A CERprisingly good way to monitor fermentation performance
Updated: Jan 30, 2021
Conventional ways to measure fermentation performance
Most brewers use gravity, plato, or brix to measure the amount of sugars extracted from the malt during mashing (or added as extract/adjunct). Brewers usually also use these methods to measure fermentation progress. Gravity is simply the amount of dissolved solids in the mash water; in the wort this would mostly be saccharides. It is important to remember that gravity is measuring the density of the wort - an original gravity (OG) of a 1.050 means that 1 mL of that wort at STP (standard temp and pressure) would be 1.050 grams, which is equivalent to 5% heavier than the same volume of water. This density is relative to the water (prior to mashing) and is temperature dependent. You can measure the gravity or density of your wort using a hydrometer (most commonly used by brewers) or a balance (if you can measure down to a precision of ~2 mg or less).
Plato is another measure of the extract content of your wort and is a measure of the dissolved solids in a solution by weight; a 12.5 degrees Plato (ºP) wort corresponds to 12.5% dissolved solids (an OG of 1.050). Brix is a similar measure to Plato and is usually taken with a refractometer - a device that measures the content of sugar in a liquid sample by the refraction of light, since the density of a solution increases proportionately with its refractive index. Brix and degrees Plato are close but not quite equivalent for several reasons, but in part because brix refractometers are calibrated for sucrose (a disaccharide of glucose and fructose), which makes up a relatively small fraction of brewing wort with the other sugars having a different influence on the refractive index.
When a brewer measures the decrease in the gravity (density), Plato (% weight dissolved solids), or Brix (sugar content) during the fermentation, she/he is measuring the decrease in the density, % weight of dissolved solids, or sugar content of the wort. Density of the wort at any point in time isn't a measure of instantaneous activity because it isn't a per unit time measure. One way to solve this is to measure gravity point decrease (or Plato decrease) per unit time (hour or minute). This rate (dissolved solids consumed / unit time) would then be a proxy of activity in the fermentation process at any given point in time. However, measuring this can be tedious if using a hydrometer (not to mention potentially wasteful at the homebrewing scale) or expensive if using a Tilt hydrometer (~200 USD).
What is CER and how can we use it to measure fermentation performance?
Another method one could use to get an instantaneous measure of fermentation activity, which has the benefit of also being less invasive, is measuring carbon dioxide evolution rate (CER, or sometimes called carbon dioxide transfer rate CTR) in the off-gas from the fermenter. To fully explain what this is, we need to go back to basics about fermentation and microbiology.
Brewing yeast ferment 1 mol of glucose (a 6 carbon monosaccharide) to 2 mol of ethanol (a 2 carbon alcohol) and 2 mol of carbon dioxide (a 1 carbon naturally occurring atmospheric gas). One of the outcomes when we measure the CO2 evolved from a fermentation is that we can accurately estimate the amount of ethanol formed AND the amount of sugars consumed (in mol). Measuring total CO2 produced would be akin to tracking gravity (but sort of in reverse because CO2 is accumulated rather than consumed). However, if we track the instantaneous rate of CO2 production (CER/CTR), we can closely estimate the instantaneous rate of ethanol formation and sugar consumption. Measuring CER/CTR is essentially measuring the instantaneous rate of fermentation. Lots of professional big time breweries already do this but most homebrewers don't.
In Saccharomyces fermentations, CER/CTR tracks very closely with growth rate, glucose consumption, and metabolic activity in general (Scalcinati et al 2012 and Anderlei et al 2004 have good figures showing this). You can see in the figure below (from Anderlei et al 2004) that the CER/CTR increases rapidly during fermentation as glucose is consumed and ethanol is formed and drops very rapidly when the glucose is depleted. This is what makes CER/CTR so useful for tracking fermentation and metabolic activity. In engineering, CER/CTR is known as a good instantaneous measure of metabolic activity and is used to predict, with very high accuracy, when a culture has consumed the available substrate and entered stationary phase (i.e. starvation, see Sauvageau et al 2010 for an example of this). As an instantaneous measure of metabolic activity, CER/CTR is also more dynamic than gravity changes - CER can vary orders of magnitude whereas gravity measures do not.
How to measure CER; measurement assumptions and caveats
I use a small, very accurate CO2 sensor produced by sensirion (SCD30) connected to a microcontroller (arduino uno) and then to a raspberry pi. This sensor is very accurate (+/- 30 ppm!), very sensitive (0ppm to 40,000ppm) and very inexpensive (~65 USD). The sensor uses nondispersive infrared to measure CO2 and can be used to measure even atmospheric levels of CO2 (~410 ppm). The sensor is mounted to the side well of my fermentation fridge where my fermentation vessel (FV) sits. The fridge is (nearly) a closed system in terms of gas flux and of a known volume - so I can measure not just CER but also the total CO2 produced during fermentation.
