Hemdeep (00:09)
Welcome to Big Ideas in Microscale, the podcast where we explore groundbreaking research happening at the microscale where microinnovations makes a big impact. We're excited to showcase the incredible work being done by our users from around the world who are pushing the boundaries of microfluidics, lab on a chip, organ on a chip, and beyond. Through these conversations, we hope to learn from their experiences, uncover their insight, and bring their big ideas to wider audience. So whether in a lab, On the go, or just curious about the future of microtechnology, join us as we dive into big ideas at Microscale.

Hemdeep (00:59)
Welcome back to Big Ideas at Microscale. I'm Hemdeep, your host and co-founder of Creative CADWorks, CADWorks 3D, and ResinWorks 3D.

Robin (01:09)
And I'm Robin, the co-host and technical writer on the marketing team.

Hemdeep (01:12)
In today's episode, we're going deeper into our conversation with Greg Norton and Adam Bulley from Brigham Young University. Last week, we explored the early challenges they faced in microfluidic device fabrication and how they used 3D printing technology to overcome them. If you missed those episodes, be sure to go back and check them out. We'll be right here when you're ready to dive into this one.

Robin (01:35)
This week, we're going into the specifics of their core innovation, a custom-built 3D printer designed from the ground up for high-resolution microfluidic applications. Instead of relying on commercial systems, Greg and Adam built their own, achieving unmatched control of their system and its different printing parameters. We'll explore the unique features they've integrated, like pixel-level light modulation, and discuss their unique perspective that 3D printing for microfluidics is all about voids and creating negative space. So, let's jump right back into big ideas at Microscale.

Hemdeep (02:10)
What about on the printer side? What did you see from the commercially available machines and what modifications did you guys start putting in place to the machine that you had created yourself?

Adam (02:21)
I'm going to say a little bit first, mean, Greg's the one who really built it. I mean, I would say when we went into this and when I say we, it's primarily Greg and his students. It wasn't like, I wonder what the commercial system has. This was a if we were to make this from the ground up, what would it be like? And I think the the resulting features that are just completely radically different from commercial systems. I think one of the key features is we have the ability to control all aspects.

of the 3D printer. So if we want to shine a different amount of light in one pixel versus another, we can do that. If we want to do multiple exposures on the same layer, we can do that. And commercial 3D printers, that's something that certainly at the time we started this, this was not something you could do. And so just having that ability to really lift up the hood on the instrument and say, hey, I want to do this. Hey, we can do that. I think that's one of the key differences. And of course, Greg will speak to the all the other things that went into the design.

Greg (03:23)
Yeah, it's been a very long multi-year journey to get to where we're at. And just as kind of a teaser in terms of where we're at, I'm just going to hold up this little device here. This is a test device that we recently started using. It has 1,600 valves on it. Each valve is 150 microns in diameter. And this device right here that I'm holding in my hand, all of the valves work except for two. And those two, we can trace back into defects of the film. at the bottom of the resin tray. So we've gotten to the point where we can just do this incredible uniformity, repeatability, reliability, and there has been so much work that has gone into that. I mean, it's not like you can just come up with a design, turn on some 3D printer, plug in some material, hit a button, and voila, this thing comes out perfectly. That is not the story. ⁓ Even though that's where we're driving to.

And so with the very first 3D printer that we built ourselves, it was going back to, know, hey, I want some projection, an optical engine that projects the most pixels I can possibly get. And so with the TI Micromirror array devices, that meant we had to go for a particular model, which of course was not the least expensive model out there. And so that was a 2560 by 1600 pixel or micromirror. projection device and then we wanted, you know, very high resolution, so one-to-one imaging optics. And so we bought that optical engine with the optics from a vendor in Germany. And then we needed to do, you know, all the build platform and its motion and the resin tray and how we're going to lift off from the bottom of the resin tray because we're building bottom up. And so we went with this sort of OEM thing from a company in Taiwan. And so Putting all of that together with our own custom software that controlled everything. That was our first shot at just seeing what can we do with this format. And that's where all of our early results came from was from this. We took optical posts and beams and mounts and other things that we had lying around in the lab from previous optical projects and used that as the structure to mount everything to.

