In this first attempt I tried to extract the aromatic principles of roughly 0,5 kg Bourbon vanilla beans (from Madagascar) by hydrocarbon percolation. For this purpose I used n-hexane Scharlau 95% (120 euros 5 L canister). The extraction product is called an “oleoresin” (resinoid or concrète) because it carries a volatile fraction and odorless waxes and resins (I suggest you to read S. Arctander’s monograph on this subject: Vanilla oleoresin).
I bought extraction grade vanilla beans: water content is kept low.
I was quite satisfied with the overall “standard” odor quality of the starting material: “senza infamia e senza lode” (nor too bad nor too good). I only perceived a faint fermented type, volatile off-note when unpacking the beans. Apparently it was distilled off when evaporating the solvent, since it is not detectable in the oleoresin anymore (when disassembling the Soxhlet extractor after the process I was able to smell an acetic acid-like note on the top of the Dimroth condenser, while the recovered hexane did not carry such note).
In this first photo you can see the Soxhlet type apparatus I’ve set up for the extraction process:
Here’s a video of the soxhlet extraction process:
Comminuted Bourbon vanilla beans before extraction:
Comparison between exhausted (left) and fresh (right) vanilla beans. From this photo you can guess the extractive power of hexane: the vanilla on the left is drier and paler in color. It still carries some aromatic principles (not hydrocarbon-soluble material I presume).
A Soxhlet apparatus is designed in order to recirculate the solvent throughout the botanical material. The temperature at which the extract is exposed is the boiling temp. of the solvent used. At standard pressure n-hexane boils at 68° C (relatively low). The raw material is not directly in contact with the boiling solvent, thus preserving it a little bit from excessive heating.
In this first attempt I let the solvent recirculate 6 times (2 hours refluxing). The first run is, not surprisingly, the most charged with volatile materials, waxes and pigments (it is the darkest in color). The fourth is almost colorless.
I let the extraction liquor stand for a day and then filtered under gentle vacuum. This process results in a small loss of solvent. n-hexane is very volatile and vapors arise easily from the filter paper. It is the most troublesome phase of the process, when you are in strict contact with the solvent. I carried out this operation on the outside avoiding breathing too much vapors.
I then evaporated the solvent under vacuum (pressure between 150-300 millibar) at ~30° C. It is really tricky to evaporate all the solvent. The last traces are the most difficult to eliminate: the temperature increases and there is the risk of spoiling the extract.
A video I recorded of the evaporation process:
A rotatory evaporator would give better results: solvent recovery would be faster and more efficient.
I used a water vacuum pump: I can reach a pressure as low as 8 millibar, but in normal working conditions (heat, arising vapors) it is stable at around 150-200 millibar. A membrane pump would be more reliable.
An ice trap would also improve solvent recovery and avoid vapor leakage in the air (I used a Dimroth condenser and recovery was very good, I registered only minor losses through the vacuum pump).
The yield was quite good: 31 g oleoresin from 471 g of crude vanilla beans (6.7%). I was able to recover most of the solvent (major losses were due to filtration and adhering solvent on exhausted vanilla).
Vanilla oleoresin is a viscous, dark brown liquid not completely alcohol-soluble. It possesses a true-to-nature, rich and creamy vanilla aroma, with woody and licorice undertones and a tobacco nuance. It is not as powerful as a vanilla absolute, since it still contains odorless waxes and non-volatile material, but it possesses good tenacity.
I don’t appreciate headnotes from this oleoresin very much: they are quite smoky-bitter, overly sweet to my nose. It may be a factor of concentration (too high to fully appreciate the extract) or there could be solvent traces (a gassy nuance? Not quite sure). I assume that during solvent recovery the temperature increased too much in the last minutes (bitter, smoky, burnt odor detected?).
An hydrocarbon extraction followed by an hydro-alcoholic one should give better results: not complete alcohol solubility, but a much more complex and rich aroma quality.
Tightly controlling the temperature is essential to achieve satisfactory odor quality.