What assumptions do I make during measurement? Well, the first is that CO2 measured above baseline (~600 ppm in my flat in Vienna) arises solely from fermentation. This is a pretty good assumption. I also assume that the inside of the fridge is a well mixed environment in terms of gas dispersion. Over longer time scales (hours), this is pretty close to true, but over the span of minutes it isn't quite true. However, if I mount the sensor in the same spot every time and put my FV in the same place (and don't change the volume of the bitter wort) then it doesn't really matter if my absolute measurements aren't 100% accurate, they will still be comparable from fermentation to fermentation on a relative aspect. So my CER measurements on a very short timescale may not be absolutely correct but they are relatively comparable between batches.
What the setup looks IRL
The raspberry pi sits on top of my fermentation fridge, connected to an arduino uno via usb cable (that's how the arduino is powered). The sensor is connected to and controlled by the arduino. The raspberry pi uploads the instructions to the arduino uno and logs the data. I use the SparkFun SCD30 library that you can find in the library manager within your arduino IDE.
The SCD30 sensor is simply taped to the side of my fermentation fridge, relatively close to the airlock. The sensor is always taped to the same place and the fermentation vessel (FV) is also always placed in the same location (the fridge is almost the same width as the FV, so there isn't much extra width).
A closeup of the sensor next to the airlock. The sensor is pretty small, not much larger than a US quarter coin (or a Canadian loonie). Tape serves as a very secure, expensive, and advanced method of mounting the sensor near the FV.
The arduino uno controlling the SCD30 CO2 sensor. Connecting them is as simple as this:
1) connect the 5V pin on the arduino to the VIN pin on the SCD30,
2) GND pin on the arduino to the GND pin on the SCD30,
3) analog pin 4 (SDA) on the arduino to the SDA pin on the SCD30, and
4) analog pin 5 (SCL) to the SCL pin on the SCD30. That's it!
Potential criticisms of CER
Is CER better than measuring gravity or brix/plato (or can I replace gravity/brix measurements using CER?)? I don't think so. CER is a different proxy for measuring fermentation progress and tells the brewer something different than a drop in gravity. However, it does have advantages. First, it isn't invasive - so you don't waste any beer. Second, it's an instantaneous rate and not an absolute number so it tells you how "fast" the fermentation is going right now with one measurement, whereas with gravity you need two consecutive measurements to calculate a rate. Some may say this is similar to a tilt but the CER tells you something different. Even after you achieve terminal gravity, metabolic activity continues as yeast consume diacetyl and other byproducts (like acetaldehyde). The common 1-2 day rest after reaching terminal gravity is due to this and with CER, you can measure even these subtle changes in metabolic activity. Once the CER drops below a certain level, you know the fermentation is really done because the yeast aren't doing any more fermentation or respiration.
One criticism brewers may have is "but doesn't CO2 continue to offgas from the green wort for a while after fermentation?". This doesn't really happen if the temperature doesn't change; at the beginning of the fermentation the wort has only a bit of CO2 dissolved in it (corresponding to the partial pressure of CO2 in the atmosphere above the wort). During the fermentation, the wort becomes saturated with CO2 and any excess that can't dissolve bubbles out. Once fermentation is done, the solution remains around saturation and the CO2 doesn't partition out of solution unless you change the temperature (thereby changing the solubility of the gas in your green/bitter wort). So no, this is not a large downside. However, be aware if you heat your wort (for whatever strange reason), then you will observe more CO2 being emitted. However, I should note that this amount of CO2 is still much, much lower than produced during biological fermentation.
Analysis of my own CER fermentation data with a lager fermented using w-34/70
So, let's look at some actual data! Here is the CER plotted over time along with gravity on a Munich Dunkel I did using w-34/70 Saflager from Fermentis at 14 degrees Celsius (with a ramp to 16 Celsius by 75 hours for the vdk rest):
The starting gravity in this fermentation was 1.046; the gravity is depicted in closed red circles and the CER corresponds to the closed blue circles. In the first hours of the fermentation, the Saccharomyces are in lag phase, allowing their membranes to adjust to the sugary liquid wort, synthesizing the essential proteins needed to import and ferment sugars, and starting to prepare for asexual reproduction. By ~13 hours, there is already a detectable drop in the gravity and the yeast have likely begun to divide, grow, and ferment sugars. The fastest drop in gravity per unit time is apparent at about 25-50 hours after inoculation - this also likely corresponds to the fastest sugar utilization/fermentation. Sugar consumption is starting to slow down by 60 hours and the fermentation hits terminal gravity at ~75 hours and doesn't change any more at 104 and 121 hours (please note that the gravity measurements at ~104 and 121 hours are somewhat obscured behind the blue CER points).
