A new approach to multi-material additive manufacturing
The first semester of my second year at Columbia University, I began work with the Layered Assembly Team of Professor Hod Lipson’s Creative Machines Laboratory. Layered Assembly is a voxel-based additive manufacturing method in which pre-manufactured voxels serve as the feed stock for producing multi-material parts. The placement of voxels in the Layered Assembly process is carried out by electroadhesion—a type of astrictive prehension, which is an elegant, scalable, low-power, and solid-state gripping technique with many applications. The multi-material capability of Layered Assembly allows for the production of complex electromechanical parts, such as robots, smart phones, and other devices, which have previously been impossible to make with current additive manufacturing methods. My work with the Layered Assembly Team has recently been published in the Journal of Additive Manufacturing (link at bottom of page) and has provisional patent-pending status. The Layered Assembly process is demonstrated in further detail in the following video:
ELECTRODE SIMULATION AND DESIGN
My biggest individual contribution to the project is one of utmost importance in the Layered Assembly process: electrode simulation and design. The electrodes that make up the electroadhesive gripping mechanism are fully responsible for both picking up and placing voxels in the proper position, which is crucial for successful Layered Assembly. The performance of the electrode is highly dependent on its geometry. An ideal electrode geometry is one that maximizes electric field magnitude across the top of the voxel thus maximizing gripping force and simultaneously maintains a high degree of manufacturability, enabling short manufacturing time and low expense. In initial testing, the Layered Assembly team used only two different electrode geometries, the comb and spiral (shown above), which had proven their effectiveness in other research papers. My teammates and I wanted to know more about the electrical properties of these electrodes, rather than placing all of our trust in prior research. To achieve this goal, I built simulations of the original electrodes in COMSOL Multiphysics and measured their average electric fields. The COMSOL model electric field results reflected our gripping test results nearly perfectly, showing that the electric field average of the comb electrode was greater than that of the spiral when it out-performed the spiral in experiment. The reverse was also true.
After simulating the comb and spiral geometries, I was inspired to design some novel geometries in attempt to further maximize the gripping force of the electrodes. My guiding principle in these designs was to maximize the length of the channel between the high voltage and ground terminals in the electrode while maintaining a constant distance between them, because the channel creates a high electric field flux through the surface of the voxel, which results in a high average electric field magnitude. Several of my designs were unsuccessful, either decreasing the average electric field output of the electrode, or increasing it a negligible amount. Three of my designs, however, were highly successful, resulting in 20-35% increases in average electric field outputs. These designs are detailed below:
The Six-Mil Comb design is an adjusted version of the comb electrode from my team’s original testing. In my first attempts at electrode design, I had attempted to increase the length of the channel separating the two terminals without changing any other variables. But from my training in electromagnetic physics, I knew that the electric field magnitude between the two terminals would increase if the terminals were moved closer together. I then did some research regarding the limitations of circuit board manufacturing and found that I could move the electrodes as close as six mils (six thousandths of an inch) to one another without increasing cost or fabrication time. I found that I could reduce the width of the electrode terminals to six mils also, allowing more space for the channel between the terminals. The result of these ideas and the subsequent research was the Six-Mil Comb design which produced an average electric field magnitude of 1.66 MV/m, a nearly 30% increase over the best electrodes used in initial testing.
The design of the Hilbert Curve electrode is inspired by the third iteration of mathematician David Hilbert’s famous space-filling curve. The idea was proposed by a colleague of mine who had used a Hilbert Curve path for a laser food cooking pattern. This electrode produced an average electric field magnitude of 1.56 MV/m at its surface, a 20% increase in performance from Layered Assembly’s best performing original electrodes.
The Square-Spiral design is an original design of mine, inspired by the circular spiral electrode my team used in our initial testing. Looking at the circular spiral design, I noticed that it left much unused space in its 3x3 mm bounding square. My solution to this issue was to replace the circular nature of the spiral with a straight-line, inward-winding pattern, which would match the shape of the boundary and maximize the use of available space. This design has been the most successful electrode to date, producing an average surface electric field of 1.77 MV/m, a nearly 40% increase over the best electrodes in initial testing.
A crucial component of any additive manufacturing system is the gantry, the machine which controls the motion of the printing apparatus. My team and I had to tackle the unique challenge of designing a gantry that is optimized for Layered Assembly. This gantry needed to achieve two main goals: minimization of stepper motor vibration and precise delivery of voxels to electrodes. The first goal is necessary because excess vibration in the Layered Assembly system can cause electrodes to drop their intended voxels, and the second because misalignment of electrodes with respect to their intended voxels results in a significant decrease in gripping capability. We solved the vibration issue using small flexure mounts to attach the electrode array to the gantry. The flexures absorbed and dissipated vibrations from the gantry steppers, preventing them from propagating to the electrodes. The problem of alignment was solved with our alignment jig design. This jig has four depressions to hold voxels in place along with two guide rails which align the corner of the electrode board to its proper position, guaranteeing consistent and nearly perfect electrode to voxel alignment.