Matt Angle:
Hi, everyone, and welcome back. Today, we’re going to talk about an often underappreciated aspect of neural engineering, which is packaging and materials. Now, just by way of example, this is a pretty good phone. It’s lasted me a couple of years and I expect I could probably keep it for a few more years, but if I were to drop my phone in the ocean, it wouldn’t last about 30 to 60 minutes. That’s because it’s really hard to run sophisticated electronics in a saltwater environment. You have to do a lot to protect them. That’s exactly the challenge that BCI designers face when trying to build BCI as a medical device. I mean, it’s really important to consider that the ideal number of neurosurgical procedures for anyone to undergo is zero. And the next best number is one. So, if you think about the way people work with consumer electronics, like cell phones, we might upgrade our phone every one or two years, but when upgrading a brain-computer interface involves a neurosurgical procedure, you have to take a longer view on things.

Matt Angle:
And that’s why you really want brain-computer interfaces that can last in the harsh environment of the body for many years. One of the macro challenges of building long-term brain implants is that the technologies to build next-generation brain-computer interfaces in research lab come from the semiconductor industry. And many of them have not been designed, let alone proven for use in a chronic saltwater environment. On the other hand, traditional medical device manufacturing hasn’t changed very much in the last 20 years and really doesn’t afford the capabilities that you would want in a high data rate brain-computer interface. So, how do we bring together the best of both worlds? And how do we know when a device is good enough? These are the kinds of questions I’m going to ask today of my guests, Loren Rieth, who is an associate professor at the Feinstein Institute. Stuart Cogan, who is a professor at the University of Texas at Dallas. Thomas Stieglitz, who is a professor at the University of Freiburg. And Vanessa Tolosa, who is one of the founding members of Neuralink, and now runs her own consulting practice in the area of neurotechnology.

Matt Angle:
I hope you enjoy today’s discussion. I also hope that if you have a little bit of time, you can click through some of the links and learn more about the material science of brain implants. The guests that I have on the program today are really top in their field and the depth of their work is much more than I can showcase in just this podcast. So, please have a listen and later have a look. Thank you. To start off, it would be nice if everyone could give a brief introduction to the audience, introduce yourself and tell them a little about where you are now.

Stuart Cogan:
I’ll go first then, Matt. My name is Stuart Kogan. I’m a professor of Biomedical Engineering at the University of Texas at Dallas. I’ve been there for six or seven years now, and I came to academia from the private sector. I worked for a small company for about 33 years before coming to UTD. I’m very interested in neural interfaces. And I spend a lot of time thinking and doing research on how to electrically interface two neural tissue in a way that allows us to stimulate and record that tissue without damaging electrodes, and also of course, without damaging…

Vanessa Tolosa:
So, I’m Vanessa Tolosa. I’ve been working in neurotech field for about 10 years. I started as a postdoc at Lawrence Livermore National Lab in a small group that was run by Sat Pannu, working on flexible polymer devices, mostly. And then, we also were doing a lot of packaging for that technology. I was there for almost seven years, and then, helped start Neuralink three years ago. And so that was… I figured, if you’re living in the San Francisco Bay Area, you got to try out the startup life at least once. And this was one I couldn’t turn down. So, I did that for about three years and left earlier this year. And I’m now consulting for several neurotech startups.

Loren Rieth:
Okay. I can go next. My name is Loren Rieth, I’m a associate professor in the Institute for Bioelectronic Medicine at the Feinstein Institute of Research. I’ve been working on neural interface technologies for a little over 15 years, started out, moved from background research in solar cells, and thin-film materials, and analysis. And then Utah corrupted me with the Utah Electrode Array. There was so many interesting problems with the device that could do a lot of fun experiments with it. And so, spent the next several years, more than 10 years working in continuing to work on developing the Utah Array as a penetrating electrode array for both central and peripheral targets.

Loren Rieth:
And then at the Feinstein Institute, I’ve been working really hard on some polyimide electrodes. So, combining a lot of the materials I developed iridium oxide and such for the Utah Array and combining that with polyimide electrodes to go after like mice models of vagus nerve stimulation and the whole sort of greater field of bioelectronic medicine, trying to treat diseases by modulating the autonomic nervous system.

Thomas Stieglitz:
Yeah, I’m the last one on the row. My name is Thomas Stieglitz. I’m professor for Biomedical Microtechnology at the University of Freiburg. And Freiburg that’s an medieval city in Germany, and I’m working since 1993 on polyimide-based neural interfaces and thin-film technology. And it’s some… Well, some detours in implant, packaging and manufacturing. And my goal is to make all those neural implants long-term stable and to minimize the material tissue rich in both sides that the body neither eats up the implants nor the body waltz out the implant that both sides like to play with each other.

Matt Angle:
Thank you. Before we start getting into the hard hitting materials and packaging questions, I think for a lot of early researchers and just people who are early in their career in general, one of the most terrifying things is to have to give a talk or have to do some public speaking. And I was in the spirit of normalizing failure. I was curious, did any of you have a story about a bad talk that you gave early in your careers or a moment of public embarrassment that you would want to share?

Thomas Stieglitz:
I can share my experience when I was for the first time on international stage on one of those famous NIH, Neural Prosthesis Meetings, I was standing there in front of my poster. It was not talk, it was just the poster. And Gerry Loeb came by and the older ones that you might like know the attitude of Gerry Loeb. And he stopped at my poster and looked at me, started to smile, said, “Hey, young chap, I did that 20 years ago and it will never work.” And then, he went on sharing his wisdom with the others.

Stuart Cogan:
I don’t recall a single instant where I felt sort of embarrassed and awkward in a presentation, but I do have a Gerry Loeb story, so that is maybe a digression to the neural prosthesis program. But very briefly, so it must’ve been about 1992, right? So Gerry Loeb is they’re giving now… Was it Jerry lobe? No, was Phil Troyk giving a presentation on wireless telemetry. And Gerry Loeb just got up and laid into him. And Troyk laid into him back and they were working on the same program, they were colleagues on the same ground. So, that’s perhaps one of the more remarkable things that I’ve seen at least at NIH workshops.

Vanessa Tolosa:
I hope one day, Stuart, I can say that I don’t get nervous. I still get nervous at every single talk I give, even though at Livermore, I think I gave them most talks of the team. I keep volunteering for them because they scare me so much. I feel like-

Stuart Cogan:
Mm-hmm (affirmative).

Vanessa Tolosa:
… one day, if I do it enough, I’m going to stop getting so nervous, but-

Stuart Cogan:
Yup.

Vanessa Tolosa:
… it still happens. In fact-

Stuart Cogan:
Yeah.

Vanessa Tolosa:
… my last one, so mine were recent. It weren’t that long ago.

Stuart Cogan:
Mm-hmm (affirmative).

Vanessa Tolosa:
My last one, my last talk at Livermore was at a NANS conference. The first time I went there, a big conference and I was like, “I’m going to go out with a bang. It hadn’t been announced yet that I was leaving for the startup.” And first two… Within the first two slides, I realized that they had put my draft slides up and it just kind of went downhill from there. I looked really stiff.

Stuart Cogan:
Mm-hmm (affirmative).

Vanessa Tolosa:
People were kind, but, yeah. It’s still,..

Stuart Cogan:
Yeah. I mean, yeah. I’d give them plenty of incomprehensible and disorganized classroom lectures in the last five years.

Loren Rieth:
Yeah. And I think for me sometimes, too, when you’re giving talks outside your field a little bit, so at Feinstein I’m obviously surrounded by a lot of people that are much more interested in molecular medicine and have no idea what engineering is. And so, trying to keep that audience engage is always a fight and always sort of trying to be prepared, but like Vanessa, I also have a big struggle with feeling nervous about some things. The more it is in your core area and the more comfortable you get with that material, then that that helps. But still, it doesn’t take much to perturb the ruffle the waters for me anyway.

Matt Angle:
Since I’ve asked all of you to share an embarrassing story, I’ll share one as well. I remember when I was doing my postdoc with Nick Melosh at Stanford, he asked me to sit in for him at an American Chemistry Society meeting in San Francisco. And I’m not a chemist. I didn’t think anyone would be interested in what I was doing and I didn’t really care what they thought of what I was doing. And so in my hubris, I under prepared for the talk. And when I arrived in the room, which was only a group of about 20 people, I saw standing in front of me the hard hitters from my field. I was working in Nano-Bioelectronics, and I saw [HongKun Park 00:11:19] and [Charlie Leaper 00:11:20] and [Vinshal Schwae 00:11:21] and [Jacob Robinson 00:11:23], all people that I see at all conferences. And I had these sweeping statements in the beginning of my talk about how I didn’t think the field was going in the right direction.

Matt Angle:
And I found myself just paralyzed with fear as I realized that many of their own papers were cited up there in that talk. And then I was wholly unprepared to give it. So, yeah. For-

Vanessa Tolosa:
Did they give you a hard time or…

Matt Angle:
… No. I was trembling like a leaf and that they were also gracious to me. Particularly, I remember I’ve [Vinshal 00:12:00] taking me aside and being like, “Oh, don’t worry about it. It happens to all of us.” Yeah. So, I think people have had some form of that experience in their early career. Okay. Since you’re all experts in the material science of neural implants, let’s start digging into it. Maybe just to frame the problem initially for the audience. Can we talk a little bit generally about the challenges of putting active electronics into the body? What are the challenges that pacemakers or neuromodulatory devices existing implantable electronics have been solving for the last decades?

Stuart Cogan:
Well, who wants this?

