Recipe Optimization and SRF Test of Cu-compatible Nb3Sn Films by DC Magnetron Sputtering from a Stoichiometric Target

Recipe Optimization and SRF Test of Cu-compatible Nb3Sn Films by DC Magnetron Sputtering from a Stoichiometric Target Dorothea Fonnesu INFN Legnaro National Laboratories Davide Ford INFN Legnaro National Laboratories Eduard Chyhyrynets INFN Legnaro National Laboratories Sebastian Keckert Helmholtz-Zentrum Berlin für Materialien und Energie Jens Knobloch Helmholtz-Zentrum Berlin für Materialien und Energie Oliver Kugeler Helmholtz-Zentrum Berlin für Materialien und Energie Matteo Lazzari INFN Legnaro National Laboratories Giovanni Marconato INFN Legnaro National Laboratories Alessandro Salmaso INFN Legnaro National Laboratories Aleksandr Zubtsovskii University of Siegen Cristian Pira INFN Legnaro National Laboratories Article Keywords: Posted Date: September 17th, 2025 DOI: https://doi.org/10.21203/rs.3.rs-7516819/v1 License:   This work is licensed under a Creative Commons Attribution 4.0 International License.  Read Full License Additional Declarations: No competing interests reported. Recipe Optimization and SRF Test of Cu-compatible 1 Nb3Sn Films by DC Magnetron Sputtering from a 2 Stoichiometric Target 3 D. Fonnesu1,*, D. Ford1, E. Chyhyrynets1, S. Keckert2, J. Knobloch2,3, O. Kugeler2, M. 4 Lazzari1, G. Marconato1, A. Salmaso1, A. Zubtsovskii3, and C. Pira15 1Istituto Nazionale di Fisica Nucleare (INFN), Laboratori Nazionali Legnaro, 35020 Legnaro, Italy 6 2Helmholtz-Zentrum Berlin f¨ ur Materialien und Energie GmbH (HZB), 14109 Berlin, Germany 7 3Naturwissenschaftlich-Technische Fakult¨ at, Universit¨ at Siegen, 57076 Siegen, Germany 8 *f[email protected] 9 ABSTRACT 10 The development of modern particle accelerators such as FCC-ee requires improved energy efficiency. On the SRF cavity side, the intermetallic compound Nb 3 Sn is a promising alternative to niobium: its higher critical temperature (18.3 K) results into a BCS surface resistance at 4.5 K comparable to the one of Nb at 2 K, potentially allowing improved performance and reduced cryogenic costs while maintaining operation at 4.5 K. However, its brittleness makes bulk machining impractical, restricting its application to thin-film coatings. This study presents Nb 3 Sn thin films deposited on copper substrates via DCMS using a single stoichiometric target. The optimization of the deposition parameters via the evaluation of the critical temperature, morphology, elemental composition and crystalline structure of the films is outlined. A niobium buffer layer is implemented to prevent copper-tin interdiffusion, and plays a key role in the film quality. The results demonstrate Nb 3 Sn films deposited at ≤650 °C on copper substrates pre-coated with a 30 µm niobium buffer layer which exhibit a critical temperature ≥ 17 K. The RF test of a film deposited via the same recipe on a bulk Nb QPR sample yielded an RF surface resistance of 23 n Ω at 4.5 K, 20 mT and 400 MHz. These findings open the way to a scalable approach to high-performance Nb3Sn/Cu cavities. 11 Introduction 12 Cryogenics is one of the major cost factors associated with the superconducting radio-frequency (SRF) systems in modern 13 particle accelerators 1 . Due to its superconducting (SC) transition temperature ( Tc = 18.3 K ), twice as high as that of niobium 14 ( Tc = 9.2 K ), the A15 compound Nb 3 Sn has the potential to achieve quality factors Q0>1010 at the same operation temperature 15 of the established Nb/Cu technology (4.5 K), which is more than twice as high as the one needed for bulk niobium ( 4.5 K 16 versus 2 K ), resulting in the reduction by approximately a factor three of grid power required for cryogenic cooling 2 . However, 17 the intrinsic brittleness of Nb 3 Sn makes its application in bulk form impractical, requiring its employment as a thin film 18 deposited onto a substrate. The state of the art technique Nb 3 Sn cavity production is Vapor Tin Diffusion (VTD), which has 19 demonstrated quality factors of 1010 up to 20 MV m−1 at 4.4 K3 . However, this technique is limited to bulk niobium cavities, 20 eliminating the advantages of using copper for the main cavity structure. The coating of SRF copper cavities with a thin niobium 21 layer via the physical vapor deposition (PVD) technique direct-current magnetron sputtering (DCMS) is a well-established 22 practice, successfully implemented for LEP-II 4 and LHC 5 . Nb/Cu cavities offer significant cost reduction for large-scale cavity 23 production, due to the lower cost of copper compared to bulk niobium, as well as higher thermal conductivity at the targeted 24 operational temperatures. Also, the copper substrate may open the way to alternative cooling methods beyond liquid helium, 25 such as conduction cooling via cryocoolers, which have recently demonstrated a capacity of up to 9 W at 4.2 K6,7 . Finally, 26 Nb3Sn on Cu is of high interest for quantum sensing, particularly for dark matter search using haloscopes8–12.27 This study presents the application of DCMS to the deposition of Nb 3 Sn films on copper, toward the optimization of the 28 deposition recipe in terms of morphology, SC properties and radio-frequency (RF) performance. However, achieving the correct 29 A15 phase of Nb 3 Sn requires high temperatures (> 930 °C ) to prevent the formation of spurious phases (Nb 6 Sn 5 , NbSn 2 ) 13 . 30 This represents a challenge when using copper as a substrate, as its stress response transitions from the elastic to the plastic 31 regime at about 400 °C . In a study performed on 6 GHz elliptical copper cavity prototypes, the experimentally determined 32 upper temperature limit before structural changes occur was observed to be approximately 650 °C14 . Another major challenge 33 is Sn-Cu interdiffusion at the film-interface, which causes an imbalance in the stoichiometry (Nb:Sn ratio) and alters the SC 34 properties of the film. Therefore, optimizing deposition parameters and implementing strategies to mitigate Sn-Cu interdiffusion 35 are crucial aspects for achieving the correct Nb3Sn A15 phase on copper substrates.36 The long-term objective of this research is to establish a scalable process for coating copper cavities with Nb 3 Sn films that meet 37 RF requirements. In the short term, the goal is to obtain films that exhibit:38 1. the correct A15 phase and stoichiometry, demonstrated by a Tc near 18.3 K , elemental analysis (EDS), and X-ray 39 diffraction (XRD) measurements;40 2. a homogeneous, compact, and crack-free surface morphology, as these characteristics are essential for the film perfor41 mance in RF environment, assessed via scanning electron microscope (SEM) imaging of both the surface and the cross42 section.43 According to the research strategy established in this study, these two conditions are baseline requirements for further 44 engineering of the film toward its final application as an SRF cavity coating. In parallel, an additional goal must be pursued:45 3. keeping the process temperature as low as possible, to ensure the process is scalable to large copper structures such as 46 SRF cavities.47 Standardized procedures with high level of control and reproducibility are essential to identify factors influencing film quality. 