title image: MicrobeResearch.Com

Maintaining the CO2 concentration inside a vinyl anaerobic chamber

Image: Vinyl anaerobic chamber, Microbe Research LLC

Vinyl chambers for working with oxygen-sensitive microbes are often operated using a mixed gas comprising between 5% to 20% carbon dioxide. Usually the gas is blended by a commercial vendor and delivered to the lab in cylinders. The cylinders are connected to the airlock of the chamber and become the source of the gas within. As chambers are typically operated the CO2 percentage inside the chamber will be lower than that of the supporting gas mix. How much lower is usually guesswork because CO2 inside anaerobic chambers is seldom monitored.

In the interval separating the setup and first use of a new vinyl anaerobic chamber, I had a chance to monitor the CO2 level within. Over a period of 5 days the chamber lost CO2 continuously. The loss could be explained only by diffusion of CO2 through the vinyl. Intending to maintain a targeted CO2 level, I decided to model chamber operations over time to better understand the physical factors controlling CO2.

Here I review some chamber basics, show outcomes of the model, review why CO2 is used in anaerobic chambers, and describe how I monitor and control CO2. The model projects that loss of CO2 through the walls of a vinyl chamber can lower the maintainable CO2 level by 0.5 to 3.0 percentage points, or even more, depending on airlock usage parameters and composition of the supporting mixed gas.

Dilution of CO2 in the Airlock

Table 1. Airlock vacuum depth selection factors.

An airlock evacuated to a depth of -26 inches of Hg (in. Hg) still contains 13.1% of whatever gas it held at the start. If that were air or purge gas, and the refill was with mixed gas, then the concentrations of CO2 and H2 in the refilled airlock would be 86.9% of their concentration in the mixed gas supply.

Table 1 shows this calculation for different airlock vacuum settings using a gas supply comprising 10% CO2. The middle‑right column shows the maximum CO2 concentration that could be maintained in the chamber via programmed airlock operation were vinyl impermeable to CO2. Operators moving warm, foamout‑prone liquids through the airlock, unless under seal, will be forced into the shallower vacuums.

If the CO2 level in the chamber is lower than what the airlock will deliver, then airlock operation will fortify the chamber. A chamber that is gaining CO2 from each airlock passage could nonetheless see declining CO2 over time owing to the CO2 permeability of vinyl.

Modeling CO2

While monitoring the CO2 level in a new Coy Type‑C chamber, I observed a drop in the CO2 level from 10% initially to 9.15% over five days. The airlock was not used during this time and the chamber was empty save for equipment. Hydroxide or carbonate were not present. The chamber was restored to 10% CO2 and monitored again for 5 days, this time observing 0.2% loss per day, and on a third time also 0.2% per day (average of 0.19%). According to one source (26) vinyl is about 5‑times more permeable to carbon dioxide than to oxygen. These observations prompted me to model chamber operations over time to better understand the combined effect of CO2 loss through the vinyl and CO2 gain through the airlock. The model calculates the percentage of CO2 in the chamber over 60 days, plotting separate lines for 1-, 2-, 3- and 4 airlock cycles per day using the following additional inputs:

  • the starting CO2 concentration in the chamber
  • the CO2 concentration in the mixed gas supply
  • the airlock vacuum depth used

Each day of each plot begins by subtracting an amount of CO2 that was lost by diffusion over the preceding 24 hours. The value of "lost" is extrapolated from the average value I had measured between 10% and 9% CO2 using the formula {% CO2 that was in the chamber yesterday x 0.019}. For example, CO2 in the chamber would be 5.886% the day after it had been 6%, and 5.774% the day after it had been 5.886%. The addition of CO2 from the day's 1- to 4 airlock cycles is then added to arrive at the day‑ending CO2 level [A, B].

Chamber operating projections for CO2

Click on any plot or else here to see all of the projections.

