Dimethyl Sulfide

Dimethyl sulfide (DMS), sometimes spelled "dimethyl sulphide" , is the simplest type of thioether , which are sulfur containing oils that are generally considered off-putting in beer . The flavor and aroma of DMS have been characterized as being like cooked sweetcorn, tomato sauce, celery, or sauerkraut. In beer, it is sometimes confused with methyl thiocetate, ethanethiol, and dimethyl trisulphide. DMS in beer originates from malt-derived precursors, S-methyl methionine (SMM) and dimethyl sulphoxide (DMSO), and to a lesser extent can be formed during fermentation by certain microbes . Small amounts of DMS have also been found in hops, which is volatilized during boiling . The flavor threshold of DMS is 30-50 µg/liter. Low levels above threshold between 30-100 µg/liter are considered acceptable and even beneficial to some lagers in the United Kingdom (but not in Germany) (~25 minutes in), and maybe some traditional farmhouse ales that are not boiled (see Considerations for Historical Examples of Raw Ale). However, amounts above 100 µg/liter are generally considered offensive for any beer. Ales typically have below the flavor threshold of DMS . The basis of the understanding of DMS and it's creation in beer was uncovered in the late 70's and early 80's. DMS is a common compound found throughout nature, including having an importance in cycling sulfur in ecosystems involving algae and other microbes, helping the navigation of seabirds, and is found in many foods such as corn, cabbage, parsley, asparagus, potatoes, beef, Camembert cheese, fish (carp), tea, cocoa, milk, wine, rum, beetroot, black truffles, and seafood .

Production from Malt

While malted barley contains small but not insignificant amounts of DMS (more than 10 ppm) , the primary source of DMS are the precursors S-methyl methionine (SMM) and dimethyl sulphoxide (DMSO), both of which are present in malted barley . When listed on malt analysis sheets they are usually listed as a combined value as "DMS-P" or "DMSP". This value should be between 5-15 ppm for pilsner malts, and less for fully modified malt .

SMM Precursor

The primary source for DMS in beer (as well as cooked vegetables) is caused by the decomposition of SMM into DMS. This decomposition is caused by heat above ~80°C. Levels of SMM in raw barley are initially low, but as the barley is malted the SMM precursor is formed inside the malt. Many factors influence the amount of SMM found in malted barley. SMM amounts are correlated with nitrogen amounts. The longer the barley is stored before malting, the more SMM will be produced. The majority of SMM in malted barley, however, is determined by how the malt is kilned. During kilning temperatures above 70°C the SMM is partially broken down into DMS and homoserine (isothreonine). Some of the DMS is driven off by the high temperatures of kilning due to it's high volatility, but the DMS present during kilning can also be oxidized into DMSO. The lower the temperatures are during kilning (such as for pilsner malt), the more the SMM precursor is retained in the malted barley .

Mashing and Boiling

During mashing, small spikes of DMS have been reported. This has been proposed to be due to the volatility of DMS existing in the malt rather than being converted from SMM (mash infusion temperatures are too low to convert significant amounts of SMM into DMS). When mashing in a closed system, evaporated DMS condenses and falls back into the mash. The small amount of DMS that is produced during the mash is volatilized by the early stages of boiling. Decoction mashing also introduces DMS due to the boiling of the mash and the resulting conversion of SMM into DMS. SMM from the malt is easily dissolved into the wort during mashing .

Boiling and cooling have the most effect on levels of DMS in beer. At boiling temperatures, SMM is decomposed into DMS. Wilson & Booer showed that SMM's half-life is about 35 minutes at a pH of 5.4, meaning that it takes ~35 minutes to reduce half of the SMM present into DMS . pH plays a role in the reduction of SMM to DMS, with a higher pH reducing the half-life of SMM. Dickenson showed that at a wort pH of 5.2, SMM had a half-life of 38 minutes, but at a pH of 5.5 the SMM has a half-life of 32.5 minutes . It has long been reported that the half-life of SMM doubles for every 6°C cooler, meaning that at 95°C the half-life is ~70 minutes (see the table below) . If the wort is held at a perfectly uniform temperature (which may not reflect real brewery conditions) then the half-life doubles more quickly as the wort cools . During the boil, the converted DMS is evaporated off due to its low boiling temperature of 37.3°C and the convection currents of the boil.