48 Accordingly, the Nb 3 Sn thin-film samples produced for this study were prepared following a standard routine, described later 49 in the text. The films are 1 µm thick and deposited in argon atmosphere using a commercial, stoichiometric sputtering target 50 as Nb 3 Sn source material. The employed substrates include sapphire, copper and copper pre-coated with a niobium buffer 51 layer of varying thickness (from now on, these substrates are addressed as ’NbBL’, e.g. NbBL-1, NbBL-10, etc. denote a 52 copper substrate pre-coated with a niobium buffer layer of 1 µm, 10 µm, etc., respectively), and bulk niobium. The deposition53 process was initially followed by a 24 h annealing step. This annealing step was later eliminated, as discussed in the following 54 Sections. After deposition, the samples are characterized by measuring their critical temperature, inspecting their surface 55 and cross-section, analyzing their elemental composition and crystalline properties, and finally, measuring their RF surface 56 resistance. To meet the conditions set by the short-term goals of this study, flat samples were used for deposition parameter 57 optimization. The RF test was also conducted on a flat sample, whose geometry is customized for a so-called quadrupole 58 resonator (QPR) device15.59 This work provides a comprehensive description of the R&D process established at INFN-LNL for optimizing the deposition 60 recipe of Nb 3 Sn films on copper. The most relevant results on film quality and initial RF tests are presented. The effects of 61 deposition temperature, annealing and NbBL thickness on the final properties of the Nb 3 Sn film are discussed, along with 62 the current lower limit found for the process temperature. The manuscript is structured as follows: after this Introduction, 63 the Results Section presents the main achievements of this study. Finally, the Methods Section details the experimental setup 64 employed for deposition, the sample production routine, and all characterization tools.65 Results and Discussion66 Dependency of the Tcand Deposition Rate on the Deposition Parameters67 Current applied to the sputtering target and pressure of the sputtering gas68 At an early stage, the dependence of the deposition rate and of the Tc of the films on the target current and process gas pressure 69 was studied for samples deposited on sapphire substrate, used as reference for the later stages of the study. The deposition 70 parameters for each run were selected in tight feedback with the measured Tc of these samples, iteratively tuned until sufficient 71 data was gathered to identify trends.72 The dependence of the Tc on the surface current density applied to the target was checked first, as the sputtering process was 73 carried out in current-driven mode. The result is shown in Fig. 1a. These samples were deposited on sapphire at 630 °C and 74 annealed for 24 hours at the same temperature. Two data sets are presented, corresponding to depositions carried out at two 75 different argon pressures, 2 × 10−2 mbar and 3 × 10−3 mbar . In both cases, the Tc of the samples are observed to degrade as 76 the applied current increases. The set deposited at a higher pressure exhibits a systematically higher Tc compared to the set 77 deposited at lower pressure.78 Further insight in this regard was obtained by looking at the deposition rate as a function of the target surface current density, 79 as shown by the data in Fig. 1b. Two data sets (triangles) are shown, also corresponding to films deposited on sapphire 80 at 2 × 10−2 mbar and 3 × 10−3 mbar , with a deposition temperature of 630 °C and annealed for 24 hours at the same temperature. 81 2/19 The linear fit of the low current data, indicated by the full line, is constrained to cross the axes origin. The dashed line extends 82 from the fit line to visually highlight the separation of the higher current data from the low-current linear behavior. The data 83 was line-fitted up to a current density value of 1.3 mA cm−2 (corresponding to a power density of 0.4 W cm−2 ). This does not 84 represent a hard current limit, but was taken as a conservative upper limit based on the available data. 85 Assuming that sputtering remains the only target material extraction mechanism during the deposition process (e.g., no thermal 86 evaporation occurs), the deposition rate is generally expected to increase linearly with the current density 16 . However, tin is 87 a low-melting-point material, likely to introduce a thermal evaporation component to the process. For low currents, thermal 88 evaporation of tin, if present, contributes negligibly, so that the dependence of the deposition rate on the target current stays 89 linear. At higher current densities (hence higher target temperatures), tin evaporation is no longer negligible, as suggested 90 by the deposition rate data departing from the low-current linear trend. Tc is a first indicator of the long range crystalline 91 order of a conventional superconductor such as Nb 3 Sn, with off-stoichiometry Nb-Sn composition being a major factor 92 contributing to its degradation. The information, obtained from the deposition rate and supported by the data on the Nb/Sn 93 atomic content ratio, that target surface current densities higher than 1.3 mA cm−2 can cause tin evaporation from the target, 94 combined with the Tc trend resulting into higher values for decreasing current density and higher argon pressure, supports the 95 choice of 2 × 10−2 mbar and maximum 1.3 mA cm−2 as process gas pressure and applied target current, respectively. 