In-depth resistance analysis of REBCO tape joints with indium insert and solders

Joints between REBCO (rare-earth barium copper oxide) tapes with low joint resistance are crucial for many superconducting applications. Joining REBCO tapes with indium insert (In-joint) is a promising joining method to fabricate low resistive joints at low temperatures (20–120°C). This study investigated the joining conditions of In-joints such as pickling, surface roughness, joining time, and temperature. The joint resistivity (product of joint resistance and joint area) was successfully reduced to 22–30 nΩcm2 at 77 K in self-field. The constitutive factors of the joint resistivity were analysed separately along with the crosssectional observations. In this study, the interface resistivity of the REBCO tape was measured as 8.5 nΩcm2 for one REBCO tape by the previously proposed method. The resistivity of the joining interface Cu/In was calculated as <3 nΩcm2 by subtracting the other resistivities from the entire joint resistivity. This result reveals the lower limit of the joint resistivity: the sum of the resistivity (nΩcm2) of indium (measurable by thickness), the resistivity of Cu/In (<3nΩcm2), and the interface resistivity of the REBCO tape (measurable beforehand). Furthermore, we demonstrated a lower and less varied joint resistivity of In-joints than those of the soldered joints.


Introduction
Joints between REBCO tapes with low joint resistance are crucial for various superconducting applications such as power cables for railway systems [1] and fusion magnets [2]. For years, the authors of Tohoku University have been developing the joining method with indium insert (In-joint) [2][3][4]. The joint resistivity (product of joint resistance and joint area) of In-joints ranges 25-60 ncm 2 [4] at 77 K in self-field, which is smaller than those of the soldered joints (30-1000 ncm 2 [5,6]). In order to know how the joint resistivity can be reduced, the resistance purely due to joining has to be evaluated because the joint resistivity itself depends on the interface resistivity inherent to REBCO tapes [5]. Therefore, in this study, we separately evaluate the resistance factors of the joint resistivity, including the resistivity at the joining interface and the interface resistivity of the REBCO tape. The objective of this study is to further reduce the joint resistivity of In-joints by investigating the joining conditions such as pickling, surface roughness, joining time, and temperature. The resistance factors of the joint resistivity were separately evaluated for a reason mentioned above. The crosssections of the In-joints were also observed to discuss what reduced the joint resistivity. Furthermore, the joint resistivity of In-joints was compared with those of soldered (In52Sn48 and Pb37Sn63) joints to demonstrate the excellent performance of In-joints.

Investigation of joining conditions
In this study, copper-stabilized 4-mm-wide YBCO tapes (SCS4050-AP, Ic>91 A at 77 K in self-field, SuperPower Inc., Schenectady, NY, USA) and 100-µm-thick indium foil were used for the In-joints. The tape has a layer structure of Cu (20 µm) / Ag (2 µm) / buffer layer (~1 µm) / YBCO (1 µm) / Hastelloy-C276 (50 µm) / Ag (2 µm) / Cu (20 µm). In-joint were fabricated in the procedure shown in Figure 1(a). The tapes were set in a face to face manner with indium insert in a jig and joined by using a pressing machine shown in Figure 1(b). The joining pressure was fixed at ~90 MPa because the joint resistivity does not change with the applied pressure over 50 MPa [4]. All the samples were joined with overlapped lengths around 7 mm. We conducted three experiments (Ex. 1-3) to find useful parameters for resistance reduction. Injoints were fabricated with different joining conditions in the experiments. The conditions of pickling, surface roughness/joining time, and joining temperature were changed in Ex. 1, 2, and 3, respectively. Table 1 describes the details of the joining conditions in Ex. 1-3. For pickling, a commercial flux (SUSSOL-F, Hakko corp., Osaka, Japan) containing ZnCl (35-45%) and NH3Cl (<10%) was used for the Cu surface of the REBCO tapes, and 10% HCl was used for indium foil.
After joining the tapes, the joint resistance was measured at 77 K in self-field by the four-probe method with a current up to 100 A, and the critical current was confirmed not degraded. Additionally, three In-joint samples fabricated in Ex. 2 were heated at 95C for three or four days in an electric furnace with an applied pressure of 29 MPa, and the joint resistivity before and after heating was measured.

