Applied Microbiology and Biotechnology

, Volume 97, Issue 1, pp 135–142

Controlling autonomous underwater floating platforms using bacterial fermentation


    • Chemistry DivisionUS Naval Research Laboratory
  • Lisa A. Fitzgerald
    • Chemistry DivisionUS Naval Research Laboratory
  • Erinn C. Howard
    • Chemistry DivisionUS Naval Research Laboratory
    • The Scientific Consulting Group, Inc.
  • Emily R. Petersen
    • Nova Research, Inc.
  • Preston A. Fulmer
    • Chemistry DivisionUS Naval Research Laboratory
  • Peter K. Wu
    • Department of PhysicsSouthern Oregon University
  • Bradley R. Ringeisen
    • Chemistry DivisionUS Naval Research Laboratory
Biotechnological Products and Process Engineering

DOI: 10.1007/s00253-012-4296-5

Cite this article as:
Biffinger, J.C., Fitzgerald, L.A., Howard, E.C. et al. Appl Microbiol Biotechnol (2013) 97: 135. doi:10.1007/s00253-012-4296-5


Biogenic gas has a wide range of energy applications from being used as a source for crude bio-oil components to direct ignition for heating. The current study describes the use of biogenic gases from Clostridium acetobutylicum for a new application—renewable ballast regeneration for autonomous underwater devices. Uninterrupted (continuous) and blocked flow (pressurization) experiments were performed to determine the overall biogas composition and total volume generated from a semirigid gelatinous matrix. For stopped flow experiments, C. acetobutylicum generated a maximum pressure of 55 psi over 48 h composed of 60 % hydrogen gas when inoculated in a 5 % agar (w/v) support with 5 % glucose (w/v) in the matrix. Typical pressures over 24 h at 318 K ranged from 10 to 33 psi. These blocked flow experiments show for the first time the use of microbial gas production as a way to repressurize gas cylinders. Continuous flow experiments successfully demonstrated how to deliver biogas to an open ballast control configuration for deployable underwater platforms. This study is a starting point for engineering and microbiology investigations of biogas which will advance the integration of biology within autonomous systems.


Clostridium acetobutylicumHydrogenPressureBallastFermentation


The biogenic production of gases such as methane, oxygen, carbon dioxide, and hydrogen has shaped the current form of our biosphere starting from the earliest stages of life on earth (Fowler et al. 2009). Biogenic gases are viable replacements for fossil fuels in both heat and power applications (Weiland 2010). In particular, hydrogen can be produced from carbohydrates by several strains of bacteria, with members of the Clostridia genus responsible for generating hydrogen from soil and wastewater samples (Kengen et al. 2009; Hung et al. 2011). Due to the high levels of hydrogen generated from cellulosic materials by Clostridium, members of this genus can be used for direct applications in the biofuels’ field. The application of these gases as a source of ballast, however, has just recently been reported for the first time (Wu et al. 2011). The calculated amount of renewable biogenic gases (complete oxidation to CO2 and H2) generated from 1 g of glucose would be approximately 2.2 L which at a 10-m depth underwater (21–26 °C) would provide a 22.7-N buoyant force in seawater (density = 1.025 g/mL).

The motivation for using biogenic gases for ballast is derived from the energy and power demands of an autonomous system. The overall efficiency of the autonomous technology will allow for longer deployments and reduced maintenance. For example, clandestine submerged devices for underwater applications can be designed to exploit marine resources to maximize its operational duration and minimize its size and weight. For submersible sensors and crafts, the surfacing of the device for radio frequency (RF) communications with airborne assets is an essential component to autonomy but is one of the largest power drains. Onboard communication using high bandwidth, lower energy RF, or ultrahigh frequency transmissions are impossible for submerged devices as RF transmissions do not propagate well underwater and acoustic communication and wired systems have logistic shortcomings (Chitre et al. 2008). This requires most underwater sensors to be interfaced directly by either ship or diver to retrieve data during deployment at a large cost to the duration and price of the overall system.

