Measuring the Relative Effectiveness of Methods to Control Ethyl Carbamate in Craft Bourbon Production

Funded: $6000
Jon Brown

Castle and Key Distillery

Brad J Berron, PhD (Co-Investigator)

Current Occupation / Title:

Jon Brown – Quality Manager at Castle and Key Distillery

Dr. Brad J. Berron – Director of Research at the University of Kentucky’s Beam Institute for Kentucky Spirits; Associate Professor of Chemical Engineering at the University of Kentucky

Please provide how your background experience will provide the foundation for your research.

Jon Brown has been focused on the analysis of alcoholic beverages for over a decade. Under his guidance, Castle and Key Distillery has been monitoring EC levels and distillery operating parameters for over 700 production batches. This existing data set enabled a preliminary analysis of the biggest drivers for product EC levels, focused our analysis, and highlighted the need for further understanding of how craft scale distilleries can cost-effectively monitor and regulate EC levels.

Dr. Berron has taught several engineering courses at the University of Kentucky on the theory and design of distillation-based unit operations. He also designed and taught the University’s first course on Bourbon Production Engineering. Dr. Berron’s research lab designs and executes research studies for regional distilleries. His team recently presented a distillery-view report on the formation and elimination of EC in a bourbon production facility based on exhaustive literature review and interviews with distillers from Beam-Suntory, Buffalo Trace, Wild Turkey, Wilderness Trail, and Castle and Key Distilleries [1]. The presented work extrapolates from literature on other distilled spirits and incorporates reaction rates and volatility data to track the ideal path for EC and its precursors to exit a continuous distillation system.

Abstract

American craft distillers are increasingly producing for markets which regulate the content of ethyl carbamate (EC) through international sales or contract distilling. Craft distillers often leverage local raw materials in their signature products, and these local grains have dramatically elevated levels of EC precursors. The EC literature places little emphasis on American whiskeys, and there is a complete lack of literature on EC levels in bourbons and ryes produced on craft scale, continuous stills. Based on findings from existing literature for EC levels in other spirits at large commercial scales, this team proposes to quantify the effect of methods to manage EC levels in craft scale American whiskeys. This research team has done a preliminary analysis of existing data from Castle and Key Distillery to develop general trends that are in line with the findings found in other distilled spirits. The resources provided by this research grant would allow the collection of additional process data, a detailed statistical analysis to determine how strongly these parameters alter product EC levels, and a process analysis of these effects. These analyses will enable the craft distiller to confidently meet challenging EC targets in craft products that use local, EC precursor-rich malted barleys.

Background

Ethyl carbamate (EC) is a potential carcinogen that is regulated in many global markets. EC originates from glycosidic nitrile content in the grains used for distilling. While some malts provide lower EC levels in the final product [2-4], many craft distillers prefer the broader range of unique flavors available from other grain sources. Distillery operating parameters regulate EC levels through 1) controlling the rate of EC formation, 2) removing EC through nonvolatile streams, and 3) potentially removing volatile precursors through incomplete condensation. The major precursor compounds and steps are involved in the formation of EC are glycosidic nitrile -> isobutyraldehyde cyanohydrin -> hydrogen cyanide -> EC [2, 5].

Rapid conversion of hydrogen cyanide to EC drives the loss of the non-volatile EC through waste streams, while slow conversion allows hydrogen cyanide to reach the barrel where it slowly converts to EC [6]. The conversion of hydrogen cyanide to EC is catalyzed by copper ions, and distilleries rely on routine cleaning of abundant copper surfaces to accelerate catalysis [7]. Alternatively, the hydrogen cyanide has the potential to escape from the low wines condenser under common operating conditions.

While many of these phenomena are known, the quantitative impact of distillery operations on these potential interventions is poorly understood and not described in any scientific literature. There is little targeted guidance available for a craft distiller to decrease EC levels for an American whiskey product. The following research plan will determine how EC levels are controlled by specific changes in distillery operating conditions.

Materials and Methods

Month 1-8. The production team will identify up to 50 upcoming batches that will have the same mash bill and yeast to control for the following parameters.