Loren Rieth:
I can definitely jump in. So, definitely, I have a long love-hate relationship with packaging. And so, they have obviously solved this problem on the larger scale of devices for pacemakers, and now moving towards making smaller devices like the micros and the things like that. But there are limitations to this sort of fully-hermetic packaging. That is the basis for all these technologies as you scale it smaller and all these companies want a technology that they can test. They want a testable product. That’s actually one of the key factors that makes them use for medic encapsulation technologies. And as it gets smaller, it gets harder to drive helium inside the things and to be able to bomb test them. And so, that is one of the limits. And then, that’s sort of a top-down approach, but a lot of us are trying things like bottom-up approaches, where we want to use the amazing density of integrated circuit technology to be able to better interface with neurons.

Loren Rieth:
But the challenge is is that there is few things that transistors hate more than salty water. And so, trying to have… And if you’re trying to get these micro-electro technologies where you’re trying to keep the transistors as close to the neurons as possible, that means that whatever materials you’re using to protect the transistors from the body and the body from the transistors has to be thin. And by its thinness, means it has to have incredible biostability so that it does not get eroded over time. And that’s my encapsulation of some of the other big challenges.

Matt Angle:
So, some of the people listening today, won’t be familiar with the helium leak test and may not understand. Can someone explain a little bit about what it means to be a hermetic package? How does that differ than just being sealed up? And how do people verify that with helium

Thomas Stieglitz:
Much earlier and to say that the major challenges that we have to put all the bioelectronics in a tank of saltwater, that is the human body, and this is something that a lot of other disciplines want to avoid. You give boundary conditions where you say, “Well, my TV screen at home works under the roof, in dry environments, and you throw it away after three years because you get a bigger one or larger one.” And this is something that isn’t boundary condition that is not spoken out very much. So, we need a long time, especially if we go into humans for chronic applications, that means for a lifetime. And in Western Societies, we can predict the lifetime as engineers over the thump as a hundred years. So, any experience with electronics it works for a hundred years, right?

Thomas Stieglitz:
And therefore, we have to see that we put the electronics in a dry environment and dry means that the electronics is already dry before we put them in a can, otherwise water that might be in there is somehow packed inside the can. And one option to do that is to put helium inside the can to prevent any other gas to be there. And helium is not that much present in our environment. That means, if we have a detector that can detect helium at small amounts, and we have a lot of helium inside a package, we can be sure, or we predict that the helium comes outside the package. And this is an indicator that there is a hole. The other way of arguing is not always correct. So, I mean, if you do not detect helium leaking out of the package, we have two options. The good assumption is that the package is water and gas-tight. The bad assumption is that it’s already empty.

Stuart Cogan:
Yeah. Yeah. Maybe stepback a little bit and address this question of hermetic, right? Because I think the word hermetic gets misused a little bit in this field in terms of encapsulation, right? So, hermetic means gas barrier, right? And so, gas barriers in hermetic implantable devices, the so-called implantable pulse generators in this field, right? They are cans typically would have some empty space inside and yes, they will be helium leak checked. And the idea is that if you can keep the gas out, you can keep water out and you keep the contents dry and so forth, but there is also non-hermetic and maybe non-hermetic encapsulation is to understand that non-hermetic means that the encapsulation strategy really is not to have a coating that is necessarily, and, Loren, I think we can go backwards and forwards on this, too, but it’s not necessarily a gas barrier or not necessarily even impermeable to water, right?

Stuart Cogan:
So for example, we can take a silicones and silicones are excellent encapsulations for medical devices. They can go directly on integrated circuits. They can go around wire bones and so on. And silicone, this is a silicone elastomer and polymer that will hydrate in the body within a few hours, right? But that hydrated polymer provides excellent long-term encapsulation of devices. And that’s currently in some clinical devices, electronics are directly encapsulated in silicone. And in our lab or in my lab, we encapsulate devices in silicone and use them long-term in the animal. This is encapsulating integrated circuits and so on. And in fact, all the way through to some clinical devices that we’re now working on. But the point is that the silicone saturates with water where the protection happens is at the interface between the silicone or whatever it’s poured onto or what it’s cured on.

Stuart Cogan:
So, and it’s all about bonding at that interface and making sure that you don’t nucleate liquid at that interface. And that can provide a remarkably good level of encapsulation. And I’m very optimistic for good old silicone that’s been around for three decades providing us quite a degree of protection of implanted electronics. And then, on the other end is the very thin film barrier layers that the silicone nitrates, the silicone oxides, the silicone carbides of this world which we hope are good barriers to the transporter of water, but they are deposited directly onto the electronics or whatever it might be. And the problem with those guys is that there’s no internal volume to determine whether or not they’re good enough in terms of a coating. So…

Matt Angle:
Yeah. Where… I was going to say, would it be fair to summarize the traditional medical device approach to protecting electronics into two camps. One being, put it inside a package that is gas-tight, and the other being, clean the electronics very well and coat them in something like silicone that may be… May not be her medic, but nonetheless prevents degradation. Would those be essentially the two strategies that are used now in clinical implants?

Stuart Cogan:
Well, I would say yes, except that vastly the implanted pulse generator titanium or ceramic package is favored very few. And in fact, I can only think of one neuromodulation device that actually uses silicone as an encapsulation. And I don’t know if it’s the same device that Thomas is thinking about. Which one are you thinking of?

Thomas Stieglitz:
I’m still thinking about the old UK Brindley.

Stuart Cogan:
Yeah, that’s exactly right. The…

Thomas Stieglitz:
We could go ahead. There are new ones. They probably explained the working mechanisms differently. But if we go, for example, in implantable blood pressure transducers or glaucoma monitoring, they end up in the same encapsulation strategy, because they need to be transparent with a silicone rubber in artificial interocular lenses and stuff like that.

Stuart Cogan:
Mm-hmm (affirmative).

Thomas Stieglitz:
But sometimes is the terminology is misleading and persons who are not that deeply involved come up with statements like they have a hermetic silicone rubber, which is completely nonsense. They have a well adhering silicone rubber-

Stuart Cogan:
Yeah.

Thomas Stieglitz:
… that prevents water [condensation 00:22:30] but those details-

Stuart Cogan:
Uh-huh (affirmative).

Thomas Stieglitz:
… are really arts to teach and hearts to communicate.

Stuart Cogan:
Mm-hmm (affirmative).

Matt Angle:
And so, going with the strategy that’s used by most medical devices now the sort of metal or ceramic can with macroscopic dimensions and millimeter plus thick walls. One of the challenges of those types of devices is then getting signals in and out of them. Is… Could someone maybe comment a little bit about feed throughs and the state of the technology for getting signals in and out of a package like that?

Vanessa Tolosa:
I just want to add a little bit on the silicone conversation, with that, so non-hermetic or near hermetic. I think a part of it too is showing, especially, since there aren’t a lot of traditional devices right now like with the IPGs showing the risk factor, the risk benefit also is going to be… It is one of the barriers, like you’ll have to do more testing, but I’m hoping that as more of those results come out that more people will be able to use them because it is about function, like is it working right now. I think a lot of people just automatically go to, “We have to have a hermetic device,” but if you take a stand back like, “Do you? Is there really a risk for that? But we have to show that.”

Stuart Cogan:
Yeah. Vanessa, we’re starting a clinical trial for an intracortical vision prosthesis. So, these would be wireless stimulators implanted in the occipital cortex. And they have integrated circuit and wire bones and the metalization, it’s just silicone encapsulation, no more, no less. And so, of course, it’s gone through all of the preclinical safety testing, but I just wanted to mention that in part, because of something that Matt brought up, which was cleanliness, right? I think that we can all agree that without exceptional cleanliness, the best laid plans of anybody’s encapsulation are going to go out the window, right? And that’s… But, yeah. So, vanessa, on the vision prosthesis, fingers crossed, right? It’s worked well in rat. It’s worked well in dog. It’s worked well in all the environmental torture testing that we do. What is it going to do in a human, right? And you could lose sleep over that, because you never know. You never know.

Vanessa Tolosa:
I’m a big fan of not overengineering. So, it’s nice when people [crosstalk 00:25:04]

Stuart Cogan:
Well, it is simple, right? You just pass the silicone and you just cure it and you’re done.

Vanessa Tolosa:
That’s all.

Stuart Cogan:
And, yeah. Well, you wash the parts of it first, right? But honestly, yeah, you’re right. It’s a little bit more difficult. There’s some vacuum casting. And so, if you want to get sophisticated, you may even do curing under pressure, vacuum casting and do curing under pressure. But on balance, it’s pretty, pretty straightforward. I do think for conventional IPGs or conventional plans, it’s the sort of tyranny of the number of interconnects, right? And how to make all of these connections, especially, since you guys are going to such a high channel counts, right? So, UTD, 16 electrodes is pretty good, right? But you guys went 256 or even 10,000. And that’s where it really, really… I mean, I think, well maybe you could figure out how to do an IPG with 10,000 connections, but it wouldn’t be individual wires going through, it’d be something much more sophisticated.

Vanessa Tolosa:
I think Matt is getting there. He’s getting to the feed throughs, but the part, and that used to keep me up at night. But now what keeps me up at night is actually the interconnect. Okay. I think there are enough feed through technologies that I can rely on. I have no interconnect/installation technologies to rely on as we get to higher channel counts.

Matt Angle:
Can you expand on that a little bit, Vanessa, just for people who are listening to understand how… what you mean by… Maybe you can tell people what a feed through is, and then, explain the classic feed through technologies? And then, also how you would connect to a feed through like that?

