Development of a large nanocrystalline soft magnetic alloy core with high μ′pQf products for CSNS-II

Bin Wu· Xiao Li· Zhun Li · Chun-Lin Zhang · Yang Liu ·Wei Long · Xiang Li,3 · Jian Wu,3

Abstract A waterproof nanocrystalline soft magnetic alloy core with a size of O.D.850 mm × I.D.316 mm ×H.25 mm for radio frequency acceleration was successfully developed by winding 18 μm 1k107b MA ribbons. The products reached 7.5, 10, and 12 GHz at 1, 3, and 5 MHz, respectively. The products of the MA core(O.D.250 mm × I.D.100 mm × H.25 mm) manufactured using a 13 μm MA ribbon further increased by 30%.Detailed improvements on the MA core manufacture process are discussed herein. Continuous high-power tests on the new MA cores demonstrated its good performance of waterproofness, particularly its stability of high products. The MA core with high product and large size can operate under a high average RF power, high electric field,and in deionized water,which will be used in the China Spallation Neutron Source Phase II (CSNS-II).

Keywords Large size magnetic alloy core · High insulation and low stress coating · Waterproof structure ·Transverse magnetic field annealing · High power MAloaded cavity

1 Introduction and background

The nanocrystalline soft magnetic alloy (MA) has the characteristics of a high saturation flux density, high permeability,and high Curie temperature[1],enabling it to be widely used in common mode chokes, radio-frequency(RF) switching power supplies, and microwave absorption[2].In recent years,the MA core prepared by winding MA ribbons has been used as an inductance loading material in coaxial resonant cavities to achieve a high acceleration gradient and wide operating frequency band;it has become a key technology for high-power proton/heavy ion synchrotrons such as the Japanese Proton Accelerator (JPARC) and the Proton and Antiproton Research Device(FAIR) [3, 4]. Compared with the traditional coaxial resonant cavity loaded with ferrite cores, the MA-loaded cavity can have a smaller volume, wider operating band range, and higher acceleration gradient. The long-term stability of the entire RF system is significantly improved as a complex tuning system is not required for MA-loaded cavity.

An increase in the proton beam power from 100 to 500 kW is planned for the rapid cycle synchrotron (RCS)of the China Spallation Neutron Source Phase II Project(CSNS-II). Three MA-loaded cavities will be installed in the RCS to dilute the space charge effect of the proton beam and increase the beam capture efficiency [5].According to the beam dynamics design, the MA-loaded cavity must provide the fundamental and second harmonic voltages. The average power density of the MA cores will exceed 0.3 W/cm3,and the maximum acceleration gradient of the MA-loaded cavity will be up to 40 kV/m. A direct cooling method was adopted for MA cooling [6]. Table 1 shows the comparison of RF parameters of the MA loaded cavity in CSNS-II/RCS and J-PARC [3, 7].

However, when the MA-loaded cavity operates under a high electric field gradient and high average power density,local ignition, ablation, and deformation may occur in the MA core owing to the problems of interlayer insulation breakdown and increase in eddy current loss[8].Therefore,the MA core should be especially fabricated, such as treating the MA ribbon with an insulation layer,increasing MA core size and applying a magnetic field during annealing, and encapsulating the MA core for special cooling environments. However, these special treatments will increase the internal stress of MA materials, resulting in a decrease in the high-frequency characteristics of the MA core [9]. Currently, MA cores with high performance and large size that can operate under a high average power,a high electric field gradient,and in water are produced by Hitachi LTD, Japan [10], but the MA cores and core preparation technology are subject to a technical blockade and embargo owing to export control laws.

In this study, with respect to the application requirements of CSNS-II/RCS, the key technologies to prepare a large-size and high-performance MA core using domestic MA ribbon were studied. The second part of this paper introduces the research in detail, including the development of high insulation and low stress coating, large MA core winding and transverse magnetic field annealing, as well as the waterproof curing encapsulation of the MA core.The research results are evaluated and discussed.The third part describes the building of a high-power test system to evaluate the stability of the new MA cores under actual operating conditions for 31 days.

