The stepping stones of analysis and a cry for help (· · · – – – · · ·)

I’ve slowly read every major book on distillation and probably 150 journal articles in the last few years. The punch line is that just like fine wine was made possible by the laboratory (just like the kings of Napa, Mike Grgich and Warren Winierski, would tell you), craft spirits will also be a product of the lab, and not many new distilleries are running labs. This unfortunately means that only the big guys are craft but that doesn’t have to be the case.

My idea has been to slowly explore spirits analysis in little steps and build a valuable skill set as well as share everything to elevate the new distillery movement. One thing I’m seeing is that for many small distilleries to stay open in saturated markets, they will have to rely on their tasting rooms for revenue. In a tasting room, it will soon be apparent that a portfolio of three products probably won’t cut it. Tasting rooms will need elaborate cocktail programs and legally will have to fabricate small batches of products like orange liqueurs to show off the core products. This orange liqueur among many others will never be sold out the back door (saturated market) but rather just be used in house and possibly be sold out the front door because many tasting rooms can operate like liquor stores.

So, for an orange liqueur to be made in small batches, some competitor analysis has to be performed on the likes of cointreau and grand marnier such as sugar content, alcohol content (before sugaring), and the weight of the aroma. The liqueur will then be assembled in a robust, paint by numbers process where a great product is made without tons of man hours or tying up people for tasting panels. Years ago I figured out elegant ways to measure alcohol and sugar content (via hydrometry without sacrificing a sample) but what about the weight of the aroma? and how the hell do you standardize your charge of peels when the oil content varies so much? That is the skill set we need to be returned to common knowledge!

So far, the answer to finding the weight of orange aroma in an orange liqueur is liquid-liquid extraction using intense, hazardous, organic solvents like iso-octane, hexane, and dichloromethane. The exotic solvents require a fisher scientific account and clearance to ship them so not everyone can play with this stuff. They also require lots of reading and safety training to handle properly (though it is nothing too extreme). The same solvents can be used again for measuring oil yield of botanicals using a piece of glassware called a soxhlet extractor which is a priceless skill for a distillery lab.

I’ve been exploring this and spending considerable money in the hopes that it will launch a lot of ships. What I don’t know and need help with is the full potential of liquid-liquid extraction. You see it in a lot of spirits research papers because it is used for sample preparation for GC-MS and it is described in a lot of the modern advanced texts but not in any specific detail or with vouched for Modernist Cuisine style best bets which is what we all need.

First a tiny overview of liquid-liquid extraction. Powerful orangic solvents with very low boiling points that are immiscible in ethanol* and water* are mixed with a spirit. The organic solvents will mix just like oil and water but their solvent power will pull congeners out of the ethanol and water. Eventually the organic solvents can be separated with a separatory funnel. Their boiling points are so drastically different from the congeners dissolved in them that vacuum distillation (solvent recovery) or sometimes just putting a fan on them (expensive sacrifice!) is enough to separate and isolate the congeners. The asterisk is for organic solvents that form an azeotrope and suck up tiny amounts of water or ethanol, but simple methods can be used to “dry” them.

What I just described is the batch process and its pretty easy but has some limitations. So far little globules of emulsion (which I need to identify but are probably aromatic oils) cling to the sides of the glass and need to be rinsed out with more solvent (expensive!). Smaller size glassware tailored for the batch might minimize this cling via reducing the surface area available to cling. Some descriptions of the batch process (not in a spirits context) use multiple iterations to extract as much as possible which is not a big deal if you can recover your solvents in a vacuum still.

Most spirits chemists isolate congeners to prepare samples for GC/MS analysis. Once the solvent is evaporated they don’t need to perfectly remove every bit from the glass and can work with batch sizes as small as 30 mL. My idea is that if the batch size is scaled up dramatically to a liter, what is extracted from say an orange liqueur can be weighed with a jewelers scale that does 1/100 of a gram. Just knowing the weight of the dissolved orange essence will get you squarely in the ball park. The problem is that no researchers use my scaled up 3rd world method because they have PhD’s, big budgets, and are in the GC/MS era. What I have going for my hunch is that I am the guy that figured out you can even measure carbonation with a kitchen scale!