96 Deposition temperature and annealing 97 At the beginning of the study, standard deposition runs included a 24 h annealing step, implemented at the end of each deposition 98 with the aim to assist the formation of the A15 phase. However, it is also known that prolonged exposure to high temperatures 99 can promote tin diffusion, so that keeping the deposition temperature as low as possible and shortening (if not removing) the 100 annealing step, may be beneficial toward preserving the film stoichiometry for samples deposited on copper and NbBL. Hence, 101 the effect of both the deposition temperature and the presence/duration of the annealing step on the Tc of the films was also 102 investigated. 103 In Fig. 2a, the Tc of the samples deposited on sapphire is plotted as a function of the deposition temperature. Results for 104 different annealing times are shown. In each case the annealing temperature equaled the deposition temperature. These samples, 105 as discussed in the previous section, were deposited at a pressure of 2 × 10−2 mbar and at an applied target surface current 106 density < 1 mA cm−2 . The samples annealed for 24 h are shown by the full triangles, the samples which were not annealed 107 by the empty triangles. Two individual data points are also shown, corresponding to samples for which an annealing step of 108 intermediate duration ( 5 h and 2 h as indicated in the graph) was performed. With increasing deposition temperature, the Tc of 109 the samples increases toward a Tc of 18.3 K for both sets of samples ( 24 h and no annealing) in a comparable fashion. However, 110 the samples which were not annealed exhibit a higher Tc than the ones annealed for 24 h at lower deposition temperatures. The 111 samples to which a short annealing was applied do not differ significantly from the non-annealed ones. This suggested that the 112 annealing of the films post-deposition, if not detrimental for the achievement of the correct A15 phase, was neither necessary 113 nor beneficial. Based on these results, the annealing step was removed for the following deposition runs. 114 Substrate type 115 Once satisfying values for the target current and the argon pressure were found, and the annealing step removed, thanks to 116 the results obtained with the samples deposited on sapphire, the focus of the study moved to the effect of the NbBL thickness 117 on film properties. At first, in addition to sapphire, the film samples were deposited also on copper and NbBL-1 substrates. 118 1 µm was the chosen starting thickness for the buffer layer, based on previous DCMS studies which implemented a 1 µm -thick 119 layer of tantalum as buffer layer 17 . The Tc of these samples, deposited under the same conditions as the samples on sapphire 120 shown in Fig.1a, was also measured as a function of the target surface current density (not shown here). The results showed 121 a Tc which remained constant between 12 and 14 K up to a threshold surface current density of about 3 mA cm−2 , beyond 122 which superconductivity was not observed, regardless of argon pressure, for the samples deposited on both copper and NbBL-1. 123 However, films on NbBL-1 consistently exhibited slightly higher Tc (by 500 mK ) than those on copper, thus indicating that the 124 NbBL partially mitigates tin diffusion and helps preserve stoichiometry, as expected. Following this result, the thickness of the 125 NbBL was gradually increased. 126 In Fig. 2b, the Tc of films deposited on copper and NbBL of thickness ranging from 1 to 40 µm are shown as a function of the 127 deposition temperature (following the development of the deposition recipe, these films were not annealed, and were deposited 128 under the same conditions as the samples in Fig. 2a). It can be seen in this case that, again, the Tc of the sample deposited 129 NbBL-1 did not show any significant improvement with respect to the sample on copper, neither an increase of the deposition 130 temperature seemed to be beneficial. On the contrary, for the samples on NbBL-10, NbBL-30, and NbBL-40 Tc increases with 131 both increasing NbBL thickness and increasing deposition temperature. The samples deposited on NbBL-30 and NbBL-40 132 reach Tc⪆17 K at 600 °C and maintain Tc > 17 K at 650 °C , suggesting that a NbBL thickness of 30 µm is sufficient to achieve 133 optimal superconducting properties. 134 3/19 Structural and Morphological Properties of the Nb3Sn Films135 Film morphology via SEM and chemical composition via EDS136 The surface morphology of the NbBL and bulk Nb substrates, and of the Nb 3 Sn films, was examined by SEM, as shown in 137 Fig. 3("SEM micrographs", top section). The first row displays, left to right, NbBLs of various thickness (NbBL-1, NbBL-10, 138 NbBL-30) and the bulk Nb substrate. The second and third rows display Nb 3 Sn films deposited on these substrates at 600 °C139 and 650 °C respectively.140 The NbBL and bulk Nb substrates exhibit a homogeneous, crack-free surface. NbBL-1 presents a fine grain structure, while the 141 surface of the NbBL-10+ and bulk substrates show well formed, larger grains, so that increasing the thickness of the buffer 142 layer correlates with Nb grain growth. The surface of the bulk Nb substrate appears smooth and exhibits even larger grains. The 143 Nb 3 Sn films show a distinct surface morphology depending on the NbBL thickness and deposition temperature. At 600 °C , the 144 film on NbBL-1 exhibits a regular, homogeneous grain pattern. On the films deposited on NbBL-10+ and bulk Nb, isolated 145 structures of marked geometrical shape, addressed in this context as islands, are observed. At 650 °C , the films deposited on all 146 NbBLs show a regular, homogeneous grain structure, with larger grains compared to those deposited at 600 °C . The grain size 147 and appearance evolve with the thickness of the NbBL. The amount of Sn-rich islands increases with increasing buffer layer 148 thickness (highest for Nb bulk, absent in Cu) and decreases with increasing substrate temperature (at 650 °C , the presence of 149 islands on NbBL-30 is almost suppressed, and significantly reduced on Nb bulk).150 The lowermost row in Fig. 3displays into more detail some morphological film features. An SEM micrograph of the ion-milled 151 cross section of a Nb 3 Sn sample, deposited at 650 °C on NbBL-30, is shown in Fig. 3a. The grain structure appears dense, 152 void-free and homogeneous. A magnification of the islands of a film deposited on bulk Nb at 650 °C is given in Fig. 3b, where 153 