and created this ad hoc 3D printer. And because of the driver board in the OEM optical engine, we had to use a Microsoft Windows PC to drive the thing, which of course drove us crazy. And with our version two, which Hua, this PhD student of mine that Adam has mentioned, which he designed because long story I won't go into, but at any rate, so he designed this version two. And one of the things we desperately wanted to do was get away from Windows PCs. And so we figured out how to bypass some of the internal circuitry of this OEM optical engine and directly access the I2C interface and just talk to all the hardware directly ourselves. And so that enabled us to get rid of the Windows PC to put everything on a run it off of a little Raspberry Pi, so Linux based system and do all of the hardware control from the Raspberry Pi. and throw up a web interface so anyone could control the whole system just through a browser window. That software has gone from that very first version and we've refined it countless times over the years since to where it's now a very sophisticated and modular and pluggable piece of software that runs all of our 3D printers of whatever configuration we've created. So that's worked out really well. But as Adam said, Because we write our own software, we do all of our own hardware, we interface to whatever sensors we want to use in our systems, that means we have complete control. So we can try any crazy idea any of us has, which is fantastic because we've tried a lot of crazy ideas. And at some point here when we're ready to go public with kind of the latest thing that we've done, mean, people are going to look at us like we're crazy. But that actually has led to...

this result of 1600 valves that work and it wasn't difficult to print that. So at any rate, that's just been a long haul with the soccer side. But the hardware side, there've been numerous generations of developments and those are branched off into different directions to evaluate and try out different ideas that we've had. But the long story short is now we're at what we call our HR 3.3 series of printers, HR meaning high resolution. The first three means this is the third generation of developed printers that we've done. And then the point three is this is the third iteration of that third generation. And the reasoning behind this whole development process has been with version one, yes, we got some great results out of that and we learned a lot of really useful things. But we also found what the pain points were on the hardware and the software side and the process side. And so version two was an attempt to address some of those pain points and make it easier and more consistently able to have successful prints. And then we learned more with that version two and then that initiated doing the version three, trying to minimize some of the pain points that were still present and likewise with the other iterations. And so now we're at a point where with this HR 3.3 series, we routinely train new students from freshmen to PhDs on how to 3D print, and they're able to be successful ⁓ quite rapidly. Of course, the smaller you try to make your negative features, the more challenging it becomes, but the more experienced students who develop some significant skills, they're also able to have those turn out pretty routinely. So it's just been kind of a long haul there, but it's turned out just extremely well, I think.

Adam (09:34)
to circle back to the power of being able to control different things. I mentioned controlling individual pixels in different intensities, but there's also crazy parameters. Who would have thought that the acceleration rate for moving your print away from the stage? mean, we can even control that. And those small things just have an enormous impact on your ability to make high quality 3D prints. And so you get a conventional 3D printer, and it's like, this is the layer thickness you're going to do.

this is the, can control the exposure time for the entire print and this is the resin you're gonna use. And you're like, okay, with the printers that the Gregness students have developed, we just had so much flexibility to fiddle around with the small things. And those are really the things that have allowed us to go from, you know, kind of standard, well, with commercial 3D printer, can make channels for about 200 by 200 microns on a good day down to just some of the, you know, this crazy stuff with. know, 1600 valves in a single 3D print. that's really the, it's just so powerful.

Robin (10:36)
Well, how'd you both not get so lost in tweaking all these different little print settings and and in general like because you've developed like the hardware the printer hardware the software the resin What would you say is kind of like the most important to actually get? Good microfluid devices for example. Is it really in the 3d printer? Is it in being able to change all these little details in the software or is it like a combination of all three together.

Greg (11:07)
It really comes back to a couple of key ideas. And one is, what is the size of the negative, of the smallest negative features you're trying to create in relation to the pixel size and the layer size and the layer size. I just have to say, parenthetically, people think that the layer size is the Z resolution, you the layer thickness, and that is absolutely false. The Z resolution is the optical penetration depth. And then

Once you know what the optical penetration depth is, where you've designed it, then you set your layer thickness accordingly. And that's discussed in our papers as to how you do that. But once you've got all that set, then it really comes down to, are the negative features you're trying to create many, many pixels and many, many layers big? If they are, then it's not that big of a deal. You should be able to do it. But if they are relatively few pixels in size and relatively few layers,

in size, then that's where all the special sauce comes in and that's where all the real finesse and the difficulties arise. And so over literally years now, we've just developed such a backlog of experience of how you deal with these things and how to think about these things and what additional knobs you need to turn and how you turn them that we can get those things to turn out pretty routinely. But if you approached things on that size scale,

