Control Beer Oxidation from Kettle to Bottle
Originally Published by George J. Fix - Brewing Techniques (Volume 6, Number 6)
The effects of the introduction of oxygen to wort or beer on flavor stability have been widely
debated over the years, and the subject has been the target of much recent research. The
compounds that contribute to staling flavors cannot easily be determined because of the complex
nature of beer chemistry, but one fact is clear: Brewers can exercise a great deal of control over
flavor stability by observing proper brewing techniques.
John Maier, head brewer at Rogue Ales and 1988 American Homebrewers Association
Homebrewer of the Year, was once asked what he reared the most as a brewer. His answer was
pointed and direct: “It is the A, B, Ds.” The “A” on his list stands for air, which is the subject of
this article. (The interpretation of “B” and “D” is left as a homework exercise.)
Successful brewers like John invariably display sensitivity to oxidation problems, and they also
have the skill to effectively deal with them. The goal of this article is to identify the flashpoints,
that is, the critical areas in the brewing process where care is needed to prevent oxidation. I will
provide some suggestions for operating a low-oxygen system, be it on the homebrew or
commercial level.
The major difference between brewing science and the biological and chemical sciences is that in
the former, flavor is the final arbitrator of what is important and relevant. This is a particularly
delicate point as far as oxidation is concerned, since there is a definite variation in how people
perceive oxidized flavors. Thus, the first thing I will do in this article is to establish criteria that
characterize oxidized beer. This material is a sequel to Scott Bickham’s recent excellent article in
his Brewing Techniques column, “Focus on Flavor” (1).
The rest of the article will deal with the parts of the brewing process where oxidation is likely to
occur. A distinction is made between hot-side aeration (HSA) and cold-side aeration (CSA). The
hot side includes events from mashing through wort chilling. When the wort is cool, oxygen is
added to the wort, or more commonly these days to the yeast, before the yeast is pitched. At this
point in the process, oxygen serves as an invaluable yeast nutrient. Once the fermentation has
started, however, oxygen returns to being a negative element, and what is referred to as the cold
side starts at this point and continues until the beer is consumed.
Perception of Oxidized Flavors
Following C.D. Dalgliesch’s well-known article on flavor stability, three stages in beer flavor
evolution can be defined as shown in Figure 1 (2). Stage A is the period of stable, “brewery-
fresh” flavor. Stage B is a transition period in which a multitude of new flavor sensations can be
detected. Dalgliesch cites a decline in hop aroma, a decline in hop bitterness, an increase in
“ribes aroma” (or sometimes “catty” flavor) and an increase in sweet/toffee-like/caramel tones.
The terms “ribes” (or “currant”) and “catty” are widely used in the United Kingdom and
Scandinavia. The terms are intended to recall overripe or spoiled fruit or vegetables. Some tasters
cite a “black currant” tone (3). In truth, these terms describe a wide spectrum of negative flavors
when beer is in Stage B. Toffee or caramel flavors can come from many sources, but those
associated with staling will invariably have unattractive cloying notes. These effects are
enhanced by residual diacetyl, and also by excess heat treatment of wort (4). Boiling in pressure
cookers will develop flavor tones like these along with other negative ones (5).
The Stage C products are the classic flavor tones involved in beer staling. These flavors are
described in subsequent sections of this article in conjunction with the part of the brewing
process in which they are created.
Although it is not treated in the references cited above, a Stage D, also exists. In this phase, Stage
C flavors evolve into a kaleidoscope of flavors, which in very special formulations (Rodenbach’s
Grand Cru comes to mind) recall the subtlety and complexity of great wines. It is to be
emphasized that this process takes years, not months, of maturation. This effect deserves a paper
all its own, and hence falls outside the scope of this article.
Do we prefer stale beer? Barry Axcell and Phill Torline, in an interesting if provocative article,
argue that most beers are consumed during Stage B (6). During this period beer flavors are
undergoing discernable change, and the authors suggest that the changes are at the root of
consumer dissatisfaction. Among other things, they cite the so-called import paradox as partial
evidence of this theory, the paradox being that a definite proportion of the beer-consuming
population actually prefers beers in Stage C. (Here, “import” means any beer consumed at a
significant distance from where it is brewed.) These authors noted the stability of flavor in Stage
C and “learned prejudices” (such as prestige of the beer and packaging) as the keys to this
paradox. Statements like “If Pilsner Urquell has HSA, then I want my beer to have HSA,” often
seen in the internet discussion groups devoted to beer, illustrate the apparent validity of this point
(7). Stage B flavors; on the other hand, appear to have very few advocates.