the structure difference between the islands and the base film is visible. Finally, Fig. 3c shows the in-depth wedge-like structure 154 of the islands. It can be observed that islands start nucleating at around 500 nm thickness, then rise parallel to the columnar 155 growth of the film and protrude at the surface. For this image, a film on sapphire (also presenting island structures) was chosen 156 as this substrate can be cracked, allowing to break the film in a way that the three-dimensional features of the islands can be 157 visually enhanced. It is appropriate to specify that, although not shown here, all Nb 3 Sn samples deposited on sapphire exhibit 158 the largest number of islands compared to the several substrates tested in this study.159 Further insight is given by the Nb and Sn atomic content percentage measured via EDS at several point locations on the films160 deposited at 600 °C on NbBL-10+ and bulk substrates is shown in Fig. 4a for the film base surface and in Fig. 4b for the islands. 161 The average composition of the base film, on one hand, stays constant, with small variations, for all the considered substrates, 162 with a ratio Nb/Sn close to 3, expected for the correct A15 stoichiometry. On the other hand, the composition of the islands 163 reveals a significant deviation from this ratio, showing a systematic excess of tin, with corresponding Nb depletion, with respect 164 to the underlying film. Despite the larger fluctuations of the composition measurements on the islands with respect to the ones 165 on the base film, on average the island composition stays also constant for all the substrates.166 DCMS from single stoichiometric target has the intrinsic advantage of not producing sub-stoichiometric phases, which are 167 instead typical of the VTD technique. However, the nucleation of Sn-rich islands observed here appears similar to that already 168 reported in literature for Nb 3 Sn via VTD, driven by the mobility of Sn along grain boundaries 18 . The mobility of Sn depends 169 on temperature and also on the total volume of grain boundaries. The latter will depend on the NbBL: the larger its thickness,170 the larger the grain size, the smaller the volume of grain boundaries near the interface with Nb 3 Sn. Therefore, one should 171 fine-tune the deposition temperature and NbBL thickness to find a proper compromise between maximizing Tc and limiting the 172 dynamics behind the nucleation of the islands.173 Structural Properties via XRD174 The crystallographic properties of the Nb 3 Sn films deposited on different substrates were investigated via XRD spectroscopy. 175 Selected diffractograms for samples deposited on bulk Nb, NbBL-30, and NbBL-1 at 650 °C are shown in Fig. 5a. The 176 measured data is presented over the 2θ range covering the three most intense peaks of the Nb 3 Sn pattern, corresponding to 177 the (002), (012) and (112) crystallographic planes, as indicated by the gray labels in bold placed next to the respective peak. 178 The peak positions for Nb 3 Sn 19 , Nb 20 and Cu 21 sourced from the XRD reference data, are indicated by the vertical lines. The 179 XRD data extracted for the samples shown in Fig. 5a and for samples deposited on more substrates are provided in Table 1. 180 The patterns in Fig. 5a confirm the presence of the Nb 3 Sn phase across all samples, with slight variations in peak intensity 181 and width depending on the substrate. All three Nb 3 Sn peaks are shifted to lower angles relative to the reference positions 182 for samples deposited on NbBL-1 and NbBL-30, though the effect is less pronounced for NbBL-30. In contrast, for samples 183 deposited on bulk Nb, the (002) peak is shifted to lower angles, while the (012) and (112) peaks shift to higher angles. The 184 peak shifts observed for the sample on bulk Nb are the smallest among the three. Shifts of this kind, according to a preliminary 185 analysis, are an indication of a strain condition of the film. The shift toward lower 2θ values in the samples on NbBL-1 and 186 4/19 NbBL-30 indicates macrostrain, likely due to growth conditions and thermal expansion mismatch with the substrate. The latter 187 is mitigated by the increased thickness of the NbBL, and the corresponding shift is consequently smaller. However, the shifts 188 observed for the sample on bulk Nb, with its first main peak shifted to lower angles, while the other two shifted to higher angles, 189 suggest a more complex, anisotropic stress state with an overall lower strain with respect to the other two samples. 190 A broadening of the most intense peak for the samples deposited on NbBL and copper with respect to the one deposited on 191 sapphire is also observed, with the sample on sapphire exhibiting the narrowest peak and the one on copper the broadest peak, 192 as confirmed by the FWHM data in Table 1. Peak broadening can suggest the presence of local variations in lattice spacing, due 193 to defects such as dislocations, grain boundaries, vacancies, etc. A smaller broadening may be associated to a lower defect 194 density and improved microstructural quality, as in the case of the NbBL-40 sample listed in Table 1. Peak broadening requires 195 a dedicated analysis to separate the crystallite size effects, and no clear interpretation can be given to this data at this stage of 196 the analysis. However, thicker NbBLs appear to mitigate macroscopic film stress and may be having a positive effect on defect 197 density, potentially influencing Sn processing and, therefore, the final film composition. 