few pixels, few layers, like you did the things that are many pixels, many layers, you'll never be successful with it. And so this goes back to a set of ideas that we've tried to introduce where most people in being familiar with 3D printing think about, okay, we have this CAD design, so a design in virtual space, and we want to turn it into an actual physical embodiment of that design. And so 3D printing is you take this CAD design and you slice it up into these equal thickness layers, create one image for each layer, present each image sequentially in the 3D printer to your layer by layer resin setup that you're doing, and then expose each layer from the same exposure time, and voila, now you've created your print. And that does work well for large features, large in terms of many pixels, many layers. as long as you have the other stuff right, like the optical penetration depth, et cetera. But when you go to high resolutions, a small number of pixels and layers, the game just totally changes, and you need additional flexibility. And so we published in 2021 in Nature Communications a paper that basically introduced a notion of a generalized 3D printing approach. And by generalized, what I mean is that for every Z position in your print in your design, you can do the following. You can have a series of more than one image at that Z position. And you can polymerize different regions within the image area for different lengths of time. And you can do this by presenting different images that maybe they overlap, maybe they don't, different exposure times for the different images. And so the bottom line is that within one Z position, one layer, you can have a wide variety of different optical doses that you've delivered on a pixel by pixel basis. And then instead of equal thicknesses, when you're trying to do these really high resolution things, often you need to mix that up and have different thicknesses for different limited areas. And so with our generalized 3D printing approach, that's just no problem. I mean, you can do whatever you want in terms of Z positions and number of images, what the images look like, exposure times. It just gives you a much broader range of spatially distributed doses than the normal conventional 3D printing approach. So this generalized 3D printing approach is just really a crucial innovation in terms of being able to get these super high resolution negative features.

Adam (15:19)
And I think one of the key things that, you know, I mean, I have students who are non-engineers, undergraduate students who use the 3D printer and can turn those knobs. And part of the reason that they can do that so effectively to optimize 3D prints is that Greg and his students have done a really good job of connecting each of those parameters with some property of the resulting 3D print. So we know that if you change exposure time, layer thickness, ⁓ step height, number of sequential images, all of these things. We have a good understanding of what that does. And so that allows us to sort of logically and rationally choose which parameters to adjust. And, you know, we do have undergraduate students who are not engineers who can do that and create really nice 3D printed objects.

Greg (16:05)
One of things that we would love to do is spread this knowledge to everyone. mean, so anyone and everyone can do this kind of thing. We don't want it to be localized only here at BYU and things that we do with our collaborators. What we want is for everyone to be able to have this kind of 3D printing capability. And the unfortunate thing is in the commercial offerings, there just aren't printers that give you the flexibility to even see the knobs and

know that they exist, let alone change them. And so one of the things we did early on, I think our first version of this was maybe 2017, 2018, probably 2018. We put up on GitHub and open sourced a print file specification that basically allows you in that specification to set all these different things on however fine detailed basis you want. You can make it so it just does normal 3D printing, that's easy.

Or you could go in layer by layer, z position by z position, and tweak whatever you want. That's available up on GitHub. Anyone can use it. I think it's an MIT license. unfortunately, no other manufacturer has adopted something like that. We'd love to see them do it. We have evolved that to that spec. It's based upon a JSON file format. But we've evolved that spec to now it's really quite sophisticated. Some way to do list to have one of my students update to the latest spec that we use internally so that that's available externally.

Robin (17:40)
was gonna actually talk about, because you use the term negative features or negative space law. And when I'm going through a lot of literature, that terminology doesn't actually come up a lot. It's monolithic devices with closed channels. So I'm kind of very curious how you even came across that terminology in the first place. Did you hear it from somewhere else? Is that something you both developed on your own?

Greg (18:07)
I don't recall hearing it from somewhere else, but positive features are kind well-known thing. so with microfluidics, when you make a microfluidic device, of course, it's all about the material that's not there in the device, because that's where the fluid is going to go. And so you're basically creating these negative features in some kind of a bulk material. And so rather than ⁓ positive features that typical 3D printing focuses on, Microfluidics, the whole name of the game is these negative features, know, channels. That's a void. That's a region in which there is no material surrounded by polymerized material or a valve or a pump. You know, it's the same thing. It's just the shape of what's not there is different. And so the 3D printing has to be tuned and focused on making those negative features so that you can... have very small regions in which there is no polymerized material because that's where ultimately you want the fluid to be.

Adam (19:10)
I mean, if you think about most conventional 3D print models, they'll show something like the Eiffel Tower or a boat or a fly or a guitar or something like that. And I guess, know, the negative features, so the Eiffel Tower, the positive features are, of course, all the metal and the negative features are the air. And we're interested in the air. Once we've made this print, then we can fill that with liquids and we can control the flow and we can use that to carry out biochemical assays and do some really powerful experiments.

