Microbial Fuel Cells for Robotics: Energy Autonomy through Artificial Symbiosis

DOI: 10.1002/cssc.201200283

Microbial Fuel Cells for Robotics: Energy Autonomy through Artificial Symbiosis Ioannis A. Ieropoulos,* John Greenman, Chris Melhuish, and Ian Horsfield[a] The development of the microbial fuel cell (MFC) technology has seen an enormous growth over the last hundred years since its inception by Potter in 1911. The technology has reached a level of maturity that it is now considered to be a field in its own right with a growing scientific community. The highest level of activity has been recorded over the last decade and it is perhaps considered commonplace that MFCs are primarily suitable for stationary, passive wastewater treatment applications. Sceptics have certainly not considered MFCs as serious contenders in the race for developing renewa-

ble energy technologies. Yet this is the only type of alternative system that can convert organic waste—widely distributed around the globe—directly into electricity, and therefore, the only technology that will allow artificial agents to autonomously operate in a plethora of environments. This Minireview describes the history and current state-of-the-art regarding MFCs in robotics and their vital role in artificial symbiosis and autonomy. Furthermore, the article demonstrates how pursuing practical robotic applications can provide insights of the core MFC technology in general.

Introduction A robot can be defined as a machine that performs tasks automatically by following a set of rules in a computer programme. An automaton, on the other hand, is a moving mechanical device that is usually built to imitate humans and can self-operate. This term can be used to describe robots and in particular autonomous artificial agents. Building automata is certainly not something new. Numerous examples of automata and complex machines were developed by the ancient Greek civilisations, with perhaps the first recorded example being that of Heron of Alexandria, who in the first century AD constructed a self-moving cart driven by a counter weight attached to the wheel base.[1] The first modern example of an autonomous system is William Grey Walter’s tortoise “Elsie”.[2] This was an electromechanical mobile system (looking like a plastic tortoise) that could follow the light and home into its nest, where a battery charger would automatically make contact with the robot to recharge its batteries. Autonomy is defined as the ability of a robotic agent to behave and operate with minimum or no human intervention. Such a robot must be able to compute and select between actions that will keep it viable without reaching lethal limits.[3] This implies that energy collection and management forms part of the agent’s behavioural repertoire. It therefore suggests that a truly autonomous robot must be capable of collecting its energy from its environment and managing it in a viable manner. Ideally, an autonomous robot would operate over a wide range of environments if it could utilise energy that is naturally available, such as solar radiation, wind and hydro-power. There are environments on the planet, which have restricted or no access to the aforementioned energy sources, but do offer—in great abundance and distribution—access to organic materials that are the main feedstock of many biological systems on the

1020

planet. A plausible strategy for an autonomous robot would, therefore, be one that allows it to extract its electrical energy from this widely available fuel source (wet biomass). One technology that is capable of performing this transformation at ambient conditions is the microbial fuel cell (MFC).

Microbial Fuel Cells MFCs represent a promising technology for sustainable energy production and waste treatment.[4] They are bioelectrochemical transducers, which convert biochemical energy from organic fuels (including low grade wastes such as anaerobic sludge, landfill leachate, urine, insects and plant materials) into electricity through the metabolic reactions of microbes growing as biofilms around the electrodes. MFCs often consist of two compartments, the anode (negative terminal electrode) and the cathode (positive terminal electrode) that are separated by an ion exchange membrane (IEM). In the anode chamber, bacteria anaerobically oxidise organic substrate (wet fuel) generating electrons and releasing protons. In the presence of an electrode and under the pressure of redox potential difference and consequent electrophilic attraction, the microorganisms interact with the electrode and make it part of their natural anaerobic respiration, that is, directly or indirectly transfer electrons onto the electrode. A connection between the anode and cathode electrodes therefore facilitates the path for electrons [a] Dr. I. A. Ieropoulos, Prof. J. Greenman, Prof. C. Melhuish, I. Horsfield Bristol Robotics Laboratory Department of Engineering, Design & Mathematics University of the West of England, Bristol T-Building, Frenchay Campus, BS16 1QY, Bristol (UK) Fax: (+ 44) 1173283960 E-mail: [email protected]

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemSusChem 2012, 5, 1020 – 1026

