Significant improvement of Jc in small Ds RRP® wires through heat treatment changes and phase control Using Nausite to our advantage η ε Charlie Sanabria1, Michael Field2, P. J. Lee1, Hanping Miao2, D. C. Larbalestier1 and Jeffrey Parrell2 Nausite Good evening everyone, my name is Charlie Sanabria, I come from the Applied Superconductivity Center at the National MagLab, of The Florida State University, and I have been studying the RRP® wires from Oxford Superconducting Technologies in the last couple of years—using metallography and microscopy—in an attempt to optimize their properties. And today I will be showing you a significant improvement of Jc in small Ds RRP® wires through heat treatment changes and phase control. 1 Applied Superconductivity Center, NHMFL, Florida State University, Tallahassee, FL 32310, USA 2 Oxford Superconducting Technology 600 Milik Street Carteret, NJ 07008, USA September 8th, 2016

Outline ε η 28% Jc (16T) increase! The ‘big picture’ Current RRP® limitations Hi-Lumi and FCC demands Heat Treatment “revelation” Ian Pong, et. al. (2013) The “Nausite membrane” The Good, the Bad and the Ugly Key Findings The 215°C dwell is useless Nausite growth is strongly dependent on temperature Cu diffusion is weakly dependent on temperature Conclusion Promoting Cu diffusion while inhibiting Nausite growth can increase Jc Our new heat treatment improved Jc (16 T) in small Ds wires by 28% (preserving RRR) Useless! η ε Nausite 28% Jc (16T) increase! Now, the outline of the talk is the following. First of all, I will obviously start with the big picture in mind, what is currently holding RRP® back, and what are the targets for the Hi-Lumi and the FCC Then we will talk about a heat treatment “revelation” that has happened in the last few years, starting with a paper from Ian Pong, et. al. followed by some of the things we have found since then like what we call “the Nausite membrane”….and the good the bad and the ugly about this After that we will talk about a few key findings like the fact that the 200°C dwell is useless among other things, And finally we will conclude that: Promoting Cu diffusion while inhibiting Nausite growth can increase Jc Our new heat treatment improved Jc (16 T) in small Ds wires by 28% without affecting RRR So, let’s jump onto the first section, the big picture… Picture by Luca Bottura

There is a dramatic drop in Jc as the subelements get smaller Different wire architectures at the same size Same wire architecture at different Sizes Field et al. IEEE Trans Appl Supercond 24, 1–5 (2014). Ø 1.5 mm e.g. You see, currently, there is a dramatic drop in Jc as the subelements get smaller—as shown in this graph. However, let me give you some examples so we understand what is going on: What you have here are different wire architectures at the same wire size, and each one gives you a Jc that is strongly dependent on subelement size For example at 0.7 mm this 61 stack with a subelement close to 70µm very often gives you about 1,500 A/mm2 at 15 T, however at that same wire size a 217 stack with a subelement closer to 35µm would give you a much lower value. This effect is also seen for the same architecture at different sizes. For example this 127 stack, at 1.5 mm in wire diameter can produce and outstanding 1,700 A/mm2 at 15 T, however when that same wire is drawn down to 0.5 mm its Jc drops down to less than 1,000 A/mm2. Of course these are just examples, but you get the idea, and this plot captures the essence of it: there seems to be a threshold at which the characteristic properties of RRP® wires no longer hold, and in fact they drop quite dramatically. Now let’s take a look at what the collider demands are: e.g. e.g. Ø 0.7 mm e.g. Ø 0.5 mm 108/127 54/61 90/91 108/127 132/169 108/127 198/217 108/127 108/127 108/127

Current Hi-Lumi LHC requirements Final Targets for FCC Conductor* Current Hi-Lumi LHC requirements for LARP Quadrupoles A. Ballarino, presented at the FCC week 2016. S-HiLumi-doc.40; Rev. No.: Original Release; Date: 05-May-2015 As far as the Hi-Lumi goes, RRP® is already there. The wires are made and the magnet manufacturing should start very soon. However, looking further in the future, the final targets for the FCC are much more demanding! And we have a pretty long way to go Field et al. IEEE Trans Appl Supercond 24, 1–5 (2014). * Values presented at 16 T (1500 A/mm2) Kramer extrapolation to 15 T = 1865 A/mm2

