JSA Graduate Fellowship 2008-2009 Final Report

Daniel Bowring∗

1 Introduction

This paper presents a synopsis of the work doneunder my 2008-2009 JSA Graduate Fellowship.Prior to the summer of 2008, I had been pursuingseveral parallel research projects1. The receipt ofmy JSA fellowship coincided with a decision be-tween me and my advisors2 to pursue multilayerthin films for superconducting radio frequency(SRF) cavities as the basis for my PhD disserta-tion. The 2008-2009 JSA Graduate Fellowshiptherefore supported the first part of my disser-tation research. That work culminated in thefall of 2009 with a poster presented at the 14thInternational Conference on RF Superconductiv-ity (SRF 2009) and a successful dissertation pro-posal to my advisory committee.

2 Work completed in ’08-’09

The basis for my work under the JSA Gradu-ate Fellowship, as well as for my dissertation,is a paper published in 2006 by Alex Gurevich[1]. The lower critical magnetic field Hc1 of anSRF cavity may be increased by coating the cav-ity interior with alternating thin layers of super-conductor and insulator. Such multilayer filmsscreen the cavity field from the bulk supercon-ductor, as shown in Figure 2. The overall effectis a higher effective Hc1 and a higher quality fac-tor Q than an equivalent bulk Nb cavity (Figure2). This would allow higher accelerating fields ina given cavity without quenching the supercon-ducting cavity walls. Higher accelerating fields

∗dbowring@jlab.org1e.g., see http://conferences.jlab.org/tfsrf/2Larry Phillips at JLAB and Blaine Norum at UVA

0 100 200 300 400 500 600 700 800 9000

0.2

0.4

0.6

0.8

1

1.2

H/H

0

Multilayer film depth (nm)

Figure 1: As in [1], magnetic field at cavitysurface d = 0 nm is screened from the bulk(d ≥ 560 nm) by a thin layer of NbTi (white).

would mean more efficient particle accelerators,since fewer cavities (and therefore less liquid he-lium and tunnel space) would be required for agiven final beam energy. To date there is no ex-perimental confirmation of this theory.

During academic year 2008-2009 I designed anexperimental program to verify and parameter-ize the performance of multilayer films. Exper-iments are conducted on small samples, whichare quick and cheap to prepare. (This approachalso decouples multilayer film performance fromthe more complex and variable cavity fabrica-tion process.) Briefly, 4 cm square samples aredeposited via sputtering, using facilities at TJ-NAF’s Test Lab. A microstrip disk resonator3

is laminated to the sample surface and appliesa known magnetic field. Hc1 of the sample isthen measured by ramping the field supplied by

3Disk resonators are commonly used as planar mi-crowave filters. See e.g. [2].

1

0 50 100 150 200 250 300 350 4001.8

1.82

1.84

1.86

1.88

1.9

1.92

1.94

1.96

1.98

2

Hc1

(mT)

Q0 (

109 )

Q0 for 3 resonator configurations

Nb3Sn

NbMultilayer

Figure 2: Unloaded Q and Hc1 for a bare Nbcavity (red), a bulk Nb3Sn cavity (blue) and amultilayer Nb3Sn/alumina/Nb cavity (green).

the disk resonator while monitoring the overall Q

of the system. As suggested in Figure 2, a sharpdegradation in Q indicates thermal runaway andtherefore that Hc1 has been exceeded. Compar-ison with single-layer sample performance thenshows the efficacy of the multilayer approach.

A key feature of the experimental design isthat multilayer samples are deposited simultane-ously with single-layer control samples in orderto eliminate disparities in film quality betweenthe two. This allows for a direct comparisonbetween multilayer and control films. The twosamples are also tested side-by-side in the samecryostat for similar reasons. For more details,see [3].

The rough dimensions of the resonator andsamples can be obtained from Maxwell’s equa-tions. However, simulation work was required tomodel the input lines that couple power to thedisk and to study radiation losses. An exampleof such a simulation is shown in Figure 2.

3 Current Work

With the design work completed, I expect tostart taking initial measurements before March2010. Work is underway to accomplish that goal:

Figure 3: CST Microwave Studio finite ele-ment simulation of the magnetic field from a mi-crostrip disk resonator. The disk resonator isshown at center, flanked on either side by rect-angular input feedlines.

• Procurement began once the simulationwork was finished and the sample design wasfinalized. At this point, all the major com-ponents (cryostat insert, low-level RF con-trol systems, mechanical coupling tuners,etc.) for the experiment have been pur-chased, found, refurbished, or built.