Microbial Fuel Cells for Robotics to flow and generate current; electrons travel to the cathodic compartment via this external circuit and protons migrate through the IEM. Protons and electrons combine at the cathode, reducing oxygen (electron acceptor) to water. Fuel diversity and oxygen availability An autonomous robot will require a microflora inside the onboard MFCs that will utilise a diverse range of fuel substrates to operate in as wide a range of environments as possible, thus, in a way, mimicking opportunistic behaviour. Microbial communities originating from anaerobic sludge or sediment samples, after inoculation in MFCs, can evolve into anodic biofilms that retain high diversity[5] such that most types of soluble organic carbon (sugars, acetate, lactate, amino acids) can be utilised by the bio-electrode system. More complex feedstocks (mixed substrates, macromolecular, particulate, microalgae, plant materials, dead insects and microbes) may require pre-digestion coupled with longer residence times in an artificial stomach, for example, of the type described for EcoBot-III. In highly diverse bacterial communities, the majority of species can “substitute” for one another thus maintaining functionality of the microcosm (that is, hydrolysis/utilisation of available carbon–energy substrates and electron transfer to the electrode). Oxygen, especially for the purposes of robotic autonomy, is potentially the most effective electron acceptor in a MFC due to its high redox potential, availability, low cost and zero chemical waste products (the only by-product being water). Provided that the microbes are frequently fed and there is constant oxygen supply at the cathode, a MFC will continuously produce power; then, onboard a self-sustainable robot, the MFCs will continue to provide the energy needed to keep the artificial agent operating. Small MFCs and stacks as onboard power supplies The performance of an MFC is a function of microbial growth and metabolic rate, which can be limited by the ability and rate of electron transfer for respiration. These microorganisms form a stable semi-solid matrix on the electrode surface, which becomes permanently stuck, robust and resistant even at relatively high flow rates. New daughter cells or other microbes, which have no access to the electrode, are released and quickly flushed out under hydrodynamic flow conditions. The microbial reaction rates in MFCs are generally slow when compared with conventional combustion reactions within chemical fuel cells. Individual MFC units can, therefore, only produce low levels of absolute power (ca. 100 s of mW—1 s of mW) and with a thermodynamic limit of 1.14 V (max. open-circuit voltage). Thus, it becomes clear that a plurality of MFC units connected together will be necessary to step-up electrical output levels. A very important feature of MFCs is the inherent link between electricity generation and waste (sludge or urine) breakdown. This means that the higher the energy output levels, the better is the waste substance breakdown and the higher is the production of water at the cathode (two incoming elecChemSusChem 2012, 5, 1020 – 1026

trons and two protons per single water molecule). Although different workgroups worldwide employ different techniques, designs and approaches for the MFC technology, resulting in significant levels of energy density as reported in the literature, the challenge of scaling up for practical applications remains largely unsolved. It has nonetheless been shown that higher energy density levels and optimum biofilm/electrode surface area-to-volume ratios, reside within smaller scale MFCs. In conjunction with the need for plurality, one valid approach to scaling up would be through miniaturization and multiplication. The advantages of using small-scale units are that such systems are more dynamic, less resistive (internally), and have shorter migration paths for substrate input and proton (and thus electron) output in addition to their footprint payload for a mobile robot. Ringeisen et al.[6] showed increased power outputs from small (1.2 cm3) MFCs employing Shewanella oneidensis DSP10 in batch culture mode. Current outputs of up to 1.1 mA were recorded using anthraquinone-2,6-disulfonate as the synthetic electron mediator in the anode and potassium ferricyanide as the catholyte. This was a single MFC, which nevertheless demonstrated that higher energy densities could be achieved from smaller MFC units. In historical terms, Cohen was the first to connect a plurality of MFCs together and treat it as a stack.[7] The first report of a practical application powered by an MFC stack was that of the Gastrobot (also known as Chew Chew train) from Wilkinson in 2000,[8] three years before the first EcoBot[9] was reported (see the section on autonomous robots below). Aelterman et al. also developed a stack of six 60 mL MFCs that used ferricyanide cathodes, which were connected both electrically and fluidically, and reported increased power output densities when compared to individual MFCs.[10] The EcoBot series of robots has been an on-going real demonstration of how collectives of MFCs can be connected to power those robots, taking into account the challenges of fluidic interconnections and their effect on conductance and leakage currents.[11, 12] It is in fact the integration of multiple MFCs with an electronic control system—involving energy harvesting–management–distribution and peripherals[13]—that can allow the powering of any practical application and it is the interaction between the abiotic (mechatronics) and biotic (microbes) elements that has been described as artificial symbiosis. The sections that follow describe in detail the development of autonomous robots, and provide new data from on-going EcoBot experiments. Autonomous robots The first example of a robot collecting its energy—in the form of organic matter—from the environment, was produced by Kelly et al.[14] and called “Slugbot”. This was the world’s first attempt at creating a robot that could collect real organic food from its environment, that is, slug pests from muddy fields. Although this robot never reached the stage of utilising the caught prey, it nevertheless illustrated the potential of energy extraction from naturally occurring organic sources and conse-

