CVD SUPERCONDUCTING RF CAVITIES:

PAST, PRESENT and FUTURE

•PAST: CVD niobium• PRESENT: CVD RF Cavity

•FUTURE: Thin Films

Thin Films Workshop INFN (Padua)ITALY

October 9-12, 2006

L.N. Hand Physics Dept., Cornell University Ithaca, NY, USA

I do not see any intrinsic reason why a CVD process cannot be engineered into a relatively low-cost and technically superior mass production method for Phase II or III of the ILC.

But who will do this engineering?

Consider the potential advantages of CVD fabrication for SRF cavities :

•CVD Cavities are seamless and require no electron beam welding•Reactor grade niobium (low RRR) can be used

•High tolerances are possible

•Surface characteristics can be controlled ; little or no etching, high pressure rinsing, electropolishing or expensive annealing.

Past: CVD niobium

•#1: hydrogen reduction of niobium pentachloride.

The reaction is 5H2 + 2NbCl5 10HCl + 2Nb

•#2: thermal reduction of NbI5 (iodide process)

2Nb + 5I2 2NbI5 2 Nb +5 I2 (hydrogen-free process)

900 oC

400 oC 900 oC

(Historically, #1 is the way niobium was first isolated in 1864.)

Well-known ways to deposit niobium by CVD:

Present: First CVD SRF cavity Ultramet* recently fabricated this CVD solid Nb cavity by method #1:

The next slide shows how this cavity was made.

(* www.ultramet.com)

+Barrier film* to prevent contaminating the niobium with the graphite

Graphite mandrel

SRF cavity

mandrel removed*

CVD Process

There is too much carbon in this niobium! (0.7 at.%)

Mostly on surface?

*Proprietary process

This lack of development is caused by the lack of funds to pursue this R (Research) as opposed to D (Development).

Almost no US $$$ are available for SRF research. The reason is:

The ILC Phase I will be developed with existing technology.

This cavity has not been tested with RF. Yttrium or Titanium treatment has not been tried.

No industrial engineering design exists for cost reduction with mass production, or work on improving RRR or on the surface treatment appropriate to CVD cavities.

Film-based cavitiesYou can save a lot of money, since you only need ~10 penetration depths (=40 nm for niobium). A 2006 review by Sergio Calatroni of CERN, in PhysicaC, discusses the problems, and efforts to cure them. Some of the problems of film-based cavities may be:

•Defects within or 2 of the surface or on the surface.Insufficient attention has been paid to this topic, including trapping of impurities like oxygen in defects.

•The grain size for the CERN Nb/Cu films is 100 nm. This is 10,000 times smaller than for conventional SRF cavities, (for which grain sizes are > 1 mm and are not important).Grain boundaries are themselves one-dimensional defects. Grain boundary diffusion is much faster than diffusion in the bulk Nb.

•Local(microscopic) thermal conductivity of the film itself (has it ever been measured?) may be poor compared to bulk Nb.

•Kapitza resistance at two interfaces: Nb/Cu and Cu/LHe(II)

The microscopic film structure is very important.

THE FUTURE OF CVD SRF IS:My opinion is:

•Seamless titanium tubing with a c-BN barrier film a few nanometers thick. The superhard c-BN will allow dissolving the mandrel in HF or perchloric acid Ti etch without interaction with, or contamination of, the niobium.

This mandrel type can be mass-produced cheaply and with precision in multicell units by flow or hydro-forming. Titanium has a CTE close to that of niobium. Ti-6AL-4V alloy can be polished to a few angstroms roughness (Rq).

High quality large grain high quality superhard c-BN (cubic boron nitride) films can now be grown by CVD. A thin diamond or alumina seed layer may be necessary to grow the c-BN, however.

Proposed new type of mandrel:

The barrier film is critical for success, because it controls the growth and structure of the niobium film. Diamond or carbon nitride could be better for 5-10 nm barrier films.

The substrate/barrier film is as important as the superconducting film itself, because it determines the nanostructure of the superconducting film!

Also, the stress/strain of the superconducting film is determined by the substrate.

A15 Compounds:A3B

There is experience with Nb3Sn cavities, but none with SRF Nb3Al (or Nb3Ge). In Padua, there have been experiments with A15 SRF and other compounds such as MgB2.

Nb3Sn cavities have not been very encouraging, but these are “early days”. Q at low fields is much higher than with Nb, however. The superheating H is about the same as for Nb, so there is no improvement in the maximum Eacc.

I want to look into Nb3Al. Why?

My reason for being interested in Nb3Al:

•We began by co-sputtering Nb and Al onto a silicon <111> substrate.

The result was: a 230 nm epitaxial film of Nb3Al on the first try! BUT…

No one else is working on it! And, we could learn more about SRF materials.

•DC Nb3Al conductors are made by the “jelly-roll method”. Over 13,000 papers have been written on Nb3Al, but I have found none on its RF superconducting properties!•Nb3Al might be made by a CVD “jelly-roll” process using trimethylaluminum (TMA) and an niobium organometallic compound.•BUT….TMA is very dangerous. It is pyrophoric, explosive and poisonous. Special equipment is needed to use it.

Lithium aluminate <100> is a close match (0.35%) to the lattice of Nb3Al. I will try this--- LiAlO2 substrates are commercially available.