• Test Lab VTA crane certification for thecryostat insert is underway.

• The sputtering chamber at the Test Lab hasbeen recently renovated and is in the finalstages of re-commissioning. I have procuredsputtering targets in anticipation of a func-tioning chamber.

• The computer control of the mechanicaltuners is nearly finished.

• Once the mechanical issues related to in-put power coupling are resolved, I can begintesting samples.

In September of 2009, the work described herewas presented as a poster at the 14th Interna-tional Conference on RF Superconductivity inBerlin, Germany [3]. Travel during that confer-ence was supported by my JSA Graduate Fel-lowship. The above work was also presented to

2

my advisory committee for my dissertation pro-posal. That proposal was accepted on October30, 2009.

4 Future Work

Work in the coming year will be supported bythe 2009-2010 JSA Graduate Fellowship. Thenext significant near-term goal is the sputteringof samples, followed immediately by cryogenictesting. The two long-term goals for 2010 are(1) presenting a poster at IPAC 2010; and (2)writing and defending my dissertation.

References

[1] A. Gurevich, Applied Physics Letters 88,(2006).

[2] A. Jenkins et al., IEEE Transactions on Ap-plied Superconductivity 7, 2793 (1997).

[3] D. Bowring and L. Phillips, in Proc. 14th In-

ternational Conference on RF Superconduc-

tivity (Helmholtz Zentrum, Berlin, Germany,2009), No. TUPPO043.

3

A METHOD OF EVALUATING MULTILAYER FILMS FOR SRFAPPLICATIONS ∗

D. Bowring† , University of Virginia, Charlottesville, VA, U.S.A.L. Phillips, JLAB, Newport News, VA, U.S.A.

Abstract

The lower critical magnetic field of an SRF cavity maybe increased by screening the interior fields from the cav-ity bulk. This screening could be accomplished by coatingthe interior of a cavity with alternating layers of insulatorand superconductor, each of which is thinner than the Lon-don penetration depth of the superconductor. This idea hasbeen proposed by Gurevich [A. Gurevich, Appl. Phys. Lett.88, 012511 (2006)]. We have developed a method for mea-suring the behavior of such a multilayer system. A super-conducting disk resonator is deposited on top of a smallmultilayer sample. The small sample approach allows forflexibility in the evaluation of many different materials andconfigurations. The onset of magnetic flux penetration inthe superconductor can be observed from the change in res-onator Q and the detection of flux flow voltages. Since thedisk resonator applies a known field to the multilayer sys-tem, the lower critical magnetic field of the system may bemeasured. This paper presents design analysis for an ex-perimental program for multilayer film evaluation.

INTRODUCTION

Gurevich [1] has proposed a method for raising the ef-fective lower critical magnetic field Hc1 of a superconduct-ing radio frequency (SRF) cavity by means of a multilayerthin film coating on the inner surface. The coating consistsof alternating layers of insulating and and superconduct-ing films, where the superconducting layer thickness is lessthan the London penetration depth. In principle, this typeof coating on an elliptical cavity could increase the peakmagnetic field (and therefore the accelerating gradient) pastthe ∼180 mT limit for bulk Nb. This paper presents theexperimental design for an evaluation of the multilayer ap-proach.

An overview of [1] is presented in this section. We treatthe example of a multilayer structure composed of layersof Nb-Ti and aluminum oxide. The superconducting layersscreen the cavity fields from the bulk, as shown in Figure 1.The field decays exponentially across the superconductingfilm, establishing a current imbalance between the top andbottom and skewing the free energy per unit length G/L

∗Work supported by the United States Department of Energy and Jef-ferson Science Associates. Poster supported by a Student Travel Grantfrom SRF 2009.

† dbowring@jlab.org

0 100 200 300 400 500 600 700 800 9000

0.2

0.4

0.6

0.8

1

1.2

H/H

0

Multilayer film depth (nm)

Figure 1: As in [1], the normalized field H/H0 as a func-tion of sample depth for alternating layers of Nb-Ti andAl2O3. λ ∼ 250 nm and the layer thickness d ∼ 0.75λ.

for flux vortex motion [1, 2]:

G/L =φ2

0

4πμ0λ2ln

[d

1.07ξcos(

πu

d)]− φ0

∫ d/2

u

J(z)dz.

The first term represents the kinetic energy of a movingvortex, and the second term is the Lorentz force contribu-tion from the net current across the film. u is the depth ofthe vortex in the superconducting layer, ξ is the coherencelength (≈ 0.7 nm for Nb-Ti), and we assume a penetrationdepth λ ∼ 250 nm [2]. This skewing of the free energy isshown in Figure 2.

