It feels next would be trying isolate the component that make CO2. Try to use smaller sample. Put them under microscope, etc.

I might be dumb, but I figured they would just continuously subdivide the sample until they isolated it. Maybe CO2 sensors wont be sensitive enough?