Metal additive manufacturing has a speed problem, and the usual fixes come with a catch: crank up the deposition rate and resolution tends to suffer. Researchers at Rochester Institute of Technology's AMPrint Center are betting that the way out is the same trick that made desktop document printing fast and cheap -- put more nozzles on the head. A newly detailed multi-nozzle molten metal droplet jetting printhead, an array that fires molten metal droplets much like an inkjet, is designed to push throughput up without giving back the fine features that make the process worth using in the first place. The work circulated through the trade press in June 2026.
How molten metal jetting actually works
Molten metal jetting (MMJ) sits in the drop-on-demand family of additive processes. Instead of melting a powder bed with a laser or pushing a wire through a hot nozzle, an MMJ system holds metal as a liquid and ejects discrete droplets only when and where they are needed. Each droplet lands, solidifies, and bonds to the layer beneath it. Because the droplets are small, the process can resolve fine features -- the resolution is set by droplet size and placement rather than by the width of a melt pool or an extruded bead.
That is the appeal and, historically, the constraint. A single jetting nozzle can only fire so many droplets per second, and if each droplet is small enough to give good resolution, then a single nozzle simply cannot move enough material to print anything large in a reasonable amount of time. The RIT team frames its work explicitly around the throughput limitations common to single-nozzle droplet jetting systems. You can have small droplets and good detail, or you can have speed -- but with one nozzle, not both.
What the multi-nozzle design changes
The RIT printhead attacks that trade-off head-on by multiplying the nozzles. According to the reporting on the work -- described as among the first demonstrations of molten metal droplet jetting with a multi-nozzle array -- the design uses a communal reservoir feeding three piezoelectric actuator pistons, with each piston driving its own nozzle. The drive waveform for each nozzle is independently addressable, allowing precise control over drop placement for raster printing of arbitrary layer shapes. The piezo actuators are what make the droplets fire on demand: a voltage pulse drives a piston, the pressure spike ejects a droplet, and the system repeats. Running that across multiple nozzles in parallel means the head can deposit material at a far higher aggregate rate while keeping each individual droplet small.
The concept is worth dwelling on. The researchers describe parts built from millions of tiny molten metal droplets deposited and fused as they cool, with the nozzles firing in parallel to lift the aggregate deposition rate well beyond what a single nozzle can manage. Crucially, the resolution does not have to collapse to get there, because the small droplet size is preserved across every nozzle. In other words, the multi-nozzle array multiplies output without coarsening the print. That is the central claim, and it is the right one to make: throughput and resolution have been the two axes metal AM has struggled to optimize at the same time.
Parallelizing nozzles also addresses the throughput ceiling that has limited single-nozzle systems. A single nozzle can only fire so fast; spreading the deposition work across multiple independently driven channels is the same logic that lets inkjet print heads run fast and keep going through the occasional nozzle dropout. The shared reservoir feeding the actuated pistons is the architecture that makes that parallelism practical rather than just bolting several independent printers together.
The recycled-feedstock angle
There is a second thread running through this work that matters as much as the speed story. The RIT effort is tied to a broader push, backed by a nearly $3 million National Science Foundation grant, to advance metal 3D printing using recycled feedstocks. By separating the melting and deposition stages, the approach is meant to run faster and accept lower-cost inputs -- including recycled metals and machining scrap, and even wire, rod, or ingot feedstock. The trade coverage of the printhead frames it explicitly around making scrap metal usable as printable input for higher-volume metal additive manufacturing.
That pairing is not incidental. Powder-bed metal processes are demanding about feedstock: they want tightly specified, spherical, gas-atomized powders, which are expensive and energy-intensive to produce. A jetting process that holds metal as a liquid in a reservoir is, in principle, more forgiving about where that metal came from -- including scrap that would otherwise be downcycled or discarded. Combine a feedstock path that can ingest recycled material with a printhead architecture aimed at higher throughput, and you have a plausible route toward metal printing that is both cheaper to feed and faster to run. Those are two of the biggest cost levers in the whole field.
What It Means for Makers
A few caveats before anyone clears bench space. This is research-stage work coming out of a university additive manufacturing center, not a product you can order. The verified material describes the printhead architecture, the resolution-preserving rationale, and the connection to NSF-funded recycled-feedstock research -- it does not promise a ship date, a price, or specific alloys validated for production. Treat it as a signal of direction, not an announcement of availability.
That said, the direction is one worth tracking. For makers and small shops, the long-standing barriers to metal AM have been cost and access: powder-bed systems are expensive, the feedstock is pricey, and throughput on detailed parts is slow. A multi-nozzle jetting approach that decouples speed from resolution, and that is explicitly being developed alongside recycled-metal feedstock, points at a future where metal printing could be both faster and cheaper to operate. If the multi-nozzle array scales -- more nozzles, reliable parallel jetting, qualified materials -- it is the kind of architecture that could eventually move metal AM from low-volume specialty work toward genuinely higher-volume production.
The bottom line: RIT has demonstrated a credible technical answer to the throughput-versus-resolution bind that has held metal jetting back, and they have wired it into a feedstock story that addresses the other half of the cost equation. It is early, but it is the right shape of progress -- and the kind of fundamentals work that tends to show up in commercial systems a few years later.
