Researchers at the University of Notre Dame just 3D printed working blood capillary networks with channels under 10 micrometers across. That is thinner than the finest human hair. The work, led by Professor Yanliang Zhang with collaborators at Harvard Medical School, landed on the cover of Nature Chemical Engineering, and it was published on August 11, 2026. Lab grown organs are still years away, but two things inside this study matter right now to anyone who prints parts or buys them: how the team combined two different printing processes, and how they used machine learning to stop guessing at print settings.
What Notre Dame actually printed
The team built a hybrid bioprinter that runs two processes in one machine. An extrusion head lays down the soft, gel-like matrix that mimics real tissue, one layer at a time. Then an aerosol jet printing head deposits ultra fine threads of gelatin inside that matrix. When the whole structure is dunked in warm water, the gelatin liquifies and washes out, leaving behind clean open channels where the threads used to be.
Aerosol jet printing uses a sheath of gas to focus the material stream, and by adjusting the ink flow rate and the sheath gas flow rate the team could print channels from hundreds of micrometers down to just a few micrometers. They then seeded the channels with endothelial cells, the cells that line real blood vessels, and the cells attached, spread along the inner walls, and formed a barrier that held without leakage. More than 100,000 people in the US are waiting on an organ transplant, with a new candidate added every 10 minutes, so the stakes behind this research are enormous. The group has already secured new NIH funding to build an even more capable version of the printer.
My prints keep failing when I change materials or geometry. How do I find the right settings without wasting a whole spool?
The fastest reliable way to dial in print settings is structured testing on small calibration parts, and the cutting edge is now machine learning that finds optimal parameters automatically. This is the part of the Notre Dame study that should make every maker sit up. Tiny changes in ink flow and gas flow changed the channel size, so instead of trial and error, the team built a machine learning framework that converged on the right parameter combination for each target size. Professor Zhang said the autonomous optimization delivered a large boost in efficiency compared with conventional trial and error. Production shops do a manual version of the same thing every day: test coupons, one variable at a time, and a locked profile once a material passes. The lesson transfers directly. Whether the machine costs five hundred dollars or five hundred thousand, guessing is the expensive option.
I need a part with tiny internal channels for fluid or air. Can 3D printing actually do that?
Yes. 3D printing is the best available way to make parts with internal channels, because the channels are formed during the build instead of being drilled in afterward. Notre Dame's sacrificial gelatin trick is an extreme example of a technique that already works at everyday scale. Manifolds, coolant paths, air distribution channels, cable raceways, and internal lattices are all routine in engineering prints, and they are exactly the features that machining struggles with. For hot, chemically aggressive, or sterilization heavy environments, high performance polymers carry those designs further: our PEEK and ULTEM printing service builds channeled parts that survive autoclaves, engine bays, and process equipment. If the part you need to replicate already exists, our 3D scanning and reverse engineering service can capture it, rebuild the CAD, and add the internal features the original never had.
Why one printer was not enough
The quiet headline of this study is that no single process could do the job. Extrusion printing builds soft structures fast but cannot hit single digit micrometer detail. Aerosol jet printing hits incredible resolution but cannot build a bulk scaffold. The breakthrough came from letting each process do what it is best at, inside one coordinated build. Real world manufacturing works the same way. The right answer to a hard part is often a chain: scan, then print, then machine a critical face, then heat treat. Shops that treat processes as teammates rather than rivals ship better parts. That is the philosophy behind our build and ship workflow, where the process chain is picked around the part instead of forcing every job through one machine.
What this means for buyers and makers
Capillary scale bioprinting will not touch your workbench this year. The direction of travel still matters. Multi process hybrid machines, sacrificial materials that vanish on command, and self tuning parameter optimization are all moving from research labs toward production floors, and each one lowers the cost of the complex internal geometry that used to be impossible. If you are an engineering buyer, the practical takeaway is that internal channel designs you shelved five years ago are worth pricing again. If you are a maker, the takeaway is to steal the method: change one variable at a time, keep records, and let data pick your settings.
If your company makes 3D printers, scanners, filament, or accessories and you want an honest hands on review from a working shop, we would love to hear from you at info@dcadditivepros.com.
Frequently asked questions
I heard scientists can 3D print blood vessels now. Is that real?
Yes. Notre Dame researchers published a peer reviewed study in Nature Chemical Engineering showing 3D printed capillary networks under 10 micrometers in diameter, lined with living endothelial cells that formed leak free vessel walls.
I need a plastic part with small internal channels. Can a print shop make that for me?
Yes. Internal channels are a standard capability in FDM and resin printing because they are formed layer by layer during the build. Send a drawing or sample and a shop can confirm the minimum channel size for your material.
My 3D prints fail every time I switch filament brands. What am I doing wrong?
Nothing is wrong with you, every filament runs differently. Print a small temperature and flow calibration part first, change one setting at a time, and save a named profile per material so you never re-guess a dialed setting.
Can 3D printed parts be used in medical or lab settings?
Only when the material and process are qualified for that use. Research bioprinting like Notre Dame's is not a consumer service, but lab accessories such as racks, holders, and enclosures are printed every day, and high temperature polymers like PEEK tolerate repeated autoclave cycles.
DC Additive Pros is not affiliated with, endorsed by, or sponsored by the University of Notre Dame, Harvard Medical School, Brigham and Women's Hospital, Nature Chemical Engineering, or the National Institutes of Health. All trademarks belong to their respective owners. Research findings summarized from the University of Notre Dame's published announcement and study.