The carbon dioxide evolution rate (CER; in mol CO2/min) looks great here. The CER is very low during the first 12 hours (lag phase) and rapidly increases as the yeast start to ferment sugars in the wort. The peak CER corresponds exactly to the peak gravity consumption (31-55 hours) and drops very rapidly as the fermentation nears FG. This data looks great! I'd even venture so far as to say that the CER is more sensitive than gravity measurements and reflects a clearer picture of the activity of the yeast at any point in the fermentation. For example, terminal gravity (1.010) is reached at 78 hours but the CER is still elevated above baseline. This is likely due to the yeast doing the final cleanup at the end of the fermentation and consuming intermediates that leaked out of the cells during the fermentation; the best known example of this is likely alpha-acetolactate which is an intermediate in the synthesis of keto-acids and amino acids in yeast and leaks out of the cell during active fermentation. The typical diacetyl rest would be 24-48 hours at an "elevated" temperature (I ramped to 16 Celsius) to allow the yeast to clean up. And this is reflected in the CER data - 24 hours after we hit FG the CER now drops to baseline levels indicating the the fermentation is finished.
I should say that w-34/70 is an absolute beast - finishing this fermentation within 3 days at 14 degrees Celsius. I should note that I mashed pretty warm here (68 Celsius) to get a more dextrinous wort in an effort to have the 1.010-1.016 final gravity that's appropriate for the Munich Dunkel style. With an apparent attenuation of ~83% for w-34/70, the final gravity would be too low and the beer may have come out too dry if I mashed cooler (at 65 Celsius, for example).
How reproducible is the data?
I ran three fermentations using w-34/70 on worts with an OG of ~1.046 at 14 Celsius (ramped to 16 Celsius for vdk rest) and plotted the data below:
In the top panel, we see the three individual fermentation CER profiles (3 replicates depicted in red circles, blue triangles, and green squares). The CER is plotted on the Y-axis and the time on the X-axis. You can see that the data is very reproducible from one fermentation to another when the yeast, gravity, and temperature stay the same. If we combine the three replicates and smooth the line using local regression fitting (loess method), the data looks very consistent and the confidence interval (in grey) is rather small! These lagers, using w-34/70 at 14-16 Celsius reliably finish at ~120 hours (5 days). This is also consistent with the Fermentis data (https://fermentis.com/en/rediscover-saflager-w-34-70/).
So far so good, CER looks promising to monitor fermentation. As a last check, I plotted the maximum instantaneous gravity consumption (the maximum slope of gravity decrease) from 13 fermentation runs using 8 different yeast strains vs maximum CER (the peak in CER). All of the fermentation were run using wort of OG 1.042-1.050 and grains were mashed at 65 Celsius. Theoretically, the maximum gravity points per minute consumed (MGPPM) and the maximum CER should be proportional to one another and positively correlated.
These are the results! The blue line is the linear trendline, with the grey area being the 95% confidence interval. Each point corresponds to a wort fermentation (2.5 gallon scale) with a particular yeast, the x-axis is the maximum slope of gravity consumption (max gravity points per minute or MGPPM) and the y-axis is the maximum peak in CER. The R number tells you how well correlated the two variables are and the the p-value tells you if this is a statistically significant (or statistically discernible) result. The R value is very high - 0.88 meaning that the two variables are very linearly correlated and the p-value is 0.000085 meaning that the fit is highly statistically improbable based on random chance if the variables aren't actually linearly correlated. This is an incredibly satisfying result and shows that CER tracks very well with sugar consumption (i.e. drop in gravity over time).
The following 8 yeasts were used for this regression: Mangrove Jacks M44 West Coast ale, Mangrove Jacks M15 Empire ale, Mangrove Jacks M31 Belgian Tripel, Lallemand Nottingham, Fermentis w-34/70, Lallemand Voss kveik, Omega Hornindal blend, and Omega Lutra.
Conclusions:
So, taken together, I would conclude that measuring CER during fermentation is a very sensitive, easy, cheap, non-invasive, and reliable method to get an insight into fermentation progress and yeast physiology. CER also strongly linearly correlates with gravity drop over time. I will be using this method a lot in the future to determine when my fermentations are done and when I compare performance/kinetics between different yeast.
References:
1) Tibor Anderlei, Werner Zang, Manfred Papaspyrou, Jochen Büchs. Online respiration activity measurement (OTR, CTR, RQ) in shake flasks. Biochemical Engineering Journal, Volume 17, Issue 3, 2004, Pages 187-194, ISSN 1369-703X. https://doi.org/10.1016/S1369-703X(03)00181-5.
2) Scalcinati G, Otero JM, Van Vleet JR, Jeffries TW, Olsson L, Nielsen J. Evolutionary engineering of Saccharomyces cerevisiae for efficient aerobic xylose consumption. FEMS Yeast Res. 2012 Aug;12(5):582-97. doi: 10.1111/j.1567-1364.2012.00808.x. Epub 2012 Apr 30. PMID: 22487265.
3) Sauvageau D, Storms Z, Cooper DG. Synchronized populations of Escherichia coli using simplified self-cycling fermentation. J Biotechnol. 2010 Aug 20;149(1-2):67-73. doi: 10.1016/j.jbiotec.2010.06.018. Epub 2010 Jun 25. PMID: 20599574.