An early pioneer in the field of additive manufacturing (AM), the story of Carl R. Deckard, PhD, ME, is an example of the University of Texas motto: “What Starts Here Changes the World.”
Deckard invented and developed selective laser sintering (SLS) while an undergraduate and later a graduate student in the 1980s at the University of Texas in Austin. SLS technology melts pre-heated powders together in very thin layers using a scanning energy (laser) beam to build up parts, layer by layer, until the final parts are completed and are separated from the unmelted powder. SLS was one of four competing technologies being developed independently of each other during the mid to late 1980s. Fused deposition modeling (FDM) is an extrusion-based process. Stereolithography (SLA) is a liquid photopolymer-based process. Laminated object manufacturing (LOM) is a sheet-based process. SLS, SLA, and FDM have become the foundation of the rapidly growing AM industry. These technologies were enabled by the availability of 3-D CAD. The availability of these technologies then accelerated the adoption of 3-D CAD.
Deckard, perhaps, as much as any individual, was destined to be the developer of an innovative and transformational technology. “As far back as I can remember, I wanted to be a scientist. That all changed when my father took me to the Henry Ford Museum when I was about eight years old,” Deckard recalls. “I decided I wanted to be an inventor. My fantasies were all about inventions, and I was always looking for things that needed to be invented. That’s not an easy thing to do, because you’re looking for something that doesn’t exist yet. You’re looking for a hole that needs to be filled.”
Deckard pursued his dream of becoming an inventor through his public school years by studying the lives of the great inventors, studying the patent process, and working on a number of inventions. When Deckard was getting ready to leave high school, he chose mechanical engineering as his major because, he felt, it was the closest thing to majoring in invention.
Deckard believes wholeheartedly that chance does indeed favor the prepared and open mind. In his case, he credits his experiences working a summer job in a manufacturing plant the summer after his freshman year for exposing him to real-life manufacturing challenges. The timing couldn’t have been better. In the 1980s, computers and 3-D CAD were about to change the way parts and the tooling that produces them are designed, and if Deckard would have his way, be manufactured.
“The hype in the early days of 3-D CAD was that you could go from a 3-D CAD model to a CNC program in some sort of automatic way,” Deckard says. “That really didn’t happen until some years later, but that’s what all the buzz was about.” From that kernel of thought, Deckard, who was 20 years old at the time, looked around his summer job in a manufacturing plant and its products for the oil patch, including centrifugal pumps, hammer drills, bits for hammer drills, check valves and the like and thought: “You could go from a computer model to a machining program, but the raw part starts out as a casting, and the shape for the casting comes from a pattern that is made by hand from a 2-D drawing made by a skilled craftsman. I thought if I could make casting patterns from a computer model, people would pay money for that capability.”
Deckard developed the concept that became SLS over the next two-and-a-half years. The first thing he realized was that the process had to be additive. “With a subtractive process there are too many geometric constraints,” he says. “If you machine in one area it affects another area, because you have to have tool access. Next, I decided the process had to be an incremental additive process in a regular sequence. Next, I decided that I wanted to do real 3-D parts, not just 2.5-D casting patterns. This required being able to do overhangs. This required a support phase.” On his way back from a job interview during his senior year, he came up with the idea of building the parts from powder. “My first concept was to use a binary powder approach. I was going to lay down one powder where I wanted the part to be and a different powder where I didn’t want the part to be, like a layered sand painting.”
After some experiments with sugar and salt at home, it didn’t take long for him to decide that getting the detail needed with the two-powder approach would be very difficult. “That’s when I decided to lay down one powder and hit it with a directed energy beam, and that I would control the energy beam with the computer,” Deckard says. “I didn’t know how I was going to control the energy beam with the computer, but I knew it could be done. I chose a thermal approach [melting or sintering] over a chemical approach because it is applicable to a wider range of materials.”
This was Deckard’s Eureka Moment.
“At this point I realized that I could make this work, and that this was not just a thought experiment. This was something really worth putting effort into,” he recalls. “I had conceptualized many inventions by that point in my life, but never had gotten to the point where I said I have to build this. By that time I had been accepted to graduate school. I realized that this would make a great graduate research project. That is when my vision for my future really came together.”
Deckard approached Dr. Joe Beaman, a UT professor he had as an undergraduate, with his proposal: “to build three-dimensional objects from a computer model by building them in layers of powder by melting the powder together with a directed energy beam.” Applications would be first prototypes, models, tooling, and low-production runs of high-value parts. He identified three major challenges: curling caused by thermal stresses; edge definition; and powder deposition. Beaman became his advisor and gave him the go-ahead to build it.
The timing could not have been better for acquiring the necessary equipment. The Mechanical Engineering Department had just moved into a new building, and there was an equipment budget available. Deckard was told to spec-out the equipment that he would need for the project. “I spec’d out the equipment and gave it a dollar value. I had spec’d the system with a 2-W laser scanning at 30 frames per second,” Deckard says. “Intuitively, the laser felt too small, but after looking back over my notebooks many times, I couldn’t find an error in the calculations. I knew something didn’t look quite right and I kept looking back at my calculations. Then I found the error. It wasn’t the calculations. It was one of the physical constants used in the calculation. I had copied the heat of fusion of iron, incorrectly from one page to the next and had lost three orders of magnitude. When I went back and re-spec’d the system with a 100-W YAG laser and galvanometers, I was still within budget. Fortunately, the dollar value was right.”
