Contact-Free 3D Hybrid Printing with New High-Performance Materials

Ed Brown

Javier Ramos, CTO, and his team from Inkbit Corporation, Medford, MA, along with researchers from MIT and ETH Zurich, have developed a 3D inkjet printer that uses contact-free computer vision feedback to print hybrid objects with a broad range of new functional chemistries.

Complex structures such as this lattice cube can be 3D printed using Inkbit’s contact-free technology. (Image: Courtesy of Inkbit Corporation)

Tech Briefs: Can you give me an idea of how you got started with this idea of using computer vision with feedback to do 3D printing.

Javier Ramos: I'll preface it with a little bit of the history our company, Inkbit. We started out of MIT, where I was part of a team working on a research project to advance multi-material printing. As we started developing the project and innovating in that space, one of the things we found was that the existing 3D printing technology, especially existing inkjet 3D printing technology, had some severe limitations with regards to accuracy, precision, and the type of materials that could be printed.

That was primarily because those machines rely on mechanical planarization for control of the print process. The systems had rollers or scraping devices that would planarize every layer as it was printing. So, we thought there could be a better, more computational, advanced way of doing it. Instead of doing it very crudely with mechanical devices, we could do it digitally with machine vision. If we could scan every layer and see each one, we could understand what the printer was doing. We could then take that information and adaptively control the print process. So, at MIT we did an initial proof of concept of that idea, combining inkjet 3D printing and 3D machine vision. That was the kernel of how we got started.

Tech Briefs: Did you have the idea of using this to print with hybrid structures and materials right from the beginning?

Ramos: While hybrid structures weren’t our initial focus, as we started exploring the benefits of machine vision, we converged to the idea of creating hybrid objects. We did some demos back at MIT — the project was called MultiFab — where we embedded components into printed parts. That wasn't the original idea, but as things evolved, we converged into that capability. That also led us to learn about new chemistries that could only be printed with systems that have no mechanical contact. Printing without mechanical contact with the substrate, is what enabled the new kinds of chemistries that we're printing today.

Tech Briefs: In what way did it enable new chemistries?

Ramos: That is at the core of Inkbit: the enablement of new chemistries. When you're printing layer by layer with inkjet printing, which is the deposition technology we use, you're essentially jetting little droplets of material about half the diameter of a human hair onto the substrate. All these droplets coalesce and form a film of fluid before the material is polymerized.

Typical 3D printing systems today flatten these layers to ensure accurate geometry. You cannot print without any sort of feedback control, either mechanical planarization, which is very crude, or machine vision control, which is what we do. When mechanical systems contact the material, they’re essentially scraping it flat. So, what happens is that with a material that's highly reactive, let's say a two-component chemistry that cures over time, the material sticks to the roller, polymerizes it, and gums it up — it essentially glues the roller in place. So, you're limited in the kinds of chemistries you can use to some that are less reactive and are compatible with mechanical flattening devices. And those chemistries usually tend to be less functional.

However, when you remove that mechanical constraint, you can print materials that are highly reactive and cure over time. For example, in our epoxy chemistry we have we what's called a cationically cured, dark-cure, reaction. That means that on top of UV polymerizing, our material also cures over time thermally. You cannot print those kinds of chemistries with mechanical planarization.

Most of the 3D printing industry uses polymer plastics, mainly thermoplastics and thermosets. Basically, all thermosets today in 3D printing are acrylic chemistry because of the mechanical constraints. By not needing mechanical planarization suddenly you can get into more functional chemistries like epoxies, thiol-enes, and COPs, which are emerging types of chemistries that are very difficult or impossible to print with the other technologies.

Tech Briefs: I read that you made a robotic hand with tendons in it. Can you talk about that.

Ramos: This is a project that's a collaboration with the ETH Zurich soft robotics group, who designed the robotic hand. It was meant to showcase the capabilities of our technology, our ability to incorporate both rigid and soft areas. We wanted to showcase something that was multi-material and since the body is the ultimate multi-material, we decided on this anthropomorphic design as one of the examples. The hand is perfect because it has a lot of detailed rigid bony components, and it also has the soft elements that are critical for its flexibility and grasping functions. And then, apart from hinges in the articulations, there are a lot of other dynamic elements like the tendons that are used to actuate the articulation. So, it's just a fantastic way to showcase the power of the of the technology — that was the inspiration for the hand example.

