Superconducting NbTiN Thin Films for Superconducting Radio Frequency Accelerator Cavity Applications

Transcription

1 Superconducting NbTiN Thin Films for Superconducting Radio Frequency Accelerator Cavity Applications Running title: Superconducting NbTiN Thin Films for SRF Accelerator Cavities Running Authors: Burton et al. Matthew C. Burton a), b), Melissa R. Beebe, Kaida Yang, Rosa A. Lukaszew a) The College of William & Mary, Department of Physics, Small Hall, 300 Ukrop Way, Williamsburg, Virginia Anne-Marie Valente-Feliciano a), Charles Reece Thomas Jefferson National Accelerator Facility, Newport News, Virginia a) American Vacuum Society member. b) Electronic mail: Current superconducting radio frequency (SRF) technology, used in various particle accelerator facilities across the world, is reliant upon bulk niobium superconducting cavities. Due to technological advancements in the processing of bulk Nb cavities, the facilities have reached accelerating fields very close to a material-dependent limit, which is close to 50 MV/m for bulk Nb. One possible solution to improve upon this fundamental limitation was proposed a few years ago by A. Gurevich 1, consisting of the deposition of alternating thin layers of superconducting and insulating materials on the interior surface of the cavities. The use of type-ii superconductors with Tc > Tc Nb and H c > Hc Nb, (e.g. Nb 3 Sn, NbN, or NbTiN) could potentially greatly reduce the surface resistance (Rs) and enhance the accelerating field, if the onset of vortex penetration is increased above Hc Nb thus enabling higher field gradients. Although Nb 3 Sn may prove 1

2 superior, it is not clear that it can be grown as a suitable thin film for the proposed multilayer approach, since very high temperature is typically required for its growth, hindering achieving smooth interfaces and/or surfaces. On the other hand, since NbTiN has a smaller lower critical field (H c1 ) and higher critical temperature (T c ) than Nb and increased conductivity compared to NbN, it is a promising candidate material for this new scheme. Here we present experimental results correlating film microstructure with superconducting properties on NbTiN thin film coupon samples while also comparing films grown with targets of different stoichiometry. It is worth mentioning that we have achieved thin films with bulk-like lattice parameter and transition temperature while also achieving H c1 values larger than bulk for films thinner than their London penetration depths. I. INTRODUCTION Particle accelerator facilities, like for example those at the Thomas Jefferson National Accelerator Facility (JLab) in Newport News, VA, rely on superconducting radio frequency (SRF) cavities made of bulk niobium. SRF cavities benefit from their ability to be continuously operated with low power dissipation, have lower microwave surface resistance and sustain higher maximum accelerating fields than conventional copper cavities. Recent advances in Nb cavity technology have increased the maximum breakdown field (~50 MV/m) to the point of approaching the theoretical limit given by the thermodynamic critical field H c (0) = 200 mt, at which the RF field induces the 2

3 depairing current density at the surface of the cavity. Thus, further enhancements require a new approach. Due to the fact that the SRF phenomenon is active at short distances into the superconducting material, one can feasibly exploit thin films with the goal of improving cavity performance. Thin films offer several advantages over current bulk materials, since their microstructure can be tailored and can also be employed in a multitude of material combinations that can result in cavities that could surpass current individual bulk properties. Therefore, in order to push the accelerating fields of SRF cavities beyond the current material limitations, it is possible to utilize thin film deposition on the interior of cavities using various new materials and their combinations, as well as improved deposition conditions compared to previous attempts. 2 A possible model to achieve RF field enhancement considers alternating layers of superconducting-insulating-superconducting (SIS) thin films deposited on the interior of cavities to shield the innermost surface from magnetic field vortex penetration thus enabling the possibility to even double the maximum field gradient achievable with bulk Nb. 3 In such a model, it is predicted that shielding of the inner surface is possible using layered superconducting materials, with higher critical temperatures (T c ) and lower critical field (H c1 ) values than bulk Nb, such that magnetic field vortex penetration and propagation is reduced or even suppressed resulting in enhanced cavity quality factor, Q. Here, the insulating layers (I) are needed between the superconducting layers (S) in order to (i) suppress the Josephson currents between the screening superconducting layers so that magnetic field (H) at the interior surface of the Nb cavity is smaller than the bulk H c1 150 mt of clean Nb at the operating Temperature, and (ii) to decrease losses resulting 3