Figure 2: As in [1], the normalized vortex free energyG/G0 as a function of displacement u from the center ofa superconducting Nb-Ti film. The red line represents thebarrier to flux motion arising from a net current across thefilm.

As the current work is an evaluation of the multilayer

Proceedings of SRF2009, Berlin, Germany TUPPO043

04 Measurement techniques

309

approach, we use Nb-Ti thin films for their relatively lowvalue of Hc1. The objective is to quantify the performanceof the multilayer approach by exceeding Hc1 ≈ 50 −100 Oe in small multilayer samples, rather than to achievefields comparable to those in elliptical high-gradient cavi-ties. This paper presents the experimental design for sucha proof-of-principle study.

EXPERIMENTAL DESIGN

Any RF structure designed to evaluate multilayer filmsmust satisfy several basic requirements. First, the appliedfield must be parallel to the sample surface. Hc1 for athin film is significantly higher for parallel than for obliquefields [3]. Parallel magnetic fields also mimic the field dis-tribution for the TM011 accelerating mode of a typical el-liptical cavity.

In addition, the onset of flux penetration in the sam-ple should be the only factor influencing resonator per-formance. We therefore require the magnetic field to bestrongest on the sample surface, ensuring that any dissipa-tion from RF vortex motion occurs within the multilayerstructure. Equivalently, the sample must have a lower Hc1

than any other part of the experimental apparatus. Samplegeometry should be flat, with minimal surface area to facil-itate film quality control and analysis. And finally, multi-pacting should be minimized so that the onset of flux pen-etration and subsequent vortex motion are relatively strongcontributors to RF losses. The issue of vortex dissipationwill be discussed in more detail below.

To satisfy the above requirements, we have designeda microstrip disk resonator that mounts on top of inter-changeable flat multilayer samples.

Disk Resonator

The disk operates in the TM01 mode, such that the mag-netic field is everywhere parallel to the sample. Figure 3shows the magnetic fields from a simulation of this configu-ration for a 4 cm diameter disk operating in the TM01 modeat 2.9 GHz. Note that the magnetic field vanishes at thedisk edge, as do the corresponding electric currents. Thisminimizes radiation losses from edge effects [4]. Since thedisk is flat, field emission is minimized. Also, since thedisk is immersed directly in liquid helium with no cavityenclosure, multipacting is minimized. The disk is madefrom high-RRR bulk niobium, so that it can reliably pro-duce magnetic fields well above the lower critical field ofthe Nb-Ti multilayers while itself remaining free of fluxpenetration.

Sample Design

Samples are 8 cm square and of varying thickness, de-pending on the penetration depth of the superconductorused. For Nb-Ti (λ ∼ 250 nm) the film and dielectric lay-ers are each 190 nm thick. To obtain a value of Hc1 which

Figure 3: CST Microwave Studio finite element simulationof the magnetic field from a microstrip disk resonator. Thefield is entirely tangential to the disk surface, with no ra-dial or longitudinal components. Stripline for capacitiveRF coupling is seen on either side of the disk.

is significantly lower than that for the bulk Nb disk res-onators, these films are sputtered rather than using the en-ergetic deposition methods available at Jefferson Lab [6].At least initially, film surface roughness and texture are oflittle concern.

Each multilayer sample is, via masking, deposited simul-taneously with a control sample on the same substrate, asshown schematically in Figure 4. The control sample is asingle thin layer of Nb-Ti, identical to that of the multilayersample, but lacking the bulk Nb ground plane. This effec-tively forms a traditional disk resonator, as in [5]. Deposit-ing both multilayer and control samples simultaneously inthe same chamber eliminates variations in film quality andthickness, allowing for a direct comparison of, for example,Q vs. E curves.

Figure 4: Schematic of the experimental apparatus. Mul-tilayer and “control” samples are deposited simultaneouslyon the same substrate.

RF Coupling

To allow the simultaneous evaluation of multilayer andcontrol samples, two high-RRR bulk Nb disk resonatorsare positioned side-by-side on a movable Teflon sheet, as

TUPPO043 Proceedings of SRF2009, Berlin, Germany

04 Measurement techniques

310

in Figure 4. RF power is capacitively coupled to the res-onators via Nb stripline [7]. Variable coupling is achievedby mounting movable Nb plates above the gap between res-onator and stripline, creating a variable gap capacitance.The Teflon sheet is used to help match the relatively largeedge impedance of the disk resonator to the 50Ω input andoutput couplers. In this arrangement, RF power can be sup-plied to one or the other disk resonator without removingthe samples from the liquid helium or disturbing the cou-pling.