Outline ε η The ‘big picture’ Current RRP® limitations Hi-Lumi and FCC demands Heat Treatment “revelation” Ian Pong, et. al. (2013) The “Nausite membrane” The Good, the Bad and the Ugly Key Findings The 215°C dwell is useless Nausite growth is strongly dependent on temperature Cu diffusion is weakly dependent on temperature Conclusion Promoting Cu diffusion while inhibiting Nausite growth can increase Jc Our new heat treatment improved Jc (16 T) in small Ds wires by 28% (preserving RRR) η ε Nausite Now let’s go into the heat treatment revelation we had recently

“There is limited Sn diffusion outwards” during the 400°C dwell – Pong, et. al. (2013) η voids Pong, et al. Pong, et al. “There is limited Sn diffusion outwards” during the 400°C dwell due to a barrier of the so-called Nausite (Sn-Nb-Cu phase) It all started with a paper by Ian Pong in 2013 when he was studying RRP® wires and other similar wires. He noticed that, there is limited Sn diffusion outwards during the 400°C dwell due to a barrier of the so-called Nausite phase Although he didn’t call it as such back then. I think I’m the one pushing for the name to stick after Michael Naus, who discovered it as a graduate student back in 2001. Pong also pointed out that pores are observed above 415°C in the filamentary region, which was very surprising because it showed that the Cu migrates to the core before the Sn diffuses out into the filament pack. “Pores are mostly observed above 415°C in the filamentary region” I. Pong, et al. Supercond. Sci. Technol., 26 (10) Oct. 2013

The 400°C dwell is where all of the interesting kinetics happens Quartz tubes Argon gas ~13 cm long pieces This talk will focus on the fact that at ~ 400°C: A Nausite ‘membrane’ forms Cu diffuses through the Nausite membrane into the core Nb3Sn This called our attention, and following Ian’s footsteps we set up an experiment, just like his, in which we took small pieces of wire inside quartz tubes with a string on them so I could pull them out and “quench” them as the heat treatment ran. Once quenched, we mounted and polished them, to see what phases had formed up to that point. And we did the same at different points in the heat treatment. As such... Now let’s stop here for a moment and observe these. Let’s appreciate the incredibly different morphologies we have here. How did this mess turn into this beauty? And then into that!? What happens inside RRP® wires is pretty confusing, and I don’t think anybody has fully understood what goes on during each one of these dwells… Either way, once we looked at the 400°C dwell in more detail we started learning a lot of interesting things. And so this talk will mostly focus on this dwell and in the fact that: A Nausite membrane forms (which was shown by Pong) But also that Cu diffuses through the Nausite membrane into the core Now!...these two kinetic mechanisms produce three effects inside the wires… a good one, a bad one, and an ugly one… η Cu Cu Cu η Three things happen… Cu ε Cu Sn Cu ε Cu Nausite

The Good, the Bad and the Ugly Cu diffusion into the core Nausite growth A good one, a bad one, and an ugly one… The good is the Cu diffusion into the core The bad is the Nausite growth And the ugly is the liquefaction of the η phase Liquefaction of η Drawings: © Orlando Aquije 2008 atixvector.deviantart.com

