3D Printed Microfluidics

3D printing is a critical enabling technology for many cutting-edge applications, particularly ones that require a very large number of components to be assembled in large arrays, whose physical size must span to macro-scale. In particular in artificial muscles, 3D printing offers the ability to upscale microcapacitors to macro-arrays that produce artificial muscle fibers and fiber bundles. Any other fabrication technique, such as layer-by-layer fabrication plus subsequent assembly, would either not work technically or be prohibitively expensive, or both. Hence, there is a huge boon to engaging 3D printing in artificial muscles.

This great promise comes with great challenges as well. Artificial muscles require wiring at the microscale, so that each individual microcapacitor is properly connected and actuated. The wiring requires a conductive material, while the dielectric part of the capacitor requires non-conductive material. If both are to be printed at the same time in the same device, the 3D printer must have the capability to print both. However, there are no such printers yet. The difficulty is that conductive materials are typically metals, which can be 3D printed by sintering, but the required temperatures would melt and burn the nonconductive resin. Until recently, this was an insurmountable obstacle.

We found a way to solve this problem by employing 3D printing to produce microfluidic channels, which can later be filled with conductive material to produce wiring. This takes a big difficult problem and divides it into two smaller problems: 3D printing of embedded microfluidics and subsequent deposition of wiring inside those microfluidic channels.

The 3D printing of embedded microfluidic channels in itself is very challenging. Both of the two dominant 3D printing technologies, PolyJet and SLA, have trouble

With PolyJet, the sacrificial material is a wax-like substance that is typically removed by a combination of heating, sonication, abrasion, and chemical treatment. To work properly, this approach requires enough fluidic access to the wax. For positive features, namely features made from resin, this is usually easy enough, as much of the volume is sacrificial material that is generally accessible from the outer surfaces of the overall build. However, in microfluidics, the desired features are negative and embedded deep within the build. Any access is typically only available through a few small input/output ports, so the wax is particularly difficult to access. 
 

The above figures show an SLA 3D-printed demonstrational build in the shape of a rook. The inner features are relatively easily accessible from the outside, allowing for efficacious removal of the liquid sacrificial material. In contrast, the right-most picture shows a microscopy image of microfluidic channels printed in wax inside a resin device. Note that the only access for these long but narrow channels is through a port at the right end of the picture. 

We developed a procedure to solve this problem. Briefly, it involves heating the devices in an oven over a few hours, followed by cooling down and flushing a solution of NaOH through the microchannels. The heating makes the wax outgas a significant amount of mass, which opens cavities that allow the subsequent flushing.

The above figure shows microscopy images of 3D printed microfluidic channels at different stages through the clearing procedure. The leftmost photo shows a device after printing but before heating. The mid-left photo shows a device after heating and cooling. The mid-right photo shows a device early in the flushing cycle. The rightmost photo shows a device after a few cycles of flushing. The opened volume looks darker, because the microchannel cross-section is convex and the refractive index of air is lower than those of resin and water. 

The clearing procedure was systematically studied and optimized. Quantitative analysis was based on integrating the cleared area and dividing it by the total area of the same channel on the same photo, to produce a clearing percentage. Analyzing the same microchannel multiple times generated statistics for the mean and standard deviation of that microchannel, so all the data from the same microchannel could be succinctly represented by a single datapoint and associated error bars. 

On the other hand, it was also useful to measure the relationship between microchannel size and clearing percentage. So, the data was assembled and represented as a single plot showing clearance percentage as a function of measure microchannel width. The results are presented below

In the above figure, the left image is a microscopy photo of a microchannel. It shows the channel is mostly cleared as it appears darker than the surrounding resin, due to the convex profile of the microchannel cross-section and the refractive index of air being lower than the refractive index of the resin. The right image on the above figure shows the plot of clearance percentage vs measured microchannel width. Overall, clearance is ~75% for the most channel sizes. The smallest channel that was successfully cleared was 200 microns across.