1Full-screen View

(Color online) Vertical test results of the eight dressed cavities. The solid triangle represents the Q₀ value, while the hollow triangle represents the radiation dose. The Q₀ values are corrected by 0.8 nΩ loss of stainless-steel flanges. The three cavities with significant radiation dose were cleaned again by high-pressure rinsing before delivery to the cryomodule assembly, without additional vertical testing

Cryomodule assembly

The cryomodule assembly includes clean assembly of the cavity string in a class 10 cleanroom, the cold mass, and the final assembly, typically completed within approximately two months. Before string assembly, all cavities underwent an outer surface rinsing process, after which they entered the cleanroom for more thorough cleaning, including wiping the outer surface, particularly the helium vessel bellows. The surface was blown with clean nitrogen gas to verify contamination levels met class 10 cleanroom standards. The assembly process began with the installation of fundamental power couplers, followed by intercavity bellows from upstream to downstream. Figure 2 shows the cavity-string assembly of the SARI cryomodule (CM) in a class 10 cleanroom.

Fig. 2Full-screen View

(Color online) Cavity string assembly in class 10 cleanroom

During assembly, nitrogen gas was vented into the cavities at 1 slm to maintain slightly positive pressure, preventing contamination from entering. Once the string assembly was completed, positive-pressure leak detection was conducted to identify major leaks. The cavity string was then evacuated into a vacuum, followed by a second leak test under vacuum conditions. Residual gas analysis assessed the cleanliness of the entire cavity string. After completing these tests, the cavity string was backfilled with nitrogen to a pressure of 1050 mbar, slightly higher than atmospheric pressure, for protection, and subsequently transported out of the cleanroom for further assembly.

Degaussing and cooling down

The cryostat was first degaussed in an east-west direction at SARI. During cryomodule assembly, two flux gates were mounted along the beam direction at the cavity 1# and 5# slots between the two layers of magnetic shields outside the cavity helium vessel. After the cryomodule was installed on the horizontal test stand, the entire cryomodule was degassed at room temperature, as shown in Fig. 3a. The magnetic field falls from 2.6 mGs to 0.3 mGs and from 0.9 mGs to 0.7 mGs after degaussing for the cavities 1# and 5#, respectively, as shown in Fig. 3b.

Fig. 3Full-screen View

(Color online) (a) The coils wound for cryomodule degaussing in the horizontal test stand; (b) In-situ cryomodule degaussing at room temperature before cooling down. The two flux gates are mounted at cavity 1# and 5# slots, between the two layers of magnetic shields outside the cavity helium vessel, as shown in the upper-right inset image

The SARI cryomodule cooling process, from 300 K to 2 K, follows a procedure similar to that of the European XFEL [29] and LCLS-2 [30], typically lasting for 3~4 days. From 300 K to 45 K, the average cooling rate has been controlled at around 6 K/h, followed by a “stand-by mode” lasting for about 12 hours to allow the entire cryomodule cold mass to stabilize at the 45 K temperature level. Fast cooling from 45 K to 4.5 K is required for magnetic flux exclusion in high-Q cavities, improving RF performance [31]. Once liquid helium accumulates along the cavity string and in the two-phase pipe, most of the cold mass is cooled to approximately 4.5 K. Final cooling is achieved by depressurizing the saturated helium vapor from around 1.2 bar to 31 mbar, reaching the operating condition at 2 K superfluid helium. Figure 4 shows the first cooling step of the standard SARI cryomodule.

Fig. 4Full-screen View

(Color online) Cooling curves of the SARI cryomodule from 300 K to 2 K

A key method to achieve high Q₀ performance at 2 K is to maintain an instant liquid-helium mass flow rate, creating a sufficient thermal gradient for magnetic flux expulsion as the cavities pass through the superconducting transition point (T_c) of approximately 9.2 K [31, 32]. Before the Q₀ test, the cryomodule was warmed up to 45 K and then quickly cooled again to release the magnetic flux trapped during previous quenches, which reached their maximum gradient or were induced by multipacting. For the SARI cryomodule, a maximum flow rate of 41.6 g/s was achieved when the cavities passed through the T_c. Figure 5 shows the thermal difference between the top and bottom of the dressed cavities during the horizontal test, where thermal sensors are mounted outside dressed cavities 1#, 5# and 8#.