Vanessa Tolosa:
Yes. So, a classic feed through would be ceramic with platinum, either pins or paste that’s been cofired in there as a conductor. So, the purpose of the feed through is to connect the electrode arrays, that’s going to be on one side to the electronics that are on the inside that have to be within the hermetic can that we’ve been talking about, this hermetic package. So, that feed through that middle layer making that connections. Once you have that, you still have to connect your electrodes to those feed throughs. Each of those pads-

Matt Angle:
Mm-hmm (affirmative).

Vanessa Tolosa:
… on the feed throughs and I… All of us here work on thin-film technologies. And so, they get to higher densities, as you get to higher densities, it’s how are you bonding this? There’s a lot of… I think the two biggest challenges on this particular component is that it has to be biocompatible. Because this is now on the part that could be exposed to the body. And it has to be as hermetic as possible. So, this is where it’s not actually hermetic near hermetic or non-hermetic, but the…

Vanessa Tolosa:
Actually hermetic, near-hermetic, or non-hermetic. The issue is, you can have shorts as liquids get in through. So you get to higher channel accounts, how do you insulate that as well? Bonding is one issue. Finding the right materials, and temperatures, but then insulating is the other. Thomas you wanted to add?

Thomas Stieglitz:
Yeah. I want just to add that people get probably in better idea of what that means. The feedthroughs are going inside, outside. The challenge is that neighboring adjacent feedthroughs should not should short circuit each others. We take a lot of effort to put an insulator in between those, let’s call it pins, or lines. Then they are outside the package, and we are in the saltwater. It comes to our electrode arrays. Then the cardiac pacemaker, the beauty is that you have one wire. If you have a sophisticated one, you have two wires, because you have a bipolar electrode. That means there’s not much what you have to insulate, electrically, against each other. If we talk about 1,000 channels, you could not, well, probably you can. I cannot imagine 1,000 wires going in parallel or being twisted, somehow, and being insulated.

Thomas Stieglitz:
Then it’s really a solid post, and no longer a flexible cable. The challenge is, if you go towards the micro, to what’s the micro or nano world. Call it whatever you like. We have an array. This array should get connected to the package. Then we have also to insulate the pins on the electrode array. Normally, they don’t grow monolithically out of a rock, or they are carved out of one piece. They come to non-compatible technologies that you have to bring to each other, which is called the bonding. Then it’s quite often underestimated that you get them insulated at the connection part. That they don’t drown in saltwater. This is one of the largest underestimated challenges. You could say, for example, we do it monolithically, but then it comes in serials manufacturing.

Thomas Stieglitz:
One of the reasons the whole world is not riding a Mercedes Benz S-Class is the price. If you want a manufacturer-approved medical devices, they should not be a million per piece. That if you would make them, for example, on the brain 10 centimeters, or four inch by four inch. You would put them to the electronics monolithically, that would mean you would get five devices on the 300mm microelectronics wafer, where one costs, I don’t know, I would say half a million U.S. dollars. Then you can divide that by four to have just the net price, without any testing, verification, approval, whatever.

Thomas Stieglitz:
You have to make the chips small, manufacture them at high volumes, and then bring them together with the cheaper parts of the story. I think this is really the challenge to go from prototypes, and first in humans. Oh, it works. The principle is fine to something that is cost-effective, efficient, and that is affordable.

Matt Angle:
Loren, Can you tell us a little bit when the Utah array started, when the Utah crews started thinking about longterm implantation, and packaging. How did they approach this?What do they have now? What are some of the different avenues that have been explored in terms of packaging the Utah array?

Loren Rieth:
Yeah. I can definitely go into that. Just before I do, sorry, I’m going to go back to the packaging an IPG thing for just one second. I want to emphasize that one of the important things with the helium leak test is not just that you have it, but as I understand it, companies test a hundred percent of the devices. That’s actually really, really important. I’ve talked to some people from these companies, and it’s like, “Well, it’s not that we can’t do the non-hermetic packaging. It’s just, we can’t test it. And we can’t sell a device we can’t test.” It was really important. I didn’t really appreciate it for even a little while after, that testing for it. You can’t inspect and quality in these things. It’s a fundamental challenge.

Loren Rieth:
Or, maybe it’s not. If you get enough confidence in something, those approaches are more tractable. It was emphasized in a way from that QA that you had to be able to demonstrate that the package was going to survive. It’s really hard to do without the Calvin and Hobbes, “Well, how do you test a bridge? You drive over the bridge until it fails, and then you rebuild a bridge exactly like it was.”

Loren Rieth:
Okay. To the Utah array side of things. We’ve tried a lot of different things. We… Myself, I only speak for myself. I was naive. I have definitely been one of those people that calls silicones hermetic. This is our new hermetic encapsulation, that’s parylene plus silicone. Eventually you learn, maybe that’s my embarrassing moment. I don’t know.

Loren Rieth:
For the Utah array, we have tried to two main approaches. One is to try to connect it to a can and just have enough basically Bal Seal connectors. Some ideas where we worked with other partners to make implantable versions of Omnetics connectors, which are in hindsight, probably pretty questionable. To hook it up to standard cans and just try to push it up. That was part of revolutionizing prosthetics. Phase One, all the way back in the 2003 timeframe. Or, No, 2005.

Matt Angle:
That’s running into what Vanessa was describing as the Utah array, as very manufacturable. The feedthroughs probably existed at the pitch that you needed, but the ability to bond them together in a way that would interface.

Loren Rieth:
It was hard. At that time, again, this is 15 years ago. Getting a standard can and working with the technology partners we had. We were trying to push up to 96 feedthroughs. At that time it was hard. There’s some stuff that came out since then, with the Brown, they have their large can. They had a polyimide interposer layer on the outside of it. That was another approach. For us, we were trying to use off-the-shelf stuff, because DARPA wanted things quickly, and that worked. We were trying for a low-risk technology. That’s one approach, just trying to scale that. That was pretty challenging. Then you had the whole wires in between, but that’s another long story on the Utah array side of things.

Loren Rieth:
The other approach we did with the Fraunhofer Institute in Germany, was to try to directly flip chip bond an ASIC on the back. It was a really fun, exciting approach. You’re using gold, eutectic tin, and the Fraunhofer could bump it on in really nice densities. You had basically one amplifier, or stimulator cell per electrode. Got some nice results in terms of the integration, but we ran out of time and money. Didn’t have that guidance of Nick Donaldson’s approach of making super clean surfaces. All the great mojo that Stuart has been able to pull together for his inner cortical device.

Loren Rieth:
We had a lot of challenges with getting a leakage, shorting, and things of that. We got some, we took them through preclinical, animal experiments, some of those things, but just always suffered reliability issues. In terms of the connections for it… The Utah array has a relatively unique and interesting approach. It has problems that aren’t well-documented, but they use insulated gold wires as wire bonds for it. They have made 1,000 channel version of this thing that’s being used by a lab in Europe for, I believe, a visual prosthesis. Or maybe it’s… No, it’s a motor prosthesis.

Thomas Stieglitz:
That was a visual one. There is a visual one.

Loren Rieth:
There is a visual one too, but is this one v-… Okay. Yes, yes. It’s basically 10 Utah arrays, attached to a larger pedestal. It’s a big pedestal. You’re starting to have planarity issues. Is the patient going to tolerate this really large plug on top of their head? Does it truly scale, or have we done a heroic experiment to show that we could, and pat ourselves on the back?

Loren Rieth:
It did, when you think about it, leverage that the cortex is spread out a little bit. It’s not that we’re trying to pack a thousand wires into one cable, which becomes a tree trunk. You can spread it into individual strands that you start to tile across. That gets you something, but it doesn’t get you towards the 10,000 channels, or the 100,000 channels or the 1,000,000 channels that NESD was trying for. That’s where you need those more directly interfaced… again, like I said earlier, how do you get the electronics as close to the neurons as possible with as little in-between as possible?

Vanessa Tolosa:
Do you remember the pitch of the thousand-channel one, with the gold wire bond?

Loren Rieth:
It was 10 arrays. It was a standard 400 Micron Pitch arrays. The Utah array, we’ve made the high density version as well. You can scale it down to 200 microns. I don’t know that we’ve ever… In fact, I do know that we have never bonded 400 wires to a high density, Utah array. We’ve always made smaller versions of it. Part of the stuff that I haven’t published yet on the [inaudible 00:38:49] program is the helical lead. That actually made the leads dramatically improved, in terms of their mechanical characteristics, stretchy, softer, and stuff like that. There’s room to scale, it’s just not orders of magnitude. If you’re really going after that 1,000 or tens of thousands of channel-device these tricks, aren’t going to get you there.

Matt Angle:
That’s a good segue to some of the higher channel-count devices we’re seeing now are based on a totally different thread of technology development. We’ve been talking about extensions of existing medical device paradigms, but a lot of neural engineering now is coming from the semiconductor industry. Building things lithographically, building things, infant films. Vanessa, that is your expertise. Can you give us a little bit of an overview of what are some of the Michigan-style devices being made now? What are the materials that you can make these probes out of? Thinking of the work that you did with Loren Franks, what are some of the limitations of those approaches, in the context of lasting a long time and connecting?

Vanessa Tolosa:
You’re talking about the probes, but I’m still stuck on the interconnects. Even to this question. The challenges are at that interconnect, and we’re starting… We had to move to MicroFab to do that, because if you look at ceramics, like the tightest pitches they can really do is like three under 400 microns, 375. Glass is becoming more popular, and they can do about 180 microns. That still wasn’t enough when we needed to go to 1,000+. This is where, actually, Thomas already described the whole thing in his. I’m just going to be repeating what you just said, where the issues, which is… Though we found a way. Using in this scenario, it’s a flex polyimide probe.