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemsuschem.org

1021

I. A. Ieropoulos et al. quently deviation and independence from batteries and photovoltaics. In the same year, Wilkinson[8] developed “Gastrobot”, which for the first time demonstrated the consumption of table sugar by Escherichia coli and utilisation of the resultant nutrient-rich fluids through a stack of fuel cells for the recharging of the onboard batteries that were powering the various actuators. On the one hand, “Slugbot” had direction (foraging for slugs) and could collect its own ‘food’, but could not use it and on the other hand, “Gastrobot” had no direction (forward movement) and was ‘fed’ to recharge batteries. The first example of a robot that was solely powered by MFCs was EcoBot-I in 2003.[9] This robot (see Figure 1) employed eight MFCs con-

Figure 1. EcoBot-I fully assembled. Left: Front view showing the two motor actuators on either side and a balance wheel shown in the middle; Right: top view showing the eight-MFC stack split as 2  4 units and the electronic control circuitry. The two long twisted wires with flat ends are the two photodiode ‘eyes’ that were controlling phototaxis.

nected electrically in series as a stack that used Escherichia coli, methylene blue (anode) and ferricyanide (cathode) and could perform phototaxis. It was manually fed with table sugar and the electricity produced from the MFC stack was accumulated into a bank of six electrolytic capacitors with a total capacitance of 28.2 mF. The introduction of capacitors and the time involved in charging them (depending on the magnitude of current) to a pre-set threshold voltage have introduced a pulsated or intermittent behaviour that emphasises the essence of energy accumulation and management and the value ‘of doing nothing’ until sufficient energy has accumulated to continue with the task. EcoBot-I would therefore remain idle for 30 s and move towards the light for 3 s. EcoBot-I operated repeatedly as long as there was sugar fuel fed into the eight MFC anodes and the anodic methylene blue and cathodic ferricyanide mediators were maintained fresh, but there were no endurance experiments performed and its longevity has not been studied. In contrast, EcoBot-II demonstrated a more autonomous behaviour—albeit primitive—by operating over 12 days in the same pulsated manner without human intervention.[15] This robot (see Figure 2) employed eight MFCs with mixed sludge culture anodes and open-to-air cathodes without any mediators (anode and cathode) or catalysts. It was the world’s first robot to perform (i) sensing (ambient temperature), (ii) processing (onboard microcontroller), (iii) actuation (phototaxis) and (iv) communication (wireless transmission of temperature) all energised by the onboard MFCs. Furthermore, unrefined feedstock substrates such as

1022

www.chemsuschem.org

Figure 2. EcoBot-II fully assembled. Left: Front view showing the two motor actuators on either side and a balance wheel shown in the middle; Right: top view showing the eight MFCs around the circumference and the electronic control circuitry. The two photodiodes for controlling phototaxis are shown on the top left and right, whereas the long cable in the middle is the 1-wire temperature sensor.

rotten fruits, prawn shells and dead flies were shown for the first time to be consumed inside the 25 mL anodes of the onboard MFCs.[15] The duty cycle of this EcoBot-II was slower than that of EcoBot-I, as it would take 14 min to charge the onboard 28.2 mF capacitor bank to produce a 3 s energy burst for all the aforementioned tasks. Even though stacks of MFCs are more powerful than individual units, they are hitherto still only suitable for powering devices that require the lowest power. Such devices must be able to manage the energy—both produced and spent—by incorporating high efficiency “electron-harvesting” electronics, which extract electrons by adjusting the load and current output to persistently charge a capacitor (accumulator) at maximum sustainable power transfer. The examples of EcoBot-I and -II demonstrated this technique to power the robots in a “pulsed behaviour mode”,[9, 13, 15] but with no foraging behaviour and with simple electronics. EcoBot-III (see Figure 6) on the other hand was powered by a stack of 48 small-scale MFCs (two tiers of 24 MFCs each one on top of the other) and demonstrated a different level of energy autonomy by collecting food and water from its environment, digesting it, distributing the digest through the onboard MFCs using mechatronic units and getting rid of its own waste.[12] This required the generation and storage of sufficient energy to run a total of five motors ( two for locomotion and three for fluid distribution) and four pumps (three impeller-based and one peristaltic) in addition to the more complex electronic controller. The robot maintained microcontroller operation continuously (low-power mode) whilst carrying out periodic (high-power mode) tasks such as feeding, hydration and locomotion. EcoBot-III employed a 0.816 F (120  6800 mF at 6.3 V) bank of electrolytic capacitors, which allowed for an energy budget of 2 J to be exchanged. The voltage operating range (discharging voltage = 2.96 V; charging voltage = 1.9 V), was dictated by the symmetry around the intersection point between the actual capacitor charge curve and its first derivative (see Figure 3). Ingestion, digestion and egestion One of the main features of EcoBot-III was the design and development of mechanism(s) to allow the intake and processing

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemSusChem 2012, 5, 1020 – 1026

Microbial Fuel Cells for Robotics facilitate the excretion of this material in an effort to rid the microflora in this digestion unit from the accumulation of inhibitory waste by-products.