Recently, Wu and coworkers developed and demonstrated a novel zero power-consuming ballast control system for water column underwater applications (Biffinger et al. 2011; Wu et al. 2011). The operating principle was centered on the generation of buoyant force using the continuous production of biogenic gases from Clostridium acetobutylicum. This control mechanism was ultimately developed to incorporate biological fuel cells as the sole power source for water column applications. Current deployable assets using microbial fuel cell (MFC) power sources are limited to the sediment (Tender et al. 2002; Reimers et al. 2006; Rezaei et al. 2007) since practical power densities from MFCs (1–10 mW/m2) (Osman et al. 2010; Qian and Morse 2011) preclude their use to power ballast control circuits, pumps, and valves that typically have power demands of 1–2 W. The replacement of active electrical feedback circuits for surfacing and submerging the platform with manipulating bacterial gas generation rates was a necessary step to make a biological power source viable for water column applications.

A second mechanism for using biogenic gas as ballast would be the pressurization of gas cylinders for gated release. Some bacteria commonly manipulated at atmospheric pressure can be grown and thrive under intense pressures (>100 atm) and temperatures but show significant changes in metabolic rates and pathway usage (Hazen et al. 2002; Sharma et al. 2002; Margosch et al. 2006). C. acetobutylicum is typically cultured at atmospheric pressure though there are some studies that have investigated the growth and metabolism under elevated hydrostatic pressure (Kalchayanand et al. 2004). Intermittent gas release tests (similar to the current study except that a large headspace pressure was not allowed to build and liquid cultures were used) have been performed by various groups over the past decade to analyze hydrogen production (Oh et al. 2009) or even to improve hydrogen yields (Valdez-Vazquez et al. 2006).

The current study will describe the application of microbial gas fermentation for renewable ballast generation in a closed and an open system. C. acetobutylicum was utilized in various agar matrices with glucose as the main carbon source to create gas liberation gels. The quantity and composition of vent gases indicated that significant amounts of hydrogen were produced throughout the experiments. Finally, the biogenic gases that were produced in the laboratory were deployed in an autonomous, continuous-release, floating platform to demonstrate a true autonomous, zero power-consuming, ballast control mechanism underwater.

Materials and methods

Growth of Clostridium and sample preparation

C. acetobutylicum (ATCC 824) was grown in an anaerobic chamber (Coy Laboratory Products, Inc.) under a nitrogen atmosphere (with 2 % hydrogen) at 30 °C in reinforced clostridial medium (RCM, BD 218081) with shaking at 100 rpm. After the culture reached a stationary phase (72 h), the 1 × 106-CFU/mL culture was used to inoculate gas collection experiments. Since all experiments were designed for gas evolution from solid supports, viable cell counts from agar gels were not attempted as all experiments were performed at the stationary stage of bacterial growth and inoculations were made from the same stocks for each experimental series.

Biogenic gas collection and quantification

A variety of chemically defined and heterogeneous carbon sources were incorporated into RCM at 5 % w/v. The carbon sources utilized were glucose, dextran, Ficoll, and pectin (Sigma). To monitor the composition of gases evolved, 1 mL of the medium with each carbon source was added to 2-mL sterile glass vials with septum seals and inoculated with 1 × 106-CFU/mL culture of C. acetobutylicum. Vials were sealed and placed in an anaerobic chamber between measurements. Headspace gas composition was measured on a Varian 450-GC at 24-h intervals. Vials were vented inside the anaerobic chamber after each measurement to reset the gas composition within each experiment. To determine total gas production via continuous gas release, 50 mL of RCM with 5 % w/v added carbon source was similarly inoculated with C. acetobutylcium, sealed in 125-mL Erlenmeyer flasks, and gas effluents were captured and measured by water displacement (at 22 ± 1 °C) over 100 h. Results of the total gas output were compared to experiments containing no additional carbon source.

Pressure measurement experiments

A RCM agar gel (3.8 g/L) was used as the base support for the growth of C. acetobutylicum under an inert atmosphere with 10 % w/v glucose. Glass 1 neck pressure tubes (100 mL, Chemglass) were filled with 70 mL of the sterilized hot agar medium. The sterilized medium was allowed to degas and solidify in the anaerobic chamber. Each tube was inoculated by dipping a sterile inoculation loop into a 1 × 106-CFU/mL culture and stabbing this into the agar supports five to seven times. The tube was then sealed with gas-tight pressure seals and connected to an analog pressure gauge. Time and pressure measurements were recorded manually. Experiments were performed in triplicate at 22 ± 1 and 35 ± 1 °C using a constant-temperature water bath. The pressurized gases were vented into nitrogen-purged, valved 500-mL gas sampling bags (Fisher Scientific) for gas analysis. Gas samples were injected into a Varian 450-GC using an in-line gas sampling valve connected directly to the sampling bag with a quick-connect Swagelok valve.

