How often has one of your designs been a trade-off of weight versus strength? Too heavy, and you waste material. Too light, and parts might fail. Well, here's good news. A recent technology fabricates components that are strong and lightweight. Called electron beam melting (EBM), the technique accelerates electrons to half the speed of light onto powdered metal to melt and weld the material, one layer at a time. As with other additive methods, EBM builds parts that can fill arbitrary volumes and makes for a great way to generate “lattice structures,” arrangements of repeating patterns with engineered stiffnesses. There is often no other practical way to fabricate some of these geometries.

In the medical area, lattices can replace material in implants. The resulting structures cost less as well as help facilitate bone in-growth. In general, lattice structures can reduce weight, transfer heat, absorb impact, dampen vibration, and be engineered to a specific stiffness.

Electron beam machines (e-machines) have a build envelope around 200 × 200 × 180 mm and a build platform usually made from stainless steel. Since the melted parts have a different thermal expansion than stainless steel, they just pop off without cutting or sawing. Titanium and cobalt chromium alloy work well with EBM and there is a continuously growing list of other materials that work as well. Arcam AB in Molndal, Sweden, (arcam.com) which invented the technology, says its versions of the alloys show no remaining layering effects or weld lines from the build process and that material microstructures still feature a normal grain structure.

“Lattice structures are actually any porous geometry or what we call non-stochastic foam,” says Denis R. Cormier, associate professor of Industrial and Systems Engineering at North Carolina State (NC State) University in Raleigh, N.C. “In 2003, NC State became the first institution in the U.S. to purchase an e-beam machine. We started experimenting with metals as a natural extension of having worked with rapid plastic-prototyping methods since 1996.”

EBM is relatively fast compared with other metal processes such as laser sintering because electron-beam energy couples well with metals, says Comier. “The melt goes fairly quickly because there is no optical reflectivity. Melting aluminum, with other methods, for instance, would make a mirror that reflects a lot of the energy back. This is not the case with an electron beam.”

Because e-machines typically generate solid structures, knowledgeable users ‘trick’ machines, so to speak, to produce lattices by tweaking processing parameters to sinter the metal in certain areas and melt it in others. “Sintering is about 70% to 75% of the material's melting temperature,” says Comier. “Here's a good way to compare sintering and melting: Imagine mud after it dries — it is a cake of dirt. Scratch the surface with a fingernail, and particles fall off. This is like sintering. But if the mud melted though, it would have turned into solid rock. This, of course, is melting.

Design and safety considerations

Cormier says for EBM, the finer the powder, the better the surface. “Fortunately, e-machines melt parts in a vacuum. Exposed to oxygen, fine metal, and even plastic, powders are explosive. In fact, aluminum powder is used as a rocket propellant. Thus, should a little bit of this material spill on the floor, never clean it up using an ordinary shop vacuum. It is necessary to use a special, explosion-proof vacuum cleaner to sweep up the material.”

Changing the powder size and shape also changes the electrical and thermal conductivity. Thus, process parameters such as electron-beam current, how fast the machine is tracking, and what direction the melt goes must be changed too. “Of course, Arcam prefers users purchase materials from it because the company has established the best settings to run each alloy,” says Cormier. “But there are no objections to your working with another powder provided you let the company know. It must service the machines and does not want to expose maintenance personnel to potentially dangerous materials.”

Of the parameters, melt direction is quite important. “The direction has to do with how uniformly the heat is distributed,” says Cormier. “Think of EBM in terms of a wave. When the beam scans from left to right, for example, it generates a little wave of molten titanium. Always scanning in that one direction might produce a little lip on the right-hand side of the part. But scanning from one side to other and then stepping over a little bit and scanning in reverse pumps a lot of heat into both sides of the scan bed and less in the middle. Uniformity of heat is important in the scan bed, so the e-machine software randomizes direction using a proprietary method.”

With enough experience, though, users can look through the e-beam window while the machine is running and qualitatively tell whether the build is working, says Cormier. When the melt pool gets a certain glassy look and a certain color, everything is going well.