Tech Briefs: Yeah, I can see it's a great showcase. It must be hard to activate all those different parts in an actual robot.

Ramos: Yes, this is an extreme example, with extreme complexity. A lot of industrial manipulation tasks are a lot simpler, a lot more constrained. There could be a simpler version of this device with the same inspired elements, articulation, some tendons, and other elements that would be much more practical. This hand has not been designed for practical use, but as a demonstration of what's possible with our technology.

Tech Briefs: Are you producing any practical things right now?

Ramos: Yes, at Inkbit, we offer the ability to have functional 3D printed parts for customers via a print service, and we also sell and install our printing systems. We have a few machines in the field with customers who are in the robotics space. And then from a service perspective, we have customers in a wide range of application areas: medical models, functional prototyping, product development — a wide range of application areas, the common theme being that the parts are usually elastomeric or multi-material.

Tech Briefs: I also read that you're using something like 16,000 nozzles. Is that typical or did you have to design in this feature in addition to the non-contact vision feedback?

Ramos: When people think of nozzles, they are usually thinking of an extrusion type of nozzle that you typically see in 3D printing. However, we’re referring to the nozzles in the inkjet print heads that we use — these nozzles are microscopic in diameter. We're talking about a 50-micron diameter nozzle, about half of a diameter of a hair. This is a similar technology to what's used in your home printer, but instead of CMYK inks, we use different polymers with different properties to achieve the multi-material combination. We use an industrial version of the technology you have in your home printer.

Our machine can be set up with as many as 16,000 nozzles. The system we used for printing the hand was set up with the full 16,000 nozzles. And that means that we have the machine set up to print four materials. That's typically the support plus the three goal materials, so each material has 4000 nozzles in the print monitor. It's a highly scalable industrial technology, if you want more nozzles, you can use more print heads.

Tech Briefs: I would imagine coming up with the software is quite a challenge.

Ramos: Yes, it is very interesting because 3D printing is a convergence of many disciplines. You have the hardware, the software, and you also have the chemistry and materials development.

The software is a critical piece of the whole puzzle, and that was one of the main challenges. We're doing the printing with the 3D scanning in the loop, in real time. So, when you look at the amount of data we're capturing, because these are very high-resolution scans, they're about a 30 Micron XY pixel resolution, and a resolution in the Z direction, in the third dimension, of about 8 microns. It's very high-density, high-speed data. One of the key challenges is how do you capture this data so quickly, how do you reconstruct it in real time, how do you compare that information in real time to your desired geometry, and then generate the print data for the next layer.

So, executing this full feedback loop in real time is a is a big technical challenge from a software perspective. We had to develop some new algorithms and processing pipelines to enable this to happen. We don't want to print a layer, scan it, and then have to wait for 10 seconds to print the next layer. We want to print a layer and then be able to print the next layer very quickly. So, we have about two seconds to do all the capture and processing of the data. We had to leverage new algorithms and new compute hardware. We are quite extensively using graphical processing units (GPUs), which are becoming very common now, for example with crypto mining and AI. Model training GPUs are now exploding with NVIDIA and other companies aggressively targeting the market. We're using the same technology but for our machine vision processing pipeline. The emergence of low-cost and high-performance GPUs over the past 5 to 10 years has allowed us to do this real time compute, along with our custom algorithms.

Tech Briefs: What are your plans for the future?

Ramos: Our vision for Inkbit is to reshape how the world thinks about production, from design to execution and make our technology readily available. The big opportunity with 3D printing is how to disrupt the world of manufacturing — that’s what we're focused on.

From a technical perspective, there are still a lot of open challenges around developing new materials and chemistries that are more functional, and to enable new kinds of application areas. Producing the design, for example of the robotic hand, required extensive PhD-level work. Once we produce the complex designs, the challenge is how do we provide the tools for people to leverage the technology for industrial use. So, a goal for our future is to create software and design tools to enable new kinds of devices and products.