Analytical methods for gas composition

Gases generated from both continuous and pressurized experiments were analyzed on a Varian 450-GC with a custom pneumatic valve and column assembly (Custom Solutions Group, LLC) coupled with a thermal conductivity detector. The gas analysis was performed with custom pneumatic valve switching between a 6 ft × 1/16 in. HayesepQ 80/100 mesh column and 8 ft × 1/16 in. MolSieve 5A 60/80 mesh column. Gas samples were injected directly onto the column. Ultrahigh purity argon was the carrier gas using an isocratic oven temperature of 100 °C. Standard gas mixtures (CH4, CO, CO2, H2, N2, and O2) were used to calibrate the method for biogenic gases.

Field test of floating platform

Three experiments were deployed in the tidal bay at a depth of 3.5 m at 24 °C. The field demonstration of the continuous release of biogenic gases was tested with a zero power-consuming ballast control system described previously (Biffinger et al. 2011; Wu et al. 2011) (Fig. 1a). The zero power-consuming ballast control deployed for this exercise was protected in the upper, larger chamber, as shown in Fig. 1b. The demonstration was performed using 700-mL gas chambers (custom fit with gas-tight, quick-connect valves). Two different agar matrices were tested containing either 0.37 or 1.5 % agar (w/v) with 3 g of glucose in 500 mL of growth matrix. Gas generation chambers were inoculated with 10 mL of active (1 × 106 CFU/mL) C. acetobutylicum cultures across three separate injections throughout the length of the chamber. Chambers were incubated at 23 °C for 36 h prior to venting and attachment to the zero power-consuming ballast control system. The inoculated gas liberation gel was enclosed in the middle chamber. The temperature and depth data were recorded onboard the platform using a titanium water-level and temperature data logger (HOBO, Inc.), and the data collected during the deployment were downloaded postdeployment. The sensor was contained in the lower chamber which also contained additional weight to orient and make the device neutrally buoyant. The assembled device was then connected to the 25-ft tether (with a physical stop at 10 ft) in the water column.
Fig. 1

a General schematic of autonomous floating test platform with integrated microbial gas tank and b photograph from field deployment exercise (Tyndall AFB, photo by Guinevere M Strack)


Utilization of carbohydrates or polysaccharides by C. acetobutylicum for biogas production

The total gas generated over 96 h for several carbon sources (pectin, dextran, glucose, and Ficoll) were compared to growth without the addition of a carbon source (control, Fig. 2a). A minimal amount of gas was liberated from the control experiment considering all media contained the RCM base, which includes 5 g/L glucose. The glucose data were similar to previously reported data showing that within 24 h, the hydrogen production was ~70 % (Zhang et al. 2006). In general, glucose and dextran generated the most gas, but dextran showed significantly higher concentrations of hydrogen after 4 days (Fig. 2b) compared to the control experiment. Additionally, there was a decrease in hydrogen production from day 1 to day 4 when comparing pectin to dextran. This could be attributed to the known decrease in enzymatic activity with the two different substrates (Montoya et al. 2001). There was a higher concentration of hydrogen 1 day after inoculation than at the completion of the experiment.
Fig. 2

Comparison between the amounts of biogenic gas generated by C. acetobutylicum with selected carbon sources (5 % w/v) for a total biogenic gas release and b hydrogen composition, 100 h after inoculation

Since glucose was one of the carbon sources which produced the most biogas from C. acetobutylicum, the fermentation of glucose was examined further using different concentration of glucose in the media (Fig. 3). All experiments were normalized to the amount of gas generated from the control experiment (0 %). Over 96 h, C. acetobutylicum generated >400 mL of hydrogen at a 5 % w/v and at >5 % glucose; the ratio of hydrogen generated per mole of glucose approached 1.1. The experiments described in the current study were performed at 25 °C, and thus, molar ratios of 3.0–4.0 were not observed (Davila-Vazquez et al. 2008).
Fig. 3

Total volume of biogenic gas generated with 0–10 % (w/v) glucose over 96 h at 25 °C with percent hydrogen yield per mole of glucose indicated