Cross-sectional observation and joint resistivity evaluation
The cross-sectional of the In-joint fabricated by the previous joining process [4] (without pickling) and the sample fabricated in Ex. 2 (with pickling) were observed with the scanning electron microscope (SEM) and energy dispersive X-ray spectroscopy (EDS). The samples for observation were prepared by cutting and milling with an Ar ion beam at 90C or 100C to avoid the deformation of soft indium. The joint resistivity rjoint (the product of joint resistance and joining area) was separately evaluated from each resistance factor, as shown in the following equation: where rCu and rAg are the resistivities (ncm 2 ) of Cu and Ag layers of a REBCO tape. rCu/Ag and rAg/REBCO are the interface resistivity, rIn is the resistivity (ncm 2 ) of In, and rCu/In is the resistivity at the joining interface Cu/In. We did not consider rCu and rAg because they are negligibly small (~0.1 ncm 2 ). As for rCu/Ag and rAg/REBCO, we measured the sum of them as 8.5 ncm 2 per one REBCO tape by using the contact-probing current transfer length method [7]. rIn is the product of the volume resistivity of indium (18.8 nm, measured by the four probe method at 77 K) and its thickness measured with a micrometer or by SEM observation. rCu/In is calculated as the difference between the joint resistivity and the other resistance factors.

Comparison with soldered joints
Three samples each of InSn (In52Sn48) and PbSn (Pb37Sn63) soldered joints were fabricated to compare the joint resistivity with that of In-joints. We pressed the solder into a foil-shape before joining and then inserted it between the connected tapes with flux (SUSSOL-F) applied on tape surfaces. We adopted this process because it is superior to the pre-tinning process for low joint resistance [6]. The joining temperatures of InSn and PbSn joints were 140C and 190C [8], respectively. After reaching the joining temperatures, a pressure around 10 MPa was applied and held until the jig cooled down to ~70C.   Table 1. Two samples were prepared for each condition. Figure2 (a) indicates that the joint resistivity decreased by ~10 ncm 2 by pickling REBCO tapes and In, compared to without pickling (previous process). Pickling the tape reduced the joint resistivity because flux containing NH3Cl and ZnCl can remove copper oxide on the tape surface. However, the effect of pickling In was minor, implying that In oxide can be destroyed mainly by the deformation of In during pressing. Figure 2 (b) shows the joint resistivity of the samples made with different conditions of surface polishing and joining time. The joint resistivities of all the samples were almost the same around 30 ncm 2 , meaning that these two parameters do not influence the joint resistivity under the condition of the pickling process. Figure 2 (c) shows the details of the joint resistivity of the samples joined at different temperatures. In this figure, rIn and rCu/In have an error of ~1 ncm 2 because the thicknesses of indium were measured with a micrometer that has an error of ~5 µm. Generally, rIn decreased with the increase of the joining temperature because indium gets softer and thinner (easily pushed outside the joint area) at higher temperatures. Note that minimizing rIn is not essential at very low temperatures where the volume resistivity of indium is negligibly small. Also, in Figure 2 (c), rCu/In was sufficiently small (<3 ncm 2 ) and did not change over the different temperatures in a short joining time of 15 min. Similarly, the joint resistivity after heating at 95C for three or four days did not drastically change, as shown in Figure 2 (d). In short, pickling and joining temperature are useful, but surface roughness and joining time are not to reduce the joint resistivity.   Table 1. In (c), rIn and rCu/In have an error of ~1 ncm 2 because of the measurement error of ~5 µm by a micrometer. In (d), Sample 1 was heated at 95C for 96 hours. Sample 2 and 3 were heated for 68 hours. These samples are marked in (b).