Fluidic isolation

Figure 3. Capacitor (0.4 F)-characteristic charge curve (black) and its first derivative (grey); ideal charge curve (dashed).

A sequential fluid (food and water) distributor was included onboard EcoBot-III to break the continuity of the liquid from the common artificial stomach to the multiple MFCs. This was an indexing-like mechanism (carousel), which was driven by one of the motors to increment its state by one position at a time so that all MFCs could be fed and hydrated in an isolated manner (Figure 5). On top of the carousel unit, there were

of food and evacuation of the waste products such as recalcitrant and inorganic matter. To this effect, a digestion unit was designed (see Figure 4) which incorporated a conical hat with

Figure 5. Top: computer-aided design (CAD) snapshots of the carousel feeding mechanism: left) complete feeding mechanism; right) mechanism uncovered showing the two separate fluid channels—one for feedstock and the other for water. Bottom: Photograph image of the actual carousel distributor (in situ).

Figure 4. EcoBot-III ingestion module designed for liquid as well as fly insect feeds. The conical hat is designed in such a way that liquids can be pumped into a catchment ‘lip’ for a liquid feedstock ‘diet’. When on fly insect ‘diet’, UV light is emitted for fly visual attraction, and a small amount of fly Z-9-tricosene sex pheromone is released for chemo-attraction. The pheromone pocket (0.5 mL, not shown) is located inside the stomach. The conical shape helps heavy-weight solids to accumulate at the bottom of this artificial stomach, where a peristaltic pump is used to periodically evacuate such waste. The underside of the conical hat (not shown) is black and the stomach has transparent windows to ensure that flies remain trapped.

added features (UV light, pheromone pocket and liquid collection lip) to allow the ingestion of either liquid food or fly insects. With regard to the latter, the rate of fly digestion was monitored in a separate artificial stomach, using time-lapse photography, where a known number of flies was shown to be fully digested within 1–2 weeks. Flies were collected outdoors using commercial fly-traps with the Z-9-tricosene pheromone attractant. Furthermore, the bottom part of the digestion unit onboard EcoBot-III was designed to allow the sedimentation of heavyweight particles and was connected to a peristaltic pump to ChemSusChem 2012, 5, 1020 – 1026

two additional smaller motor-driven distributors (one for food and one for water) that allowed the distribution of fluids over four quartiles at the same time. The amount of fluid flowing per feeding and hydration was intentionally excessive so that food (from the anode outlet) and water (from the cathode tray) of the MFCs on the top tier would overflow into the corresponding four MFCs on the bottom tier. The fully assembled EcoBot-III is shown in Figure 6. It must be emphasised that the whole robot, including the MFCs but excluding the electronics hardware, was fabricated by using a rapid prototype technology. This was extensively used throughout the EcoBot research programme. As explained above, the robot was designed to run on two ‘diets’: flies and liquid food from a feeder mechanism on the side-wall of the test bed arena (EcoWorld; see Figure 7). Although flies were manually fed in a sludge broth, which was then added into EcoBot-III’s artificial stomach (in a way starting the fly experiment on a ‘full stomach’), the experiment in which the robot attracts and catches its prey has not been performed and is currently on-going. The two feeding strategies nevertheless demonstrate the diversity of feedstock that can be utilised in MFCs running in a practical application.

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemsuschem.org

1023

I. A. Ieropoulos et al.

Figure 6. Left: EcoBot-III in its final state running autonomously in the EcoWorld. Two MFC tiers are shown in the middle; a Nanocure resin was used to 3D-fabricate the MFCs; the distance between the two tiers is used to avoid shunt losses. The electronic circuit board stack is not shown as it is behind the artificial stomach. The peristaltic pump is bolted to the bottom of this digester header tank to enable the removal of waste. The chassis is made from yellow acrylonitrile butadiene styrene and semi-translucent polycarbonate (ISO). The robot, when filled with fluid weighing 5.88 kg, was moving along a ‘railway track’ to maintain position when at the feeding or water stations. The duty cycle was 30–40 min of charge time and 10–15 seconds of actuation. Right: CAD drawing of the robot with the various parts labelled.