The largest contribution of DMS from SMM is after boiling the wort, and during the chilling process. SMM continues to breakdown into DMS after boiling and before the wort is completely chilled. DMS formed during this time is mostly retained in the wort due to the wort being still, especially in a closed cooling system where evaporation is prevented completely. Once the wort reaches a temperature of 80-85°C, the decomposition of SMM into DMS is greatly reduced . It has been shown that a longer boil will help decompose the SMM and drive off DMS , however if the level of SMM in the malt is high (3-8 µg DMS equivalents/g malt) and more than 50 µg DMS equivalents/liter of SMM survives the boil, then reducing the time in the whirlpool where the wort sits above 80°C can help reduce the amount of DMS in the finished beer. SMM that is not decomposed into DMS during the boil/whirlpool and survives going into the fermenter is not metabolized by yeast, but is also not decomposed into DMS (typical brewing conditions result in little SMM going into the fermenter) .

Temp°C	SMM half-life at ph 5.2 (min)	SMM half-life at ph 5.5 (min)
100	38	32.5
94	76	65
88	152	130
82	304	260
76	608	520
70	1,216	1,040

DMSO Precursor

Dimethyl Sulphoxide (DMSO) is the second precursor to DMS, and is also present in malted barley. Conversion of DMSO to DMS in beer is a function of microbial activity. DMSO is formed in malted barley during kilning at temperatures above 60°C (ale wort can contain more DMSO than lager wort because of this ). Drying the green malt before kilning also increases DMSO (and SMM). DMSO is readily dissolved into water during mashing, and with a boiling point of 189°C, it survives mashing and boiling temperatures. Wort generally contains 200-400 µg of DMSO per liter, with wort made from higher kilned malts containing more DMSO .

Saccharomyces species convert less than 25% of DMSO into DMS as a side effect of an enzyme whose primary function is to reduce methionine sulfoxide to methionine . In a lab setting with simple glucose-salts and DMSO added, ~13% of DMSO is converted to DMS. However, in wort only ~5% of DMSO is converted to DMS , which generally equates to about 5-10 µg/L of DMS . The percentage of DMSO that is converted to DMS does not change as DMSO levels increase, so although low percentages are converted, high amounts of DMSO can still contribute significant DMS. With high levels of DMSO in the wort, a slight increase in DMS from DMSO precursor can be observed towards the end of fermentation from yeast metabolism. This increase in DMS from yeast metabolism has been observed during the conditioning of fermented beer and surprisingly under cold temperatures (0°C in one report), so if yeast is left in the beer then it can convert DMSO to DMS in the packaged beer .

Yeast species/strain, temperature, pH, wort composition, and open/closed fermentation vessels contribute to how much DMSO gets converted into DMS. For example, S. uvarum (potentially reclassified to S. bayanus) produces less DMS than S. cerevisiae, as does S. pastorianus . DMSO is converted to DMS by yeast more readily at lower temperatures than warmer temperatures with five times as much at 8°C than at 25°C. Higher gravity worts (1.033 vs 1.060 in the linked reference) also produce more DMS from DMSO during fermentation. A higher pH of wort also leads to more DMS production; for example lager wort pH is typically 5.4-5.7, while ale wort pH is typically ~5.1. This might explain why DMS is present more in lager beers .

Spoilage Organisms and Spontaneous Fermentation

Many types of microbes are capable of producing DMS from DMSO as a secondary metabolite of fermentation. Microbes that can produce high amounts of DMS include gram-negative, facultative anaerobes in the Enterobacteriaceae family, which includes species of Klebsiella, Citrobacter, Enterobacter, Obesumbacterium, Proteus, Salmonella, and Escherichia, as well as gram-negative aerobic bacteria such as Pseudomonas aeruginosa . Gram-positive bacteria can also produce high amounts of DMS, such as Bacillus subtilis . Many other bacteria such as species of Clostridium, Streptococcus, and Staphylococcus produce only small amounts of DMS (much less than S. cerevisiae even) . All bacteria that can produce DMS from DMSO do so using a different enzyme than yeast, which might account for the ability of some bacteria to convert a higher percentage of DMSO to DMS than S. cerevisiae. The DMS production by facultative anaerobic bacteria is encouraged by the lack of oxygen .

