Electrical Wiring for Biofuel Cells through T-channel Architecture

A functional 3D-printed biofuel cell must satisfy multiple requirements: it must be printed in 3 dimensions with a large degree of arraying and large-scale integration; it must have microchannels to contain the bacteria; it must have electrical wiring within microscopic distances from each bacterium; the electrical wiring must be biocompatible and in-built within the 3D printed volume. These constraints are very challenging to satisfy, particularly in their combination.

We have already resolved the 3D printing challenge and the sacrificial material clearance challenge. What remains to do is biocompatible wiring inbuilt throughout the volume of the 3D-printed cell.

Ideally, such wiring should be produced by hybrid 3D-printing wherein both the non-conductive bulk material and a conductive material are 3D-printed together seamlessly throughout the same device. Unfortunately, such printing capability does not yet exist. Typically, 3D printers print non-conductive resin and sacrificial material. The most advanced printers, e.g. our Stratasys Object500 PolyJet, have the ability to print three materials at the same time in the same build: two different resins and a sacrificial material. However, none of these is conductive.

Currently, there are just two options for printing conductive materials. The first one is printing metal by sintering of metal dust. This is done by a focused laser beam heating up the metal dust until it partially melts and essentially welds together. This requires temperatures that are too high for resin to avoid burning. So, sintering and resin printing currently appear to be fundamentally incompatible.

Second option for printing conductive materials is to spike the resin precursor with conductive microparticles, e.g. silver, then print as usual. This works reasonably well when just the spiked resin is printed, but current printers cannot do this with both spiked and unspiked resin at the same time. Furthermore, viscosity and particle size seriously limit the achievable resolution in such prints. Finally, while spiked resin constructs can in principle be printed on top of already printed resin substrate, the result is essentially a 2D circuit grafted onto a 2D substrate, which loses most of the advantages of 3D printing in terms of packaging and integration.

Our approach circumvents these problems as it steps away from direct printing of conductive material but instead chooses to use microchannels as wiring by building conductive material inside the channels by fluidic means. The tradeoff in difficulty is that our approach would require this deposition to reach all parts of the fluidic matrix, so that the emitted electrons from all bacteria in the biofuel cell can be collected.
 

Specifically, we developed a technique based on self-organization by surface tension. The technique is presented schematically in the figure above. Briefly, the microchannel cross-section profile is one of a main channel and smaller side channels, called flanges (see A). Hydrophobic conducting fluid, such as uncured resin spiked with conductive microparticles, is fed into the channel (see B). it fells the channel easily, because the channel walls are hydrophobic as well, and thus are well wetted by the fluid. In fact surface tension would make the filling particularly easy in the flanges. Then an aqueous solution (e.g. water or buffer) is fed into the channel (see C). Because the channel walls are hydrophobic, the water will have trouble entering the flanges, due to surface tension, while it will displace the fluid from the central section much more easily. Hence, the central section will be flushed and filled with water, while the flanges will retain the hydrophobic fluid. Then evaporating the water or flushing it slowly with air will produce the final configuration (see D), in which the hydrophobic conductive fluid is only present in the flanges.

This method is inherently self-organizational, because it requires no control from the outside and thus can be done throughout the fluidic network of a microfluidic device. What was described above was the sequence for a particular channel, but if the entire network is made of such flanged channels, the same effects would apply everywhere. 

In the initial embodiment of this technique, the channel had only one flange, giving it a distinct cross-section of a T-junction. That is why this method and technique were called “T-channels”.

To produce an experimental confirmation, we built chips with arrays of flanged microchannels, then used oil with Prussian Blue organic dye to represent the hydrophobic fluid. 

The microscope images above show different examples of the correct formation of the T-channel effect. In each pair of images, the top is the same channel filled with oil completely, while the bottom is the same channel after water flushing. The images show the retainment of the oil and dye in and around the flanges, while the central section is cleared. The numbers in the bottom right corners are the intended widths of the main channel / flanges, in micrometers.

While the effect works, applied pressure must be optimized and chosen well, so that it is high enough to flush the central section, but not high enough to flush the whole of the channel. If the pressure is too high, even surface tension would not be sufficient to keep the hydrophobic fluid inside the flanges. Examples of “failed” self-assembly are shown below.

As expected, excessive applied pressure would push out the oil, break it up in spherical bubbles, and flush it out. Dosimetry of the applied pressure would solve this issue, while the optimal pressure values will be a function of the channel dimensions and overall geometry of the fluidic network of the overall device.

In this experimental proof of principle for the self-organization effect, the hydrophobic fluid was not conductive yet, but the results still establish that the overall technique is valid and the self-assembly is achievable. What remains is to repeat the same experiments with a suitably designed conductive hydrophobic fluid that also is biocompatible, e.g. carbon microparticles carried by uncured resin.

The use of uncured resin is particularly attractive, since it will would allow the in-situ curing of the resin inside the flanges after the final step (see D) in the conceptual schematic above. This in principle should produce a robust system with in-built wiring extending to all parts of the fluidic network, which can then be used as the reservoir for the power-generating bacteria suspended in appropriate aqueous medium.