My experiment began last month. There were several stalled efforts because I couldn't bring myself to sacrifice my freshly brewed espresso to science. Eventually I settled on using Starbucks to prepare the coffee, which does seem to offer perhaps an edge in reproducibility.
My aspiration was to use a minimal amount of equipment. In terms of cost:
300ml borosilicate beaker ($5.95+tax)
Digital Scale 300 gram Max - 0.01 gram precision ($10.95+tax)
Step 1: Solid-Liquid Extraction
This is the easy step, fancy nomenclature for brewing coffee as usual. The beaker fits easily under my espresso machine:
I weighed the beaker empty, then weighed it again after adding ~60ml (two shots) of espresso. (Picture shows my home-brewed double shot just for illustration. A "store bought" Starbucks doppio was used in actual fact.)
Step 2: Dehydration of Extract
This is the tedious step, removing the water from the brewed espresso. I did this as gradually as possible with incremental heating in my 1200w microwave, a few seconds at a time on high followed by blowing over the beaker to remove the water vapor.
As the volume of liquid reduces it is necessary to limit the heating time to mitigate splashes from the boiling of confined solution. The picture shows that even with a couple of dozen repetitions of just a few seconds, splashing within the beaker occurred. Indeed a few drops escaped the beaker and landed on the microwave rotating platter, so a better method of dehydration (freeze drying) would be an attractive alternative. Once the espresso residue was fully dried, I weighed the beaker + dried brown residue.
Step 3: Sublimation of Caffeine
Once the removal of water appears complete, we take the final step of removing the caffeine that remains. The key physical chemistry property of caffeine is that its boiling point (352 °F at atmospheric pressure) is lower than its melting point (441 °F or higher).
I preheated my oven (on bake) to 355 °F and put the beaker with the dried residue of espresso on an aluminum pie pan in the oven (center rack). After a matter of seconds a light gray vapor formed in the beaker. I opened the oven door to allow this vapor to blow away. After repeating this twice the light gray vapor stopped forming, and I removed the beaker to cool for a final weighing.
Conclusion and Notes
The total caffeine present is measured by the difference of the last two weighings, as the weight of caffeine driven out by sublimation. This figure is 0.19 grams or 190 milligrams plus or minus the scale's precision of 0.01 grams (10 milligrams). This agrees roughly with the figure of about 150mg cited by CaffeineInformer for caffeine in a Starbucks doppio espresso.
We mention some sources of experimental error besides the 0.01g precision of the affordable Pocket Scale™ pictured below:
One source of error, which would tend to reduce both the final residue and the sublimated caffeine, is splatter from the drying process noted earlier. Another is possible moisture retained after Step 2, which would tend to exaggerate the weight of caffeine by confounding it with water weight. Also the weight of caffeine we want is small in relation to the weight of the beaker, which presents a subtractive cancellation issue.
In an effort to minimize the effect of these errors, I chose a "single vessel" procedure, so that the relatively small size of the experimental sample is not unduly prejudiced by transfers. I plan to repeat the experiment and will update if any discrepancies are observed. In particular after thorough washing the weight of the beaker empty came back to 112.49g.