Diacetyl: Formation, Reduction, and Control


By George J. Fix (Brewing Techniques)
Diacetyl can smell like the artificial butter used on movie theater popcorn
Diacetyl — the compound responsible for buttery or butterscotch flavors that sometimes arise in beer — can be controlled if you understand the mechanisms that contribute to its production. This review of the basic processes behind diacetyl formation and reduction will help you understand how to keep the diacetyl level in your beer at or below the acceptance threshold for the style. 


Diacetyl and 2,3-pentanedione are important contributors to beer flavor and aroma. Organic chemists classify both as ketones, and diacetyl is usually called 2,3-butanedione in the literature. Sometimes these two ketones are grouped and reported as the vicinal diketone (VDK) content of beer.
Brewers’ awareness and acceptance of both diacetyl and 2,3-pentanedione have changed dramatically over the past four to five decades. A 1952 report, for example, stated that the average diacetyl level of American commercial beer was 0.33 mg/L, more than three times the flavor threshold of 0.10 mg/L. Today the average is near 0.05 mg/L. Some notable exceptions exist. Some stouts can have levels as high as 0.60 mg/L, and a few British pale ales have diacetyl levels near 0.30 mg/L. Diacetyl levels in beers brewed by microbreweries and brewpubs tend to vary considerably, ranging from 0.03 mg/L to more than 1.0 mg/L in cases I have investigated.
The presence of diacetyl is usually indicated by a buttery or butterscotch tone. In fresh beer the flavor can be confused with that of caramel malts. Given time it is easy to distinguish the two; diacetyl tends to be unstable in most beers and can take on raunchy notes. The flavoring imparted by caramel malts, on the other hand, tends to be stable.
The flavor threshold of 2,3-pentanedione is near 1.0 mg/L; it usually imparts flavors that recall honey. This compound can be found well above threshold levels in sonic Belgian ales, where it is considered a natural flavor constituent for this style. It occurs less often in other beer styles and is regarded as a defect. 

Diacetyl Formation and Reduction

The study of diacetyl and beer began with Pasteur’s fundamental work in the 1870s. Using microscopy, Pasteur found that what we know today as lactic acid bacteria were responsible for many off-flavors in beer. The term sarcina sickness is used to describe this effect. Apparently, the involvement of diacetyl in sarcina sickness was discovered early, but it was not until 1939 that Shimwell linked this compound with the taste and smell of butter. Earlier studies got the organic chemistry right but were wide of the mark in terms of flavor chemistry. Even today it is estimated that 20% of beer drinkers do not detect the presence of diacetyl even at rather high concentrations.
During the early period, the only known mechanism of diacetyl formation was bacterial infection caused by unsanitary conditions. Practical brewers believed that some other factor must be involved, because buttery tones occasionally showed up in beer brewed in impeccably clean environments. Major breakthroughs occurred during the 1950s and early 1960s. J. Owades developed an effective technique for measuring diacetyl and used this method to study the fate of the compound in brewing. This work pointed to culture yeast as a major player in both the production and the reduction of diacetyl. Inoue and his colleagues at the Kirin Research Laboratory in Japan also made a major contribution by identifying acetolactic acid as the precursor to diacetyl. This work was followed by a large number of papers in which various facets of diacetyl formation and reduction were studied. Wainwright’s excellent 1973 review article contains 149 references.
Although a number of factors affect diacetyl formation in beer. The dominant carbon flows down the leftmost branch, leading to ethanol production. Most beers contain between 30,000 and 50,000 mg/L of ethanol, so a significant amount of pyruvate is processed in this manner. Because the flavor threshold for diacetyl is 0.10 mg/L, a slight diversion of the carbon flow to the middle pathway can profoundly affect the flavor of finished beers. Note that this pathway also competes with yeast assimilation and the utilization of the amino acid valine. The practical significance of this is discussed in the section “Wort and Proteins” below.
A similar pathway is involved in the production of 2,3-pentanedione, except that different compounds are involved. The precursor is acetohydroxybutyrate, and the competing amino acid is leucine.
A number of factors lead to diacetyl formation, but only one reliable method can reduce diacetyl levels: enzymatical reduction by yeast (Figure 2). Acetoin, the intermediate product, has a rather unpleasant, musty taste, but because it has a flavor threshold of 3.0 mg/L its effect is not nearly as damaging to beer flavors as an equivalent amount of diacetyl. The final product, butanediol, is neutral as far as beer flavor is concerned. 