Beta glucosidase levels in the yeast. Determined by Lallemand for each lot. Only data from batches with levels within 10% of the mean will be used for analysis.
Mash bill. Only mash bills consistent within 1% will be selected for this study. • Glycosidic nitrile levels in each mash bill component. The University of Kentucky will measure levels according to established assays in the literature [11]. Only data from batches with levels within 10% of the mean will be used for analysis.
We will systematically vary and record the following parameters during production:

Number of batches since caustic cleaning
Number of batches since citric cleaning
Feed tray location
High wines proof
Low wines condenser temperature setpoint
We will also record the EC level for each batch. Distillate samples will be aged in vials for 30 days for complete conversion of hydrogen cyanide to EC [10] before being mailed for EC testing at ETS Laboratories.

Month 8-12. Castle & Key and University of Kentucky teams will relate findings to practical guidance on distillery operation for low product EC. This analysis will include potential congener impacts of each operational change. We will also discuss how these findings qualitatively inform distilling approaches for EC management in other craft spirits, including malt whiskey and fruit brandies. These results would be compiled into an article in the Process section of DISTILLER magazine or a forthcoming ADI scientific journal.

Budget and Justification

$1,000 – Testing materials of glycosidic nitrile levels of grains. $500 for beta-glucosidase digestion reagent. $500 other reagents including buffers, glassware, colorimetric indicators. $1,000 – Statistical software and consulting for analysis of Castle and Key EC data. $500 – Yeast beta glucosidase analysis by Lallemand $100 each and ~5 lots. $4,500 – EC testing of up to 25 samples at ~$200/sample. If additional samples are needed for statistical power, up to 25 additional EC assays will be collected and paid for by Castle and Key Distillery.

Observations and Significance

We expect to provide the craft distiller a clear map of the relative importance of methods for EC management. This will enable the craft distiller to adjust their operations to deliver low EC product derived from unique, regional grains. Additionally, this will prepare the craft distiller for any potential changes in domestic EC regulations.

Based on our preliminary findings and the literature, we hypothesize that the following parameters will regulate product EC levels.

An increase in the number of batches since caustic cleaning is expected to strongly increase product EC levels due to decreased copper contact in the still.

An increase in the number of batches since citric cleaning is expected to strongly increase product EC levels due to decreased copper contact in the still.

The feed tray position controls both the amount of copper surface area in the rectifying section and the residence time of materials in the rectifying section. First, adding the feed to a lower tray on the distillation column will increase the copper surface area per wash that the HCN contacts as it goes up the column. Secondly, adding the wash to the lower stage will increase the time materials spend in the rectifying section. An increase in the number of rectifying trays from 4 trays to 5 will increase the surface area and amount of time for HCN reactions by ~25%. Both aspects of a lower feed tray will encourage HCN to convert to EC, and EC formed in the column is cleared out with the stillage. This effect has not been previously documented in the literature.

High wines proof is simply a measure of how well the overall process is separating ethanol and water. Since EC is less volatile than water, we would expect EC to largely stay with the water. If we are allowing more water into the distillate, we are likely also allowing more EC into the distillate. This has not been reported in any of the EC literature, but it is an easy parameter for craft distillers to control. Student researchers have confirmed this with computational preliminary process modeling, but real-world data is needed to fully evaluate its practical importance.

The low wines condenser temperature governs which compounds in the low wines are condensed and which are vented out of the process. HCN is the highly volatile precursor of EC. If the condenser is running hot, the HCN will not condense and will be vented out of the process. This common process parameter is usually modified to refine flavor or yield, but the impact of the low wines condenser temperature on EC content has yet to be explored.
Reviewer comments and suggestions

Overall: Why is this important to the craft distilling industry?

Craft distillers often leverage local raw materials to provide a value-added distinction from mass market distilled spirit products. Given this heightened regional emphasis in craft distilling, local grains are a common factor in the marketability of craft products. American malts can have significantly higher levels of EC precursors than European varietals developed for distilled spirits. In some cases, these precursor levels are 10-fold higher than AMBA guidelines. The proposed work seeks to guide craft bourbon distillers on the best operational practices for minimizing EC levels in signature products that use local, precursor-rich malted barleys and grains.