Figures 1b and 1e. Projected CO2 while operating on 10% CO2 supply gas. Figures 1a‑d and 1f show model projections for CO2 levels in chambers operating at airlock vacuum depths ranging from -20 to -26 in. Hg on mixed gas comprising 10% CO2. The starting point of each plot is the concentration of CO2 that could be maintained in the chamber if there were no diffusion loss. As the CO2 concentration declines from diffusion, the rate of diffusion also declines. Since the amount of CO2 gained per airlock cycle remains constant within the plot, a lower CO2 value is approached where daily gain balances daily loss. Solid lines show the concentration of H2 the airlock can deliver at the indicated vacuum setting using mixed gas comprising 5% hydrogen.

Figure 1b. The model projects that a Type‑C chamber being entered twice per day at -22 in. Hg of vacuum would stabilize at about 6.2% CO2 using a mixed‑gas supply of 10% CO2.

Figure 1e. The model projects that in order to maintain 10% CO2 in a chamber that is entered twice per day at -22 in. of vacuum, the airlock would have to be delivering CO2 at 11.9%. To do that, the CO2 concentration in the supporting mixed gas would have to be 16.2%.

The higher the CO2 concentration of the supporting mixed gas supply, the more dramatic is the effect of diffusion loss on the maintainable CO2 level in the chamber. For instance, the model projects that a Type‑C chamber operating at -24 in. Hg of airlock vacuum on mixed gas comprising 20% CO2 could maintain only 14.3% CO2 under duty of 3 airlock cycles per day (plot not shown).

The model assumes an empty airlock and empty chamber of volumes 1.67 ft3 and 31 ft3. Anything that reduces the airlock volume relative to chamber volume should lower the maintainable CO2 level relative to the model's predictions. The act of partially evacuating a chamber and refilling by manual injection of mixed gas to correct a low‑H2 condition, would of course elevate the CO2.

CO2 gained via microbial metabolism

Table 2. Projected CO2 gain from hexose fermentation

Table 2 shows the calculated rise of CO2 level in a chamber from total consumption of 15 grams of D‑glucose by the branched fermentation pathways of Clostridium butyricum (8, [C]).

For the small chamber (+ 0.43%), the model projects that fermenting 15 g of dextrose every week could raise the CO2 level of a lightly-used chamber by half a percentage point (Figure 1f versus Figure 1b).

Why use CO2 in anaerobic chambers?

Nearly a century ago it was posited that first, all bacteria, yeasts and molds require carbon dioxide for their growth, and second that the CO2 is used as a source of carbon (18). We now know that carbon from CO2 is fixed into certain cellular building blocks using chiefly bicarbonate as the reactive species [D]. Two important examples are the biosynthesis of purine bases (25, 11) and the biosynthesis of carbamoyl phosphate (20, 22). Carbamoyl phosphate donates its carbon to the pyrimidine bases and to arginine (20, 11), and to the important "t6A" modification of certain tRNAs (5). The bicarbonate-dependent carboxylation of pyruvate or phosphoenolpyruvate to oxaloacetate (22) supports the biosynthesis of six aspartate-branch amino acids (11, [E]). Fatty acid biosynthesis consumes bicarbonate in the making of malonyl-CoA by acetyl-CoA carboxylase (22), but without net carbon fixation because CO2 is released in each cycle of chain elongation (11).

The rate at which CO2 can be spontaneously hydrated to bicarbonate and delivered to enzymes can be growth limiting at low CO2 concentrations such as in air (0.04%) (14). Diverse prokaryotes can cope with this by encoding the enzyme carbonic anhydrase (CA), which catalyzes the reversible hydration of CO2 to bicarbonate (21). Bacteria lacking functional CA, whether mutated or naturally absent, have been unable to grow in lab studies except if supplemented with CO2/bicarbonate or in some cases with nutrients (14, 10, 15). The lowest initial gas-phase CO2 concentration that could support full growth of an E. coli CA mutant growing in Luria broth was about 1.5% (24). The most common general recommendation that is made about CO2 in anaerobic chambers and jars for heterotrophic growth is to use 10% in the gas phase.