The Good, the Bad and the Ugly The Ugly: Any remaining η will liquefy and produce large amounts of Nausite The Bad: The Nausite membrane grows with time The Good: The Cu diffusion (facilitated by the membrane) consumes the low melting point phase η Time Time How do we rescue this Nb??? Upon liquefaction… Nausite Liquid Sn ε η η ε So let’s get to know our characters a little better, As far as the good goes, the Cu diffusion (facilitated by the membrane), consumes the low melting point phase (η) Here we have the subelement undergoing the 400°C dwell, with its Nausite membrane activated. Now, as I told you, the Cu flows through the membrane into the core. This means that, in time the Cu content increases in the core—forming more of the ε phase, which is a high melting point phase. This is good! Then, the bad: the Nausite membrane grows with time… Once again, the subelement here: And as time goes by, the membrane tends to get thicker… The problem is that this membrane is 20% Nb, and we have noticed that this Nb CANNOT form fine gran A15 in the end. It forms a porous structure unfit to carry current. So this Nb is sequestered, and is bound to form disconnected Nb3Sn at the end of the reaction… So too much Nausite growth is bad And now the ugly, is that any remaining η will liquefy and produce large amounts of Nausite Here is the subelement one more time, notice in this example there is some η left behind… and if we ramp the temperature up this eta will liquefy… And when that happens, this happens! Wherever the Nb is in contact with liquid, large chunks of Nausite grow in a matter of minutes! And is the same situation as before, this is sequestered Nb, bound to form disconnected A15 and unable to contribute to current transport. So how do we rescue this Nb??? Liquid Sn Cu 66.8 at. %Sn 20.7 at. %Nb 12.5 at. %Cu Sequestered Nb, bound to form disconnected Nb3Sn Sequestered Nb, bound to form disconnected Nb3Sn Cu η Cu η ε Cu Cu Nausite Nausite Cu Nausite

Outline The ‘big picture’ Current RRP® limitations Hi-Lumi and FCC demands Heat Treatment “revelation” Ian Pong, et. al. (2013) The “Nausite membrane” The Good, the Bad and the Ugly Key Findings The 215°C dwell is useless Nausite growth is strongly dependent on temperature Cu diffusion is weakly dependent on temperature Conclusion Promoting Cu diffusion while inhibiting Nausite growth can increase Jc Our new heat treatment improved Jc (16 T) in small Ds wires by 28% (preserving RRR) Useless! Nausite This is where we move to the key findings:

There is virtually no degradation of properties when skipping 215°C The 215°C dwell is useless, and it can be skipped without affecting strand properties Ø 0.85 mm Ø 0.85 mm Ø 0.778 mm Ø 0.7 mm Ø 0.6 mm Ø 0.85 mm 108/127 132/169 The first one is kind of a digression, but I want to get that out of the way. The 215°C dwell is useless and it can be skipped without affecting strand properties You see, we grabbed a bunch of billets at different sizes and different architectures, And we heat treated them with the standard heat treatment as well as a heat treatment skipping the 215°C dwell. And after measuring their Ic we found out that the difference was minimal. The average change in Kramer negligible, as well as the average change in Jc and n value. There is virtually no degradation of properties when skipping the 215°C! and you are saving about 50 hours of heat treatment times!   With this, we can forget about the 215°C and focus on the interesting one, the 400°C dwell. Skipping 215°C dwell Standard There is virtually no degradation of properties when skipping 215°C Δ Hk ≈ 0.06 T Δ Jc (12 T) ≈ -38 A/mm2 Δ n-value ≈ 0.04 (when skipping 215°C dwell)

Nausite growth is strongly dependent on temperature Power law growth 𝑥(𝑇,𝑡)=𝑘 𝑡 0.27 Arrhenius 𝑘= 𝑘 0 𝑒 − 𝑄 𝑔 𝑅𝑇 Activation energy 𝑄 𝑔 =−98.274 kJ/mol First we found that that The Nausite membrane growth-rate is strongly dependent on temperature Here’s is your average subelement at the beginning of the 400°C dwell. Now as I’ve shown you before, if you let it run, you can see the Nausite layer getting much thicker as time goes by right? Now here we have a plot with a least squares fit of the Nausite layer thickness as a function of time at 398°C. Then we tried 390°C, 380°C, and 370°C… With these least squares fits we can calculate the power law exponent as 0.27 in which k is a function of the activation energy and the temperature as Arrhenius equation tells us. So when we plot k as a function of the inverse temperature for all these, we can find the activation energy. And why would I want to confuse you with all this? Well because now we can predict values! Which can be very useful… But overall, reducing temperature seems to be beneficial if we are to prevent Nausite formation (as a layer)—“the bad” @ 398°C **Predicted values Reducing temperature seems to be beneficial if we are to prevent Nausite formation (as a layer)—“the bad” N 16 hours 24 hours 32 hours 40 hours 48 hours 8 hours