Vanessa Tolosa:
The answer to the interconnect problem, is to get rid of the interconnect. Best connector is no connector. Which, as you mentioned, what we did was a monolithic buildup, starting from the feedthrough substrate. We did that from scratch, then start to build the array onto it. Including additional adhesion layers, and barrier layers, which is okay for a specific type of company. The real problem is the manufacturing and scaling up. How do you get other people to do something like this? Especially when microfabrication facilities for an underquadding management doesn’t really exist for most of… I think that’s a huge barrier, for neurotech. Even though in R&D, we’re all moving to MicroFab, it doesn’t leave R&D. There isn’t anything affordable for a company, especially a startup.

Matt Angle:
Yeah. But also, Thomas, you’ve done a lot of characterization of polyimide and parylene-thin-films. Those films don’t last for 100 years, even when they’re processed correctly. I don’t know, are we… Even if you build a monolithic device with no connectors, are we there yet? Do we even have the materials? Or is that still a basic research problem?

Thomas Stieglitz:
I don’t give the promise for 100 years, but I learned now that the technologies that we have, my cochlear implants, cardiac pacemakers, they would not last for 100 years either. We are now coming with a cochlear implants, at lifetimes, where we see them fail. After they really boomed, and got a success story, we learned more about failure nodes. I believe the main challenge in thin-films is that they are so thin. Yeah, that’s stupid, but let me give an example. If you are in cochlear implants, for example, the metal there, let’s say it’s about 100 microns thick, and I believe it’s thicker. You really have a metal plate, or a metal ring, which is really thin. You don’t want to do that by hand. This is really pain in the neck, but imagine you would have corrosion. That’s something where you can work with, or it’s still conductive. It’s still holds mechanically.

Thomas Stieglitz:
If you start with 200 nanometers or let’s start with really thick, thin-films of one micrometer. You cannot afford a lot of corrosion, because then your whole electrode, and the cable that you have, is gone. That’s really one of the challenges. Therefore, I’m really skeptical also of insulation layers. If people say, “We go below two microns of substrate at insulation layers. We’re manufacturing in a cleanroom.” Cleanrooms are defined by the number and the size of dust particles. Normally, if you are in I take the old terminology in a class 100 cleanroom, you have less than 100 dust particles in the cubic foot that half the diameter, or diameter of 50 microns.

Thomas Stieglitz:
That means if you would have an insulation layer of one micron, and you would have a dust particle of two microns diameter, that friendly dust particle will sit on top of your wafer in the manufacturing process. Then you pure at some 100s of degrees centigrade. Your insulation layer, the dust particle, will burn, and will leave a hole behind. That hole is the entry part for water, for salt. If this is the voltage carrying line, you can electrolyze water into gases, and then it’s like an explosion.

Thomas Stieglitz:
This is something that is completely underestimated to my impression, when going smaller and smaller, that you would need the thickness for two good reasons. One, is to have it thicker than the dust particles in your manufacturing line. The other, is that you should remember that normally not the engineers or the neuroscientists in implant those things, but medical doctors, surgeons, neurosurgeons. If you are in the OR you have to wear your rubber gloves. Try it out. Put on a pair, or two pair, of rubber gloves and try to handle a one or two micron thick polyimide foil or parylene foil. That doesn’t work. You need to make those things search improve. That means you have to get the balance between the least possible thickness for structural biocompatibility, and the maximum thickness for robustness and reliability in a moving environment.

Loren Rieth:
I’m just going to add one thing to that before I have to drop off. I totally agree with that. Honestly, I think it is that we don’t have an answer to how thick it needs to be at this point in the game, is a bit of a hole that needs to be filled still. The one thing I was going to plug here is the need for test structures. Going back to maybe Nick Malosh. They just had their nano roots approach. Here we have these very, very thin layers of polyimide. When I talked to him, I was like, “So have you tested these with test structures?” And he’s like, “What? What kind of test structures?? I’m like “Test structures, where you don’t open the encapsulation around the electrode site.”

Loren Rieth:
Anytime you open that electrode site, you have a low impedance path. How can you ever hope to electrically measure the characteristics and the leakage paths of your polyimide threat electrode, and how it changes over time, when you have a short circuit that has six orders of magnitude lower impedance right next to it? You just can’t perform that measurement. We have a real need for more test structures. In my lab, I call them… I have two different kinds of electrode test structures, and interdigitated electrode structures. Just two different styles that give you a little bit different ways to measure how the encapsulation is. You need to have structures that measure how the encapsulation ages by itself, and hopefully back that up with some physical characterization and chemical characterization to go with it, as well as the electrode sites and how those age and things like that.

Loren Rieth:
The last thing I’ll say is in support and amplification of what Thomas says is, the reason that we’re doing this is not only are we trying to take the size of these… Let’s say cochlear implant electrodes, which are a certain size. We’re trying to make them smaller, which is increasing the current density. Increasing the impedance, which means we have to drive them harder. We’re going to higher voltages. So we’re pissing them off, making their life harder. And now we have less than one micron of material to sacrifice it. So it’s like, “We’re squeezing it from all sides, and Scotty is not happy.”

Stuart Cogan:
Oh, I got it.

Loren Rieth:
Okay.

Stuart Cogan:
I am vastly more sanguine about thin-films.

Loren Rieth:
I didn’t say I wasn’t sanguine. I’m still doing it. I haven’t given up, but there are big challenges.

Loren Rieth:
No, no. In fact, what are we learning? As the field of electro structures have been studied over the last few years, one of the things we find is that when we put very small things into the brain, we get very minimal adverse foreign body response. We all agree, foreign body response is a bad thing, blah, blah, blah, blah. Now we’ve been talking about that for 20 years. When you put things in that are really tiny, you must get a foreign body response, but it doesn’t seem to be particularly adverse response. We are highly motivated to make devices that are really tiny, that go into the brain, to avoid foreign body response. Then the question is… That means, I think, only thin-film devices of some kind is going to do that.

Loren Rieth:
To all of your points, yeah. You do have problems with manufacturing. You do get dust particles. Actually, not even in the cleanroom. It’s just the dust particles degenerate in the deposition systems. They’re real boogeyman there. But the payoff can be very significant, if you can learn to do that. I have so much to say. To your point about not being able to test devices, you do have to have 100%, in my view, testing going through. You do have to have some strategy to test all your devices that are non-hermetically encapsulated. Whether that means saving, soaking and stimulating for a little while before you package them up, or whatever it might be. I’m not sure. In any case, in my experience with thin-film structures, you can design them, and make them, so they can provide a really remarkably good level of long-term protection of devices.

Loren Rieth:
It gets to be a different way of thinking about it, because these devices are so small, your encapsulation is the device. Your encapsulation isn’t on the device, it is the device. Then you have little metal wires or however you’re going to make them. Of metal traces, I should say, this is no wire-wire inside the encapsulation. By design, perhaps we can do very well here. To take the example of a good old University of Michigan probe. They’ve been around since what? 1985? 1990? And they have a three layer passivation release. They used to, they have a silicon oxide, silicon nitride, silicon oxide. Yep.

Loren Rieth:
We know that silicon oxide is rubbish as an encapsulation layer. We know that silicon nitride dissolves rapidly in the body. I think it’s totally fortuitous, or it’s serendipity, that the silicon oxide, silicon nitride, silicon oxide, tri-layer structure, which is pretty thin, actually lasts pretty well.

Matt Angle:
What does pretty well mean for the audience? Is that…

Loren Rieth:
All right. So what does pretty well mean? Pretty well means that I’m not aware that University or that Michigan-style pro, from a company like NeuroNexus, for example, I’m not sure that they have any clinical use at this point. They do very well for the lifetime of the rat in some people’s hands, [crosstalk 00:52:55]

Matt Angle:
But also Michigan probes are classically bad for longterm in vivo physiology. They do great for the two-day experiment the graduate student does, before they sacrifice the animal. I’m not really aware of Michigan probes being used for chronic, non-human, primate work. I think that they’re not preferred.

Loren Rieth:
Sure, but I’m not actually talking about Michigan probes as the probe themselves. Just talking about the passivation layer. Just the [crosstalk 00:53:30]

Matt Angle:
I’m just suggesting though, that the reason why you don’t have those data points of material failure, is because no one ever bothers to use them for very long.

Loren Rieth:
Yeah. Although we have studied the long-term, but we were going down a rabbit hole we don’t need to. We have looked at the long-term stability of the side of dialectics, and they can do pretty well. I think Michigan probes fail for other reasons. They make an excellent comparison point for new technology, because you can pretty much expect them to fail over a fairly well-defined course of time. I want to reemphasize how enthusiastic I am for all thin-film structures, that embody our vacuum deposited encapsulation. Perhaps combined with the ceramic, the oxide or the carbide dielectrics, maybe with a very thin polymer layer on them. I just wanted to go back to something that Vanessa said, which is, “The best way to have a really good interconnect is not to need one.”

Loren Rieth:
I think that where we’re going here, is the electronics is shrinking and getting closer and closer to the electrode arrays. We’re very interested in wireless. We don’t have to have transcranial wires and cables, and pack everything in as close as possible to the electrodes, then capsulate that. We can get rid of the cable, get rid of the connectors. That’s where we’re going. That’s how we’re going to succeed, when it comes to a very large number of electrodes. So that’s my sanguine view of [crosstalk 00:55:29]

Loren Rieth:
And I guess, sorry. My last thing, and then I’ve really have to drop off, sorry, is that I obviously I’m still working in thin-films, so I have not given up. Just identifying, I think, some challenges for the field. Then the only thing from my side, is that long-term is a really big change. As part of the auditory nerve project now, I go and talk to Madell, and they’re like… I say, long-term, they were like, “What does that mean?” Anything less than 10 years, and even 10 years, that’s the bare minimum of something they want to deal with.