Cross-sectional observation and joint resistivity evaluation
To discuss the effect of the pickling process, we observed the cross-sections of the samples with and without pickling. Figure 3 shows the cross-sectional SEM images and EDS elemental maps of the sample without pickling. This sample was fabricated by the previous process [4], in which the REBCO tapes were joined at room temperature after surface polishing by a #240 abrasive paper. At the joining interface, we observed an intermetallic compound CuIn2, which forms below 148C [9]. Also, numerous voids smaller than 1 µm generated in the In region near the joining interface probably because of uneven CuIn2 formation accompanied by volume shrinkage. This volume shrinkage is calculated as 7.15% by the following equation: where V CuIn 2 and V Cu+2In are the molar volumes of CuIn2 and Cu+2In, mCu and mIn are the molar masses of Cu and In. The densities of CuIn2, Cu, In are  CuIn 2 =8.2 [10], Cu = 7.31, and In = 8.96 g/cm 3 at room temperature. Figure 3(b) describes the details of the joint resistivity of the observed sample. The joint resistivity was mainly shared by the interface resistivity rCu/Ag+rAg/REBCO (17 ncm 2 ), the In resistivity rIn (16.9 ncm 2 ) and the joining interface resistivity rCu/In (6.5 ncm 2 ) that includes the resistivity of CuIn2. Although the growth of CuIn2 can increase rCu/In, the joint resistivity after long heating did not drastically change as already shown in Figure 2 (d). The volume resistivity of CuIn2 will be discussed in future study.
Subsequently, we observed the cross-section of the pickled sample marked in Figure 2 (b). In this sample, the CuIn2 layer formed thickly and uniformly probably because the pickling process decreased Cu oxides that hindered CuIn2 formation. Besides, the voids near the joining interface decreased considerably compared to those in the sample without pickling (Figure 3(a)). Figure 4(b) shows the detail of the joint resistivity of the pickled sample. The resistivity at joining interface rCu/In was evaluated as <1 ncm 2 . This significant reduction of rCu/In can be attributed to the decrease of the voids and the uniform formation of CuIn2 in this pickled sample. Therefore, we conclude that the pickling process sufficiently reduced the resistivity of In-joints.
(a) (b) Figure 3. (a) SEM images and EDS elemental maps, (b) The details of the joint resistivity of the sample fabricated by the previous process [4]. In the previous process, the tapes were joined at room temperature after surface polishing by a #240 abrasive paper without pickling.   Figure 5 compares the joint resistivities of In-joints, InSn joints, and PbSn joints. The joint resistivity of In-joints showed smaller and less fluctuated values than those of soldered joints. Besides, In-joints can be fabricated at lower temperatures of 20-120C, which reduces the joining time and the risk of Ic degradation of the REBCO tape compared to the soldered joints.

Conclusion
This study investigated the joining conditions of In-joints with the analysis of the constitutive factors of the joint resistivity and the cross-sectional observations. Mainly by pickling the Cu surface of the REBCO tape, we successfully reduced the joint resistivity to 22-30 ncm 2 at 77 K in self-field with joining temperatures of 20-120C. The joining temperature contributed to reducing the resistance (thickness) of In. However, joining time and surface roughness did not affect the joint resistivity. The constitutive factors of the joint resistivity were separately evaluated to find that the joint resistivity has been reduced sufficiently. The interface resistivity was measured as 17 ncm 2 for two REBCO tapes and the resistivity at the interface Cu/In was calculated as <3 ncm 2 with small fluctuations. This result reveals the lower limit of the joint resistivity: the sum of the resistivity (ncm 2 ) of indium (measurable from thickness), the resistivity of Cu/In (<3ncm 2 ), and the interface resistivity of REBCO tapes (measurable from [7]). The cross-sectional observations revealed that CuIn2 forms at the joining interface but not increases the joint resistivity severely. The volume resistivity of CuIn2 will be examined in future study.
Finally, we compared the joint resistivity of In-joints with those of InSn and PbSn joints. The comparison indicated that In-joints have a reproducibly low joint resistivity by a simple process and lower joining temperatures than those of the soldered joints.