2Fabrication of high-Qf product and large MA cores

Table 1 Comparison of RF parameters of high power and high gradient MA cavities of CSNS-II/RCS and J-PARC

The key processes of a MA core manufacture are shown in Fig. 1a. In this study, an MA ribbon with a width of 25 mm and thickness of about 18 μm was used; the model was 1k107B with the composition of Fe84Cu1Nb5.5Si8.2B1.3(at%) and provided by Advanced Technology & Materials Co.,Ltd.(AT&M).To prevent the MA material from being oxidized during the annealing process, we added an inert gas to the annealing furnace for protection.

2.1 Insulation treatment of MA ribbon surface

Four types of coating method were studied. The advantages and disadvantages are summarized in Table 2.Only the SiO2coating prepared using the sol–gel process satisfied the requirement of low stress and high insulation used for high-performance MA core fabrication. Figure 1b shows the sol–gel coating facility and process. The thickness of the wet film was controlled by controlling the spacing of Group 1 rollers. Three small rollers were installed in the oven to prolong the standing time of the ribbon to ensure that the surface of the MA ribbon was completely dried before winding.

Figure 2a and b shows the cross-sectional observation of the sol–gel/SiO2coating on the MA ribbon before and after annealing, respectively, using a Hitachi S-3400 N highresolution scanning electron microscope.Before annealing,the sol–gel film was a coating with an irregular morphology,and the element proportions of Si and O were 50.33%and 37.07%, respectively. A dense SiO2inorganic coating was formed with an average thickness of about 2 μm after the annealing at 500–600 °C (Fig. 2b). The element proportions of Si and O were 50.85% and 35.99%, respectively, which were consistent with the element proportions before annealing, indicating that the sol–gel/SiO2coating had good thermal stability. An insulation tester (Fluke 1550C) was used to test the voltage hold-off on the annealed MA ribbon. The selected positions of the MA ribbon samples were located at the inner diameter,middle,and the outer diameter areas of the MA core, as shown by the symbols B1, B2, and B3, respectively, in Fig. 3a.Testing was conducted along the 2.5 cm width of the MA ribbon. The results of the inner, middle, and outer regions are indicated by the black, red, and blue scatters, respectively, in Fig. 2c. The symbols a1 and a3 represent the areas 5 mm away from the edge of the MA ribbon, and a2 represents the middle region. The average breakdown voltages of regions a1, a2, and a3 were 214, 253, and 278 Vdc, respectively. Figure 2d shows a histogram of the frequency statistics of all the results.The test data satisfied the Gaussian distribution, whose median and mean were 247 and 250.3 Vdc,respectively.The MA core used in the CSNS-II/RCS MA-loaded cavity has a lamination coefficient of 0.75. According to its definition, the induced RF voltage is about 2 V with the ribbon spacing of 6 μm[15].The test results indicated that the developed sol–gel/SiO2coating can satisfy the RF voltage insulation requirement of CSNS-II/RCS MA core and with a sufficient engineering margin.

Fig.1 (Color online) Key processes of manufacturing high-performance MA cores a; Sol–gel coating facility and process b

Table 2 Comparison of different coating methods for the insulating treatment of the MA core

Fig.2 (Color online) Cross sections of the sol–gel coating on the surface of the MA ribbon before and after annealing a and b; Ultimate breakdown voltage of the samples in the different regions of MA core c;Frequency statistics of all ultimate breakdown voltage measurements d

Fig. 3 (Color online) Appearance of the MA core with the size of O.D.850 mm × I.D.316 mm × H.25 mm after annealing a; Comparison of the measured and theoretical values of radial DC resistance of the MA core b

A thermal shock experiment was performed on the coated MA ribbon to evaluate the possibility of coating cracking in long-term engineering use, as the difference between the thermal expansion coefficients of SiO2and MA ribbon was large, which were about 3 × 10–6/K and 10.6 × 10–6/K, respectively [16]. The coated MA ribbon samples with a length of 5 cm were heated from room temperature to 120 °C within 3 min; the ribbon was maintained at 120 °C for 1 h, and then, it was cooled to room temperature within 3 min.This process was repeated 200 times. The results showed that the appearance of the MA ribbon coating was complete without apparent peeling off, and the insulation performance and energy-dispersive spectroscopy element ratio of the MA ribbon coating were almost the same as those before the thermal shock test,which further proved the excellent thermal stability and adhesion of the sol–gel/SiO2coating.