4 from vortex propagation near the interior surface of the cavity. Proposed candidate superconducting materials for this approach include: NbN, NbTiN, Nb 3 Sn, MgB 2 and possibly other more exotic materials such as the recently discovered superconducting pnictides. The ultimate goal of the SIS structures is to reduce or even eliminate strong dissipation of vortices in a thin film geometry for which no thermodynamically stable position of a single vortex exists up to H H sh, the so called superheating field which is expected to exceed the critical field by 10-20% at low temperature. An important feature to take into account is that it is well known that superconducting films thinner than their London penetration depths exhibit enhanced H c1 values in the parallel field geometry compared to bulk H c1 values. However, concerning screening ability, this enhanced H c1 would be relevant primarily for a special metastable state in which the trapped field in the insulating layer is equal to the applied field H, since a thermodynamically stable vortex scenario does exist for films thicker than the London penetration which would compromise the proposed screening due to strong RF vortex dissipation. Thus, it is possible to tailor a lower critical field at the inner cavity surface by varying the layer s constitutive materials and their thicknesses thereby allowing the optimization of overall cavity performance. Experimental implementation of the SIS model will test the theoretical prediction of shielding at the inner surface, which may result in cavities with much larger maximum accelerating field gradients and improved cavity quality factor than presently available with bulk niobium. However, before this model can be used we must first understand the relationship between the thin film geometry, the material microstructure, the ultimate surface 4

5 morphology and the resulting superconducting properties of the materials considered in order to have adequate control of the process for greatest benefit. Various possible deposition techniques available may be explored in order to determine the optimal method that yields the best films and that can also be adapted to successfully coat the interior of cavities with high quality films. Work is ongoing to successfully test the SIS model in coupon samples, and some promising results have been reported recently on test cavities 4. However, despite this modest success, the latest experimental results on coupon samples suffer from reproducibility issues due to uncontrolled surface roughness and microstructure defects whose effects are magnified in the thin film geometry. Because SRF is a surface phenomenon, acting at about one micron from the surface into the active material, a decreased surface quality can have a detrimental impact on the resulting cavity performance. In fact, even bulk Nb cavities must undergo extensive surface treatments to achieve their ultimate quality and SRF performance. Since thin films have one dimension that is highly constrained, achieving films with bulk-like properties is not a trivial exercise and is also highly dependent on the material in question, the underlying substrate, the deposition parameters, and the deposition method employed. In this paper, we present our recent results correlating the microstructure of NbTiN thin films with their resulting DC superconducting properties. We also compare the results of samples grown using DC magnetron reactive sputtering using a 70/30 (%wt) NbTi target to samples grown using an 80/20 (%wt) NbTi target. II. EXPERIMENTAL 5