198 The (011) peak of niobium is visible in samples deposited on NbBL, appearing as a distinct peak for NbBL-30 but not fully 199 resolved from the Nb 3 Sn (012) peak in the NbBL-1 sample. Given the X-ray beam penetration depth of about 750 nm at 200 3°incidence angle, and the Nb 3 Sn film thickness of 1 µm , these likely originated from the underlying NbBL substrate. This 201 Nb peak was also present for the film deposited on bulk Nb, but it is not as evident in the pattern shown in Fig. 5as the main 202 orientation of the bulk substrate (hence, the most intense Nb peak) corresponded to the (022) plane, placed at a 2θ value out of 203 the range displayed in the given graph. The (111) peak of copper is also observed in the NbBL-1 sample, likely coming from 204 the substrate or from the washer print present on the sample, as visible in Fig. 10 in Methods - Sample Production.205 Position (°) Shift (°) FWHM (°) ↓Substrate (002) (012) (112) (002) (012) (112) (002) (012) (112) Cu 33.6865 37.7724 41.5251 -0.1835 -0.2396 -0.2759 0.2387 0.2656 0.2697 NbBL-1 33.6964 37.7806 41.5388 -0.1736 -0.2314 -0.2622 0.2454 0.2590 0.2688 NbBL-10 33.7884 37.9183 41.6823 -0.0816 -0.0937 -0.1187 0.2233 0.2314 0.2476 NbBL-30 33.8242 37.9945 41.7764 -0.0458 -0.0175 -0.0246 0.2129 0.2330 0.2335 NbBL-40 33.8517 37.9956 41.7839 -0.0183 -0.0164 -0.0171 0.2207 0.2162 0.2407 bulk Nb 33.8618 38.0196 41.8206 -0.0082 +0.0076 +0.0196 0.2357 0.2369 0.2459 Sapphire 33.8597 38.0108 41.8164 -0.0103 -0.0012 +0.0154 0.2146 0.2153 0.2295 Nb3Sn ref. 33.870 38.012 41.801 0 0 0 - - - Table 1. Extracted values for the peak position, peak shift (with respect to the nominal line) and peak full width half maximum (FWHM) for the three main peaks of Nb3Sn samples deposited on NbBL-1, NbBL-30 and bulk Nb substrates at 650 °C. The values in bold indicate the most intense peak. Dependency of the Tcand of the Compositional and Crystalline Properties on the NbBL Thickness 206 The evolution of the lattice parameters of Nb 3 Sn ( aNb3Sn ) and NbBL ( aNbBL ), along with the Tc of Nb 3 Sn films deposited 207 at 650 °C as a function of NbBL thickness, is shown in Fig. 5b. The lattice parameter for the Nb 3 Sn films was calculated as 208 the average of the lattice parameter values extracted from the respective three most intense XRD peaks, as the one shown in 209 Fig. 5a (namely the ones corresponding to the (002), (012), and (112) planes). The same method has been used to calculated the 210 lattice parameters for the NbBLs, whose XRD patterns were acquired separately and before the deposition of the Nb 3 Sn film. It 211 can be seen that, as the thickness of the NbBL increases, aNbBL approaches the reference bulk value of 3.300 angstrom20,22 , 212 likely due to reduced residual stress. A similar trend is observed for aNb3Sn , which tends to the Nb 3 Sn A15 phase value of 213 reference 5.29 angstrom19,23 as the thickness of NbBL increases. In particular, for a NbBL thickness ≥30 µm , the Nb 3 Sn lattice 214 parameter stabilizes around 5.29 angstrom , indicating a well-formed A15 crystalline structure. For thinner NbBLs the lattice 215 parameter deviates significantly, a possible indication of lattice distortions, in agreement with what discussed in the previous 216 Sections. 217 The corresponding Tc values of the Nb 3 Sn samples, provided as labels next to the data points in Fig. 5b, are found to increase 218 with NbBL thickness. The same trends were observed for samples deposited at 600 °C , though they are not shown in this report. 219 The EDS measurements indicate that the intra-grain composition of the Nb 3 Sn films is stable, with a Nb/Sn ratio close to 220 the stoichiometric value of 3:1, independently of the thickness of the NbBL, as discussed in the previous Section (Fig. 4a). 221 Similarly, the composition of the islands also appears to be independent of the thickness of NbBL ( 4b). Therefore, the relation 222 between the NbBL thickness and the increase in Tc may be mainly due to the relaxation of the crystal lattice, which for NbB-30+ 223 5/19 grows with a lattice parameter which saturates at values close to those of the ideal A15 structure. The NbBL thickness also 224 appears to suppress the thermal stress caused by the different thermal expansion of the two materials, which has been indicated 225 by a recent study as the possible cause for the low Tc of Nb 3 Sn films deposited directly on copper via High Power Impulse 226 Magnetron Sputtering (HiPIMS)24.227 These trends are compared with data from literature 13 in Fig. 6. This data (star markers) was obtained from bulk samples 228 produced via levitation melting with controlled stoichiometry variation 23 , with Sn content varying from 18 to 25 At%. 229 Therefore, while a direct quantitative comparison with the samples produced via DCMS in this study is not possible, a 230 qualitative comparison of material properties can be made. The resulting dependence of the Tc and lattice parameter of the 231 Nb 3 Sn data from literature is an increase of both these quantities with increasing Sn content. For the samples in this study, 232 on the other hand, what was varied in a controlled way is the thickness of the NbBL. The Sn content was measured by EDS 233 in an area of size 400×270 µm2 . Hence, the obtained value takes into account the contribution of grains, islands, and grain 234 boundaries, which were not considered in the analysis shown in Fig. 4, as those values were extracted by probing the surface 235 over point-sized areas ( ≤0.2 µm2 ). For these films, the average Sn content measured via EDS ranges from 22% for the sample 236 on NbBL-1 to 25% for the sample on bulk Nb, which is within the uncertainty bars given in Fig. 4a. Unlike in the bulk case, the 237 films exhibit increasing Tc with decreasing lattice parameter. As a result, the two data sets show a converging trend in Tc toward 238 the literature A15 phase values, corresponding to opposite variation in the lattice parameter.239 Nb3Sn Film RF Properties with the Quadrupole Resonator (QPR)240 The ultimate test to assess the suitability of the produced Nb 3 Sn films for SRF applications is a cryogenic performance 241 measurement under realistic field and RF conditions. The RF properties of the optimized coating procedure, especially its 242 surface resistance RS , were studied using the QPR at Helmholtz-Zentrum Berlin (HZB). The experimental setup is described in 243 detail later in Methods - Characterization Tools. In Fig. 7a and Fig. 7b, the measured RS is plotted against the RF peak field at 244 the sample surface (at 4 K and 417 MHz ) and against the sample temperature (at 30 mT and 417 MHz ), respectively. Depending 245 on the cooldown dynamics, different values are obtained. The initial cooldown yields a minimum RS of 51 nΩ at 30 mT . To 246 estimate the contribution of residual resistance caused by the sample’s sensitivity to cooldown conditions, thermal cycles at 247 different cooldown rates were performed, with each data set corresponding to a different thermal cycle. As shown in the plot, a 248 slow cooldown ( −0.1 K min−1 ) with active heating of the sample leads to a slightly increased RS of ∼60 nΩ at 2.5 K . A fast 249 cooldown ( −18.5 K min−1 ) yields a reduced RS of ∼35 nΩ . The lowest values are obtained after a full thermal cycle of the 250 entire resonator, resulting in a uniform cooldown at a rate of −6 K min−1 . Subsequently, a minimum RS of ∼25 nΩ is recorded. 