M.C. Meilgaard, for one, is sharply critical of Stage C flavors because they are one-dimensional
(8). He states, “I think it ranks as an all-American scandal that fully half of all the interesting and
unusual packaged beers that are on the market get to us so oxidized that stale-ness and cardboard
are the main flavor tones.” I feel this is a very important point, and this article is based on the
premise that the best strategy for brewers (amateur or commercial), is to produce beers in which
Stage A flavors are as stable for as long as possible. If they, or their consumers, prefer Stage B or
C (or D for that matter), then putting the beer aside for the required time is all they need to do.
Staling, like entropy, can go in only one direction!
Cold-Side Aeration
Oxidation on the cold side is the easiest to understand, and until the last decade has occupied all
of the attention of those concerned about this issue. The major mechanisms are given in Table I.
Flavor thresholds in parts per million (ppm) were taken from Meilgaard (9).
Oxidation mechanisms: Ethanol. The first oxidation-reduction (or redox) reaction in Table I is
the start of the classic “vinegar process.” The oxidation of ethanol to produce acetaldehyde
marks the second appearance of acetaldehyde in the brewing process. Acetaldehyde is first
encountered in the main fermentation pathway, in which it emerges as a reduction product of
pyruvic acid and a precursor of ethanol (10,11). At that stage it has a flavor that recalls freshly
cut apples. As an oxidation product of ethanol its flavor changes to an “old apple” tone. (The
stronger, more negative flavor results from higher concentrations and concomitant compounds.)
It may also recall “rotten apples,” although this effect is not nearly as strong as the flavors
associated with Zymomonas infection.
Table I: CSA Mechanisms
(1) Ethanol + oxygen (14,000 ppm)* acetaldehyde (25 ppm)
(2) Acetaldehyde + oxygen (25 ppm) acetic acid (175 ppm)
(3) Hop alpha-acids + oxygen (10 ppm) valeric and butyric acids (1–2 ppm)
(4) Hop oils + oxygen various oxygen-bearing compounds (9, 10)
(5) Unsaturated fatty acids from trub + oxygen oleic and linoleic acids (< 1 ppm)
*Levels are those normally found in beer.
Acetaldehyde. The transformation to vinegar, or acetic acid, continues with the oxidation of the
acetaldehyde in reaction 2. But because of the high flavor threshold of acetic acid, that reaction
will rarely give a full vinegar taste. Generally, sherrylike tones emerge instead. Sizable amounts
of oxygen and/or acetic acid bacteria (Acetobacter) infections can, however, lead to stronger
vinegar flavors.
Hop constituents. As noted above, the loss of hop flavors (taste and smell) associated with Stage
B in Figure 1 is the easiest way to detect the start of staling. By the time stage C is reached, the
oxidation products of alpha-acids (the valeric and butyric acids noted in reaction 3 of Table I)
will start to appear. “Cheesy” is a widely used descriptor for these flavors. The use of old hops at
elevated levels will intensify this effect. The flavors from hop oils, on the other hand, will
generally just disappear in Stage B, although sometimes residual tones recalling hay, sagebrush,
or grass can be detected. Taste panels differ considerably in their response to grassy flavors (12).
Many tasters do not see these flavors as a defect, and highly valued hops like Columbus or Kent
Goldings can display these tones in attractive ways.
Fatty acids. The final CSA mechanism is often underestimated because unsaturated fatty acids
from trub in wort play a highly beneficial role in the fermentation. Their value on the cold side,
however, is the exact opposite. They are highly reactive with oxygen, and at this stage oxidize to
a broad range of products, none of which are attractive. The soapy or goaty tones from oleic
acids are examples. These effects are usually not seen in beers produced from reasonably
clarified worts. The relatively low flavor thresholds of these oxidation products are a special
concern.
Table II: Effect of Thermal Abuse and Headspace Air on Staling
Days to Staling
Headspace air (mL per 1/3 L) Storage at 86 °F (30 °C) Storage at 43 °F (6 °C)
0.5 60 120
1.0 40 100
1.5 20 80
2.0 5 70
Data obtained from test brews that were carefully controlled with respect to HSA and other
forms of CSA (5).
Preventing CSA: Clearly the most straightforward way to prevent CSA is to prevent oxygen
ingress during the postfermentation processing of beer. As far as beer transfer is concerned, the
first important step is to remove as much air as possible from the receiving tank. Purging the tank
with carbon dioxide is a start. If, in addition, the beer is transferred by applying CO2 pressure to
the sending tank and allowing a slight gas bleed in the receiving tank, one can eliminate oxygen
pickup. In fact, I have found that the use of CO2 pressure to transfer beer throughout the post
fermentation process can result in null dissolved oxygen readings, at least according to standard
meters (5). The next best option is to transfer by means of gravity flow. Even in this case, it is
crucial to keep a carbon dioxide cover over the beer to protect against oxygen ingress.