  • In a study of rumen anaerobes, 31 of 32 strains required CO2 to reach normal growth (4). Species producing succinic acid as a major end product showed an absolute CO2 requirement and demanded a higher CO2 concentration for initiation of growth and for optimal growth. Fermentation product profiles can also be shifted. Anaerobes that produce large amounts of succinate under adequate CO2 may shift to lactate production if CO2 is inadequate (9).
  • CO2-dependence can be tied to energy conservation. Some anaerobes can run CO2‑dependent PEP carboxykinase (EC in the direction of oxaloacetate, thus coupling the reaction to phosphorylation of ADP (13).
  • Elevated CO2 or bicarbonate may be required to isolate organisms that have adapted to stable high‑CO2 environments. Such organisms may be naturally deficient in carbonic anhydrase (15, 24).
  • The level of CO2/bicarbonate can affect the nutritional requirements of anaerobes as demonstrated using CA‑deficient vs. wildtype Clostridium perfringens when growing on lean medium vs. rich medium (10). Some anaerobes can require more than 5% CO2 even on medium that is highly enriched (17).
  • Levels of CO2 or bicarbonate have been reported to influence the germination and outgrowth of bacterial spores (2, 16).
  • An O2‑tolerant anaerobe could be erroneously categorized as "obligate" if testing is conducted only in air and not in air supplemented with CO2 (23). Walter Loesche, in a broad study of anaerobe oxygen tolerance included at least 10% CO2 in all tests (12, [F]).

The use of CO2/bicarbonate buffering systems inside vinyl chambers need not be elaborated here, except to remind that the initial pH of an open system may be higher than was intended if the chamber CO2 level is lower than was being counted on.

Monitoring and controlling CO2 in the chamber

Figure 2. Direct injection of CO2 into chamber.

My system for controlling CO2 in the chamber relies on using N2/H2 as the supporting mixed gas to reduce expenses, coupled with direct injection of CO2 into the chamber under manual control. An infrared sensor (NDIR) provides a readout of the CO2 level in the chamber. Gas sampling and CO2 injection occur along an external loop of tubing through which chamber gas circulates. The system is inconvenient compared to using a commercial 3‑gas mix, but a lot more economical and it could be automated.

Figure 2 shows a series of readings from the NDIR while raising the chamber from 9% to 10% CO2. The spikes occur because the NDIR sampling point is downstream of the CO2 injection point. The smaller spike was a CO2 injection of 4 seconds duration; about what my Type‑C chamber would need daily to maintain 10%. System components are:

  • an NDIR CO2 meter [G]. Mine is mounted outside the chamber, but it could be inside.
  • a laptop or PC to run the CO2 meter software
  • a 20 Lb. cylinder of CO2 with a gas regulator and shutoff valve (mine set to deliver 20 psig)
  • an external loop of vinyl tubing having a pump [H] and filter situated in‑line. Many chambers already have this.

All tubing in the system is 1/4 in. ID x 1/2 in. OD clear vinyl. Two barbed tee fittings for sampling chamber gas and injecting CO2 are inserted into the tubing loop . The gas-sampling tee is downstream of the pump and filter so that only filtered gas is routed to the NDIR, and to provide a small amount of positive pressure at the sampling tee. My choice of mounting the NDIR outside the chamber was driven by additional need for the meter outside the chamber. For inside the chamber I'd have chosen a different model of the NDIR.