Reviewer 1 commented that malt whiskey and fruit brandies are much more likely to have EC problems than bourbon.

We agree that malt whiskey and stone fruit products are more prone to high EC levels, and these products are almost the exclusive focus the current EC literature. The EC levels seen in some bourbon batches are of concern for export markets, and the craft bourbon distillers have few informational resources for addressing this challenge. We will also include a discussion in our report of how the practices learned in this study may be adapted to the production of other products challenged by high EC levels.

Reviewer 2 commented “not clear whether they’re collecting chemical data or just process parameters.”

For each batch in this study, we will record the following:

Beta glucosidase levels in the yeast strains
Mash bill
Glycosidic nitrile levels in each mash bill component
Number of batches since caustic cleaning
Number of batches since citric cleaning
Feed tray
High wines proof
Low wines condenser temperature setpoint
EC level of product after 30 days
In addition, each batch will have the standard work up grain, fermentation, and distillate chemical analysis production quality practices
We will keep the following parameters constant:

Beta glucosidase levels in the yeast strains to 10%
Mash bill to 1%
Glycosidic nitrile levels in each mash bill component to 10%
We will systematically vary the following parameters during production:

Number of batches since caustic cleaning
Number of batches since citric cleaning
Feed tray
High wines proof
Low wines condenser temperature setpoint
I think Reviewer 3 has some pushback around the cost of the stats. Regardless, I think if the objective is around statistical analysis, control needs to be factored into the design. Also, there might be some other considerations to include that could make this more value added across the craft industry.

In our prior approach, the methods required to decouple the influences of variables were not straightforward with software packages, and the pitfalls of software assumptions can have powerful implications on the validity of the results. We have redesigned our study to have more controlled variables and focus on fewer parameters. This will reduce the statistical complexity of the proposed study. We will still work with our statistician on the design of experiments to ensure sufficient power for the analysis of each variable.

GN data on the malted barley and mashes $1000 is listed for consumables, etc. through UK. Aaron MacLeod’s lab at Hartwick college is already set-up for this analysis. I have collaborated with him on a number of EC studies and he is the official partner for the American Malted Barley Association for screening. He may be interested in collaborating at gratis, if there is also a publication in it for him as well. I would ensure GN data is known/ the same batch is used for the entirety of the project.

We appreciate the reviewers looking out for opportunities to interface with other talented researchers. For this study, we would prefer to avoid the added complexity of sample mailing and research coordination with out of town labs. We are excited to look for other opportunities to work with Aaron. His lab listed the cost of the GN assay at $225/sample in January 2019, and we believe the $1000 in requested costs are appropriate for the proposed 50 batch study.

High DP malt the GN data can be quite variable, depending on varietal, maltster, whether it’s’ being blend to a spec, etc… AMBA has a guidance of 1.5 g/ MT…I have seen some varietals in the 16 range (KY grown varietals, btw). 1.5 is arbitrary though, as downstream processing can have a huge impact, that is one of the reasons why this project is exciting.

Thank you for this insight. We have revised our proposal to reflect this motivation.

ETS can perform an EC predictive analysis on distillates, I would consider that (MC EC is catalyzed by light and heat, their test capitalizes on that).

Thank you for the suggestion. We have been focusing our analysis on distillates that are aged at least 30 days. We will be sure to use this higher priced testing if we need to run distillates that are less than 30 days old.

Curious on the CIP would be interesting to see batches between caustic cleanings vs citric cleanings. In my experience, citric is much more effective at regenerating copper

We have revised the proposal to also consider this question.

A little confused about the variables still wanting to be tested. Feed tray, low wines condenser temp and high wines proof? This was a little unclear for me, so just would like to see that refined.