6 The films studied were prepared with two different composition NbTi targets, as mentioned above. The films using the 70/30 (%wt) target were prepared on single crystal MgO (100) using DC Magnetron Reactive Sputtering in a high vacuum deposition system with a base pressure in the 10-8 Torr range and a fixed target to substrate distance of 13.5 cm. While those from the 80/20 target were prepared on single crystal MgO (100), bulk Nb, and AlN (ceramic) substrates and some having AlN interlayers, to further our studies on the SIS model 5, and a fixed target-to-substrate distance of 8 cm. Prior to film growth, the substrates were cleaned in an ultrasonic bath of alternating acetone and methanol and annealed at 600 C for one hour under vacuum. After growth, the structural properties of the samples (lattice parameter, grain size etc.) were determined via X-Ray diffraction (XRD) using a PANalytical 4-circles diffractometer, while the DC superconducting properties (Tc, H c1 etc.) were measured using a Quantum Design MPMS SQUID (superconducting quantum interference device) magnetometer. A. NbTiN Thin Films A set of NbTiN films were grown using a single two inch 70/30 (%wt) NbTi target in an Ar and N 2 gas mixture with a fixed target to substrate distance of 13.5 cm. The growth parameters of the NbTiN films were optimized by varying the total pressure, mass flow rate, partial pressure of N 2, and substrate temperature while comparing the microstructure of the resulting films aiming at the most bulk-like properties. To this end, films were grown using: ~2.5 % N 2 partial pressure, a total pressure of ~4.1 mtorr, and a substrate temperature of ~600 C, since it has been shown that higher temperature growth leads to films with higher T c values 6. A second set of films were grown in a similar setup using DC Magnetron Sputtering on bulk Nb, AlN and crystalline MgO substrates with an 80/20 (%wt) NbTi target, a fixed target to substrate distance of 8 cm and with a thin AlN 6

7 interlayer grown prior to NbTiN. The films used in these experiments varied in thickness from 50 nm to 2 µm. 1. Microstructure We note that the NbTiN system has many possible phases, hence the resulting film may have different properties if any of such phases are also present in addition to the desired superconducting B1 phase. During deposition, phases present in the films can be tailored by altering parameters such as pressure. In this study, it is of utmost importance to create films in the optimal superconducting B1 phase of NbTiN. Thus, a parameter of particular interest was the nitrogen partial pressure. In order to optimize the nitrogen partial pressure, a systematic study was carried out changing only the N 2 partial pressure while holding other variables constant during growth. The films were then analyzed using XRD to determine the resulting lattice parameters, grain size, and mosaicity. From the XRD scans in Figure 1, our samples exhibit reflections from the crystalline substrates and NbTiN (002) reflections corresponding to the B1 phase; showing clear epitaxial growth of the NbTiN on the MgO(001) substrates. No other phases of NbTiN are observable in the range from deg in our samples, leading to the conclusion that our samples exhibit only the optimal superconducting B1 phase. The grain sizes of our samples were obtained from FWHM of the NbTiN (002) XRD peak and using the Scherrer equation. The grain sizes varied between 28 and 40nm, while the film with the highest H c1 value had a grain size of approximately 31nm. 7

8 Fig. 1 (Color Online) XRD scans from all of the NbTiN samples measured for this study. The letter labels correspond to those in Table 1. Samples (f)-(m), made with the 80/20 target, exhibit a clear shift to lower angles corresponding to a larger lattice parameter 8

9 than those made with the 70/30 composition target. The extra peaks present in samples in (f) & (g) correspond to the multiple phases of Nb present in the bulk Nb substrate for those samples while those in sample (m) result from the ceramic AlN substrate used. Adequate stoichiometry, to achieve the appropriate superconducting phase in a ternary compound, is very important 7, 8. From our measurements, the lattice parameter is a possible indicator of film stoichiometry. NbTiN is a ternary alloy and hence, the actual composition of a thin film depends on adequate nitrogen partial pressure as well as target composition when using a composite NbTi target. Primarily, the stoichiometry of a particular film determines its lattice parameter. Once films with close to bulkcomposition and properties have been achieved, as evident by the lattice parameters, one can begin to vary the thickness in order to improve Hc1. At this stage, effects such as grain size, surface roughness and strain will start to play a secondary role in determining the film properties. In addition, some variations are also expected, in both the grain sizes and lattice parameter, due to the strain from lattice mismatch and grain on grain growth of the films with the various substrates. We note that in order to achieve films with the desired composition, it is possible to work under deposition conditions that are near the edge for the formation of unstable phases, i.e. in an unstable reactive deposition regime. Thus, for this study, two targets of differing stoichiometry were used to investigate the dependence of the resulting films superconducting properties on the growth stability, as evidenced by the lattice parameter of the resulting films. To this end, two different target compositions, i.e. a 70/30 (%wt) NbTi target operating in the unstable reactive deposition and an 80/20 (%wt) target were tested. We note that one of the difficulties associated with controlling the composition of the resulting films is the inherent difference in the 9