  1. Allison MJ, Robinson IM, Baetz AL. 1979. Synthesis of alpha-ketoglutarate by reductive carboxylation of succinate in Veillonella, Selenomonas, and Bacteriodes species. J Bacteriol. 140(3):980-986. PMID: 533772
  2. Ando Y, Iida H. 1970. Factors affecting the germination of spores of Clostridium botulinum type E. Jpn. J. Microbiol. 14(5):361-370. PMID: 4919766
  3. Arank A, Syed SA, Kenney EB, Freter R. 1969. Isolation of anaerobic bacteria from human gingiva and mouse cecum by means of a simplified glove box procedure. Appl. Microbiol. 17(4):568-576. PMID: 4890748
  4. Dehority BA. 1971. Carbon dioxide requirement of various species of rumen bacteria. J. Bacteriol. 105(1):70-76. PMID: 5541030
  5. Deutsch C, El Yacoubi B, de Crécy-Lagard V, Iwata-Reuyl D. 2012. Biosynthesis of threonylcarbamoyl adenosine (t6A), a universal tRNA nucleoside. J Biol Chem. 287(17):13666-13673. PMID: 22378793
  6. Fine DH. 2006. Dr. Theodor Rosebury: grandfather of modern oral microbiology. J. Dent. Res. 85(11):990-995. PMID: 17062737
  7. Gai CS, Lu J, Brigham CJ, Bernardi AC, Sinskey AJ. 2014. Insights into bacterial CO2 metabolism revealed by the characterization of four carbonic anhydrases in Ralstonia eutropha H16. AMB Express. 4(1):2. PMID: 24410804
  8. Gottschalk G. 1986. Bacterial Metabolism, 2nd edition. Springer-Verlag. p.230
  9. Holdeman LV, Cato EP, Moore WEC (eds). 1977. Anaerobe laboratory manual, 4th edition. Virginia Polytechnic Institute and State University, Blacksburg, Va. p.125
  10. Kumar RS, Hendrick W, Correll JB, Patterson AD, Melville SB, Ferry JG. 2013. Biochemistry and physiology of the β class carbonic anhydrase (Cpb) from Clostridium perfringens strain 13. J. Bacteriol. 195(10):2262-2269. PMID: 23475974
  11. Lengeler JW, Drews G, Schlegel HG (Editors). 1999. Biology of the Prokaryotes. Wiley-Blackwell.  (purines and pyrimidines,  pp.139‑141)  (arginine,  p.128, p.473)  (oxaloacetate and aspartate-branch amino acids,  p.126, p.129)  (fatty acids,  pp.147‑150)
  12. Loesche WJ. 1969. Oxygen sensitivity of various anaerobic bacteria. Appl. Microbiol. 18(5):723-727. PMID: 5370458
  13. Macy JM, Ljungdahl LG, Gottschalk G. 1978. Pathway of succinate and propionate formation in Bacteroides fragilis. J. Bacteriol. 134(1):84-91. PMID: 148460
  14. Merlin C, Masters M, McAteer S, Coulson A. 2003. Why is carbonic anhydrase essential to Escherichia coli? J. Bacteriol. 185(21):6415-6424. PMID 14563877
  15. Morotomi M, Nagai F, Watanabe Y. 2012. CO2‑dependent growth of Succinatimonas hippei YIT 12066T isolated from human feces. Microbiol. Immunol. 56(3):195-197. PMID: 22469182
  16. Plowman J, Peck MW. 2002. Use of a novel method to characterize the response of spores of non-proteolytic Clostridium botulinum types B, E and F to a wide range of germinants and conditions. J. Appl. Microbiol. 92(4):681-694. PMID: 11966909
  17. Reilly S. 1980. The carbon dioxide requirements of anaerobic bacteria. J. Med. Microbiol. 13(4):573-579. PMID: 6107383.
  18. Rockwell GE, Highberger JH. 1927. The necessity of carbon dioxide for the growth of bacteria, yeasts and molds. The Journal of Infectious Diseases. 40(3):438-446.
  19. Rosebury T, Reynolds JB. 1964. Continuous anaerobiosis for cultivation of spirochetes. Proc. Soc. Exp. Biol. Med. 117:813-815. PMID: 14244963
  20. Shi D, Caldovic L, Tuchman M. 2018. Sources and Fates of Carbamyl Phosphate: A Labile Energy-Rich Molecule with Multiple Facets. Biology (Basel) 7(2):34. PMID: 29895729
  21. Smith KS, Jakubzick C, Whittam TS, Ferry JG. 1999. Carbonic anhydrase is an ancient enzyme widespread in prokaryotes. Proc Natl Acad Sci USA. 96(26):15184-15189. PMID: 10611359
  22. Smith KS, Ferry JG. 2000. Prokaryotic carbonic anhydrases. FEMS Microbiol Rev. 24(4):335-366  PMID: 10978542 (see Table 3)
  23. Socransky SS, Holt SC, Leadbetter ER, Tanner AC, Savitt E, Hammond BF. 1979. Capnocytophaga: new genus of gram-negative gliding bacteria. III. Physiological characterization. Arch. Microbiol. 122(1):29-33. PMID: 518236
  24. Watsuji TO, Kato T, Ueda K, Beppu T. 2006. CO2 supply induces the growth of Symbiobacterium thermophilum, a syntrophic bacterium. Biosci Biotechnol Biochem. 70(3):753-756. PMID: 16557001
  25. Zhang Y, Morar M, Ealick SE. 2008. Structural biology of the purine biosynthetic pathway. Cell Mol Life Sci. 65(23):3699-3724. PMID: 18712276
  26. "Tubing Selection Guide" originally published by Cole-Parmer Company. I do not know if it is still accessible. The listed permeability numbers for vinyl at 25 deg C were 79 for CO2, 15 for O2, and 4.3 for N2 expressed in units of {(cc‑mm)/(sec‑cm^2‑cmHg)} x 10e‑10.