Hemdeep (19:38)
And I guess there's also this big transition between macro printing and micro printing. And I think you guys have touched on the fact that even as you go down in scale, you now have to have that level of control in terms of pixel, in terms of your material. in terms of your background and optics, what skillset did you have from there that you found valuable as you moved into micro printing and developing the printer that you had?

Greg (20:05)
So for me, things that have been crucial are understanding the optical interaction with materials, which gets back to optical penetration depth. But when it comes to resolution in X, Y, as well as Z, it's just knowing how image formation works, knowing about diffraction, knowing about aberrations. In designing an optical system, even though I don't design the lens system that we buy that goes with our optical engine,

I know the details of how all that works and I know what the specs mean in terms of the modulation transfer function, et cetera. so just understanding in detail how the optics work has been really helpful for me just to understand on a micro scale what's going on. And in fact, I see in the literature all the time ⁓ groups publishing papers where they model an individual pixel

its projection in the image plane as a Gaussian beam. In fact, I just went through a paper two days ago that did this, and I've been seeing this for 15 years. And that's just absolute hogwash. That's not how it works. They are not Gaussian beams. So it just drives me nuts. And we have a paper that we've been meaning to write for a number of years now that lays all this out and shows actual measurement results for what you really get in real life, which are not Gaussian beams at all.

We just haven't gotten to it. mean, just haven't had the bandwidth to bring that to closure yet. That's something we'll be doing one of these days, hopefully in the next year or so. It's on the roadmap for one of my PhD students now. I think now it's actually going to get done. At any rate, that optical understanding has been really helpful.

Hemdeep (21:51)
just explain what is a Gaussian beam and

Greg (21:53)
Sure, yeah. So if you have a high quality laser pointer, the beam that it sends out is a Gaussian beam, meaning that if you project it on your hand and you look at that spot, if you were to take ⁓ just a cut through there and plot the irradiance, meaning the power, that would follow a Gaussian shape.

So it would be an e to the minus x squared over whatever the parameter is for its width. so Gaussian beams arise very naturally from laser cavity design. They are the zeroth order transverse mode that a laser cavity produces. And so most lasers, if it's a good laser, produces a Gaussian beam. And so for people in optics, you learn about Gaussian beams in the context of lasers because that's what they produce.

You learn the properties of them and how you deal with them. And for whatever reason, I don't understand people have taken this Gaussian beam notion and somehow transferred it over to image formation and image projection and modeling each individual pixel as a little Gaussian beam, which is just absolutely wrong. Instead, it's the diffraction pattern of whatever your pixel size is, modified by whatever aberrations are introduced by the optical system.

And ⁓ ideally your optical system is what's called diffraction limited, meaning that your resolution and fidelity is limited by actual diffraction of light, which you really can't get around. That's the best you can do. And so what happens is in designing an optical system, you reduce all of the aberrations so that their impact is about the same as diffraction is. And then you don't do anything more than that because that's just wasting money because it's not going to have any improvement because diffraction is the ultimate limit.

Robin (23:50)
that was a goldmine of information. I'd say that this conversation really puts into perspective the importance of understanding the interaction between hardware and software and how you need to optimize how you approach these to get the results you need. So with that, I think this is a good place to wrap up today's episode of Big Ideas at Microscale.

Hemdeep (24:12)
A big thank you to Greg and Adam for breaking down their custom built 3D printer and for sharing their innovative approach and perspective on microfluidic 3D printing.

Next week in our final episode with Greg and Adam, we'll go over their technology in action, some devices that they've and printed to test their system, and their potential applications in bioessays and molecule detection. We'll also explore the surprising discoveries they've made in some offshoot experiments that showed that their platform is really pushing the limits of 3D printing. You won't want to miss out.

Robin (24:44)
Thanks for tuning in to Big Ideas in Microscale. If you enjoyed the episode, make sure to follow us and stay up to date. You can listen on Apple Podcasts and Spotify, or watch the full video on YouTube. You can also follow us for more updates and behind the scenes content on LinkedIn, Instagram, Blue Sky, and X. We're Cadworx3D across the board. That's spelled C-A-D-W-O-R-K-S-3D. for show notes, paper references, and bonus resources on today's topic, visit our website, catworks3d.com. That's spelled C-A-D-W-O-R-K-S 3D.com.

Hemdeep (25:25)
Thank you for tuning in and as always stay curious, keep exploring and never stop asking the big questions that are shaping our world. Whether you're in the lab, on the go or just curious about the future of technology, join us as we continue to dive into big ideas at Microscale.