Chemical laboratories did not change a lot in the last century. In contrast to biological research, most of the time consuming work is still done manually and relatively large quantities have to be used. The 3 main challenges for a modernization are: Scaling down, automation and integration of sensor technology.
Our project addresses all 3 challenges, solves a real life problem of every chemist and was successfully tested on realistic chemical compounds. We chose one of the most time consuming and labor intensive work step of chemical synthesis: Phase separation has to be monitored by eye, the completion of the process is evaluated by means of intuition and the desired phase has to be removed manually.
How we addressed the challenge
First, we scaled down the reaction volume from liters to milliliters. We are currently working with standard eppendorf tubes (1.5 ml) which are common for automated lab solutions in biology. Then we mounted a high resolution camera on a translation stage - both of which are controlled by a raspberry pi which automatically transfers close-up images of the reaction tubes to a computer. The images are run through a customized pixel intensity algorithm and a profile along the length of the tube is displayed together with the exact position of the phase boundary. In a second step the phase boundary position for each reaction tube serves as a reference point for the OpenTron laboratory robot. The desired phase is automatically removed and transferred to a new reaction vessel for further (automated) processing.
Problems we ran into
Control of the translation stage for the camera movement was a time consuming issue. Full automation of the entire work-flow could not be realized due to limited access to the OpenTron.
Accomplishments that we're proud of
All steps of the automated phase boundary detection and separation are working. The image analysis yields a clear graph and the calculated position of the boundary is sufficiently precise for all tested compositions (liquid/liquid, liquid/solid, liquid/emulsion).
What's next for TwoFace
The phase detection can easily be adapted for turbidity measurement to increase precision when working with same-color liquids. Spectral analysis during image acquisition would also be a nice and relatively easy to realize add-on. Adapting the design to work with multiple-well plates.