251 Notably, a renewed fast cycle increases this value back to ∼35 nΩ . The obtained RS showed the known behavior of Nb 3 Sn being 252 sensitive to trapped magnetic flux, especially driven by thermoelectric currents, present during the superconducting transition 2 . 253 In Fig. 7this is visible as a constant offset in RS between different cooldowns, hence only affecting the residual resistance 254 Rres . As for other cavity measurements, a cooldown with minimum temperature gradient leads to lowest Rres , in this case 255 about 18 nΩ at BRF = 20 mT . By design, QPR cooldown conditions are quite different to other vertical tests of cavities, due to 256 the limited conduction cooling of the sample. Hence, a full thermal cycle of the cavity inside the cryostat and not the slowest 257 possible cooldown rate yields lowest thermoelectric currents and lowest Rres . The fact that RS data for both "fast" cooldowns 258 agree perfectly, independent of previous cooldowns, underscores that this behavior is not due to trapped environmental flux, as 259 the total amount (or at least the geometrical distribution) would have changed during the full thermal cycle of the resonator.260 For comparison, QPR data for a sample prepared by vapor tin diffusion (’VTD sample’) on a bulk niobium substrate, and at 261 temperatures up to 1100 °C25,26 , is added to Figs. 7a and 7b. The original data can be found in 25 . Before plotting, the data 262 measured on the sample produced by VTD was reduced by 12 nΩ to compensate for parasitic losses from the normal conducting 263 adapter flange which was used at that time 27 . This correction leads to similar values as the ones obtained in this study, but it can 264 be seen that the magnetron sputtered sample outperformed the VTD one by roughly a factor 2 in surface resistance. However, 265 the observed increase of RSwith increasing RF field was stronger in the present case.266 The RF penetration depth ( λRF ) of the sample was also extracted from the measured frequency shift as a function of sample 267 temperature, shown in Fig. 8a. A non-linear fitting of the data yields Tc=17.60(3)K and λ0=315(24)nm . The result for Tc is 268 in good agreement with the Tc data presented in this work. Also, the result obtained for λ0 is in good agreement with the London 269 penetration depth λL measured, in a separated study, via the microwave characterization of coplanar-waveguide-resonators 270 (CPWRs) patterned on Nb 3 Sn thin films produced according to the optimized deposition recipe presented in this work, which 271 lead to a value of 310 nm11.272 Fig. 8b shows the RF quench field of the sample measured at the first two quadrupole modes, 417 MHz and 851 MHz . The plot 273 only includes data points with BRF <50mT ( T>10K ), conditions for which RF heating is negligible. A quadratic fit of the data 274 leads to Tc=17.65(7)K , which is again in good agreement with the Tc data presented here, and a low-temperature extrapolated 275 6/19 quench limit of B0=77.5(5)mT . Again, QPR data for a sample prepared by VTD is added to Fig. 8b for comparison. The 276 original data can be found in 25,26 . The RF quench field is independent of frequency which excludes significant RF heating 277 as systematic error source. The extrapolated quench limit for B0 indicates non-ideal performance, potentially coming from 278 stoichiometry imbalances 18 . Also, it can be seen that the VTD sample showed higher values for Tc and RF quench field, 279 indicating room for RF performance improvement for the DCMS coating procedure. Nevertheless, the quench field value 280 obtained for the DCMS sample corresponds to a maximum accelerating gradient of Eacc =18.2MV/m in a TESLA-shaped 281 cavity28, which is overall a satisfying result. 282 With the QPR, quench field and frequency shift measurements allow independent access to Tc that can lead to different results. 283 This is due to the fact that, for quench field data, the possible effects due to the presence of point-like defects may be dominant, 284 while frequency shift data always depicts a volumetric average over the RF-illuminated area and the penetration depth. For the 285 sample presented here, both values showed good agreement, indicating a quench behavior that was magnetically dominated by 286 regions of average Tc, compatible with a rather homogeneous Tc-distribution and, therefore, a homogeneous Nb3Sn coating. 287 Conclusion 288 This study demonstrates the successful development of Nb 3 Sn films on copper via single-target DCMS, thanks to an optimized 289 set of deposition parameters resulting in high Tc (17 K) and good film morphology. The presence of a Nb buffer layer (NbBL) 290 of thickness ≥30 µm is crucial to stabilize film stoichiometry via residual stress reduction and enhanced crystallinity. To this 291 point, the optimal base recipe in terms of Tcof the films can be summarized as the following: 292 • 1 mA cm−2 maximum target surface current density; 293 • 2 × 10−2 mbar argon pressure; 294 •≤650 °C maximum substrate temperature; 295 • no annealing; 296 •≥30 µm minimum NbBL thickness. 297 Unlike VTD, which is prone to the formation of sub-stoichiometric phases, DCMS proves to be a promising technique to avoid 298 the latter, as demonstrated by the absence of spurious peaks in the XRD diffractograms. The SEM analysis highlights the role 299 of the NbBL in mitigating Sn-rich island formation, while XRD also confirms improved structural properties with increased 300 NbBL thickness. RF measurements using the QPR indicate promising RS values (23 n Ω at 4.5 K, 20 mT and 400 MHz ), with 301 quench fields ( 77 mT ). Up to 40 mT , the RS values are comparable or better than those of the QPR sample VTD, while the 302 quench value is lower, although still sufficient for many applications. The quench field, as well as the sharp increase in Q , 303 can be attributed to specific stoichiometric imbalances introduced by the Sn-rich islands, which this study has shown to be 304 particularly favored in PVD films grown on bulk Nb substrates, while NbBL on Cu has the ability to mitigate them. The RF 305 test also demonstrates that good RS values can be achieved already at Tc lower than 18.3 K , which makes the Tc of the films 306 a good guide value for the initial film development phase, to become not determinant for the advanced phase, in which the 307 final