Long ago I read a paper from the 1970’s where scientists were pioneering liquid-liquid extraction sample preparation for the study of gin. They were using Freon-11 (which is now banned for its effects on the ozone). These guys were concerned that their sample was representative of the gin so once they extracted all the gin congeners, they re-dissolved them in vodka and drank it! and then compared it to the original gin! If you have a reliable vacuum still, what you extract from the point of view of a chemist is drinkable! (but I’d use extreme caution, though most of the solvents smell like rubber cement which makes incomplete separations easier to spot).

A problem I’ve been having is I don’t think my samples are representative. I’m leaving way too much aroma behind in my gin test material. I took my organic solvent blend from a recent study on a gin (which might be a red herring to support a patent that should be bogus) and I think the process might call into question their results. Or I’m just new at this and am missing something. I’m using a 1:1 blend of Hexane and Dichloromethane and using the batch process with just one iteration like in the study, yet a significant amount of juniper aroma lingers in the ethanol-water of the gin.

Another options is to use a continuous liquid-liquid extractor and some are described in modern spirits texts but not with any real guidelines, best bets, or testimonials. I’d love to try one but the glassware starts to hit $500 to $1000 dollars very quickly (with no testimonial they work in this context!). Continuous extractors run small amounts of solvent through the spirit in a loop where at one point in the loop the extracted congeners are separated from the solvent by evaporation and the clean solvent is run back through the loop. These rigs take up not insignificant counter space and run sometimes for sixteen hours. I’m not afraid to pay for one but finding counter space and sixteen hours is a big challenge. But keep in mind this analysis only has to be performed once and then can be shared by the distilling community!

Once we know there is X mg of dissolved orange oil in each liter of quality, intuitive-to-use orange liqueur, we can elaborate the process we learned slightly. We can use the next extraction tool which is the soxhlet extractor. We place 100 grams or so of orange peel into our extractor and start to draw the essential oil out of it. This type of extractor also runs in a continuous loop for numerous hours unattended and there are lots of Youtube videos that show them in action. Once the solvent is separated by vacuum distillation we will know how much oil is in every 100 grams of peels which will change often dramatically with each batch of peels. If we just weigh the peels and throw them in the still, the oil yield will be wacky and the product will be inconsistent, but if we scale the botanical charge for oil yield, we will have a much better standard and it will open a lot of doors to taking on new botanical sources while hitting a higher standard of product consistency.

Two more important things. A steam distillation rig, designed to produce essential oils, can also ball park the oil yield but it is an inferior method (but we are concerned with affordable stepping stones! so do explore, especially if you have no access to exotic solvents!). The most important thing is that this procedure of measuring oil yield can scale up to all the botanicals that come through the distillery. The full botanical charge of everything from gin to absinthe to amaro to bitters can be standardized for increased product consistency and this is the skill set. What we need are best bets, testimonials, and what-ifs answers, and Youtube videos. Many distillers working with botanicals are not standardizing their charges because of a false sense of consistency. Yes, supply chain management for botanicals is staggeringly more advanced than it was decades ago, but if you want to get off the beaten path, and forage, or grow your own, or seek terroir, you need at least this very basic laboratory analysis.

1 thought on “The stepping stones of analysis and a cry for help (· · · – – – · · ·)

  1. Above the plait point at about 70% ethanol, limonene, water, and ethanol are miscible.
    http://www.sciencedirect.com/science/article/pii/S0378381204002882

    Acetone behaves similarly.
    http://www.sciencedirect.com/science/article/pii/S0021961410002077

    So extract (2 or 3 x probably better) peels with Everclear.

    Use density or refractive index of extract to estimate oil content.
    http://web.mit.edu/5.310/www/Essential_oils.pdf

    A low temperature version of Rezzoug and Louka’s DIC process using N2O or CO2 in a cream whipper/rapid infusion approach might be worth considering. Not sure what the vacuum rating of a cream whipper might be.
    http://www.sciencedirect.com/science/article/pii/S1466856409000587

    If you built a pressure rig, gas expanded liquid (ethanol) is probably a more effective solvent than ethanol.
    http://pubs.acs.org/doi/abs/10.1021/bk-2009-1006.ch001

    You probably wouldn’t want to use N2O for GXL, since it is an oxidizer.
    http://pubs.acs.org/doi/abs/10.1021/ac00069a028?journalCode=ancham

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

search previous next tag category expand menu location phone mail time cart zoom edit close