Maintaining a smooth laminar flow to avoid mixing air and beer is also important.
In commercial brewing, the above two procedures are not always an option. Carbon dioxide is
expensive, and a horizontal brewery layout precludes gravity flows. Thus, the use of pumps is a
necessity. To say that there are good pumps and some bad pumps is a mild understatement.
Operator error may also play a role. For example, the operator must adjust the flow to avoid
cavitation. Pumping beer through filters requires some special attention as well. For example, it
is important that the bell of a diatomaceous earth (DE) filter be purged of air before the flow
starts. Brewers also should take care to prevent air from entering the bell during filtration. The
ideal liquid for dissolving DE before pre-coating is beer, but its use can lead to foaming. Thus,
many brewers instead use water — but it should first be deaerated. Reference 5 describes this
process in detail.
Table III: Flavors Associated with CSA and HSA
CSA Flavors HSA Flavors
Old or rotten apples Cheesy Papery/cardboard
(acetaldehyde) (hop acids) (staling aldehydes, stage B)
Sherry like
Hay/sagebrush/grassy Leathery/woody
(low levels of acetic
(hop oils) (staling aldehydes, stage C)
acid)
Vinegary Soapy/goaty Sherry like
(w/Acetobacter) (fatty acids) (low levels of acetic acid)
In spite of these other, earlier potential sources of oxygen pickup, the greatest concern about
CSA has been correctly pointed at the packaging of beer. The deleterious effect of headspace air
has been evident ever since the human race started putting beer into small bottles. Fortunately,
modern technology has made it possible to dramatically lower oxygen pickup during filling. For
example, double pre-evacuation bottle fillers routinely keep bottle air levels below 0.2 mL air per
1/3 L beer. These levels are strikingly low compared to the situation that existed a few decades
ago, at which time air levels as high as 1.0 mL per 1/3 L were considered acceptable (13). In
fact, even most of the good homebrew hand fillers on the market today, if properly operated, can
get below 0.5 mL per 1/3 L. (Having beer foam displace the bottle’s headspace just before
capping is very important.) On the other hand, cheap homebrew fillers, and many reconditioned
commercial fillers, invariably give air levels above 1 mL per 1/3 L, and in adverse cases 2 mL or
more.
As important as headspace air is, its indicted co-conspirator in the staling process is thermal
abuse. Storage at high temperatures can dramatically shorten the beer’s shelf life, even at very
low headspace air levels. The data in Table II illustrate the synergy of the two factors as beer
progresses through Stage B. (These data were obtained from test brews that were carefully
controlled with respect to HSA and other forms of CSA [5].) It is remarkable that beers with very
low headspace air levels that have been subject to thermal abuse are no more stable than beers
that have been afforded proper storage but have the high air levels one expects from cheap
homebrew hand fillers.
Hot-Side Aeration
New thinking: The first hint that oxygen uptake during wort production could be a problem was
found in the fundamental work of N. Hashimoto, one of the leading experts on beer staling, in
the early 1970s (see reference 14, for example). His prime concern was volatile aldehydes such
as 2-nonenal.* This compound has a characteristic papery or cardboard flavor that sometimes is
accompanied by leathery or woody tones during Stage C of the staling timeline in Figure 1.
These flavor notes are of great concern because of their very low flavor thresholds, which for
most people fall in the range of 0.1–0.4 parts per billion (ppb). The defining step of Hashimoto’s
work came from the observation in his classic 1975 paper that “molecular oxygen does not take a
part [in forming nonenal] in bottled beer” (15). In later work he identified oxidative processes
during wort production as a source of nonenal (16).
*2-nonenal is a representative of a fairly large class of so-called staling aldehydes that also
includes 2-butenal and 2-pentyl-2-butenal (10).
The idea that wort production processes could contribute significantly to oxidation was a new
finding at the time. The only previous published study involving oxygen uptake during wort
production was conducted by Jean DeClerck in 1957. He found that the total elimination of
oxygen during wort production increased the finished beer’s tendency to form chill haze (17). A
follow-up study demonstrated that massive oxygen uptake prior to sparging greatly enhanced
colloidal stability (18). This study did not, however, take into account finished beer flavor, and in
fact illustrates the point that the promotion of physical stability is not always in harmony with the
improvement of flavor stability.