Custom bone implants

The NC University lab experiments with several applications, including building custom bone implants. “There are two issues with bone and metal, says Cormier. “One is the fit or shape. The other is the load. The human bone is not as rigid or stiff as titanium. This can lead to a lot of problems with implants. To illustrate, astronauts in the microgravity of a space station for a long period lose quite a bit of bone mass because their bones are not subjected to loading. When individuals walk around on Earth, they're loading their bones, which respond by getting stronger. But if an individual constantly lies around, his body concludes it doesn't need that much bone so it atrophies away.”

Similarly, because a solid metal hip implant is so much stronger and stiffer than the bone around it, the implant takes up most of the load,” says Cormier. “The bone therefore starts changing shape and atrophying away. On the biomedical side, we design lattice structures with specific geometries so the stiffness of the implant closely matches the stiffness of the human bone. And because EBM is freeform, it produces the shape needed for a good fit.”

Because metal is harder than bone, implants are not designed one-to-one size-wise, but rather strength-wise, says Andy Christensen, president of Medical Modeling Inc., Golden, Colo. “Without an accurate transfer of the load from the implant to the bone, the bone might die or be reabsorbed underneath the implant. Overall, implant design is a matter of function, form, and even aesthetics. There is also a marketing side that says structures need to look ‘cool.’”

EBM can take digital models in the form of CAT scans or MRI information of a bone structure to build a custom implant to fit it, says Christensen. “Solid and lattice parts work well for hip, knee, and shoulder joints, cranial implants, and spinal applications using lattice ‘cages,’ structures that go where the spinal disk used to be.”

Christensen says EBM works with copper and aluminum, but those materials have no direct application for implants. The company uses a cobalt-chrome alloy for parts with articulating surfaces. “Take a hip joint, for example. It has a ball and socket. Typically, the socket is made from an ultra-high-molecular-weight polyethylene cup and the head of the femur is made from cobalt-chrome because it wears well and has a low coefficient of friction. This arrangement is also used in the knee.”

In the future, lattice structures will replace coatings now used on medical implants, says Christensen. “Today, metal beads are sintered onto solid metal parts, and bone grows as best it can into the spaces between the beads. Other methods spray on flakes of titanium to roughen surfaces. Building implants with lattices, however, will provide a better mechanical lock between the bone and the implant.”

Lattice parts by Medical Modeling produces are still in the prototype stage, but could soon be commercially available. “The FDA does not approve a material or a process, per se,” says Christensen. “Rather, it approves each individual product. To my knowledge, the FDA has not yet cleared an EBM-made implant yet. The process is relatively new and there are only about 50 machines in the world. But still, EBM has made the design world a lot bigger. structure.”

The math behind EBM

There are basically two ways to design a EBM porous part, says Medical Modeling President Andy Christensen. One is to create a CAD file of a porous shape, something that looks like a tree, say, with branches. The trouble is, traditional CAD does not handle lattices well. Structures quickly get large and unruly, and file sizes can easily go to GBytes

The other method is to take the shape that needs to be porous and design it as a solid object. Then, to produce the lattice structure, tell the machine that instead of melting every layer the same, melt a little here, and sinter a little there.

“We use STL files of shapes and fill them with repeating patterns using a voxelization procedure in our software,” says Denis Cormier, associate professor of Industrial and Systems Engineering at North Carolina State University. “The basic idea is that instead of designing the whole structure, just start with the single-unit cell to be repeated. The point is to create a 3D array using that one element. The algorithm in our code makes copies of the element and checks whether or not it's inside or outside the solid, or if it is straddling the surface. Cells completely inside the part are deemed fine. Those outside the part get tossed. Cells trapped in the surface get trimmed.”

That part of the math is relatively easy, says Cormier. “But the math for bending shapes such as for skull implants is a different story. We have also developed algorithms to deform the lattices. Also, imagine a hip implant with a customized stiffness. Basically, there is a big cube full of little cubes, each of which represents a lattice cell. However, a region of the model may not need as much stiffness, and so requires a larger mesh structure. But corners of the large cells must match with the corners of the small cubes, or the fabricated part will fail. Again, we implement an algorithm that ensures different size cell nodes mate correctly.”