While waiting for the equipment to be delivered, Deckard developed a way to modulate the laser to make a 3-D shape. He did it using a Commodore 64. “I had to fit my program and all my data into 4K bites of memory,” Deckard notes. “My program was hand-assembled and was 153 bytes long. It outputted a value through the user port and a D/A converter that I compared to the position feedback signal from the galvanometer. When the two values crossed, it triggered an interrupt that would then run the program and output the next piece of data.”
The original setup to prove out the concept was located in a small darkroom and involved a box on the floor into which Deckard sprinkled plastic powder with a giant salt shaker and a laser and scanner sitting on the counter top above. “I sprinkled in some powder while it scanned, and started out making crude parts,” he says. “This was the chips and chunks stage of the project. Then I started moving the box on the floor to give the part a more complex shape. Then I implemented the laser control system I had developed. Now I was making parts with a shape that came from the computer, but the edge definition still was crude. That problem was mostly solved once I got the particle size and scan spacing right. Now I had a process that could make a part with a recognizable shape with edge definition and enough strength that it didn’t fall apart in your hand. I made a square part with a square blind hole in it. I showed it to my faculty advisor, and he told me to write it up for my master’s degree.” Deckard received his MSME in May 1986.
At this time, Deckard asked the University of Texas to file for a patent. “I approached the patent committee with a part that I had made. It was a cube within a cube, a hollow cube with a couple of holes in it to get the powder out and another little cube captive inside it,” Deckard recalls. “The committee filed the necessary patents. The first patents were filed in October 1986, and became the source of royalties over the years for me, the University, and UT colleagues who participated in development of SLS. There were a number of follow-on patents. Some of them turned out to be significant in the later patent enforcement battles, and some didn’t. The most significant was the heat patent.”
During the same time period, Deckard searched for a path to commercialization. After approaching several large companies without success, he began looking for partners to do a start-up. On the same day as the patent was filed, Deckard met with Paul McClure and Harold Blair who were looking for a UT technology to license and commercialize. “We formed a partnership and negotiated a license from UT during 1987,” he says. “1988 was our fund-raising year. The license required us to raise $300,000. We talked to a number of large companies and ended up teaming up with B.F. Goodrich. At the time, the press was calling this area desktop manufacturing because of the obvious analogy to desktop printing, so we adopted the initials DTM as the company name.”
1987 was also the year that Deckard solved the two other key technical problems: leveling and curling. It took Deckard nine months to come up with a solution for leveling the powder. “I tried a number of approaches, from vibration to electrostatic to pneumatic. They all worked until the laser was turned on,” he observes. “The laser would make the material stick together so that you couldn’t level it. Finally, I tried the counter-rotating roller and it worked better than any of the other approaches could have ever worked, if they had worked.
With the leveling issue addressed, curling came to the top of the list. “When you deposit hot material on a colder part you get thermal stresses that tend to cause the part to curl,” Deckard states. “I tried a number of ways of overcoming curling, primarily by holding the part down, but with no success. After a particularly unsuccessful experiment that had required two weeks of preparation, I decided to do a quick-and-dirty experiment. I borrowed a heat gun and waved it over the surface of the part while scanning. I already had a system that sucked air down through the part bed in attempt to hold the part down. The hot air passing around the part and down through the part bed caused the part to curl in the opposite direction. The part was more solid than anything else I had made, and I knew that preheating was the answer.
“The experimental setup evolved over a few months until it was clear that I needed to rebuild the system with a closed process chamber. Two weeks before the most important demonstration to date, I got to the point where I had to tear down the old experimental setup in order to keep moving forward with the new experimental setup. The first successful run of the new setup was during that demonstration.”
During 1988, Deckard did experiments using the new setup and completed his PhD in December of that year. Also during 1988, Deckard worked with graduate student Paul Forderhase to do the top level design of the first SLS machine to be built from the ground up. “We started with a set of requirements and produced a design that met those requirements,” he says. “This design was called Godzilla because it ended up being so big, heavy, and expensive. We then backed off of the requirements and came up with a more modest design that we called Bambi. Bambi ended up being 10′ [3.1-m] long, almost 7′ [2.1-m] tall and weighed several thousand pounds.”
During 1989, Deckard and Forderhase along with a number of other students used Bambi to further develop and demonstrate the SLS process. Among the dignitaries that came to see Bambi were University of Texas System Chancellor Hans Mark and US Senator Phil Graham. Also during 1989, Deckard worked with a small group funded by Goodrich to develop the first commercial SLS system, the SLS125. The first SLS125 was built that year at Gem City Engineering in Dayton, Ohio, and demonstrated at AutoFact ’89.
Deckard continued to develop SLS machines, materials, and process at DTM through the launch of the Sinterstation 2000 in 1993. Deckard then went on to a 3½-year stint as an engineering professor at Clemson University before coming back to Austin to work on the Deckard Engine, a hybrid rotary reciprocating gasoline engine with one major moving part. The Deckard Engine was aimed at supplanting emission-emitting two-stroke engines for small hand-held applications.
Today, Deckard continues to bring his analytic problem-solving abilities to bear on his selected field. He and a partner, Jim Mikulac, have started a materials consulting business, Structured Polymers LLC. “We’re doing material development work, and positioning ourselves to be part of the second wave in innovation that I see coming in additive manufacturing as key patents run out.”
Deckard concluded the interview by saying: “I was able to pursue the ambitious, low probability goal of creating and commercializing a new manufacturing process at the same time that I was pursuing the much more realistic and deterministic goal of earning a PhD. I hope my story will inspire young technically minded students who are struggling through their public-school years surrounded by a culture that values athletic success, physical appearance, and family wealth over intellectual curiosity.” ME
Connect With Us