In lambic production where the pH of the wort is not lowered to less than 4.5 before entering the coolship for spontaneous fermentation, Enterobacteriaceae are responsible for high amounts of DMS production. No DMS was found in the referenced study before the wort was cooled in the coolship, which might be due to the lengthy boil of the wort due to the turbid mash. After two weeks of fermentation, 450 ppb of DMS were found, far more than the 30 ppb taste threshold, and the vegetal aroma of DMS could be detected during the fermentation at this time. After two weeks the fermentation of Saccharomyces begins, and the DMS levels decline due to the formation and blow-off of CO₂. At 6 months the DMS was down to 100 ppb, and a range of 25-75 ppb of DMS found in bottles of lambic (and at 16+ months), which is a typical amount for regular ales and lagers .

Volatility of DMS

DMS is a very volatile compound. Scheuren et al. (2016) determined that there is not a significant difference in DMS evaporation in water versus wort, and came up with equations for determining the evaporation of DMS in water using the laws of thermodynamics. They presented a somewhat counter intuitive result that DMS actually volatilizes more readily as temperatures drop, until about 50°C (volatilization of DMS drops significantly under 50°C). Their calculations state that 3.2% of the total wort volume needs to be evaporated for 90% of the DMS to volatilize at 100°C, whereas only 1.3% of the total wort volume needs to be evaporated for 90% of the DMS to volatilize at 80°C (keep in mind that the rate of evaporation at 80°C is much slower, and thus it takes more time to reach 1.3% evaporation). This indicates that some amount of DMS is evaporating off at temperatures below boiling until 50°C is reached .

They also established that the volatility of DMS is the same regardless of the gravity of the wort, and that it is instead effected by temperature, atmospheric pressure, and the concentration of DMS (higher concentration of DMS slightly raises the volatility of DMS). For non-wort solutions, 10% sucrose in water greatly increased the volatility of DMS, possibly due to a salting-out effect of sucrose which leads to a higher evaporation rate for DMS molecules . A larger top surface area will allow for faster evaporation of the total DMS present in the wort, but the total DMS present in the wort would eventually be evaporated off regardless of what the top surface area of the kettle is . In order to limit DMS in the end product, it is advised to allow no more than 100 µg/L of DMS into the fermenter .

Much of the DMS in wort from the SMM precursor is volatilized off during fermentation due to off-gassing of CO2. However, if high amounts of DMS survive the boil then off-gassing from fermentation may not be enough to volatilize all of the DMS. Shape and type of the fermenter also play a role in how much DMS is volatilized during fermentation, for example Anderson et al. and Booer & Wilson showed that open fermentation leads to less DMS production compared to closed fermentation . Higher fermentation temperatures (18°C versus 9-12°C, for example) can lead to higher rates of DMS volatilization . DMS can spike towards the end of fermentation from yeast metabolizing DMSO into DMS (see DMSO Precursor) .

Short Boils and Raw Ale

Raw ale, also referred to as "no-boil", is a method of wort production that involves not boiling the wort, or perhaps by some definitions, very short boils . Although mainly a historical method of brewing, this style of brewing has recently become popular in the production of Berliner Weissbier and other styles of beer using sour worting or kettle souring methods. Many recipes for these styles of beer call for pilsner malts to be used, which can contain higher amounts of SMM precursor. An often asked question about no-boil/raw ales and wort boiled for 15 minutes or less is: are there concerns about DMS production?

Anecdotal reports of no issues with DMS in these types of beers seem to far outweigh the reports of DMS problems . The specific nature of (or lack of) DMS detection in no-boil/raw ale has not been widely explored by science. There are, however, some explanations that have been proposed. For example, when boiling smaller volumes of wort such as on the homebrew scale there is a larger surface area to volume ratio. This larger surface area to volume ratio allows for more evaporation and volatilization of DMS to occur (~30 minutes in). Smaller fermenters would also benefit from a larger surface area to volume ratio since CO2 from fermentation volatilize DMS. This may account for the general lack of DMS reported in homebrewed and small-scale farmhouse beer.