After Hashimoto’s work appeared, manufacturers of brewing equipment (notably in Germany)
funded additional research into this topic. They were concerned about the relevance of HSA to
German beers, which are quite different from the light Japanese lagers used in Hashimoto’s
experiments. Dr. Ludwig Narziss and his colleagues and students at the Technical University of
Munich–Weihenstephan were leading figures in this research. They found that, if anything, HSA
is even more important for the bigger beer styles, a fact that is not too surprising in retrospect
since HSA’s off-flavors derive from the breakdown of malt-based constituents. In any case, this
group quickly became the strongest and most influential advocates for the “low-oxygen
brewhouse.” Reference 19 contains a good survey of their findings, and additional references can
be found in references 5 and 20. See also the recent reference 21, in which the authors showed
that nonoxidized materials from specialty malts can play a highly beneficial role in flavor
stability. I am not aware of any articles in professionally peer-reviewed brewing journals that are
contrary to either Hashimoto’s findings or those of the Weihenstephan group with respect to
HSA.
This research had a profound impact on the design of brewing equipment in Europe. For
example, the evolution of Huppmann, one of Germany’s most prestigious brewing equipment
manufacturers, with respect to HSA is documented in reference 22. This company has developed
notably gentle, low-oxygen systems for transferring beer and wort between brewery vessels.
Table IV: Evaluating Thermal Stress Tests
Effect Cause
CSA flavors appear about the time predicted in Table II. CSA at fill time
CSA during processing or
CSA flavors appear before the time predicted in Table II.
possible air ingress after fill
Staling effects — particularly papery or cardboard tones —
HSA
appear before the time predicted in Table II.
Beers stored heated at 86 °F (30 °C), tasted twice a month. See Table III for a list of flavors
generally associated with CSA and HSA.
Oxidation mechanisms: The first mechanisms to be studied involved malt-derived compounds
like polyphenols and melanoidins. Oxygen uptake on the hot side immediately sets in place a
number of redox reactions. Temperature is crucial, because the speed of these reactions increases
exponentially with temperature. For example, introducing, say, 1 mL of oxygen per liter of wort
at 158 °F (70 °C) will start reactions that take place in seconds. Doing the same for wort at 68 °F
(20 °C) will have virtually no effect. At this temperature, the oxygen fraction will typically
remain an inert gas until consumed by yeast. The studies have thus shown that the oxidation of
melanoidins and polyphenols is an important consequence of HSA. These oxidized compounds
are normally held in check during fermentation by the strong reducing power of yeast
metabolism, but during maturation and Stage A (see Figure 1), these compounds undergo a
complex series of electron exchanges. The net effect is the oxidation of beer alcohols and the
creation of volatile aldehydes. This process is sometimes called “oxidation without molecular
oxygen” because these reactions can take place without oxygen being present. The aldehydes
produced do not display their presence until a lag period is over (typically 3–4 weeks). This
seems to be because of bonds between the aldehydes and natural sulfur compounds from yeast
metabolism (23). These bonds, alas, are temporary. When they are broken, a wide range of
flavors appears. All have a grainy astringency associated with them, and metallic undertones are
often present. Sherrylike notes have sometimes been identified as well, and this is one flavor
note that is commonly attributed to both CSA and HSA (11).
Unfortunately, these mechanisms do not explain how the all-important 2-nonenal is created,
because it has been demonstrated that there is no alcohol relevant to beer that is a precursor to 2-
nonenal (11). Current research (see, for example, reference 24) points to fatty acids in wort,
along with malt-based enzymes such as lipase and lipoxygenase, as being potential precursors.
Heat and oxygen stimulate the enzymatically induced creation of the hydroperoxides that are the
precursors of 2-nonenal. In any case, oxygen uptake in the brewhouse remains an obvious culprit
in the production of staling aldehydes, along with a still-to-be-determined effect from malting.
An extensive review of all of this research is given in reference 11.
Preventing HSA: HSA is easy to avoid in home brewing because it will arise only from very
sloppy brewing practice. It has been my experience over the years that advocates of such
procedures rarely stay with the hobby. Thus, on the homebrew level, HSA is a problem that
seems to take care of itself.
Some sources of HSA are easy to avoid in commercial brewing. An example is the minimization
of shearing forces from excessive raking or stirring, a remarkably important effect (5).
A more perplexing problem is the HSA created by splashing during hot wort transfer. Following
a presentation I gave at the April 1998 National Craft-Brewers Conference in Atlanta, I was
approached by someone who has been very skeptical about the importance of HSA. He correctly
noted that there are a lot of commercial systems in operation that have very high oxygen uptake
owing to splashing. He then asked if HSA is sufficiently important to require the purchase of an
entirely new brewhouse. I feel, at least for commercial brewers, that the answer is definitely yes,
and the issue is one that also should be taken up by the manufacturers of the problematic
equipment. The comparative advantage of craft brewers in the market place is the perception of a
dedication to excellence in all aspects of brewing. It is very hard for even ardent supporters of
craft beers to maintain this ardor in support of beers whose “primary flavor tone is cardboard” (to
quote Meilgaard).