Gas-liberating agar gels containing Clostridium—continuous release

The viscosity of the agar gel was used to control the rate of gas delivery from the bacterial cells (Fig. 4). After 30 h, the culture with 0 % agar generated 100 mL of gas and had a 3-mL/h rate of gas generation without the addition of thickening agent. In contrast, culture media with higher concentrations of agar (1.13 and 1.50 % w/v) showed significantly impeded gas evolution over 45 h and rapid gas release rates after that time period. For example, the overall rate of gas evolution from 1.5 % w/v agar was 1.9 mL/h.
Fig. 4

Total volume of biogenic gas collected from 10 % (w/v) glucose in various concentrations of agar over 75 h at 25 °C

Gas-liberating agar gels containing Clostridium—repressurization experiments

The initial pressurization experiments were performed on liquid RCM cultures with 5 % (w/v) glucose containing C. acetobutylicum (106–107 CFU/mL). Experiments that did not contain agar generated no observable overpressure over 200 h at 30 °C (data not shown). The overpressure was periodically vented, and the gases were collected for gas analysis (Fig. 5c). In order to build pressure using C. acetobutylicum, gas-liberating gels were inoculated and sealed over 100 h at 35 and 22 °C. At 35 °C, an overpressure of 10 psi could be consistently generated over 24 h (Fig. 5a). However, an overpressure of <10 psi was observed when the same experiment was performed at 22 °C (Fig. 5b).
Fig. 5

Pressurization and release of biogenic gases generated from C. acetobutylicum using glucose (10 % w/v) in an 11.2-g/L agar support over 200 h at a 35 °C and b 22 °C and including c average composition of the biogenic gases at both 22 and 35 °C generated during pressurization periods. Pressure tube was filled to 30 % capacity

Field deployment of zero power-consuming ballast control system

A selected time period during the deployment is shown in Fig. 6. Two different agar culture formulations (0.37 and 1.50 %) are shown for the deployment to demonstrate how gel viscosity can be used as a method to control the rate of gas evolution in the field. There were multiple surfacing events that occurred for the 0.37 % agar gel at the beginning of the deployment which were not seen for the 1.5 % agar gel (Fig. 6).
Fig. 6

Selected time period chart showing change in depth with two different solidifying agents and temperatures


The rate and composition of the biogenic gases generated by C. acetobutylicum will ultimately be dictated by the growth medium. C. acetobutylicum has been reported to generate H2, CO2, and various volatile organic compounds such as butyric acid, ethanol, and acetone (Girbal et al. 1995). The overall purpose of these experiments was to analyze the bulk properties of the biogenic gases from abundant polysaccharides which include pectin (Saissac et al. 1952), dextran (Kim and Weigand 1992), Ficoll (Montoya et al. 2001), and glucose (Kengen et al. 2009) that are typically used as carbon sources for designing gas-liberating gels. The addition of glucose and dextran generated the greatest volume of biogas (Fig. 1a), but after 4 days, dextran was shown to produce a higher concentration of hydrogen than the culture with glucose (Fig. 2b).

The generation of biogas is vital for microbial ballast, but for long-duration deployments, the time release of these biogases is required. The creation of Clostridium-containing, gas-liberating gels will extend the viability of the bacteria and allow access to using strictly anaerobic cells under air-exposed conditions by limiting the diffusion of air into the gel. This is a delicate balance because fermentation gases need to escape from the gel but air in the headspace needs to be blocked from entering into the gel. There was a higher concentration of hydrogen 1 day after inoculation than at the completion of the experiment, which is most likely due to the depletion of the added carbon source. Thus, these results indicate that glucose would be the best candidate for applied short-term experiments.

The fermentation of glucose by Clostridium has led to the rapid production of biogenic gases confirmed by these data and previous work using flow reactors (Zhang et al. 2006). Thus, the fermentation of glucose was examined based on the concentration of glucose in the medium (Fig. 3). Hydrogen production has been studied previously with the hydrogen to glucose molar ratio (above 30 °C) typically between 1.4 and 2.0 (Kengen et al. 2009). Extremophiles usually have a far higher rate of conversion of glucose to hydrogen (ratios 3.0–4.0), but the elevated temperature environments of some extremophiles are found not suitable for a system deployed for water column applications.

