Elmet Uses 3D Systems Printer to Build Hypersonic Heat Exchangers in One Piece

Elmet Technologies is installing a DMP Flex 350 Triple metal 3D printer from 3D Systems to produce hypersonic heat exchangers as single monolithic components. The system will run at Elmet's facility, targeting a 2026 production entry for parts used in hypersonic vehicle thermal management.

The heat exchangers are printed from C103, a niobium-based refractory alloy that retains strength at extreme temperatures. C103 is not a beginner's metal AM material. It is oxygen-sensitive, it oxidizes at the slightest temperature change, and the powder costs roughly $4,000 per kilogram. Those constraints have kept it out of all but the most specialized production workflows. Elmet has spent eight years developing the process parameters and qualification data needed to make C103 printing reliable.

Why Print a Heat Exchanger as One Piece

Hypersonic heat exchangers require internal channels that move coolant or fuel at high flow rates while surviving surface temperatures that would melt conventional alloys. Making these channels inside a solid part is exactly what metal AM was built to do: the printer deposits material only where it is needed, producing thin walls and high surface-to-volume ratios that are difficult to achieve with conventional fabrication.

The alternative to additive manufacturing for these components is brazing. That means fabricating individual sections, aligning them, applying filler metal, and running the assembly through a furnace. Each joint is a potential failure point under extreme thermal cycling. Printing the entire exchanger as a single part from one digital file removes those joints entirely. It also eliminates part-specific tooling, which matters when every design iteration is expensive to re-tool.

The Role of Process Control

The DMP Flex 350 Triple maintains oxygen levels below 25 parts per million inside the build chamber, with typical readings between 0 and 6 ppm. Ambient air contains roughly 210,000 ppm of oxygen. Keeping the chamber that inert allows C103 powder to be reused across multiple builds without degrading material properties, which is economically significant given the powder's price tag.

The printer uses three lasers across a 350 x 350 x 350 mm build volume. Elmet plans to use that capacity to produce heat exchangers large enough to be functional in hypersonic applications, not just lab specimens. The company's stated goal is production entry in 2026, which suggests the eight-year qualification effort has generated enough data to satisfy internal standards and the agency certifications required for defense work.

Supply Chain Considerations

There is also a supply chain dimension to Elmet's investment. Refractory metals like C103 pass through a small number of suppliers and fabricators. By bringing the printing process in-house, Elmet reduces its exposure to upstream concentration risk and positions itself as a domestic source for hypersonic thermal management components.

The broader industry context is worth noting: qualifying C103 for AM has been a multi-institution effort. Velo3D has been advancing C103 powder qualification from Amaero on its Sapphire printers, and Auburn University's National Center for Additive Manufacturing Excellence confirmed the powder met ASTM F3635 Class B standards following a 2,200-degree-Fahrenheit heat treatment. Elmet is not the only player working through C103's challenges, but it is one of the furthest along in moving from qualification to production.

What Comes Next

Elmet Technologies is positioning this installation as a step toward localized production capability for the US defense industrial base. Hypersonic vehicles remain a high-priority area for defense spending globally, and the thermal management components they require are genuine production bottlenecks. If Elmet's C103 workflow delivers consistent, qualified parts at scale, it fills a gap that current supply chains struggle to serve.

The DMP Flex 350 Triple installation is a concrete step in that direction. The real test will be whether the process data accumulated over eight years of development holds up when the machine is running production builds rather than qualification specimens.