2.2 Large MA core winding

A horizontal winding process was adopted for a large and high-filling-factor MA core winding instead of the traditional vertical winding process, which causes significant damage to the MA ribbon coating[18].The weight of the MA core was evenly distributed on the support backplane, and the tension required for the MA core winding was significantly reduced. The ribbon tension was kept as low and constant as possible to ensure that the stress distribution of each layer of the MA core was uniform by dynamically adjusting the rotational speed of the unwinding and rewinding motors. In addition, we developed a multi-core lap joint process. A large MA core was composed of several small coated MA cores as the front end of the last roll was mounted on the back end of the first roll.

Figure 3a shows a MA core with a size of O.D.850 mm × I.D.316 mm × H.25 mm after winding.We measured the radial resistance of the MA core using a digital multimeter to evaluate whether the inter-layer insulating coating of the MA core was peeled off after winding [18]. To reduce the measurement error, we measured three path lines on the end face of the MA core at 120° intervals along the core circumferential direction, as shown by the three red dotted lines in Fig. 2a.The averagevalue of the three lines at each the 25-mm interval points were considered the final radial resistance measurement results. Figure 3b shows a comparison of the radial resistance distribution of the two MA cores denoted as‘‘850–1’’and ‘‘850–2’’ and their theoretical values, in which ‘‘Theoretic value’’ is the theoretical calculation value of the radial resistance in the assumption that the inter-layer insulation of the MA core is good. We observed that the measured results could be better distributed near the theoretical value, which demonstrated that the horizontal winding process has minimal influence on the coating of the MA ribbon [18]. We also observed that the results could completely coincide with the theoretical value. The main reason is that the thickness and resistivity of the MA ribbon fluctuated typically within the range of ± 5% during production, and a measurement error was also introduced.

Table 3 RF properties of MA cores with or without coating treatments

2.3 MA core annealing with a transverse magnetic field

Fig.4 (Color online) Comparison of the products of the developed and J-PARC MA cores

2.4 MA core waterproof package

An epoxy curing encapsulation process is often adopted to avoid the long-term erosion to the MA core by water[8].However, the cured epoxy resin has a large stress, which results in a large decrease in the high-frequency performance of the MA core. Through numerical simulation and experiments, Morita et al. confirmed that the epoxy resin infiltrating an MA core is what generates a strong stress in the circumferential direction under a high temperature[8, 16]. Therefore, we designed a new type of packaging structure that can effectively prevent the epoxy resin from infiltrating the interior of MA core. The packaging structure is described as follows: The inner and outer diameter of the MA core is supported by a fiberglass-reinforced tube with a thickness of about 3 mm. The first layer on the end face of the MA core is a coating containing large particles of SiO2with a thickness of about 50 μm to fill the tiny gap between the MA ribbons, preventing the subsequent penetration of epoxy into the interior of MA core. Additionally,it provides a good insulation on the MA core end face.The second layer is a cured layer formed by high viscosity epoxy resin and glass fiber cloth,and its main function is to preliminarily shape the MA core. The final layer is a leveling layer formed by the curing of low-concentration epoxy resin and fiberglass cloth; thus, the MA core has sufficient mechanical strength and is waterproof while smoothing the MA core surface. The low-concentration epoxy wraps the entire MA core during curing, and the inner and outer diameter corners of the MA core will also be sufficiently protected. The entire curing process was completed in a vacuum chamber. After curing, the total thickness of the encapsulation layer was less than 1 mm,the thermal conductivity was about 0.5 W/m/K, and the maximum temperature resistance of the encapsulation layer was about 150 °C. Figure 5a shows the appearance of the MA core after packaging.