The design of gas-liberating gels for Clostridium was attempted using several concentrations of agar and one concentration of glucose (Fig. 4). Glucose was chosen as the added carbon source since its addition consistently generated the most gas compared to other polysaccharides (Fig. 2). Agar was used as the thickening agent due to its biocompatibility and potential use as a long-duration carbon source. The viscosity of the agar gel was used to control the rate of gas delivery from the bacterial cells. These experiments confirm that the increase in agar concentration did not block the total amount of gas released (only the rate at which it was released) and that agar concentrations >0.75 % w/v can be used to control the rate of gas evolution from the gas-liberating gel.

If the rate of diffusion can be decreased by increasing the viscosity of the medium, then overpressure could be built in a sealed chamber. Considering the low compressibility of water (bulk modulus = 2.2 GPa for a 1.8 % decrease in volume), agar gels can also be used for pressurization experiments using bacteria. Since there was no significant deviation of gas composition with temperature, the mole percentage of hydrogen and carbon dioxide from both temperature experiments was averaged and shown in Fig. 5c. In general, the gas composition was almost entirely hydrogen, although the carbon dioxide concentration increased with each successive gas release. These data confirm that the bacterial cells were still viable and active at each release and that the approximate rate of repressurization was 0.17 psi/h at 22 °C and five-fold higher at 35 °C (0.82 psi/h). The largest overpressure generated was 56 psi at 35 °C (in a stainless steel cylinder over a 48-h period) (data not shown).

The repressurization of a biogenic gas tank also has several other practical applications besides renewable ballast for an underwater device. Developments toward using hydrogen as an alternative energy source will require on-site methods to store and manipulate using renewable feedstock such as wastewater or municipal solid waste (Hung et al. 2011). In comparison to petroleum-based methods, generating hydrogen using methods involving fermentations is lower in temperature, generates less waste, and requires no additional external energy (Guo et al. 2010). The use of a semisolid support built significant overpressure and could be combined with granule-based biofilms (Zhang et al. 2008) to create highly active gas-liberating gels.

The combination of bacterial fermentation rates and device operation was brought to practice in a field deployment at Tyndall Air Force Base (Panama City, FL) in collaboration with researchers at the Air Force Research Laboratory and the University of New Mexico. One interesting facet is because the overall system is neutrally buoyant, as the gas is slowly collected in the control chamber, the platform will gradually begin to surface. For this particular system, the amount of ballast that could be collected before the ballast control vented the gas was 15 mL. For this demonstration, the rate of gas evolution was so high under these conditions and scale (>1 mL/min) that multiple surfacing events occurred for the 0.37 % agar gel at the beginning of the deployment, as predicted by the results shown in Fig. 4. This is also consistent with the lack of multiple surfacing events for the 1.5 % agar gel (Fig. 6c) which shows the effect of hindered gas evolution. These data confirm that with the same amount of glucose and different concentrations of agar, the rate of surfacing and submerging of a tethered, free-floating platform can be dictated using bacteria alone, thus validating this method of zero power-consuming autonomous control by bacteria. From a practical standpoint, any bacterium that generates gaseous by-products could be used with this method and allows for any number of conditions to be tested and evaluated.

The results of this study indicate the feasibility of using biogenic gases for renewable, autonomous, ballast regeneration for underwater devices using two unique fermentative methods. C. acetobutylicum has shown to be an amenable organism for this application. Under these conditions, C. acetobutylicum generated approximately 10 psi over 24 h, a rate that is reproducible at ambient temperature with repeated venting incidents. This is the first report of microbial gas production as a means for gas cylinder repressurization and demonstrates how agar-containing gels can be used for environmental applications. These successful field tests will be a solid foundation by which future studies can build upon and advance the integration of microbiology with autonomous system engineering.


The authors thank the Naval Research Laboratory/Office of Naval Research 6.2 program for funding and Glenn R. Johnson and co-workers (Air Force Research Laboratory, Tyndall AFB) for organizing, providing supplies and support, and assisting the field deployment demonstration. The authors thank Barry J. Spargo for his support and advice for this program. Field deployment of the zero power-consuming ballast control system was a joint collaboration between the U.S. Naval Research Laboratory (J. Biffinger), the U.S. Air Force Research Laboratory (G. Johnson), and the University of New Mexico (P. Atanassov).

Copyright information

© Springer-Verlag (outside the USA) 2012