Fig. 5Full-screen View

(Color online) Fast cooling curves for SARI cryomodule from 45 K to 4.5K

Q₀ performance

The mass flow rate method was employed to measure the static and dynamic 2 K heat load of the cryomodule [33-35]. Figure 6 shows the calibration results of the heat load via heaters for the SARI CM. The CM was first injected with liquid helium above half of the two-phase pipe and then evaporated in a semi-closed system (i.e., a cavity string with supply valves closed and the helium-gas-return pipe valve opened) under different heat load conditions ranging from 0 to 90 W. According to the law of conservation of energy, when dynamic balance is achieved, the helium evaporating mass flow rate, along with its latent heat (~23 J/g at 2 K) takes the same energy deposited on the 2 K cold mass from the outside heat load. With the above relationship between heat load and evaporating mass flow rate, a linear fit can be obtained based on the thermal heater power and flowmeter measurements. Consequently, the absolute value of the intercept point at “zero” mass flow rate gives the 2 K static heat load of the SARI cryomodule around 21.7 W, as shown in Fig. 6.

Fig. 6Full-screen View

(Color online) Static heat load measurements for the SARI cryomodule at 2 K: the overlap of circles at 20, 30, and 80 W serves to verify the data's repeatability and reliability

The dynamic heat load of each cavity was measured by subtracting the heat load of three cavities from four cavities to improve accuracy. The estimated measurement uncertainty of Q₀ was less than 10%, and the accelerating gradient was less than 5%.

To directly compare Q₀ between vertical and horizontal tests, Q₀ values with a sudden drop in three cavities were compensated for by the gap at the drop gradient. Fig. 7 shows a comparison of Q₀ at 20 MV/m for the eight dressed cavities in both vertical and horizontal tests.

Fig. 7Full-screen View

(Color online) Q₀ measured at 20 MV/m for each cavity in both vertical and horizontal tests. To allow direct comparison, the Q₀ values of cavities 3#, 5#, and 6# were adjusted to account for the Q₀ gap in vertical tests due to flux trapping after soft quenches

The average Q₀ values at 166 MV and higher voltages in CW mode were also measured for the SARI CM. The total 2 K heat load was 104.9 W at a total voltage of 166.1 MV with all eight cavities at 20 MV/m, 194.9 W at 223.8 MV, and 236.4 W at 241.3 MV, corresponding to an average Q₀ of 4.0×10¹⁰ at a gradient of 20 MV/m, 3.4×10¹⁰ at 27 MV/m, and 3.2×10¹⁰ at 29 MV/m, respectively. Figure 8 provides the measurement points of the SARI CM and compares them with other top cryomodules [9, 21, 22].

Fig. 8Full-screen View

(Color online) Average Q₀ at 20, 27, and 29 MV/m for the SARI cryomodule in the horizontal test, compared with other top-performing cryomodules [21, 22] in the world. The administration limit for LCLS-2 HE cryomodule testing is 26 MV/m for each cavity, corresponding to a maximum CM voltage of 216 MV [9]

Gradient performance

Figure 9 shows the maximum accelerating gradients for the eight dressed cavities measured in CW mode during vertical tests and in the cryomodule, where the maximum gradient is defined as stable operation for at least one minute. The usable gradient in the cryomodule is defined as meeting the conditions of being 0.5 MV/m less than the quench field, stable operation for one hour, and radiation dose less than 500 μSv/h measured by the G-M tube radiation detectors placed around 2 m from the cryomodule in the horizontal test stand. As can be seen, all cavities approach full-gradient performance except for cavity 5#, which is limited by HOM heating, as described below.

Fig. 9Full-screen View

(Color online) Accelerating gradient measured for each jacketed-cavity in vertical and horizontal tests

To accelerate a beam, long-term stable operation at the working gradient is essential for the CM. During operation, RF power can heat the FPC antennas, which are cooled through a ceramic window and CF100 flange by a 45 K intercept, a conduction-cooling braid connected to the 45 K helium gas pipes, eventually reaching thermal equilibrium. For the SARI CM, the temperature of each CF100 flange was monitored using thermal sensors. A stable operation test was performed at 220 MV for the SARI CM. The external Q values of the eight cavities were adjusted to their optimal values of approximately 6.1×10⁷ for a working gradient of 20 MV/m. Due to the cryogenic limit of the horizontal test stand, the eight cavities were split into two groups for stable operation testing. Each cavity operated at 27.3 MV/m, except for cavity 5# at 21.7 MV/m. The group with the first four cavities maintained a total voltage of 113 MV, while the other group maintained 108 MV. To reduce test time, we detuned the phase of the self-excited loop to increase reflected power, thus heating the main coupler to its threshold of approximately 150 K. Afterward, we tuned it back and awaited thermal equilibrium, or maintained it for 10 hours. Figure 10 shows the temperature behavior of the eight FPC CF100 flanges during the 220 MV stable operation test. As shown, all the temperatures of the eight CF100 flanges decreased, reaching quasi-equilibrium, with a maximum temperature of less than 120 K, below the 150 K threshold. This confirms the stable operation capacity at 220 MV for this CM.