Moreover, corrections are not always expensive. A particularly elegant “fix-up” can be seen in
the pilot brewery at Coors in Golden, Colorado. The once 100% copper brew kettle now has a
stainless steel tube leading from the original kettle inlet at the top of this vessel to its bottom.
What was once a splashing flow was replaced with a gentle laminar fill. I do not know how
much the modification cost, but it clearly is a minute fraction of the price of the kettle.
Detecting Oxidation
The amount of headspace air in bottled beer is probably the single most important data point to
determine in trying to control oxidation, particularly for home brewers, and also happens to be
the easiest. For commercial brewers producing bottled beer, such monitoring is essential.
Measuring headspace air: The most cost-effective, and the most widely used, instrument for
measuring headspace air levels in bottled beer is the Zahm & Nagel (Buffalo, New York) air
tester. Their 7000 series model provides carbon dioxide levels at the same time air levels are
measured, and the instrument itself is trivial to operate. The device pierces the cap, and CO 2 in
the beer pushes head-space gases into a caustic solution. The solution entrains CO2, but allows
air (oxygen and nitrogen) to escape into a burette, where its volume is measured. (As a rough
rule of thumb, the oxygen fraction of air is around 21 %.) This unit costs $ 900, which may be
steep for the average home brewer, but not necessarily for the larger homebrew clubs. As noted
above, an instrument like this is essential for commercial breweries.
It is important that headspace air be measured as soon after filling as possible, because it will
gradually be absorbed into the beer and react in the ways cited in Table I (page 60). In fact,
measuring attenuation of air levels as the beer ages gives a nice way of tracking its evolution by
way of staling. One important exception is a situation in which headspace air levels increase
owing to air ingress through imperfectly sealed bottle crowns (or PET bottles). This can
dominate any effect due to air levels at filling (25). Table II dramatically shows the combined
effects of high headspace air levels and thermal abuse on beer.
When planning a headspace sampling program, keep in mind that neither commercial nor home
bottle fillers fill bottles uniformly. Samples should be taken from the middle of the bottling run,
from bottles that foamed over properly when filling.
Measuring dissolved oxygen: It is also important to determine dissolved oxygen levels during
beer processing on the cold side. Respectable dissolved oxygen (DO) meters cost $400–$ 500.
The recently published Laboratory Methods for Craft Brewers gives an excellent set of protocols
for taking these measurements (26). As noted above, oxygen pickup during filtration is a special
concern. It has been my experience that it is possible and desirable to keep dissolved oxygen in
beer below 0.05 mL/L during processing.
DO meters are not suitable for measuring HSA, since on the hot side dissolved oxygen will react
with wort constituents before it can be measured. The classical Indicator Time Test (ITT) is a
good way to measure oxidation caused by HSA. The test measures the time, in seconds, for a
solution of dichloro indophenol (DCI) to become discolored when beer is added. DCI exists in a
highly oxidized state and will react with a color change when combined with a substance in a
lower oxidation state. The speed of the color change is a measure of the substance’s degree of
oxidation and varies from seconds for highly reduced solutions to hours for those having a
significant amount of oxidized components. Easy-to-follow protocols along with criteria can be
found in references 27 and 28. Special note should be made of the improved media recently
recommended for this procedure (29).
Stress tests: In my own brewing, I have found that stress tests for beer have been the most
effective means of detecting CSA and HSA. My procedure requires the knowledge of beer air
levels at fill. Samples are held at 86 °F (30 °C), and are tasted twice a month. Using the
information in Tables II, III, and IV, it is possible to approximate the source of the oxidation.
Find out What Works for You
Discussion of oxidation has been bedeviled by sweeping generalizations ranging from “The
flavor stability of beer is determined in the brew house,” to “No other oxidative effect is as
important as headspace air.” My own work, especially reference 30, has been sharply criticized,
perhaps with some justification, for overemphasizing HSA. My posture came about primarily
because of the dramatic effect that I observed when I eliminated HSA. In particular, the
consistency of the performance of my beers in competitions dramatically improved when I
eliminated HSA from my brewing process. Others may have a different experience. The best
way to deal with real and relevant problems associated with oxidation is to get the necessary data
from your own beer so you can determine for yourself what you need to do in your own system.