Fig.5 (Color online)Appearance of the MA core after waterproof curing and encapsulation a.Comparison of the μ′pQf product of MA core in each process of packaging

3 MA core test platform and continuous high power testing

A schematic of the high-power test system of the MA core is shown in Fig. 6a.The test system included a singlegap MA test cavity, RF power source, a low-level RF control system (LLRF), and temperature-monitoring system. The two sides of the ceramic acceleration gap in the middle of the MA-loaded cavity were two water tanks.Each tank could simultaneously cool three MA cores with the size of O.D.850 mm × I.D.316 mm × H.25 mm.Owing to the large size of the MA core, a deflector was installed between the MA cores to reduce backflow. The RF power required by the cavity was directly fed using an RF power source composed of two TH558 high-power tetrode tubes connected to the metal rings on both sides of the ceramic through the feeding copper bar. The operating mode of the two RF power sources was PUSH–PULL,which meant that the voltages ～V1and ～V2on both sides of the gap had the same amplitude, but the phase difference was 180°. The amplitude and phase were closed-loop controlled using the LLRF. The two RF power sources would observe the impedance of the respective half-cavities; the entire test system was equivalent to the parallel equivalent circuit in Fig. 6b. The two power sources were equivalent to current sources Ig1and Ig2, the distributed capacitance including the cavity and water is Cp.Lpand Rpare the parallel equivalent inductance and shunt impedance of the MA core, n is the number of half-cavity used MA cores, and Cgapis the capacitance of the ceramic gap,which was about 34 pF.

Fig. 6 (Color online) Schematic of the MA core high power test system a; Equivalent circuit of MA cavity test system b; The amplitudes of the fundamental, second, and third harmonics are represented by black,red,and blue solid lines,respectively,the cavity operating frequency variation is represented by red scatters, and the amplitudes correspond to the right red axis in c

Fig.7 Comparison of the Qf product of MA cores before and after a 31-day high-RF power test with an average power density of 0.33 W/cm3 and maximum accelerating gradient of 40 kV/m. The

The high-power test of the MA core under the average power density of 0.33 W/cm3was performed using the second-harmonic voltage sweeping mode used for CSNSII/RCS. Figure 6c shows the voltage amplitudes measured with the LLRF on one side of the ceramic gap. The frequency range was from 1.5 to 2.6 MHz,and the duty cycle was 15% under the condition of a 50 Hz repetition frequency. Because the TH558 high-power tetrode tubes operated in the AB1 class mode,the gap voltage contained the components of higher harmonics. Through fast Fourier analysis,we observed that the maximum voltage amplitude of the fundamental component on each tank reached 12 kV, the maximum voltage amplitude of the second and third harmonic components reached 3 and 0.5 kV,respectively, as shown by red and blue solid lines in Fig. 6c.The maximum fundamental voltage synthesized by the gap was 24 kV. The cavity acceleration gradient was 40 kV/m with a cavity length of 0.6 m.Owing to the low Q value of the MA cores, the power loss of the cavity was primarily concentrated in the MA core of the tank;thus,the power loss of the MA cores could be calculated by measuring the water temperature difference between the outlet and inlet of the tank and water flow rate. According to the theory of coaxial resonant cavity, the MA cavity length was much smaller than the wavelength of the cavity operating frequency(about 20 m at 1 MHz); therefore, the power losses of each MA core in the tank were almost equal.The thermal power calculation formula of the tank is shown in Eq. (4) [15], where ΔT is the water temperature difference between the inlet and outlet of the tank. The temperature probes were high-precision PT100 thermocouples with a measurement error of ± 0.001 °C; L is the water flow of the cavity, taken as L = 40 L/min; ρ is the water density, taken as 997 kg/m3at 30 °C; C is the specific heat capacity of water, taken as 4181.7 J/kg/K at 30 °C.MA cores 850–1#, 850–2#, and 850–3# were located in Tank 1 and the remainder were in Tank 2

4 Conclusion

Author contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Bin Wu, Chun-Lin Zhang, Yang Liu, Xiang Li,Jian Wu and Zhun Li.The first draft of the manuscript was written by Bin Wu and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.