Fig. 10Full-screen View

(Color online) Long-term stable operation test at 220 MV for SARI CM, monitoring the temperature of the FPC CF100 flanges connected to the 45 K intercept. The test was performed in two groups, with four cavities each. The phases of the self-excited loop were detuned to speed up the FPC heating to a maximum of around 150 K for FPC1, then tuned back, as shown by the turning points marked by the red dashed lines

Table 1 summarizes the CW mode performances of the eight mid-T-baked cavities in VT and CM. The Q₀ of the entire CM measurement was 4.0×10¹⁰ at 166 MV, with all eight cavities operating simultaneously at 20 MV/m. The slight difference in the average Q₀ between the individual cavities and the entire CM measurement was likely due to the uncertainty of the small heat load, as shown in Fig. 6.

It is worth mentioning that the cryomodule was also tested in pulsed mode with a repetition rate of 0.5 Hz and a 10% duty factor, where the total accelerating voltage reached 247.6 MV. In this mode, cavity 5# was powered up to 29.2 MV/m, the same as its maximum gradient in the vertical test. Additionally, no detectable field emissions were observed during any of the cryomodule tests.

Limitation factor

When comparing the RF performances of the eight cavities in the vertical and horizontal tests, cavity 5# exhibited a significant decrease in the accelerating gradient. The maximum CW operating gradient for cavity 5# was limited to approximately 25.2 MV/m by quenching, accompanied by a rapid temperature increase at the FPC-side HOM coupler, where the thermal sensor was mounted on the copper clamp at the copper sleeve of the HOM feedthrough, as shown in Fig. 11.

Fig. 11Full-screen View

(Color online) A thermal sensor is mounted on the conducted-cooling clamp of a HOM-feedthrough copper sleeve at the cavity 5# FPC-side to monitor HOM heating

To investigate the limiting factor, we tested cavity 5# in pulsed mode. With a repetition rate of 0.5 Hz and a duty cycle of 5%, we began increasing the input power to the cavity, which showed that the cavity could stably operate at 29.2 MV/m, similar to the maximum gradient in the vertical test, with a slight temperature increase at the FPC-side HOM. We then gradually increased the duty cycle in steps of 5%, while maintaining a gradient of 29.2 MV/m, and a positive correlation between the FPC-side HOM temperature and duty cycle was observed. The cavity could stably operate at 29.2 MV/m with a duty cycle of up to 40%, where the HOM temperature approached approximately 40 ℃, but quenched quickly once the duty cycle was increased to 45%. Figure 12 shows the relationship between the HOM temperature and the duty cycle for cavity 5#. It is important to note that no field emissions were observed during these measurements. Therefore, we concluded that the limitation of cavity 5# gradient is due to overheating of the HOM antenna.

Fig. 12Full-screen View

(Color online) Accelerating gradient of cavity 5# limited by FPC-side HOM heating, where the thermal sensors are mounted on the cooper sleeve of HOM feedthroughs

Conclusion

A high Q₀ cryomodule equipped with eight mid-T-baked 1.3 GHz 9-cell cavities was assembled and tested at SARI. This cryomodule achieved an ultra-high average Q₀ at the operating gradients and an unprecedented total accelerating voltage in CW mode. The cryomodule's average Q₀ was 4.0×10¹⁰ at 20 MV/m and 3.2×10¹⁰ at 29 MV/m in the horizontal test, which corresponds to a maximum CW RF voltage of approximately 241 MV. The RF performance of the cavities was well-maintained from the vertical test to the horizontal test. Furthermore, no field emissions were observed in any of the eight cavities in the cryomodule. The successful development of this ultra-high Q₀ and ultra-high gradient cryomodule demonstrates the techniques mastered from components to a completed cryomodule, marking an important milestone for CW accelerator projects, such as high-repetition X-ray FEL facilities, high-power extreme ultraviolet lithography light sources, and other future high-duty factor colliders.