The Effect of Yeast

As noted above, brewer’s yeast contains enzymes for both producing and reducing diacetyl. Various yeast strains differ dramatically in this regard. The data in Table I were compiled for a book I am writing. Three lager strains were tested. The strains W-206 and W-34/70 are regarded as excellent reducers. The third strain, W-308, is less reliable. Sterile wort was fermented in each case at 10 °C (50 °F), and the diacetyl was measured using high performance liquid chromatography (HPLC).
Highly flocculent yeast usually behave much like W-308 and can leave perceptible levels of diacetyl in beer, which is one reason why most commercial yeast strains are powdery and fully flocculate only after chilling. Yeast behavior for a given strain can also vary with reuse.
In my experience, increases in the diacetyl formed with repitching, such as those reported above, parallel increases in the level of respiratory-deficient mutants in the pitching yeast. These mutants are strong producers of diacetyl and have lost their ability to reduce diacetyl. The presence of Gram-positive bacteria also cannot be ruled out as contributors to diacetyl formation. Yeast that is free of those two defects usually displays better performance with reuse.
Diacetyl production and reduction are strongly influenced by temperature, and the rates for both increase as temperature increases. Thus, an ale fermented at 20 °C (68 °F) typically has a higher diacetyl peak than, say, a lager fermented at 10 °C (50 °F). The rate of diacetyl reduction, however, is much higher in the ale than in the lager, which is why most lager brewers prefer to get diacetyl levels below 0.10–0.15 mg/L at the end of the main fermentation. Some additional reduction occurs in cold storage, but at a very slow rate. For this reason, some brewers raise the temperature of a cold-fermented beer to 20 °C (68 °F) for a brief period following the end of the main fermentation, a practice that is usually called diacetyl rest.
One alternative is the so-called Narziss fermentation. In this procedure the first two-thirds of the fermentation is done at 8–10 °C (46–50 °F). During the final third of fermentation, the temperature is allowed to increase to 20 °C (68 °F), after which the beer is transferred to cold storage. Another alternative is to add freshly fermenting wort (kraeusen) to diacetyl-laden beer in cold storage. 

Bacterial Influences

The bacteria that can directly promote diacetyl production consist of Gram-positive cocci (Pediococci) and select strains of Gram-positive rods (Lactobacillus). The effect of using these bacteria is easy to identify — both bacteria also produce lactic acid, and the net effect is a rather raunchy butter tone with an unmistakable acid aftertaste. In both amateur and commercial brewing of yesteryear, infections from unsanitary equipment were not uncommon. Today, with the availability of highly effective sanitizers, infections tend to occur only in sloppy and poorly managed operations. It has been my experience that in modern operations, infections, when they occur, happen through pitching yeast.
It is unnecessary for fully developed pitching yeast to be sterile. Practical experience has shown that as long as bacteria relevant to beer are kept below the level of 1–10 cells per 10 million yeast cells, the finished beer will remain unaffected. The situation is dramatically different in the initial stages of yeast propagation. Here, sterile conditions are needed, as is pure culture yeast. This is particularly true when propagating yeast from slants, but it also applies to starting up semidormant liquid yeast. It is relatively easy to measure bacterial levels, particularly for lactic acid bacteria, and hence there is little justification for leaving these matters to chance.
Care should be taken when using yeast that has been held in bulk storage. If culture yeast go dormant, they tend to excrete amino acids, which bacteria can use as a source of nitrogen. Low levels of bacteria can grow to unacceptably high levels by this means. The safest course is to store the yeast under a sterile wort cover, at a temperature as close to 0 °C (32 °F) as possible, and for as brief a period as possible. 