The “BUT” means:

Our first Nb3Al film was not even superconducting at 5oK! (Tc=17-18oK for Nb3Al) It was highly strained, though epitaxial.Si<111> is the wrong substrate for this material.

At 400oC, use PECVD*.

No one has ever tried to do a “jelly-roll” PECVD process for niobium-aluminide films, but hafnium-aluminate has been produced this way.

At what temperature should we do CVD?

900oC is too high for cheap mass-production, but it is not necessary to go that high.

* plasma-enhanced chemical vapor deposition

We would like to find a PECVD process which works at low temperatures, say 400oC.

Niobium (V) halides vaporize at the following temperatures:

NbF5: 235oC, NbBr5: 365oC, NbCl5: at 247-254oC. NbI5 dec. at 270oC to NbI4, at 430oC to NbI3. NbI3 dec. > 510oC.

Try a PECVD version of the “iodide” process* with NbBr5:

2Nb + 5 Br2 + He

Of course, the “devil is in the details”. How do you get rid of the Br2? Also, you want to ionize the Nb so that you can use an electric field.Plenty more to think about!

* method #2

2NbBr5 + He*(23S1) + e(plasma)

A completely different kind of film: MgB2 (Tc=38-39 oK)

This is the “Holy Grail” of SRF thin films.

Why?

Because we could imagine running accelerators at 20 oK instead of 2 oK. This is a big saving in refrigeration costs.

And it is a cheap material.

NEXT: Magnesium Diboride

I saved the best for (almost) the last:

At Penn State, polycrystalline films have been deposited on YSZ* . They have the the lowest microwave surface resistance yet seen in MgB2. Prof. Xi will speak at this Workshop. *Yttria(Y2O3)-Stabilized Zirconia(ZrO2)

Could we try PECVD: He* + Mg(vapor) + B2H6He+MgB2 + H2 ?

Success with magnesium diboride means:

2. A mass production cost for nine-cell units of less than 10,000 (2006) Euros each.

1. At 20oK: A cavity surface resistance Rsurface

< 200 nanoOhms*, at an rf magnetic field Hrf > 2000 Oe (Brf > 200 mT, Q> 109 at 1.3 GHz.)

*to save on refrigeration costs, it has to be lower than this.

Returning to niobium films:

I would like positive or negative comments from people at this Workshop about two possible ideas for Nb films to be made at 400 oC or less:

Both ideas involve metalorganic compounds of niobium (like diethylniobium).

1. Use a UV laser to decompose the metalorganic compound and deposit a niobium film.

And/or: 2. Use the excited triplet 2S state of He gas (50 millisec. lifetime, 20 eV energy above the ground state) to do the same thing or in combination with 1.

“There are many ways to skin a cat.”—American saying.

•In the future, we need to develop seed layers for depositing films with the right SRF characteristics. This has high priority.

•Find out what causes Rsurface(residual) in films like Nb, Nb3Al and MgB2?

• We must have much better theoretical and practical understanding of Rsurface and Hsuperheating in realistic thin films, with defects and dislocations, etc. under high current conditions .

What do we need to do?

•What really determines Hsuperheating? How long does it take to form a vortex and what is the effect of pinning and de-pinning the vortices? Does Hsh scale with Hc? What is the frequency dependence of Hsh?

•A way to test films under high field RF conditions without having to build a cavity each time would help a great deal.

•I would like to find a way to make small-scale tests of different ideas for PECVD. Suggestions will be welcome! I do not have the facility to do this myself. Would someone like to collaborate on this?

Conclusion: remarks on “beyond the first phase of

the ILC”

Excellent question! We do not truly understand leptons at all. Why are there three charged leptons and three neutrinos?

I once asked Feynman: “how big is a quark?”

He replied: “how big is an electron?”

We can’t imagine what new physics will be uncovered by the ILC. But I think we will need higher energies beyond 0.5-1 TeV to do even more new physics.

Finding the Higgs is not the end of high energy physics, any more than the end of the Cold War was the end of history. Not finding the Higgs is even more challenging! Also, there is much more to understand* at the > 1 TeV scale than only the Higgs.

*I am referring to the “hierarchy problem”.

New technology requires years of patient development. We should start doing this

development now, even if 99% of the world effort goes into phase I of the ILC.

“Fools rush in, where angels fear to tread..”

--Alexander Pope, English poet, 1688-1744

Remember, it is a step-by-step slow process:

I hope that this Workshop will stimulate further ideas and research here in Padua, in Europe, Asia and elsewhere in the world.

Arrivederci and good luck!

Acknowledgements

I would like to thank the staff of CCMR (Cornell Center for Materials Research, Prof. Melissa Hines, Director) for their administrative and technical help plus encouragement over the last years.

Also, I have been greatly helped by the staff at the CNF (Cornell NanoScale Facility) for not only giving direct technical assistance, but being willing to try ideas outside the CNF core mission of nanoelectronics. Mike Skvarla and Gary Bordonaro deserve special mention, among others.

Both the CNF and CCMR are funded by the US NSF.

Shawn McNeal of Ultramet provided the information about the prototype SRF cavity built recently with a small amount of temporary DOE funding. He also provided the CVD Nb samples for analysis in 2005 and 2006.

Prof. Bill Frisken of York University in Toronto, Canada was a valued friend and collaborator from 1998-2005, when he retired.

Finally, I want to thank Larry Phillips of JLAB for many good ideas and comments, which were especially helpful for this talk.