Loren Rieth:
That’s like the bare minimum of something that they want to deal with. And I did have my eyes open. What I’ve seen on the HAPTIX Program and taking the Utah Array from 1.5 months in the periphery to 17 months in the periphery and hoping that the next one will be longer, there’s a lot going on there. And those times are still really short. And in one of the papers on the reactive aging, I showed a picture of the Utah Array after three and a quarter years in a rat… Or a cat sciatic nerve. And so again, long-term has been something that’s evolved for me and these discussions with companies, so especially for a cochlear implant where they’re putting it in kids. And we’re starting here. There’s just a lot of things to figure out at that.

Loren Rieth:
And even with Accelerated Aging Test, to try to get a sense of that, we have these nice tools that have proven successful over time, the silicones, and the platinum meridians, and the platinums. But a lot of other things have a ways to go to show that they can work. So, yeah, I think it’s big challenges, but I think like Stuart says, getting things down close, avoiding that fibrotic response will all pay huge dividends. And so, definitely exciting to do and fun to keep working on. It’s been a really great conversation. I…

Thomas Stieglitz:
[crosstalk 00:57:23]

Loren Rieth:
Yeah. This has been fun. I really…

Thomas Stieglitz:
Yeah, what you talked-

Loren Rieth:
Indeed. All right. So, I’ll take care.

Vanessa Tolosa:
I think with the thin-films, one of the issues too were… I hear a lot of confusion on, “Does it work? Does it not work?” I think part of the problem is people discuss it as just thin-films, rather than in relation to design or application.

Thomas Stieglitz:
Mm-hmm (affirmative).

Vanessa Tolosa:
And so, this is what I say when people ask me, “Well, does it last a long time?” Or, “What’s the longevity or lifetime of this?” Well, it depends on the design. Like I will design a thin-film probe based on the location of the body. I mean, you have to think about how is it thin… Especially with polymers, how is it failing? Is it flip… Failing through the thickness? Is it failing through delamination, through ingress from the side of ions? Or is it failing from degradation of the material itself?” For example, with the accelerated aging with peroxide that was initially started through [F David Pasha 00:58:20].

Vanessa Tolosa:
The way I thought about that result was in combination with what Stuart we’re saying with the size of the probe. Like, yeah, I think he was using 20 millimolar peroxide, which is pretty high, but if you have a small small probe, you’re not going to cause that kind of reaction. So, it’s not going to degrade as fast as what that study showed. So, you can’t take-

Stuart Cogan:
Mm-hmm (affirmative).

Vanessa Tolosa:
I guess the point is you can’t take one result from one type of design and then say, “This is how thin-films are going to… How they perform in all cases.” And the same way with stimulation, you have devices like DBS that are on all the time, 24/7. So, maybe some thin-film designs won’t survive that, but there are a lot of devices out there that are only on when the person is awake or in duty cycles are a lot lower and it could be used for that. So, to just blanketly say that thin-film can’t work for long-term I think really depends on application and design.

Thomas Stieglitz:
This is probably due to the case that most persons know the cardiac pacemaker, and it’s obvious that a cardiac pacemaker has to work 24/7, right? Otherwise, you might get into trouble. And this is something how will the field evolve? It started from the cardiac pacemaker, and then, people fiddled around a little bit and stuck it into the brain, the same technology, more or less. It’s the IPG for the deep brain stimulation is more or less the same than the cardiac pacemaker and the cochlear implant looks differently, but from the technology, it’s still very close to that. And so, I believe if we do not change that way of thinking, people don’t tell us, “We need the deep brain stimulator, but with 10,000 electrons.” No, that’s wrong. I mean, if you have to look from the application side, say, “Okay.”

Vanessa Tolosa:
Mm-hmm (affirmative).

Thomas Stieglitz:
“We need a hundred electrodes here, and hundred electrodes there and they have to talk to each other. Yeah. I think this is the beauty with the new technologies that we can think outside the box and get something new that we can use redundancy on purpose. And not because we are not able to make it smaller or larger or whatever, but we can really play with physiological details, try to understand them, and find the right, as you said, Vanessa, the right design, the right combination of technologies and materials for that particular application.

Matt Angle:
If we talk about materials, do we have some intuition for the lifetime of a perylene or [polyamide 01:01:05], say 10 micron film, do we think that it’s possible to push that out beyond 10 years or do we think that intrinsically, these things let in too much water and degrade too quickly?

Stuart Cogan:
Well, that’s a really hard question, because actually how a [polyamide 01:01:27] fairs and how a perylene fairs a little bit to the Nessus point, it’s sort of depends who makes it and who’s puts it on the device, right? It’s so dependent on who’s doing it. It’s very dependent on the design of the device. It’s very dependent on where the device is going in the body. My guess is if you put [polyamide 01:01:48] or perylene done well in some saline 37 degrees Celsius, it’ll last for a very long time, put it into body, it’s a different story, right? We don’t really know. But one thing that Loren brought up, which I think is critical here is the only really good long-term test we have is putting it in an animal and seeing what happens over the course of a few years.

Stuart Cogan:
And so we don’t have… We can do accelerated testing and increase the temperature. We can add some peroxide or we can increase the temperature, and add peroxide and so on, but it doesn’t really give us a very satisfactory way to predict a lifetime in the body, at least not yet. But then, from a very fundamental material standpoint, I think you can look at a material and you can look at its structure, look at its bonding and say, “Okay. This material is likely to last a long time or this material is going to be susceptible to hydrolysis or this material is going to be susceptible to stress cracking or whatever it might be.” So, while I am… I don’t want to make this sound too simple, right? I think it is possible that we can come up with a… Well, I was going to say that we can come up with encapsulations that lasts a long time.

Stuart Cogan:
Well, actually of course we have polymer encapsulations that do last a long time. And getting to thin-film encapsulation that lasts a long time is challenging, obviously, for reasons that we haven’t really touched on yet, besides they’re thin, they’re thin, they consist… Be susceptible to process defects. They have lousy coverage of surface topography. What we haven’t talked about is stress and mechanical forces on these films. So, if we’re wiggling these things backwards and forwards in the body, then we have a whole new failure mode that we have to consider that it really doesn’t crop up much with polymers or maybe with perylene and [polyamide 01:04:07] in some applications, but certainly not with conventional clinical polymers. So, that’s something that we are going to have to address. And again, I think that really, to Vanessa’s point about getting rid of things, we don’t want to use thin-film encapsulation anywhere we don’t have to.

Stuart Cogan:
And we want to design from the get-go if we are going to use thin-film encapsulation, we want the design to start off with using thin-film encapsulation in mind. We don’t want to have a design and say, “Okay. We’re going to take the polymer off and use thin-film encapsulation,” because that’s never going to work, right? So, that’s really not answering your question, Matt, because I don’t think it has an answer, but it is… Again, there are a lot of challenges. There’s no reason why we can’t overcome those challenge-

Matt Angle:
I was just going to say on our last podcasts, we had Cindy Chestek on and we were talking about some general challenges in BCI.

Cindy Chestek:
Yeah. No. If I was going to write a science fiction series where like suddenly one new technology turned up and changed everything, it would be the magic coating.

Matt Angle:
Yeah. Yeah.

Cindy Chestek:
Right? That it lets you make all of your medical devices can now be chip-scale, right? Because you have the magic coating. And so, it’d be great if we had it. I wonder if material science has been as fully engaged as they could be. Maybe it’s out there somewhere or I don’t know.

Matt Angle:
So, the magic coating would be something that you can put down at micron thickness or maybe a few micron thickness at low temperature. So, you could do it on active electronics or on a polymer substrate. And it would last as long as existing packaging technologies. Do you think that the magic coating exists? Do you think it’s possible?

Thomas Stieglitz:
First of all, I think that’s on the wishlist of a neuroscientist. And what I mean with that is that neuroscientists quite often approach the field from a way of, “I think this would be good from the biological side.” But if you want to sell that, that really a lot of patients could benefit from it. It’s not from the wishlist. It’s from the side of, “Can I manufacture it and do I get it through an approval procedure with a lot of validations?” And from that point of view, I would say a magic coating would be so magic that would… That I can not imagine that I can convince an FDA officer or whatever officer worldwide and another approval system that this magic [world 01:06:56] holds true as universal coating for each and every place in the body, under each and every mechanical and electrical load and disease case that I have.

Thomas Stieglitz:
We should keep it in mind that most of the patients are not healthy. Otherwise, they would not need a neural implant. And if I would need something, probably there might be a different metabolic state there around, and it must be stable. So, a magic coating must be so magic that I can put it on the device in a clean room, but it goes through washing validation and sterilization lies on the shelf for half a year, then it’s put in an airplane, is shipped over minus 40 degrees C at 30,000 feet heights, spends two hours in Dubai for refueling, gets up in Australia, out of the case and still has that magic property where you put it on eight months before.

Matt Angle:
Mm-hmm (affirmative).

Thomas Stieglitz:
And this is where I, as an engineer in translational research, really, I doubt that-

Matt Angle:
Yeah.