10 relative sputtering rates for NbN and TiN. Hence, the final composition will not only be dependent on the nitrogen partial pressure and power, but also on the composition of the alloy target being used. The films grown using a 70/30 (%wt) NbTi target resulted in B-phase fims with lattice parameters between 4.30 and Å and the film with the largest H c1 value had a lattice parameter of 4.32 Å. While B-phase films grown using the 80/20 (%wt) target exhibited somewhat larger lattice parameter, between 4.34 and 4.36 Å with the film with the largest H c1 value having a lattice parameter of 4.35 Å. We note here the relatively low value and large variance in the lattice parameter of the films grown using the 70/30 target compared to the samples grown with via the 80/20 target, indicating process control instabilities when using the 70/30 target. This topic will be discussed further in upcoming sections. 2. Superconducting Properties The superconducting properties of bulk NbTiN are difficult to find in the literature due to the fact that NbTiN is not a natural compound. Therefore, unlike other materials, bulk values are not readily available. One can only report on the optimal values observed for experimentally achieved NbTiN. In this regard, it has been reported that NbTiN has a highest observed Tc around 18 K 9. When comparing across our samples in Table 1, we note that the samples grown using the 80/20 target had overall better DC superconducting properties than the 70/30 counterparts. The 70/30 samples exhibited T c values between 12 and 14 K while the 80/20 samples achieved T c values up to 17 K. The H c1 values in all samples were estimated from SQUID DC measurements in the parallel- 10

11 field to the film surface geometry, that were carried out minimizing systematic errors following a standard method described in the literature 10. Table 1 Properties of NbTiN samples prepared using different stoichiometry targets. Sample Substrate Target Stoichiometry (Nb/Ti %wt) % Nitrogen Lattice Parameter (Å) H c1 (Oe) T c a MgO(001) 70/ b MgO(001) 70/ c MgO(001) 70/ d MgO(001) 70/ e MgO(001) 70/ f Nb(bulk) 80/ g Nb(bulk) 80/ h MgO(001) 80/ NA i MgO(001) 80/ NA 15.2 j MgO(001) 80/ k MgO(001) 80/ NA l MgO(001) 80/ m AlN(cer) 80/ As can be seen in Figure 2, the highest H c1 observed in the films grown using the 70/30 target was around 830 Oe while the highest for the 80/20 films was on the order of 2000 Oe. However, it is also necessary to note here that all of the H c1 values listed in Table 1 are necessarily underestimates of the true value due to geometric constraints within the measurement system, i.e. the experimental difficulty in ensuring that the sample surface is absolutely parallel to the applied DC field. The H c1 enhancement quoted is only valid in absolutely parallel DC field geometry, and any component of the applied magnetic field perpendicular to the sample surface will necessarily result in a reduction in the observed H c1 value due to early onset magnetic flux penetration; hence 11

12 our clarification regarding the underestimate quality of our H c1 measurements. Also, we note the ongoing debate within the field over the exact value measured in this scenario not necessarily being a true H c1 value, but, more accurately, the field at which flux first penetrates into the material. Fig. 2 (Color Online) SQUID magnetometry data for select NbTiN samples with H c1 values of (a)~140 Oe (b)~320 Oe (c)~180 Oe (d)~500 Oe (e)~830 Oe (l)~2000 Oe and (m)~1350 Oe. Data was obtained and analyzed using the Bohmer method mentioned above. All samples exhibited purely the NbTiN (002) phase and those with larger lattice parameters similarly had higher H c1 values. The letter labels correspond to those in Table 1. 12