The low wines condenser temperature governs which compounds in the low wines are condensed and which are vented out of the process. HCN is the highly volatile precursor of EC. If the condenser is running hot, the HCN will not condense and will be vented out of the process. This common process parameter is usually modified to refine flavor and yield, but the impact of the low wines condenser temperature on EC content has yet to be explored.
High wines proof is simply a measure of how well the overall process is separating ethanol and water. Since EC is less volatile than water, we would expect EC to largely stay with the water. If we are allowing more water into the distillate, we are likely also allowing more EC into the distillate. This has not been reported in any of the EC literature, but it is an easy parameter for craft distillers to control. UK student researchers have confirmed this with preliminary process modeling in ASPEN, but real-world data is needed to fully evaluate its importance.
The role of the feed tray discussed immediately below.
Also, it might be more value added to instead of position the variable as feed tray, to instead call out copper rectification surface areas Ii.e. x cu ft of copper per wash processed). Also, to further control, is a dephlegmator or partial overhead condenser at play at the distillery?

We apologize for not making this clearer in the first submission. We have added the following text: “The feed tray position controls both the amount of copper surface area in the rectifying section and the residence time of materials in the rectifying section. First, adding the feed to a lower tray on the distillation column will increase the copper surface area per wash that the HCN contacts as it goes up the column. Secondly, adding the wash to the lower stage will increase the time materials spend in the rectifying section. An increase in the number of rectifying trays from 4 trays to 5 will increase the surface area and amount of time for HCN reactions by ~25%. Both aspects of a lower feed tray will encourage HCN to convert to EC, and EC formed in the column is cleared out with the stillage.”

Since most distillers can only change feed tray number (not these other parameters), we thought it would be most helpful to keep the discussion focused on that parameter. Our paper will qualitatively relate these findings to the role of 1) total surface area per wash, 2) rectifying area per wash, and 3) residence time in the rectification section.

A copper dephlegmator is in line with the column, and the cooling water flow into the dephlegmator is held constant during production.

One of the things I was looking at previously was the BGL (beta glucosidase) level during fermentation…different yeast strains have varying levels, I think it would be interesting to at least control for that variable as well. Lallemand Mascoma provides that testing…depending on the yeast strain they use, this value could be known already

We have revised the proposal to control and test for this parameter, as well.

List of References/Citations

Gollihue, J. and B.J. Berron, Ethyl Carbamate formation during the production of distilled spirits, in James B. Beam Institute Industry Conference. 2019: Lexington, KY.
Cook, R., et al., Ethyl carbamate formation in grain‐based spirits: part III. The primary source. Journal of the Institute of Brewing, 1990. 96(4): p. 233-244.
Nielsen, K.A., et al., Leucine-derived cyano glucosides in barley. Plant physiology, 2002. 129(3): p. 1066-1075.
Swanston, J.S., et al., Using molecular markers to determine barleys most suitable for malt whisky distilling. Molecular breeding, 1999. 5(2): p. 103-109.
Bringhurst, T.A., 125th Anniversary Review: Barley research in relation to Scotch whisky production: a journey to new frontiers. Journal of the Institute of Brewing, 2015. 121(1): p. 1-18.
Aylott, R., et al., Ethyl Carbamate Formation In Grain Based Spirits: Part I: Post‐ Distillation Ethyl Carbamate Formation In Maturing Grain Whisky. Journal of the Institute of Brewing, 1990. 96(4): p. 213-221.
Aresta, M., M. Boscolo, and D.W. Franco, Copper (II) catalysis in cyanide conversion into ethyl carbamate in spirits and relevant reactions. Journal of Agricultural and Food Chemistry, 2001. 49(6): p. 2819-2824.
Knoch, E., et al., Biosynthesis of the leucine derived α-, β- and γ-hydroxynitrile glucosides in barley (Hordeum vulgare L.). The Plant Journal, 2016. 88(2): p. 247-256. 9. Ehlert, M., et al., Deletion of biosynthetic genes, specific SNP patterns and differences in transcript accumulation cause variation in hydroxynitrile glucoside content in barley cultivars. Scientific reports, 2019. 9(1): p. 5730-5730.
Aylott, R.I., et al., Ethyl Carbamate Formation In Grain Based Spirits: Part I: Post Distillation Ethyl Carbamate Formation In Maturing Grain Whisky. Journal of the Institute of Brewing, 1990. 96(4): p. 213-221.
Morrall, A.B. and M.A.O.G. Britain, Determination of repeatability and reproducibility of a new rapid enzyme method for the determination of glycoside nitrile in malted barley. Journal of the Institute of Brewing, 1996. 102(4): p. 245-247.