Thomas Stieglitz:
… I could find that magical thing. It’s probably too…

Stuart Cogan:
Mm-hmm (affirmative). Well, Thomas, you’re obviously a very well-traveled man. What can I say? Yeah. So, I think there are coatings that are likely to do very, very well in certain applications, in certain devices, in certain ways, whether that can be a fix or just talking about coatings that are barrier layers or encapsulation coatings, can there be a coding that’s a… I am not optimistic for that, because I…

Matt Angle:
Some people think that that’s silicon carbide.

Stuart Cogan:
Well, okay. So now that you bring up silicon carbide, now I don’t want to be self… Yeah. I don’t want to be self-serving because we use silicon carbide, of course, all the time and I highly invested in it. I mean, silicon carbide is extremely stable in the body. It’s very adaptable to thin-film processing and we’ve had it in animals for long periods of time, but it comes with limitations, right? So, you’ve got to put these… If you have a lot of surface topography, coding things that have a lot of surface topography is very challenging. So, if you ask me, “Can I take silicon carbide and put it over a wire bonded [ASIC 01:09:33]?” I would say, “Well, you’re pressing… You’re pushing your luck there”, right?

Stuart Cogan:
But on some thin-film structures, sure. I mean, I think it can be great. And again, to the point that I think have been making, you can’t use something like silicon carbide in a device design that was intended to be encapsulated with a silicone, right? It’s just completely different, right? So, I really, really liked silicon carbide, Matt. I think it’s going to find its way certainly into the clinic, but I think it’s going to have a fairly well-defined range of applications and uses in these devices. And it’s definitely not a panacea for encap…

Matt Angle:
Are you familiar with the Neuropixel device? Are all of you familiar with that coming out of Janelia Farm?

Thomas Stieglitz:
Well, We could discuss if it really comes out of Janelia Farm or the most of the technology.

Matt Angle:
IMAC. Yeah, yeah, yeah, yeah. I should’ve mentioned all collaborators there.

Thomas Stieglitz:
It’s my European centric view. Yeah. Yeah. Uh-huh (affirmative). Yeah. Yeah.

Matt Angle:
Do you think the coating exists to seal an active Michigan probe, like the Neuropixel, seal it away and be an effective barrier for 10 years? Something that could be applied on CMOS at low enough temperature that you don’t cause problems for the active electronics and some of them could be thin enough that it wouldn’t substantially change the geometry of a probe like that. Because I think that’s where the field feels like it’s going, but it does certainly seem like the barrier is the remaining question.

Thomas Stieglitz:
Well, I believe the intention of the Neuropixel probe is to gain knowledge of the structure and the communication in the brain in neurosciences. I mean, as far as I’ve seen the presentations on that talk, it was not on longevity of five or 10 years.

Matt Angle:
Oh, no, that’s not what they’re claiming. That’s exactly what they’re building it for. But-

Thomas Stieglitz:
And then, that means, so if you have… Like Stuart said that old style coating and probably a little bit more, might it now be one of those atomic layer depositions where there a lot of promises-

Matt Angle:
Mm-hmm (affirmative).

Thomas Stieglitz:
… are made or a combination of silicon, low temperature silicon carbide, which is not as good as high temperature silicon carbide from the material properties. I think the truth will be a combination of very thin materials that help to fulfill the intended use. And this is the point where we have to jump back, I guess, to one of the first questions. What is the intended use? Should it work-

Matt Angle:
Mm-hmm (affirmative).

Thomas Stieglitz:
… for a day, a week, a month? Should it be in monitor human primates for a year? And then, we have to find out what happens if. So, the beauty of small devices is that they have a small amount of material. So, it’s not that toxic because it’s not that much, right? So, it’s probably not toxic. And the second question is how does it fail? If you have 10,000 electrode sites and three of them fail, you still have more than 9,000 and you can proceed with your experiments, right? And therefore, I mean, it’s really, what is that good for? If you say, go in with 10,000 electrodes because I need five of them and I’m pretty sure whatever five I select out of the hundred, they’re good enough.

Thomas Stieglitz:
There are some statements of the BCI community that doesn’t matter that much if you’re a 10 microns above or below, if you can train a person to use a certain signal for driving an assistive device, then you have that redundancy or that plasticity inside the brain. This is different from the question. If you want to map the whole brain with thousands of signal sites to see how the signals interact with each other or if you have the certain phase dependency or whatever.

Matt Angle:
Mm-hmm (affirmative).

Thomas Stieglitz:
And therefore, I will believe that we will see different solutions depending on the intended use.

Matt Angle:
Vanessa, I’m curious to get your take on this because you’ve been very invested in building higher channel count thin-film devices that are looking at long-term use. Where do you see things going and what are you optimistic about?

Vanessa Tolosa:
Well, for the… specifically, on the silicon devices, I guess I’m not aware of anything that’s gone pretty far for deep targets. So, I just think that… I don’t really think about the silicon too much, except for the science, even though a lot of BCI work is in cortex, no one compare the [leads 01:14:49].

Matt Angle:
Mm-hmm (affirmative).

Vanessa Tolosa:
So, yeah, I always think about flexible polymers in that sense for a longer term deeper targets, which I know I sound like a broken record now, but it’s like, to me, I don’t care about a magic coating. I care about a magic interconnect, because when I work with flexible probes, it’s always, “How do I then bond it?” Whereas for the Neuropixel, they’re able to just build it all up off of a-

Matt Angle:
Mm-hmm (affirmative).

Vanessa Tolosa:
… silicon wafer. And it’s really nice. You could do wafer level.

Matt Angle:
So, do you think if you had the magic interconnect that you’re comfortable right now with the longevity of [polyamide 01:15:24] structures, it’s just intrinsically?

Vanessa Tolosa:
No. So, there was-

Matt Angle:
Mm-hmm (affirmative).

Vanessa Tolosa:
… It’s just that’s the first place I think that I’m hitting a wall where-

Matt Angle:
I see.

Vanessa Tolosa:
… I get farther with the probe, but then on the probe side, [polyamide 01:15:36] alone, I think based on… If you can… Depending what the application is, you can design it to last a long time. For example, one of the failure modes is coming from the edge of these devices, the ions and moisture migration, if you increase that path, you’ve increased your lifetime. What is the lifetime you’re going after? Is it one year, is it five, is it 20? So, that’s just one design low-hanging for design change. Of course, with these brain probes, you want to go small as possible. There, I stand on… Actually these two guys right there, they have really affected my thinking on how to… The path to a longer life polymer… [Polyamide-based 01:16:19] device. And, I mean, and silicon carbide is a big one. That’s in my head.

Vanessa Tolosa:
And again, it’s thanks to Stuart and Thomas’ work that makes me think that we haven’t really shown what that can do that it can last. How long it can last? It’s unpublished work, but at Livermore, a lot of the work we did was unreliability of [polyamide 01:16:41] and [polyimide 01:16:44].

Matt Angle:
Mm-hmm (affirmative).

Vanessa Tolosa:
And so, I’ve seen some things that making it believe… Even [polyamide 01:16:48] alone can last a pretty long time, but the conditions under which you have to manufacture are so tight that it’s hard for me to imagine that and in the near term for any company.

Matt Angle:
I see.

Vanessa Tolosa:
So, I think wanting to… Or needing an additional layer, which again is also a manufacturing issue, carbide isn’t that easy to get to with the device or the tool, and specialty in manufacturing that at a high volume.

Matt Angle:
What about when we talk about ceramic films, I have a clip from Tim Harris here that I’ll play.

Tim Harris:
Do you need single crystal films or can you plausibly argue that a thin polycrystalline film can be truly in penetrant for the years’ timescale and body temperatures?

Stuart Cogan:
All right. So, you go from single crystal to amorphous, right? And in between is polycrystal, and polycrystal now we say it’s rubbish because of all the grain boundaries and transport down the grain boundaries and so on. But an amorphous material which can last a very long time, it’s sort of all grain boundary, right? So, I think the answer is no, it doesn’t need to be… And in fact, I mean, maybe silicon carbide is an example, but… Silicon carbide, sorry. Silicon oxide, but, I mean, I don’t think it has to be single crystal, I mean, I think polycrystal, and if done right, can be very good. And I’m not sure quite why, right? Why polycrystal and silicon dioxide doesn’t work all that well is probably as much due to silicon dioxide as it is due to the grain boundaries. And so, I’m… Yeah. So, no. The short answer is, I don’t think it has to be single-

Thomas Stieglitz:
Single crystal or is a polycrystalline or amorphous, but rather what are the deposition conditions? So, what other agents do I have in the area?

Matt Angle:
Mm-hmm (affirmative).

Thomas Stieglitz:
If I have a lot of point defects with, let’s say rubbish like hydrogen, right?

Matt Angle:
Mm-hmm (affirmative).

Thomas Stieglitz:
And it might have changed… Might exchange. And then, it’s not that the material itself is bad. The Silicon dioxide is still fine, but all the other material has gone-

Matt Angle:
I see.

Thomas Stieglitz:
… it leaves open spaces there. That’s the same with Atomic Layer Deposition. Aluminum oxide. So, alumina is a material that’s beautiful, but if you deposit that in a single layer, and this is not homogeneous, and you have a lot of crap in between, when it’s eaten up by the water immediately when it gets in contact.

Matt Angle:
Mm-hmm (affirmative).

Thomas Stieglitz:
I mean, that does not mean that the material is not good, but normally in normal conversations, if you say something like point defects, people don’t… They no longer listen to you because you sound too technical, right? And now, it means it’s again all about manufacturing. But the promises of Atomic Layer Deposition that a lot of nanotechnologies work on the basic mechanisms. That means they are now a little bit overstretching, but they’ve published when they see three nano rods, and we want to work on wafer scale or at least on device scale. That means we have to deposit area sites that are far above what’s done in fundamental nanoscience. And this transition is really painful and the machines are very expensive. And then, in most labs, too many persons playing around on a single machine, and don’t tell the others what they’re using, and then don’t clean up the machine as properly as it should be. And therefore, the reliability of those layers is not as good as it would be. The [inaudible 01:20:36] machine would stand in the class, one clean room to manufacturer gazillions of microchips per year.