Wort and Proteins

The basic diacetyl formation pathway shows clearly the major role that amino acids play. Worts deficient in valine tend to lead to elevated diacetyl levels. As long as a sufficient amount of valine is present, there will be a net reduction of diacetyl after it reaches a peak level. If the valine content is depleted prematurely, however, net diacetyl production will resume, leading to what has been called the “second diacetyl peak.” Some of this extra diacetyl ultimately will be reduced, but invariably the finished beer will contain a higher diacetyl content than it would had the wort contained adequate valine. The same remarks apply to 2,3pentanedione and leucine.
Both leucine and valine are regarded as critical amino acids because yeast usually cannot metabolize adequate replacements from other nitrogen sources if leucine and valine are missing. If sufficient amounts of proteins are to be available in the fermentation, they must come from wort; this in turn depends on the amount of malted grains used. High-quality malt used with reasonable mashing systems will yield wort that is rich in all the relevant amino acids.
Brewers use a single number to characterize the size of their wort’s amino acid pool — namely, its free amino nitrogen (FAN) level. This number is to wort amino acids what specific gravity (or percent extract) is to wort carbhydrates. High FAN levels mean adequate leucine and valine pools. Conversely, low FAN levels result in inadequate levels of leucine and valine.
An all-malt wort at 12 °P (SG = 1.048), for example, typically has a FAN level in the range of 300–325 mg/L. This level is considered ideal. A 10 °P (SG = 1.040) all-malt wort, on the other hand, has a FAN level near 250–270 mg/L — generally regarded as adequate. However, if a third of the malt were replaced with an equivalent amount of unmalted cereal grains or sugar (or both), then the FAN level would fall to 165–180 mg/L, which is generally regarded as inadequate. Apart from economic considerations, this is one of the main reasons why large industrial brewers using high adjunct levels tend to favor high-gravity brewing. If the 10 °P wort with 33% adjuncts were concentrated to, say, a 14 °P wort, then the FAN level would increase to 230–250 mg/L. This is a major improvement over the dilute adjunct wort.
Reports have been published of various deficiencies in certain brands of malt extracts. It appears that, in general, the protein levels of wort produced from extracts are lower than those obtained from grain worts. These and related issues are discussed in an excellent article by Lodahl. 
Although it is important that our wort have a sufficient amino acid pool, it is also important that our culture yeast be able to use this nitrogen source. The situation for leucine and valine is particularly critical because their uptake by brewing yeast is rather slow in general and incomplete in the case of dysfunctional yeast. Respiratory-deficient mutants represent an extreme example of this. In all such cases the inevitable result is elevated diacetyl levels.
One cannot overstate the importance of using appropriate media in propagating yeast. The instruction sheet that accompanies yeast slants sold by Siebel (Chicago, Illinois) states that propagation should be done with wort that is “… of a similar original gravity and composition to the major production brand.” Many brewers use dilute, unhopped wort having a specific gravity of 1.020. Although opinions may differ about these two options, totally artificial media should be avoided. I recently did some tests with what is often called “baker’s media.” It consisted of a dextrose-sucrose solution enriched with mineral nutrients. Yeast propagated on this media showed excellent cell growth rates during propagation. Their performance in the main fermentation, however, was unacceptable. A major defect was an extremely slow and incomplete valine uptake. Diacetyl levels were always higher by at least a factor of 4 relative to yeast propagated with the Siebel procedure, and in one example the level was 10 times higher. 


The reaction
                          acetolactic ---> acid diacetyl 
is of the redox type. Acetolactic acid is oxidized to diacetyl, and other constituents (for example, various aldehydes as well as wort-derived melanoidins and tannins) are reduced. In all of the mechanisms described so far in this article, this is done enzymatically by microbes, culture yeast, and, in adverse cases, by other guests in our worts. The reaction can occur nonenzymatically, however, in the presence of an appropriate oxidizing agent. Indeed, a widely observed but little discussed phenomenon occurs when diacetyl appears spontaneously in a beer that seemed to have normal flavors. Strong evidence indicates that this can occur when marginally dysfunctional yeast have been used in the main fermentation — they tend not to metabolize all the acetolactic acid in the wort. The acetolactic acid spills over into the finished beer and later is oxidized to diacetyl. Mechanical abuse of packaged beer can promote this; headspace air is the oxidizing agent. Elevated temperatures augment the effect. I have seen cases in which wort constituents (melanoidins and tannins), oxidized on the hot side in wort production, were passed on to the final beer, only to play the role of oxidizer there.
Oxygen and diacetyl are linked in another way, in this case in a positive manner. Hoffmann has shown that inadequate oxygenation of chilled wort can lead to elevated diacetyl levels. In one of his test brews, the dissolved oxygen content of chilled wort was a mere 0.80 mg/L. The second brew had 10 times that amount (8.0 mg/L), which is a widely used value. To achieve this level of oxygenation, one typically must saturate chilled wort with direct oxygen injection. Hoffmann reported that after the seventh day of fermentation, the poorly oxygenated wort had a diacetyl level of 0.80 mg/L, whereas the level in the second brew was down to 0.20 mg/L.
Ironically, Hoffmann also reported that diacetyl levels increased again after dissolved oxygen levels exceeded 8.0 mg/L. I have found that the maximum amount of oxygen that can be dissolved in, say, a 12 °P wort (SG = 1.048) at 10 °C (50 °F) is 7.8 mg/L. The amount of dissolved oxygen decreases as either the temperature or the wort gravity increases. To get higher values one would have to use supersaturation procedures, which requires special equipment. Thus, small-scale brewers need not fear that too much oxygen has been dissolved in their chilled worts. 

A Ubiquitous Guest

It can probably be said that the extreme sensitivity that practical brewers tend to have for diacetyl flavors in beer results, in part, from the fact that virtually every aspect of brewing involves diacetyl formation or reduction. Opinions will probably always differ about the suitability of diacetyl flavor in various beer styles. Nevertheless, we should be all aware of how it can arise and how its level can be controlled. 


All contents copyright 2019 by MoreFlavor Inc. All rights reserved. No part of this document or the related files may be reproduced or transmitted in any form, by any means (electronic, photocopying, recording, or otherwise) without the prior written permission of the publisher.