The disks, Nb stripline, and Teflon sheet together makeup a movable module. Fixed SMA-type RF connectorsare used to couple power to the resonators. Using thesame fixed resonator and coupler scheme ensures that fieldstrength and coupling constants do not vary between sam-ples.

MEASUREMENTS

The RF breakdown field is characterized by entrant fluxvortices in the thin superconducting layers. The onset offlux penetration in these layers can be measured via tworelated phenomena: flux flow voltage signals propagatingin the thin dielectric layer, and a change in the Q of theresonator.

Flux ow oltage

A transport current density �J on the thin film surfacegenerates a Lorentz force per unit vortex length �F/L =�J×�φ0, where �φ0 is the flux quantum. If the Lorentz force islarger than the vortex pinning force, the fluxoid will exhibitviscous flow across the film surface, perpendicular to �J .This flux motion, in turn, induces a dissipative electric fieldparallel to �J . Essentially, the motion of flux lines across thefilm creates a propagating voltage signal in the dielectric,which may be measured at the sample edge [3, 8].

Resonator Q

Because vortex motion is dissipative, flux flow voltagesignals can be directly correlated with a drop in resonatorQ. Below Hc1 we calculate Q0 ≈ 6× 107. After the onsetof flux penetration, we can use the thermal feedback modelfor vortex dissipation to calculate the extent of Q degrada-tion [1, 2]. This approach requires knowledge of the tem-perature dependence of the thin film surface impedance,as well as the penetration depth. Therefore, measurementsmust be made for each film. The new surface impedancecharacterization (SIC) facility [9] at Jefferson Lab will beuseful in making such measurements.

CURRENT AND FUTURE WORK

Experimental design is complete and construction of theapparatus is underway. We are currently assembling thecryostat insert and the RF couplers. In parallel with this,we have begun sample preparation. Note that the Nb-Ti

films discussed here are a preliminary effort. In principle,any superconductor of any thickness less than λ may beused with this system. We intend to evaluate as many su-perconducting materials as possible.

Calculations of loaded Q and RF breakdown field de-pend heavily on the thin film penetration depth, coherencelength, and thickness. Very rough a priori estimates arepossible, but accurate predictions of these values requireTEM and SIC measurements of prepared samples. Fur-thermore, analytic calculations of the RF breakdown field[1] use a vortex image approach that assumes a microscop-ically flat film surface. We would like to model the effectof film roughness on the multilayer breakdown field.

ACKNOWLEDGEMENTS

The authors would like to thank Jean Delayen and TomGoodman for their helpful discussions.

REFERENCES

[1] A. Gurevich, “Enhancement of rf breakdown field of su-perconductors by multilayer coating”. Appl. Phys. Lett. 88,012511 (2006).

[2] G. Stejic et al. “Effect of geometry on the critical current ofthin films.” Phys. Rev. B 49, 2 (1994).

[3] M. Tinkham, “Introduction to Superconductivity, 2nd Ed.”Dover Publications, Inc. Mineola, NY. 1996.

[4] A.P. Jenkins et al. “Microstrip Disk Resonators For FiltersFabricated From TBCCO Thin FIlms.” IEEE Trans. App. Su-percond. 7, 2 (1997).

[5] A.G. Derneryd, “Analysis of the Microstrip Disk Antenna El-ement.” IEEE. Trans. Ant. Prop. AP-27, 5 (1979).

[6] A.-M. Valente, et al, “Niobium thin film cavity deposition byECR plasma.” Proc. EPAC 2004, Lucerne, Switzerland.

[7] P. Benedek, P. Sylvester, “Equivalent Capacitances for Mi-crostrip Gaps and Steps.” IEEE Trans. Microwave TheoryTech. 20, 11 (1972).

[8] Y.B. Kim, C.F. Hempstead, A.R. Strnad. “Flux-Flow Re-sistance in Type-II Superconductors.” Phys. Rev. 139, 4A(1965).

[9] B. Xiao et al. “RF Surface Impedance Measurement of Poly-crystalline and Large Grain Nb Disk Sample at 7.5 GHz.”Proc. SRF 2009, Berlin, Germany.

Fl V

Proceedings of SRF2009, Berlin, Germany TUPPO043

04 Measurement techniques

311