Matt Angle:
It sounds like something that all three of you have been emphasizing is the material itself isn’t necessarily the determining factor, but rather the process conditions for depositing that material seemed to be so critical. That seems that’s the very thing that often isn’t shared between labs or between groups and often isn’t well documented or well reproduced. What do you think as a field we can all be doing to try to sort of float the industry?

Stuart Cogan:
Well, so what we’re starting to do is be willing to share our SOPs, our standing… standard operating procedures, and what’s… And a little besides variabilities and equipment and so on. Hey, in my lab it’s, “Oh, if so-and-so makes the coating, it’s great. If this person makes the coating, it’s lousy, right?” So, when you have that thrown into the mix as well, besides the foibles that come along with any university fab, right? So, but I think that what I’m finding just in terms for my own efficiency of producing devices is that we need to have our standard operating procedures, the graduate students follow, right? And so, it’s making the thin-film coatings is not an ending itself, right? So, it’s just a tool so that we can put them in an animal, so then we do the research or whatever it is. So, I’m quite keen on in the first place providing good standing operating procedures, besides that, I’m not sure what you do. Maybe labs can be encouraged to swap samples and examine each other samples, things like that. But I’m open to suggestions on that from Vanessa and Tom-

Thomas Stieglitz:
And I think it’s… The problem is that publish and perish thing, you need certain amount of publications that you get promoted-

Stuart Cogan:
Mm-hmm (affirmative).

Thomas Stieglitz:
… or tenured or whatever. And therefore, people don’t want to listen to you. For example, if you say, “You know that it works really well, you have to clean your machine for five hours.” This is something that nobody wants to hear, because that would mean that your whole day is screwed up just by cleaning, and cleaning a machine doesn’t give you any publication.

Stuart Cogan:
But you have-

Thomas Stieglitz:
You have to add that and it gives you the final devices. And if you-

Stuart Cogan:
Right. Yeah.

Thomas Stieglitz:
… If you… Probably that’s too hard to say, but if you know that you’ve cheated a little bit, and that your yield is 10% and not the 70% that you always present on conferences, you don’t want to share so many samples because you don’t have them. And if you share them, another left mine find out that… Or you see something that you have actively ignored while…

Thomas Stieglitz:
… Thing that you have actively ignored while writing your last paper. Yeah. So, don’t give you examples because then you would know the names, but there is that tendency that you have to be fast, you have to be the first one. If you make a paper about failure modes, you don’t get it published. Right. And if you make a paper about testing methods, you get at best to the IEEE transaction with the impact factor between three and four, and then you’re really a hero. But if you want to get tenured, you have to go to advanced materials, you have to go to science of the nature. And these are not the journals who are happy to publish work that is about soak tests of five years.

Vanessa Tolosa:
I think, the underlying message here is the manufacturing and that, especially with Microfab what made the experience at Livermore, I think unique for me was that we weren’t, luckily we weren’t driven by this need to publish something cutting edge. Instead, like I said, we really focused on reliability because we were funded to provide a platform technology across many universities, many different kinds of animals, stem record. We had like every design, it was really fun to make something that had to work multiple times because, also when I joined like people didn’t really know who we were, so we had to build a reputation. So if we’re sending devices out and they’re constantly failing, it would have been really hard for us to build. So sending things out that worked really mattered. So I think that’s one option where you have a facility or group who it’s their full-time job to be working on reliable products, not trying to push the edge and make a publication.

Vanessa Tolosa:
And it’s grad students who are needing to learn something new with each year. Also, the testing to that’s involved and required. I mean, I’d say even at a government lab that having long-term testing and being able to do something multiple times, was also difficult. So I can only imagine at a university how difficult that is. So I think there are a lot more discussions on this huge gap between research and commercialization that everyone in neuro tech is recognizing, especially because they think a lot of R and D is Microfab and there’s none in the commercial space. So why is that? How are we ever going to move the med device industry beyond incremental steps, if we can’t do like 10% of the things we’re doing in R and D, which is what’s pushing the technology. And so I think these new ideas of an institutionalized way to think about tech development.

Vanessa Tolosa:
There was an article out by Adam Robinson and another author on Day One Project recommending what they called the focus research org. So it’s kind of taking the best of startups, government labs, research institutes, and academia, and then creating some kind of government funded system to help this kind of technology. I totally resonated, with what I’ve been thinking, based on my experience from a government lab and the startup. Like that startup environment was almost perfect. You have everyone under one roof with one agenda with all the resources that you need, which is not common for most startups, but that still like how do you replicate that, so more people can be creating and developing and have a chance to move the technology.

Matt Angle:
Can you speak a little bit to what it’s like, just generalities of course, like what it’s like to do some of the same work in a national lab environment versus in a startup environment, where is it easier to do what? What are some of the pros and cons?

Vanessa Tolosa:
A big difference that I saw was, at least in our particular group at the government lab, we were still one component of a bigger system. So we were building BCI’s as part of, for example, a DARPA program. You have clinicians at one university or one hospital or multiple hospitals, we had chip designers at another university, we had cables being made by a biomed company. So, it was also only funded for only, four years is a long time for academic funding. And it was a lot for academic funding, but still you can feel that there’s a rush to make the device, but it’s counter against the agenda of all the teams that are involved. Everyone is at a different institution. They’re only working on this. part-time a lot of them rely on a couple of years of funding. So they want to get enough data thinking about how to use this data for their next funding, rather than how do we build this device that we signed up for and fully thinking about that.

Vanessa Tolosa:
So, the agenda, and it’s not their fault, the incentives are just not right for that kind of system. So, I feel like when we switched to the startup environment in this particular startup, it was all under one roof. Again, it’s a very I think, unusual environment, everything about it, the funding that we started with is a huge, huge factor in how we operated, was to be able to start with such a large funding. You can start from first principles approach. So, a lot of places, a lot of startups will say, okay, we’re going to make a biomed device that’s super expensive, what are the materials that are already approved by the FDA? Let’s start with that. Well, you start from that point of view you’re already going to make an incremental step because you’re starting with the same five materials that everybody else is starting with. But if you’re in a position where you have a long runway, you can say, and if the leadership is really emphasizing tech and engineering, how do we solve this from an engineering point of view?

Matt Angle:
It’s interesting that you say that because some first principles approaches in biology, sometimes they’re phenomenally successful because they disregard a lot of kind of tribal lore that tends to accumulate in biology. But sometimes they’re remarkable failures because, of essentially biological systems have many, many parameters and many of those parameters are not well characterized. And so the successful approaches have tended to be empirical. And it’d be interesting if you could kind of speak a little to, what it’s like to try to rethink BCI based on first principles.

Vanessa Tolosa:
I mean, I wouldn’t recommend it for every startup. I think there are practical questions that a biomed or medical device company, starting with the market size with reimbursement codes, the indication, how much the cost of manufacturing, how much is this going to cost? Are people going to be willing to pay for it? Those are things that don’t really fall under first principles engineering thought. But, if we’re just looking at the engineering bubble because, we’re allowed to think that way we looked at the problem, is it true, we literally started with, can we interact with 83 billion neurons? Can we have 83 billion electrodes, for example. And then you start from that assumption and like pare it down based on either physics and then eventually through like practical reasons. But I think it was freeing in the sense that you were allowed to think about the problem.

Vanessa Tolosa:
Like, how do you really solve the insulation problem driven to that, what material could you use, if you’re okay with going through the FDA process. But even that, it’s still all… I wouldn’t say things have been solved. But, it can take a little longer, actually we do this approach, but in the end, the hope is that when you do come out, it’s something that is steps above what others are doing. And hopefully it brings up others. That’s really my hope is that what I said is most startups will say, okay, what are the five things that I already approved? Well, I’m hoping that companies like Neuralink and others who are willing and who have the money and capability to add more materials to that, will now lower the barrier for all the other neuro tech companies to jump in and start at that next level, rather than have to stick with, say silicons all the time.

Stuart Cogan:
Yeah well, Neuralink can afford quite a vision, right? So, I mean, there’s really a remarkable vision there. So, interesting…

Matt Angle:
Stuart And Thomas, I’m curious, with increasing interest and increasing kind of exposure, for the BCI community and a number of startups popping up, has that changed the way that you think about your programs? Is that directed you in any sort of different ways? Or do you feel like they’re doing their thing and you’re doing your thing?

Stuart Cogan:
No, it’s made me a little bit more anxious to stake a claim on technology, not IP, but just now I’m thinking, this is really good, we should publish this soon. I would hate someone else to beat us to it. That kind of thing. But beyond that, no, it really…

Thomas Stieglitz:
Same to me. So it hasn’t changed a lot. And I believe you give and to take, we don’t know what the future brings. I mean, if you look in the field of retinal vision prosthesis, that was a field where a lot of groups and even some companies then thought this is awesome. You don’t have to open the skull and you just go into the eye and you do a retinal vision prosthesis. That was second sight, or his second sight in the US, and was retinal implant in Germany. And they both had the medical device approval. And then a lot of patients got implanted. The device was at a high maturity level. And then it came out that the disease as such is not so well understood that patients can really benefit from the implants, but that was possible only because those implants reached that high level of maturity.