  1. By "cycle", I mean the complete sequence of airlock evacuations and refills necessary for one chamber entry, or one exit, starting from an aerobic airlock. At the end of each cycle the inner door of the airlock is open long enough to equilibrate with the chamber. Airlock volume was estimated to be 1.67 ft3 by direct measurement including the door coves.
  2. The model assumes there is zero differential in total gas pressure between inside and outside the chamber ie. the chamber is not overinflated, and the CO2 concentration outside the chamber is constant at near zero (air ~0.04%). With these assumptions the rate of CO2 diffusion through the vinyl at any time is taken to be a linear function of the CO2 concentration in the chamber. This is the basis of the model's recalculating the drop in chamber CO2 concentration for each 24-hour period in linear proportion to the average drop rate that had been measured between 10% and 9% CO2. The units in which gas permeability of vinyl were expressed (26) suggest that the method should be valid. Temperature is a factor in the diffusion rate of gases through polymeric materials but is not included in this model.
  3. The calculation uses 188 mol CO2 evolved per 100 mol dextrose fermented. Pressure is 1 atm and temperature is 294.26 deg K. "R" at these values for the ideal gas law (PV=nRT) is 0.08207 calculated using 24.15 liters as the volume of one mol of gas. Fifteen grams of dextrose at 0.5% concentration is 3 liters of liquid culture, or 85 Petri plates at 35 mL per plate.
  4. I focus on limited elements of nonautotrophic anabolic and "anaplerotic" metabolism because they explain well enough the benefit of maintaining CO2 in anaerobic chambers. There are additional important roles for CO2 and bicarbonate in prokaryote physiology (7, 21).
  5. The carboxylation of pyruvate or PEP to make OAA is a net-carbon fixation for five of the six aspartate-branch amino acids (for OAA that becomes aspartate, since OAA can have other fates). For lysine, diaminopimelate decarboxylase sheds a CO2 in the last biosynthetic step, negating the earlier carboxylation. Under anaerobiosis, OAA from pyruvate/PEP carboxylation may partially supply the 2‑ketoglutarate-branch amino acids (glutamate, glutamine, arginine, proline) via: 1) citrate and isocitrate on the oxidative side of the anaerobic TCA cycle or; 2) by the reductive carboxylation of succinate in some anaerobes (1) on the reductive side of the anaerobic TCA cycle.
  6. Loesche manipulated cultures for these tests in a steel-walled glove chamber that he had acquired from Theodor Rosebury of Washington University (19, 6). It is reported that Rolf Freter, having visited Rosebury earlier, incorporated elements of Rosebury's chamber into his own vinyl-walled design and that the steel chamber was transported to the University of Michigan (6, 3).
  7. K33 BLG 30% CO2 + RH/T Data Logging Sensor from co2meter.com.
  8. I use Gast model DOA-P701-AA.

last edit:  2023-10-22