In sour beer, there might be other compounds that make the detection of DMS more difficult. For example, 2-phenylethanol and phenethyl acetate mask the perception of DMS in beer . Additionally, some tasters might be genetically predisposed to perceive the flavor of DMS more easily than others.

Considerations for Historical Examples of Raw Ale

In the case of raw ale, and particularly Norwegian/Latvian/Lithuanian traditional farmhouse ales, Finish "sahti", and Estonian "koduõlu", there is some debate as to whether or not DMS should always be considered an off-flavor. Traditionally these beers were made with lightly kilned malts that were malted by the brewers themselves (up until about 20 years ago), and these malts may have had high levels of SMM precursor. Lars Marius Garshol offers his philosophy on off-flavors in beer in general, which is inspired by writings of Michael Jackson, and makes an argument that DMS may have been considered desirable or acceptable in farmhouse raw ales brewed in certain regions of Europe. especially considering that the flavor of DMS is desirable in other foods .

Additionally, it has been suggested that small amounts of DMS in wine can give the wine a pleasant "jammy" character. High amounts still lead to vegetal off-flavors in wine. This may or may not have a similar effect for sour beers with fruit .

Avoiding DMS

If the brewer is experiencing unwanted DMS in no-boil/raw ale/short boiled beers:

• With the lid on the boil kettle ("closed system"), avoid allowing wort to stand between 80-100°C/176-212°F (or between 80°C and your area's boiling point). If the lid is off, DMS will continue to evaporate even at lower than boiling temperatures due to its thermodynamic properties, and less will be retained.
• If the wort is allowed to stand in the above mentioned temperature range in a closed system, boil the wort rigorously for a few minutes afterwards, and then quickly cool it below 80°C (176°F).
• Use open and/or shallow fermenters .
• Increase fermentation temperature .
• Allow the beer to age longer, particularly if it contains Brettanomyces. Studies in lambic brewing have shown that DMS will volatilize over time if left in the fermenter.
• Use more highly kilned malts such as 2 row instead of pilsner malt.

DMS Prediction Models

Equations have been established for estimating how much DMS will be converted from SMM during boiling, and how much SMM will be converted to DMS during cooling the wort after the boil. These equations are well explained in "Principles of Brewing Science: A Study of Serious Brewing Issues", by George Fix, 1999, Pgs 89-93. One thing to keep in mind is that the equations assume "typical" brewing practices, which include boiling the wort at some point at a pH of 5.2 - 5.5. Because of this, the equation for how much SMM is converted to DMS during cooling may not reflect no-boil or kettle sour wort accurately. Let's look at some examples of the equations from "Principles of Brewing Science: A Study of Serious Brewing Issues":

George Fix Model

First let's look at an example that should accurately predict how much SMM is in a given batch of wort, and how much of that SMM is converted during an 82°C no-boil pasteurization rest at 15 minutes. Assuming a given pilsner malt has 5 µg of SMM per gram, and the malt concentration (weight of malt per liter of water) is 200 g/L, the total SMM content of the malt can be calculated:

Total SMM = 5 µg/g x 200 g/L = 1,000 µg/L

The half-life of SMM at 100°C is ~40 minutes, but since we are not boiling then the half-life of SMM at 82°C is ~300 . Time also has to be considered, but since the heating up time period would make the equation unwieldy, it is suggested to add the heating up time to the "boil" time (or for us, "pasteurization" time) . If it takes 20 minutes to heat up the wort, and it is held at 82°C for 15 minutes, then that gives us 35 minutes. With the time and half-life values, we can predict how much SMM will survive the heat pasteurization and how much will be converted to DMS:

SMM left over = 1,000 µg/L x 2-(35 min/300 min) = 1,000 x 2-0.12 = 1,000 x 0.92 = 920 µg/L

and

DMS created = 1,000 µg/L - 920 µg/L = 80 µg/L

As expected, a high amount of SMM survives (920 µg/L) the 82°C heat pasteurization, and a relatively low amount of DMS is created (80 µg/L). The example in "Principles of Brewing Science: A Study of Serious Brewing Issues" uses the same total SMM value for the malt, but with a 90 minute boil 210 µg/L of SMM is left over after the boil, and 790 µg/L of DMS is created during the boil. During boiling the created DMS is evaporated off, but during heat pasteurization the DMS is retained in the wort.

