Laser Focused: How Light Reflections Could Revolutionize 3D Printing of Metals

Even if you’ve never heard of “light caustics,” you’ve probably seen them. They’re the ethereal patterns of light that coat the bottoms of swimming pools and break up the shadows of glass. Anywhere light is bent or reflected by a curved surface, caustics can appear.

This trick of the light is turning out to be an important tool for 3D-printing metal, a process in which high-powered lasers melt metal powder one layer at a time. NIST scientists have found that caustic reflections from those lasers contain clues about the metal’s microscopic shape.

This research will give us never-before-seen, real-time insights into metal 3D printing.

What’s Metal 3D Printing?

Most metal 3D printers work by laying down a thin layer of metal powder, using a laser to melt that powder together into a solid shape, then laying down a new layer of powder and repeating the process.

Humans have been making things out of metal for thousands of years and have come up with ways to make metal into a specific shape: forging, stamping, bending and many others. 3D printing won’t replace them, but it has a few unique advantages that can change the way we think about what’s possible to make with metals.

First, 3D printing makes the inside and the outside of a part simultaneously. This enables 3D printing to make complicated shapes that aren’t possible with other methods. That means you can 3D-print a single part that otherwise would have required making several pieces bolted together.

Another big advantage is flexibility. Most metal manufacturing techniques are designed for mass production. They require expensive molds, dies or tools for each new part you want to build. Essentially, you need to customize the factory for each new part you want to build. That’s fine if you plan to make many identical copies of a part. But sometimes you only want to make one of something, such as a medical implant that’s shaped to fit a specific person, a unique replacement part for a nuclear reactor, or a prototype for a new invention to work out the kinks before committing to mass production.

But 3D-printed metal is a new technology, and the results don’t come out perfectly every time. Sometimes the printing process creates small defects that make the part weaker. Finding and correcting those defects while the print is happening is one of the biggest challenges for metal 3D printers.

Part of NIST’s mission is to strengthen American manufacturing through research. So, our scientists have been studying metal 3D printing up close to search for defects. Solving this problem could make metal 3D printers more consistent and reliable.

But getting a very close look at metal 3D printing isn’t easy; it’s next to impossible to see what’s going on inside the machine. That’s partly because metal 3D printers use serious hardware that needs to be contained for safety.

First, there’s the powder. Some of the powders used are toxic to breathe and touch, and some types of fine powders can be explosive under the right conditions. If the particulate is kicked up into the air, the high surface area makes the particles easy to ignite. This is a serious safety concern in places such as flour mills, sawmills and coal mines. Although you don’t typically think of metal as being flammable, many metals can burn, and they ignite more easily as a powder.

To prevent explosions and stop corrosion, the whole printing process needs to be carefully contained in a box of inert gas. The lasers themselves are powerful enough to blind and can’t be looked at directly. And, of course, the business end of the laser gets very, very hot.

Like any manufacturing process, metal 3D printing is safe with the right precautions. But these extreme conditions do make it difficult to observe what happens during the print.

There’s a lot to see. In the tiny area at the end of the laser, the metal powder melts into a white-hot, churning puddle called a melt pool. A miniature windstorm caused by the sudden heat kicks up sputtering clouds of cold powder and molten metal. Some of the metal at the surface of the pool instantly vaporizes into a gas and jets up into the air. The process is so chaotic, it’s amazing that it works at all.

“It would be very helpful to monitor how the print is going in real time,” said David Deisenroth, a mechanical engineer at NIST. “Is the part getting too hot? Are there any defects? We want to be able to adjust the printer to address these problems because it will lead to stronger and more consistent parts.”

Finding Keyholes

There is a kind of Goldilocks principle when it comes to the temperature of the melt pool in metal 3D printing. Too cold, and the powder won’t completely fuse together. Too hot, and you tend to get defects called “keyhole pores.”

A keyhole sometimes forms when vaporizing metal at the melt pool’s surface presses a pit into the metal part. As the vaporized metal shoots up, an equal and opposite force pushes the surface of the pool down.

“The keyhole in molten metal could collapse in on itself and leave a pore in the material after it cools,” explained Deisenroth, “and that's bad for the strength of the final part.”

Keyhole pores could be prevented by easing up on the intensity of the laser when they start to form, but they’re hard to catch in action. The defects are smaller than the width of a human hair. They also happen in a fraction of a second, in the middle of a blinding pool of liquid metal, under a high-powered laser, in an air-tight box of toxic powder.

If he wanted a chance of catching them in real time, Deisenroth had to get creative.

During the printing process, some of the laser light reflects off the surface of the metal. The pattern of this reflection can give the researchers information about the shape of the liquid metal’s surface.

Deisenroth has built a special metal 3D printer that can be used for experiments, called a test bed. He set out to use this test bed to record the laser reflections as accurately as possible.

Deisenroth outfitted the test bed with a hollow dome about the size and shape of a basketball cut in half. That dome, originally sold as an architectural decoration, covers the metal sample that will be melted with the laser. It has a small slit at the top where the laser can pass through. The dome was designed to catch all the light caustics reflected by the laser, in the same way that the underside of a bridge catches the light reflected off a river.

“The biggest challenge was creating a coating for the inside of the dome that would reflect the laser light only once. If the dome were too reflective, the light would bounce around many times, and it would look uniform. If it wasn’t reflective enough, we wouldn’t see any light at all,” he said.

By taking high-speed video of the inside of this dome, Deisenroth could be sure that he saw all the light reflecting off the metal melt pool.

The experiment was a success! Deisenroth found that he could use the reflected laser to determine whether a keyhole was forming and estimate how deep it was.

This experiment was just the first step in proving that this idea works. Deisenroth has plans to improve the measurement in future experiments. For example, the camera he used takes images at 60,000 frames per second. This is very fast — normal video flashes images at 30 frames per second. But even at those incredible frame speeds, the laser light reflections are still faster, and the video is a little jittery. He would like to record with an even faster camera, up to 825,000 frames per second. 

This technology isn’t quite ready to be used in a commercial 3D printer. The reflecting dome restricts the distance the laser can travel to a few centimeters, so it can’t be used with printing larger objects.

But getting the measurements as accurate as possible here at NIST has advanced the general understanding of laser reflections. This research will help with the design of new ways to use laser reflections to monitor metal 3D printers.

For example, one idea is to place little satellite dish-like reflectors in the corners of the printer to measure the reflected laser light. But tools like that wouldn’t be possible without fundamental research like this.

To put this all into context, the first 2D laser printer for paper was the size of a minivan. It cost nearly $1 million in today’s money and required a special technician to operate. That was only 50 years ago.  

A modern office printer is now a fraction of the size and 1,000 times cheaper. Ultimately, by reducing defects, Deisenroth’s technology could make metal 3D printing easier to use, less expensive and more reliable. His work could make the process available to new industries and smaller businesses.

In the meantime, Deisenroth will continue his work measuring the seemingly impossible.

“There’s something really exciting about seeing what otherwise can’t be seen,” said Deisenroth. “The lasers we use are invisible to the eye, and the reflections move so fast that you can only see them with a high-speed camera. It’s amazing to think that we can capture these caustics in action and draw meaning out of them.”