RF performance becomes of central interest, and for which the morphology and composition become essential. Overall, 308 these findings provide a first, promising step towards the scalability of the PVD recipe for Nb 3 Sn films to high-performance 309 Nb3Sn/Cu SRF cavities, supporting the development of future particle accelerators like the FCC-ee. 310 Methods 311 Experimental Setup 312 The Nb 3 Sn films object of this study were produced via the PVD technique DCMS, using a 4" ( 10.16 cm ) diameter commercial 313 Nb 3 Sn stoichiometric planar target (75% Nb, 25% Sn, 99.99% purity) as source material. The experimental setup is schemat314 ically represented in Fig. 9. The top frame shows a technical drawing of the stainless steel (SS) ultra-high vacuum (UHV) 315 chamber inside which the deposition processes were carried out. The sample holder consists of a 10 cm diameter SS plate onto 316 which the substrates to be coated were mounted, as shown by inlet on the top-left of Fig. 9. The magnetron with the Nb 3 Sn 317 target, and the sample holder plate are mounted at two opposite sides of the chamber, in front of each other, coaxially, and 318 separated by a 90 mm distance. The magnetron field is provided by two current-supplied electromagnets. The gas employed for 319 the sputtering process was 6N purity ( 99.9999 % ) argon. The substrate temperature was regulated via a set of three infrared 320 (IR) lamps ( 500 W ) placed below the sample holder. The power provided to the IR lamps was regulated via a PID feedback 321 loop which takes as input the temperature measured by a thermocouple fixed onto the sample plate itself, next to the substrates. 322 The base pressure achievable in the UHV chamber is about 5 × 10−10 mbar . The complete piping and instrumentation diagram 323 of the sputtering system is given in the bottom frame of the figure. 324 7/19 Figure 8. (a) Frequency shift measurement. The material’s penetration depth and transition temperature are extracted by non-linear fitting. (b) RF quench field vs. sample temperature measured at two quadrupole modes. The quadratic fits yield again Tc and the extrapolated quench limit at 0 K . ’VTD sample’ denotes the QPR measurement data for a sample prepared by vapor tin diffusion25,26. 14/19 Figure 9. Top: technical drawing of the stainless steel UHV deposition chamber. The 4" magnetron with the Nb 3 Sn sputtering target and the sample plate are mounted coaxially in front of each other at a 90 mm distance. The IR lamps employed to control the substrate temperature are visible below the the sample plate. The top left inlet shows a rotated perspective on the substrates mounted on the sample plate. Bottom: piping and instrumentation diagram of the DCMS system, showing the pumping stages, valves, pressure and temperature sensors, and the process (argon) and venting (nitrogen) gas supplies. 15/19 Figure 10. Schematic representation of the different substrate types employed for the deposition of the Nb 3 Sn films. The size of the substrates is also given. The thicknesses are not to scale. The bright gray layer on top represents the Nb3Sn film. A real-life picture of the Nb3Sn film side of each substrate after the deposition process is shown in the line below, with the NbBL-10+ being in this case a 30 µm thick NbBL. Figure 11. Cutaway view of the HZB Quadrupole Resonator (QPR), with real bulk Nb sample shown on the left. The interchangeable sample (shown in red) is inserted into the cavity (depicted in green) from below and positioned less than 1 mm from the pole shoes (shown in orange) which focus the RF magnetic field onto that region. The surface resistance of the sample is measured calorimetrically by employing a heater and temperature sensors attached beneath the sample. The entire cavity is immersed in superfluid liquid helium. 16/19 Data Availability 425 The datasets generated during and/or analyzed during the current study are available from the corresponding author on 426 reasonable request. 427 References 428 1. Hutton, A. Energy-recovery linacs for energy-efficient particle acceleration. Nat. Rev. Phys. 5, 708–716, DOI: 10.1038/ 429 s42254-023-00644-6 (2023). 430 2. Posen, S. & Hall, D. L. Nb3Sn superconducting radiofrequency cavities: Fabrication, results, properties, and prospects. 431 Supercond. Sci. Technol. 30, 033004, DOI: 10.1088/1361-6668/30/3/033004 (2017). 432 3. Posen, S. et al. Advances in Nb3Sn superconducting radiofrequency cavities towards first practical accelerator applications. 433 Supercond. Sci. Technol. 34, 025007, DOI: 10.1088/1361-6668/abc7f7 (2021). 434 4. Boussard, D. Performance of the LEP2 SRF system. In Proceedings of the 17th Particle Accelerator Conference, 2879–83 435 (Vancouver, Canada, 1998). 436 5. Boussard, D. & Linnecar, T. P. R. The LHC superconducting RF system. In Proceedings of the Joint Criogenic Engineering 437 Conference and International Cryogenic Materials Conference (Montreal, Canada, 1999). 438 6. Hao, X., Zerkle, B., Cosco, J. & Dausman, R. Development of a 5 W/4.2 K two-stage pulse tube cryocooler. IOP Conf. 439 Series: Mater. Sci. Eng. 1301, 012140, DOI: 10.1088/1757-899X/1301/1/012140 (2024). 440 7. Mitchell, S. SHI Cryogenics Group Releases World’s Highest-Capacity 4K Cryocooler. https://shicryogenics.com/shi441 cryogenics-group-releases-worlds-highest-capacity-4k-cryocooler/. 442 8. Sikivie, P. Detection rates for “invisible”-axion searches. Phys. Rev. D 32, 2988–2991, DOI: 10.1103/PhysRevD.32.2988 443 (1985). 444 9. Sikivie, P. Experimental Tests of the "Invisible" Axion. Phys. Rev. Lett. 51, 1415–1417, DOI: 10.1103/PhysRevLett.51.1415 445 (1983). 446 10. Marconato, G. et al. NbTi Thin Film SRF Cavities for Dark Matter Search. In 21th International Conference on RF Super447 conductivity (SRF’23), Grand Rapids, MI, USA, 25-30 June 2023, 96–99, DOI: 10.18429/JACoW-SRF2023-MOPMB014 448 (JACOW Publishing, Geneva, Switzerland, 2023). 449 11. Ghigo, G. et al. Heavy ion irradiation effects on the high-frequency properties of YBCO and Nb3Sn thin films. Supercon450 ductivity 13, 100149, DOI: 10.1016/j.supcon.2024.100149 (2025). 451 12. Vidal Garcia, P. Microwave Vortex-dynamics Characterization in Nb3Sn under High Magnetic Fields. In 11th International 452 Workshop on Thin Films and New Ideas for Pushing the Limits of RF Superconductivity - TFSRF2024, DOI: https: 453 //indico.cern.ch/event/1376902/contributions/6111702/ (Orsay, France, 2024). 