While our desiccant was being produced and shipped, we began testing our condensation system so we would be prepared to integrate the material. We needed to determine what form of Peltier plate apparatus would condense the most water out of the air, and funnel it into a container.
To conduct the experiments, we constructed an apparatus that contained a Peltier plate mounted on a larger metal plate. We divided the metal plate into nine equal sections, and mounted a different shape on the underside of each section, with two blank sections as controls. We then inserted a remote weather station inside, to track internal temperature and humidity, before insulating the entire device.
Our first step was to make sure we could create a big enough temperature differential with the Peltier plate to condense water out of the air. We set up our apparatus without the collection vessels and tracked the change in temperature and humidity inside our apparatus. It proved successful, as when we removed the insulation, we found a puddle of water on the floor of the device.

The MOF will allow us to concentrate moisture and raise the humidity in our device higher, which will further reduce the temperature differential we have to create to collect water.

In the next experiment, we set a ceramic collection vessel for each section under the plate. We wired up the Peltier, and ran the device for 24 hours, taking internal temperature and humidity readings regularly. After the 24 hours, we weighed the collection dishes to determine the amount of water they collected. We relied on the ambient humidity (averaging about 38% RH) to provide the water, and treated anything less than 0.5 grams as within the margin of error.

We found that the central heatsink rendered the most liquid (3.1 grams), followed by the whistle-like hollow tubes (0.57 g), vertical solid rods (0.5 g), and a strainer-like mesh (0.5 g), with the device collecting a total of 5.87 g of water. Further analysis revealed that the heatsink may have collected more water, not from its geometry, but rather because it was directly attached to the Peltier plate, while the other sections were attached to the metal plate, which has a much larger mass yet less thermal contact with the Peltier.
From these results, we determined that whatever shape we use will need to be directly connected to the Peltier plate, with full thermal contact. This could mean adding additional Peltier plates to chill the entire plate evenly. Further testing will be needed to narrow down between the four best geometries, however.