Thomas Stieglitz:
And I think this will happen in some other diseases, that’s painful. I really believed in that stuff, and now it’s bounced back to academia, to neurophysiology labs, to graduate schools that you look deeper into the mechanisms. And I strongly believe that will happen in some BCI groups that might happen on bio electronics medicine, that you come to a point where it can go to patients, and then you find out, oh, that hypothesis about this disease is different, and that would mean a drawback to that device but, it helps the whole field to learn more, make better devices, better therapies, better treatment options. But this is probably sometimes something where I look differently on that field, and then I know there will be back lashes and pitfalls, but that it helps to to get to the next level.

Stuart Cogan:
Well, the second sight thing is funny, I mean, maybe we’ve learned a couple of lessons there. So, the second sight approach was to use fairly traditional technology. And really their target and what they were hoping to get out of that target really needed technology that was much more fine scale than they were developing. And so perhaps with some of the more refined actually, thin-film approaches that are coming, perhaps out of Stanford, for example, will improve a great deal our ability to provide artificial vision by retinal stimulation. But, that’s one lesson, but I think another lesson which sort of gets more into the medical details with second sight is just the target, right? So, the disease was Retinitis Pigmentosa. And so two to 4,000 people a year in the United States maybe, an unsustainable market unless you have some significant funds.

Stuart Cogan:
But the point I want to make is they were also targeting an organ that was horribly diseased, hopelessly disorganized. There was [Mula cell proliferation 01:37:26] and all sorts of things going on. And this was progressive, right? And so they were targeting tissue that was becoming progressively worse. So, maybe what we should learn from this is we should not be targeting what’s bad, right. I mean, for an aside, we should be targeting somewhere in the stream where we have healthy tissue. Unless of course the thing is we need to inhibit the tissue that’s diseased and that’s a different story. So, I mean, I think second sight, I mean, that was writing on the wall, has been on the wall for 15 years. And that’s what it was not going anywhere, at least not very easily.

Matt Angle:
I see we’re getting close to the end and I could probably just pick your brains about material stuff for another hour, but I wanted to get a chance to, in closing, get a few of your thoughts on one of the podcasts that we’re going to do next, which is on the ethics of BCI. And as BCI practitioners, do you have any thoughts, suggestions, questions. What would you want to ask an ethicist who specializes in neuro ethics? Or what concerns would you like to raise about the future of the field?

Thomas Stieglitz:
Well, one thing where I’m playing here, the traveling salesman, and my environment is from the ethical point of view, do not raise misleading expectations. I think that’s very important, especially for those who have fatal diseases who look for every option. And this is sometimes difficult for persons who are really excited about their research. Might it be an engineer, a material scientist in neuro science. So, that you stay responsible in your statements. While knowing that the others’ read something in between the lines that you’ve never said.

Thomas Stieglitz:
And, I would ask that question, if you’re in that ethics field, how can you balance that out? You have to say some really things for the futures. You have to promise a lot to get funding. On the other hand, you should not promise too much to raise expectations in persons who have definitely a strong need in getting something. And how to do that and how, it’s not the question, but I would wish that more persons feel involved in that, even if they only make [inaudible 01:40:23] cables or plucks or stir silicone rubber, that’s our responsibility that we contribute to that and should communicate that in correct and an honest way.

Stuart Cogan:
Yeah. It’s an interesting question to ask what motivates participants in a first-in-human trials. Right. And so, one of the things that we’ve been doing with the cortical vision prosthesis, not me specifically, but with our ethicists slash psychologist, is trying to understand that. And you really have to, well, where I believe that then what they now think is that, these early participants really have to be motivated by altruism and that they have to have no expectation of any improvements in their disease. But just more broadly, you assess an ethical question and we want to have BCI’s that interface with our brains to treat disease. But, at some point we’re going to get to a position in which it can be argumentative rather than just a treatment. And so when do we or how do we decide is a brain machine interface to augment capabilities, when do we cross an ethical barrier?

Vanessa Tolosa:
Yeah, I’m curious. I don’t know if this is already being answered in the medical field in other applications, but how much does the patient have a say in whether they get a device implanted or not, a patient or someone that speaks for them? So say it kind of related to what Thomas was saying, say there’s a company that’s touting really great things about their device and which has gotten the patients really excited and interested, but there’s a lot of questions by experts on that. Like when does a patient have that right to say, no, I want it versus… [crosstalk 01:42:36]

Stuart Cogan:
Yeah. Are you talking clinical trial now? Or are you talking about [crosstalk 01:42:40]

Vanessa Tolosa:
As, a product. Yeah, I guess any, any brain implant, like whether a clinical trial or product. Can they demand it and is it the patients right or someone going to protect…

Matt Angle:
Potentially demand it even outside of a clinical trial or a regulatory approval as under like a humanitarian need or something like that.

Vanessa Tolosa:
I was thinking even as a product, if it’s available and maybe people don’t think it will be useful for that person, but that person is just so sold by all the advertisement, that no, I want it for me. Even if people say it’s probably not safe.

Stuart Cogan:
Yeah. Well, I think the Dobelle cortical vision prosthesis was a little bit like that and they left the United States, went to Portugal and I know there’s about 10 or 12, something like people were implanted with electrodes on the surface of the brain. And it was really a disaster, right. It was disaster for the participants. And I don’t know, but actually now that I think about that, there is a very good, very interesting I should say, not good, very interesting book written by someone called Jens Naumann, Jens Naumann, but I think it’s called In Search Of Paradise and he was a Dobelle implant recipient.

Stuart Cogan:
And you’ll learn a lot from reading that book, not only about a lot of things, about what does motivate a subject and how even just looking at a face and seeing a couple of spots of light and knowing those spots of light, were your child’s face, how rewarding that was. But you also learn that, false expectations or unrealistic expectations build up very quickly. And then you also learn that in book anyway, that the subject or the participant of the trial, they have their own agenda as well. Right. And it does get to be very tricky, but if anybody’s interested, I would really recommend…

Matt Angle:
We’ll put a link up to that book all through the conversation, every paper or every researcher we’ve referenced will have a link that the audience can follow along. So maybe just to end things on a positive note, I’d be curious to know what each of you are most excited about in the five to 10 year horizon. What do you expect to see that may not be obvious to people who are listening to this podcast today?

Vanessa Tolosa:
Mine’s pretty boring. Just seeing how much neuro tech has become a thing, over the last even five years, and there’s somebody I work with that said recently he was going to buy a laptop and the kid who was selling laptop or probably teenager or college student, not kid, but said, “Oh, what do you do for work?” And he said, “I make brain implants.” And his response was, “Oh, like Elon Musk? I don’t think these laptops can do that.” But the point is, whenever I talked about brain implants in the past, like nobody knew what that was or cared, but now like random people know it.

Vanessa Tolosa:
So the fact, I think, not only are more companies getting involved, which means we’ll see a lot, both noninvasive and invasive, we’ll see where the technology can really take us in the next five to 10 years. I wasn’t aware of really neuroscience stuff till I was in college, which was really embarrassing. But I think now people at a younger age are becoming familiar with this field, which is only I think good for the whole field because, it’s going to be many years before we solve a lot of these problems.

Stuart Cogan:
I mean, I don’t know what particularly to expect in five years, but what I do think is that we are on the cusp because, we have recognized a whole slew of problems with implementing a brain machine interfaces. And we have emerging solutions, whether it be electrode coatings, whether it be wireless transmission, whether it’s electronics and right on the implant itself. All this is just sort of in there in a sort of a nascent stage. And I expect to see over the next five to 10 years as these technologies improve and as they come together, we will have some really remarkable devices, really remarkable devices in a decade from now. Whether they’re going to be just used in neuro science research or whether they’re going to be used in subjects or participants or in patients, I don’t know, but they are going to be some pretty remarkable devices.

Vanessa Tolosa:
Hopefully better connectors.

Stuart Cogan:
Yeah. Oh, it’s such good [crosstalk 01:47:47]

Matt Angle:
Everyone listening to this, Vanessa’s very interested in better connectors.

Thomas Stieglitz:
Me too. Me too. And cables. I would be happy if we would see one or two applications in patients really, approved applications with novel technologies, at least one piece should be thin polymer stuff, not being cast or carved out of a big chunk or block of material, but really something with novel technologies. And I hope that at least one application will make it, that we don’t fall into that gardeners hype cycle trough of dissolution because, like the first wave or artificial intelligence in the fifties or sixties where people promise too much, and then all the others stayed away and it took another 15 years to come back to that. So I really pray and hope that at least one application comes through.

Stuart Cogan:
Well, I hope so too, Thomas.

Matt Angle:
I feel confident that a few applications are going to come through on that timeline. We probably represent three of those companies on the forefront, just on this call, Neuralink, NeuroOne and Paradromics. Pleasantly, pleasantly optimistic that we’re turning that corner.

Vanessa Tolosa:
It’s a fun time to be in the field.

Stuart Cogan:
Okay. Well, I actually have undergraduates waiting for me and they’ve already delayed half…

Matt Angle:
Okay. Well thank you for taking the time.

Stuart Cogan:
Oh, it was absolutely my pleasure. I had a wonderful time. Thank you, Matt and Vanessa, Thomas. Thank you. And Lauren, thank you. And it was…

Matt Angle:
Yeah, I can’t wait until real conferences exist again and we can all meet in a real pub.

Stuart Cogan:
Yes, no, that’s, that’s something to…

Vanessa Tolosa:
Directly answer your questions then.

Matt Angle:
Yeah, I can pin people down more easily. Thanks a lot. Take care.

Stuart Cogan:
Thanks everybody.

‍