Cu diffusion to the core is weakly dependent on temperature 2.4 ng 2.8 ng η 8 hours (at 398°C) 16 hours 24 hours 32 hours 40 hours ε Void 48 hours (at 398°C) The problem is that we were thinking that lowering the temperature is that it may also slow down the Cu diffusion, meaning it will affect “the good”… Well, it doesn’t!!! The Cu diffusion to the core is weakly dependent on temperature. Here we have a set of images where I’m only showing you the cores and its components, namely the η and ε Cu‑Sn phases (and some voids). Now, measuring the fractions of these phases we can approximate the mass of Cu inside the core in nanograms per meter—I know that sounds weird, but it is what it is. But obviously a single core is not representative, so we measure a bunch of cores! Doing this, we can plot the Cu content inside the core as a function of time. And it becomes very clear that temperature is not a factor here. So:

…so more Cu gets drawn in Longer heat treatments at lower temperatures draw more Cu in and inhibit Nausite Growth “The Good” is not affected by lower temperatures “The Bad” is slowed down significantly by lower temperatures From previous slide ‘The Good’ is not affected by lower temperatures, ‘The Bad’ is slowed down significantly by it… So all we have to do to take care of ‘The Ugly’ is let this run for a long time… So more Cu gets drawn in… And as long as we stay at a low temperature, ‘The Bad’ should be under control All we have to do to take care of “The Ugly” is let this run for a long time… …so more Cu gets drawn in

Outline 28% Jc (16T) increase! The ‘big picture’ Current RRP® limitations Hi-Lumi and FCC demands Heat Treatment “revelation” Ian Pong, et. al. (2013) The “Nausite membrane” The Good, the Bad and the Ugly Key Findings The 200°C dwell is useless Nausite growth is strongly dependent on temperature Cu diffusion is weakly dependent on temperature Conclusion Promoting Cu diffusion while inhibiting Nausite growth can increase Jc Our new heat treatment improved Jc (16 T) in small Ds wires by 28% (preserving RRR) 28% Jc (16T) increase! And now for the conclusion

Promoting Cu diffusion while inhibiting Nausite growth can increase Jc Standard Heat Treatment 1st attempt to control Nausite 0.85 mm 28% increase 0.7 mm 0.6 mm RRR was not affected Standard Heat Treatment Well, we tested a representative billet at different sizes and used the standard heat treatment. You can see clearly the degradation as the wire gets smaller. But knowing now who we are against, and who is on our side, we tested a new heat treatment that ends up being roughly the same length. (and this was our fist attempt, there’s definitely more room for improvement). And what happened was remarkable, the gap is much smaller here And all the traces went up significantly. At the smallest size there was a 28% increase in Jc at 16 T! Without affecting RRR 1st attempt to control Nausite Same A15 reaction T Roughly the same length

Back to the (sobering) ‘big picture’ There are two kinds of heat treatments in this world my friends, those who use Nausite to their advantage, and those who waste Nb Plenty of work ahead of us! Final Targets for FCC Conductor* Current Hi-Lumi LHC requirements for LARP Quadrupoles A. Ballarino, presented at the FCC week 2016. S-HiLumi-doc.40; Rev. No.: Original Release; Date: 05-May-2015 And now going back to our sobering reality… well we were here… And now we are here… That’s our target… meaning we have plenty of work ahead… Nonetheless, I’ll leave you with what Clint Eastwood said at the end of the movie: “There are two types of heat treatments in this world my friends, those who use Nausite to their advantage, and those who waste Nb” Don’t waste Nb… Field et al. IEEE Trans Appl Supercond 24, 1–5 (2014). * Values presented at 16 T (1500 A/mm2) Kramer extrapolation to 15 T = 1865 A/mm2

Thank you Special thanks to Arup Ghosh (BNK), Ian Pong (LBNL), Dan Dietderich (LBNL), and Lance Cooley (Fermi Lab) for fruitful discussions. This work was funded by the Department of Energy under grant: DE-FOA-0001604 The National High Magnetic Field Laboratory where the experiments were performed is supported by the National Science Foundation Cooperative Agreement DMR-1157490 and by the State of Florida. Some of the wires used in this study were made under the US Conductor Development Program. September 8th, 2016