454 13. Godeke, A. A review of the properties of Nb3Sn and their variation with A15 composition, morphology and strain state. 455 Supercond. Sci. Technol. 19, R68–R80, DOI: 10.1088/0953-2048/19/8/R02 (2006). 456 14. Pira, C. Nb Thick Films in 6 GHz Superconducting Resonant Cavities. Ph.D. thesis, Università degli studi di Padova (2018). 457 15. Keckert, S., Kleindienst, R., Kugeler, O., Tikhonov, D. & Knobloch, J. Characterizing materials for superconducting 458 radiofrequency applications-A comprehensive overview of the quadrupole resonator design and measurement capabilities. 459 The Rev. Sci. Instruments 92, 064710, DOI: 10.1063/5.0046971 (2021). 460 16. Ohring, M. Materials Science of Thin Films (Elsevier, 2002). 461 17. Ilyina, E. A. et al. Development of sputtered Nb3Sn films on copper substrates for superconducting radiofrequency 462 applications. Supercond. Sci. Technol. 32, DOI: 10.1088/1361-6668/aaf61f (2019). 463 18. Willson, S. A. et al. Impact of submicron ${\mathrm{Nb}}_{3}\mathrm{Sn}$ stoichiometric surface defects on high-field 464 superconducting radiofrequency cavity performance. Phys. Rev. Res. 6, 043133, DOI: 10.1103/PhysRevResearch.6.043133 465 (2024). 466 19. Geller, S., Matthias, B. T. & Goldstein, R. Some New Intermetallic Compounds with the “ β -Wolfram” Structure. J. Am. 467 Chem. Soc. 77, 1502–1504, DOI: 10.1021/ja01611a029 (1955). 468 17/19 20. Dry´ s, M., Sosnowski, J. & Folcik, L. Phase equilibria in the niobium-gallium-iron system at 1000 ◦ C. J. Less Common 469 Met. 68, 175–181, DOI: 10.1016/0022-5088(79)90054-7 (1979).470 21. Suh, I.-K., Ohta, H. & Waseda, Y. High-temperature thermal expansion of six metallic elements measured by dilatation 471 method and X-ray diffraction. J. Mater. Sci. 23, 757–760, DOI: 10.1007/BF01174717 (1988).472 22. Straumanis, M. E. & Zyszczynski, S. Lattice parameters, thermal expansion coefficients and densities of Nb, and of solid 473 solutions Nb–O and Nb–N–O and their defect structure. J. Appl. Crystallogr. 3, DOI: 10.1107/S002188987000554X 474 (1970).475 23. Devantay, H., Jorda, J. L., Decroux, M., Muller, J. & Flükiger, R. The physical and structural properties of superconducting 476 A15-type Nb-Sn alloys. J. Mater. Sci. 16, DOI: 10.1007/BF00542375 (1981).477 24. Rosaz, G. Nb3Sn coatings for RF cavities. In FCC Week 2025, DOI: https://indico.cern.ch/event/1408515/contributions/ 478 6514112/ (Vienna, Austria, 2025).479 25. Keckert, S., Hall, D., Knobloch, J., Kugeler, O. & Liepe, M. Surface resistance characterization of Nb3Sn using the 480 HZB quadrupole resonator. In Proceedings of the 18th International Conference on RF Superconductivity, vol. SRF2017, 481 863–866, DOI: 10.18429/JACoW-SRF2017-THPB053 (JACOW, Geneva, Switzerland, 2018).482 26. Keckert, S. et al. Critical fields of Nb3Sn prepared for superconducting cavities. Supercond. Sci. Technol. 32, 075004, 483 DOI: 10.1088/1361-6668/ab119e (2019).484 27. Keckert, S. et al. Mitigation of parasitic losses in the quadrupole resonator enabling direct measurements of low residual 485 resistances of SRF samples. AIP Adv. 11, 125326, DOI: 10.1063/5.0076715 (2021).486 28. Aune, B. et al. Superconducting TESLA cavities. Phys. Rev. Special Top. - Accel. Beams 3, 092001, DOI: 10.1103/ 487 PhysRevSTAB.3.092001 (2000).488 29. Pira, C. et al. Evaluation of cleaning process. ARIES Deliverable Report D15.1, INFN (2018).489 30. Palmieri, V., Stivanello, F., Stark, S. Y., Roncolato, C. & Valentino, M. Besides the Standard Niobium Bath Chemical 490 Polishing. In Proceedings of the 10th Workshop on RF Superconductivity (Tsukuba, Japan, 2001).491 31. Garcia Diaz, V. et al. Thick film morphology and SC characterizations of 6 GHz Nb/Cu cavities. In Proceedings of the 20th 492 International Conference on RF Superconductivity, vol. SRF2021, 18–22, DOI: 10.18429/JACOW-SRF2021-SUPCAV007 493 (JACoW Publishing, Geneva, Switzerland, 2022).494 32. Fonnesu, D. Thin Films on Copper for Superconducting RF Cavities within the Future Circular Collider Study. Doctoral 495 Thesis, University of Siegen (2023). DOI: 10.25819/ubsi/10600.496 Acknowledgements497 The authors are thankful to their collaborators within the I.FAST programme, C. Antoine, O. Malyshev, A. Medvids, T. Proslier, 498 S. Prucnal, G. Rosaz, E. Seiler, A. M. Valente-Feliciano, R. Valizadeh, W. Venturini-Delsolaro, M. Wenskat for the fruitful 499 scientific exchange. They acknowledge the work and technical support by the INFN-LNL mechanical workshop, in particular 500 by A. Battistello, T. Bortolami, A. Minarello, and F. Pasquato. The authors also wish to thank O. Azzolini, R. Caforio, A. 501 Fetaj, G. Keppel, G. Mastrotto, F. Stivanello from the Superconductivity and Surface Technology Service at INFN-LNL for 502 the advice, support, supply of equipment and facilities and for providing chemical surface treatments. Finally, the authors 503 warmly thank A. Bianchi and R. Vaglio for the scientific insight, and TESCAN (via Assing SpA) for kindly providing the SEM 504 micrographs in Fig. 3a and 3b.505 Funding506 This research was partly supported in by the European Union’s Horizon-INFRA-2023-TECH-01 under GA No 101131435 - 507 iSAS, the European Union’s Horizon 2020 Research and Innovation programme under GA No 101004730 – I.FAST, the PNRR 508 MUR project number PE0000023-NQSTI, the INFN CSN5 experiment SuperMAD and INFN ESPP project SRF.509 Author contributions statement510 D.F.*: design of work, thin-film sample production, Tc , SEM, EDS, XRD data acquisition, analysis and interpretation, all 511 figures elaboration, manuscript writing.512 18/19 D.F.: design of work, thin-film sample production, Tc , SEM, EDS, XRD data acquisition, technical support, figure 4-9-10 513 elaboration. 514 E.C.: substrate chemical preparation, technical support. 515 S.K.: QPR data acquisition, analysis and interpretation, figure 7-8-11 elaboration. 516 J.K.: scientific advice. 517 O.K.: QPR data acquisition, analysis and interpretation. 518 M.L.: Tc, SEM, EDS, XRD data acquisition. 519 G.M.: thin-film sample production, Tc, SEM, EDS, XRD data acquisition. 520 A.S.: thin-film sample production, Tc, SEM, EDS, XRD data acquisition. 521 A.Z.: SEM, EDS data acquisition, analysis and interpretation, figure 3-4 elaboration. 522 C.P.: design of work, Tc, SEM, EDS, XRD data analysis and interpretation. 523 All authors contributed to the revision of the manuscript. 524 Additional information 525 Competing interests 526 The author(s) declare no competing interests. 527 19/19