13 One important result from the work presented is the observed relationship between lattice parameter and the superconducting properties (H c1 and T c ). It is well known that superconducting properties are highly dependent on the composition and microstructure of the sample; since we have established that all films grew epitaxially, the dependence of the measured superconducting values with lattice parameters reported in Figure 3 is a clear indication of the relationship between superconducting properties and film composition or stoichiometry. We observe that H c1 and T c exhibit clear increases for increasing lattice parameter approaching bulk values, in contrast with previous reports 11. FIG. 3 (Color Online) H c1 vs Lattice Parameter vs T c Graph of NbTiN samples studied. Tc data exhibits an asymptotic approach to a bulk value while H c1 data shows a general trend towards some bulk value which is expected due to the dependence of H c1 on other variables. Large variability in sample performance showcases need for precise process control, especially when depositing films in the unstable reactive sputtering regime. III. CONCLUSION 13

14 NbTiN thin film samples were prepared using reactive sputtering and a systematic study of the superconducting properties on films with different thickness below the London penetration depth indicated that it is also possible to achieve successful thin layers with this material with lower critical field enhancement, up to 2000 Oe at the operating temperature, capable of shielding bulk Nb cavities in the SIS model. We note that the choice of deposition parameters as well as target composition have significant effect on the superconducting properties of the films, showcasing a possible path to control the superconducting layers in order to follow multilayer studies similar to the ones carried out previously in our group with NbN based tri-layers 12. However, more work must be done to identify the proper process controls needed to prepare samples with consistent quality and stoichiometry. The relatively large variance in the lattice parameters of the samples created using the 70/30 target compared to the 80/20 target showcase the sensitivity of reactive sputtering of ternary compounds when grown under the unstable reactive regime. There are yet some possible process modifications to be explored in order to better control and overcome growth instabilities, such as co-deposition of Nb and Ti in a N 2 + Ar atmosphere, as well as the use of mass flow controls with a feedback loop for real-time pressure monitoring. More work is under way in order to obtain consistent high quality films for the proposed materials that have potential to be realized on an SRF cavity. ACKNOWLEDGMENTS We would like to thank the SRF Institute at Jefferson Lab and acknowledge support from the Defense Threat Reduction Agency under grant HDTRA

15 1 A. Gurevich, Appl. Phys. Lett. 88.1, (2006). 2 S. Calatroni, Physica C 441.1, (2005). 3 W. M. Roach, J. R. Skuza, D. B. Beringer, Z. Li, C. Clavero and R. A. Lukaszew, Supercond. Sci. Tech , (2012). 4 C. Z. Antoine, J. C. Villegier and G. Martinet, App. Phys. Lett , (2013). 5 T. Shiino, S. Shiba, N. Sakai, T. Yamakura, L. Jian, Y. Uzawa, H. Maezawa and S. Yamamoto, Supercond. Sci. Tech. 23.4, (2010). 6 L. You, R. K. Singh, H. Liu, S. Y. Wu, R. Hu, D. Durand, J. Bulman, J. M. Rowell and N. Newman, IEEE T. Appl. Supercon. 15.1, (2005). 7 L. Zhang, W. Peng, L. X. You and Z. Wang, App. Phys. Lett , (2015). 8 K. Vasu, M. G. Krishna and K. A. Padmanabhan, J. Mater. Sci. 47.8, (2012). 9 J. R. Gavaler, D. W. Deis, J. K. Hulm and C. K. Jones, App. Phys. Lett , (1969). 10 C. Bohmer, G. Brandstatter and H. W. Weber, Supercond. Sci. Tech. 10.7A, A1 (1997). 11 H. Myoren, T. Shimizu, T. Iizuka and S. Takada, IEEE T. Appl. Supercon. 11.1, (2001). 12 W. M. Roach, D. B. Beringer, Z. Li, C. Clavero and R. A. Lukaszew, IEEE T. Appl. Supercon. 23.3, (2013). 15

16 Intensity (arb. units) (a) NbTiN(002) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (l) (m) MgO(002) theta (deg.)

17 (a) (b) Magnetic Moment (emu) (c) (d) (e) (l) (m) Applied Field (Oe)

18 Hc1 Tc Asymptotic Fit of Tc Linear Fit of Hc Hc1 (Oe) Lattice Parameter (nm) Tc (K)