The next equation determines how much SMM is converted to DMS during cooling, and this amount is considered more important because it is mostly not evaporated off (especially in a closed cooling system). This equation, however, uses an average between the boiling temperature and the final chilling temperature. This model has proven to work well assuming normal brewing procedures which assume the wort is boiled, but cannot be used for wort that is not boiled. This is because the half-life of SSM is doubled for every 6°C cooler (see the SMM half-life table above).

An example will help demonstrate this issue. Suppose the wort cools from 82°C to 20°C over 60 minutes. Using the accepted set of equations that predict how much SMM is converted to DMS during cooling, first an average between the starting temperature and the final temperature is computed:

Average temperature = (82°C + 20°C)/2 = 51°C

then, using 4,000 as a constant used in the equation, a time-dependent differential equation is used:

Time differential = (60 min x 51°C)/4000 min) = 0.765

and finally the 0.765 number is used to determine how much SMM is left over:

SMM left over after cooling = 920 µg/L x 2-0.765 = 920 µg/L x 0.588 = 541 µg/L

and the amount of DMS created in the wort during cooling:

DMS created during cooling = 920 µg/L - 541 µg/L = 379 µg/L

now add the DMS created during the heating and 82°C pasteurization to get the total calculated DMS:

Total DMS = 80 µg/L + 379 µg/L = 459 µg/L

Using the exact same wort composition but with a 60 minute boil, this example in "Principles of Brewing Science: A Study of Serious Brewing Issues" computes only 92 µg/L of DMS, mostly because a lot more of the SMM is converted to DMS during boiling, which is then volatilized during the boiling, and leaving less SMM to convert to DMS during cooling. Even with the higher SMM during cooling in our heat pasteurized wort example, that hardly seems fair considering that the half-life of SMM is ~300 minutes at the pasteurization temperature of 82°C.

Numerical Modeling Using Updated DMS Volatility Data

Mark Hammond from MTF used a computer program to model the conversion of SMM to DMS based on SMM half-life at any given temperature. Rather than using Fix's average half-life approach, numerical modeling divides the heating, boiling and cooling times into very small time steps (for the work below, a time step of 0.6 seconds was used), during which the temperature is approximated to be constant. The computer program calculates the amount of SMM converted to DMS during the time step, the amount of DMS volatilized during the time step, then plots the total DMS and SMM in the wort at the end of the time step. Finally, the program calculates a new temperature, which depends on whether the wort is heating up, being held at a constant temperature, or cooling. The program then loops through all the calculations at this new temperature, calculating all the same quantities for the next time step.

The new equations from "Scheuren, Baldus, Methner and Dillenburge (2016): Evaporation behaviour of DMS in an aqueous solution at infinite dilution – a review" were used to determine the evaporation of DMS during the cooling time (this assumes an open cooling system; closed cooling systems will retain this DMS). Hammond assumed a linear heating rate and used Newton's Law of Cooling with constants based on empirical data taken from his own homebrewing equipment. Evaporation rates were likewise modeled using thermodynamic equations and empirical data from Hammond's equipment. By observing these estimations, it can be seen that no-boil or "raw ale", and wort boiled for short durations, Hammond predicts less DMS than what is predicted using the traditional model.

For the DMS amounts in the following graphs, Hammond calculated the mass of DMS to be 62/164 of a gram of DMS for every gram of SMM decomposed. Since we get one molecule of DMS (62 g/mole) from each molecules of SMM (164 g/mole), we don't get one for one mass of DMS for SMM. Keep that in mind when comparing the decline in the SMM concentration graph to the DMS concentration graph .

SMM conversion to DMS during a 20 minute heat up, 15 minute pasteurization at 82°C, and 60 minute cool down to 20°C. Graph created and provided by Mark Hammond.

In the above computer generated graph based on the half-life rates of SMM, only ~60 µg/L of SMM is converted into very little DMS. Even if a closed cooling system were to be used, only ~6 µg/L of DMS would be retained. This demonstrates the very slow decomposition of SMM into DMS at below boiling temperatures.

Hammond also generated graphs for a 15 minute boil and a "0 minute" boil (wort heated to 100°C, then immediately cooled). These graphs are seen below:

SMM conversion to DMS during a 20 minute heat up, 15 minute boil at 100°C, and 60 minute cool down to 20°C. Graph created and provided by Mark Hammond.

SMM conversion to DMS during a 20 minute heat up to boiling temperature (100°C), then immediate cooling for 60 minutes to 20°C ("0 minute" boil). Graph created and provided by Mark Hammond.

In the case of the 15 minute boil, approximately 350 µg/L of SMM is converted into about 40 µg/L of DMS. 15 minutes of boiling appears not to be long enough to efficiently boil off the created DMS, however in an open cooling system much of the DMS will still be evaporated during cooling (see DMS Volatility). Even if the wort was cooled in a closed system and this amount of DMS was retained in the wort, 40 µg/L of DMS is still below the recommended threshold of 100 µg/L of DMS that should be allowed to enter the fermenter .

In the case of the "0 minute" boil, approximately 175 µg/L of SMM is converted into approximately 20 µg/L of DMS. Just as in the other examples, the DMS continues to volatilize below boiling temperatures in an open cooling system (see DMS Volatility). Even if the wort was cooled in a closed system and this amount of DMS was retained in the wort, 20 µg/L of DMS is still below the recommended threshold of 100 µg/L of DMS that should be allowed to enter the fermenter .

For comparison sake, a graph of a 60 minute boil is shown below, again with an "open cooling" system that allows continued evaporation of DMS during the cooling period:

SMM conversion to DMS during a 20 minute heat up, 60 minute boil at 100°C, and 60 minute cool down to 20°C. Graph created and provided by Mark Hammond.

Kettle Souring and Effects of pH

Normal procedure for kettle souring techniques is to boil the wort a second time after the pH has been lowered in order to kill the Lactobacillus souring culture. The lactic acid fermentation by the Lactobacillus generally results in a wort pH of around 3.0 - 3.6, at which time the wort is boiled to kill the Lactobacillus. In this case yet another consideration is the effect of low pH on the decomposition of SMM into DMS. The half-life table above demonstrates that a decrease of -0.3 pH increases the half-life by 5.5 minutes at 100°C. Unfortunately we do not have data to show how the much lower pH values that are achieved after souring wort will effect the half-life of SMM , but assuming the effect is linear then the conversion of SMM into DMS during the second boil would be greatly reduced. Assuming the effect of pH on SMM half-life is linear, and that every -0.3 pH increases the half-life by 5.5 minutes at 100°C, the half-life table could hypothetically be updated to include data that looks something like what we present in the first column below. This indicates that wort boiled after souring probably does not create significant amounts of DMS from SMM.

Temp°C	SMM half-life at ph 3.4 (min - may not be reality)	SMM half-life at ph 5.2 (min)	SMM half-life at ph 5.5 (min)
100	71	38	32.5
94	142	76	65
88	284	152	130
82	568	304	260
76	1,136	608	520
70	2,272	1,216	1,040

Updated Prediction Models

Updated DMS prediction models have been proposed in the scientific literature. These are largely based on thermodynamic laws.

• Scheuren, Baldus, Methner and Dillenburge (2016): Evaporation behaviour of DMS in an aqueous solution at infinite dilution – a review.
• Scheuren, Tippmann, Methner, Sommer (2014): calculation for wort boiled below atmospheric pressure.
• Scheuren, Sommer, and Dillenburger (2015): calculating DMS formation during mashing due to DMS volatility and thermodynamics.

See Also

Additional Articles on MTF Wiki

• Sour Worting
• Spontaneous Fermentation
• Isovaleric Acid
• Butyric Acid

External Resources

• "Update: Lab Data on Pils Malt Boil Length Exbeeriment" on Brulosophy.
• "Raw ale" article on Larsblog by Lars Marius Garshol.
• "All Grain Pale Ale 30-Minute Boil Experiments" by James Spencer on Beer & Wine Journal.
• "A pretty corny post" on Beer Sensory Science blog.