cycle ingestion!digestion!egestion. The arena along with the feedstock and water pumping-in mechanisms were externally powered and a separate (EcoWorld) microcontroller managed the inflow to the robot. EcoWorld telemetry was also recorded; for example, fluid levels of the liquid food and water containers were transmitted. This was the first attempt of establishing communication between the robot and its environment and avoiding waste of vital energy to reach the water or feeding station (goal) if this was empty. In terms of behavioural control, although simplistic in nature, EcoBot-III is programmed to remain at the feedstock distribution mechanism until the next actuation, when it can reverse and move away from the feed. It is also programmed to change its actuation sequence depending on where it is in the arena as well as on its fluid and energy levels. For example, the robot would continuously feed and hydrate the MFCs following food or water collection from the wall and once this was completed, it would then move away and towards the other side of the arena. In doing so, it intermittently recycled fluids from the collection trough and, once every 24 h, evacuated waste from the stomach. EcoBot-III experiments and longevity

Figure 7. EcoWorld arena with partially assembled EcoBot-III during initial testing. Acryllic material was used for making the arena, and the temperature was maintained at 30  5 8C. The arena dimensions were (w) 70 cm  (d) 100 cm  (h) 67 cm. Feeding and hydration occurred when the robot made contact with the micro switch attached to each mechanism.

EcoBot-III employed telemetry, through which it was transmitting its status to a remote receiver monitored by the human operator. The robot could communicate time-stamped data such as (i) which task has just been completed, (ii) what is the next task and at which direction, (iii) MFC voltage levels and (iv) fluid levels (food and water). Communication was maintained as a low-energy task so that in a case of low power emergency (approaching ‘lethal limits’) it could still be transmitting its status. The robot was designed to operate within a confined space (EcoWorld) through which it would be receiving its food and water and into which it could be excreting its waste, thus making it the world’s first to complete the

1024

www.chemsuschem.org

The same ‘stomach’ contents were fed with a further six houseflies of equal mass (i.e., twelve in total from the beginning of this experiment) and used to fill-up the artificial stomach onboard EcoBot-III to conduct the first experiment (that of a full stomach; experiment no. 1) without further liquid food collection from the arena. Once this line of experiments was completed, the robot was emptied and refilled with a fresh sludge-mixed culture that was allowed to acclimate. Then, the arena containers were filled with real wastewater (experiment no. 2), synthetic wastewater (experiment no. 3) and pure acetate (20 mm) in phosphate buffer (0.1 m, experiment no. 4). The longest run for EcoBot-III was seven days[12] when fully working and twelve days when only transmitting and maintaining the microcontroller. These limits were primarily dictated by mechanical failures on the robot. Although the MFCs were not failing, it was difficult to maintain fluidic distribution due to the various pumps failing and blocking as a result of the intermittent behaviour. This stopped the anodic feedstock and cathodic water supplies, which caused the open-to-air cathodes to dry from evaporation and the microbial anodes to starve. At the end of the twelfth day, EcoBot-III stopped transmitting and lost power to the microcontroller, that is, it went beyond its ‘lethal limits’ and came to a complete halt. Artificial symbiosis The series of EcoBots and the experiments carried out, as mentioned previously, have demonstrated the unique relationship between the live microbial ‘engines’ and the artificial mechatronic parts, in the sense that one system depends on the other. If the mechanics fail, then no fresh food will be collected and distributed to the microorganisms and no water will be supplied to the cathodes. On the other hand, if the microbes

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemSusChem 2012, 5, 1020 – 1026

Microbial Fuel Cells for Robotics die, then no electricity is produced to operate the mechanical parts of the system. This is an important relationship, which, apart from the behavioural significance that it carries in relation to the robotics field, becomes essential regardless of the application. The artificial symbiotic relationship exemplified by EcoBots has resulted in the coining of a new term known as “SymBots”.[16]

MFCs for autonomous robots There are a number of robots designed and developed with a certain degree of independence, mainly in the form computational autonomy but also energetic autonomy. For example, an indoors robot that can home in to a docking station could be seen as being energetically autonomous. Similarly, an outdoors terrestrial robot powered by photovoltaics could also be considered to be energetically autonomous. There are also examples of robots used in space exploration missions that are powered by radioisotope thermoelectric generators (RTGs), for example, Viking to Mars, and Pioneer, Voyager, Ulysses, Galileo and Cassini missions to the outer solar system.[17, 18] All these examples, under the boundaries of the environment they operate within, could be considered to be energetically autonomous. However, a robot powered by MFCs that allow the opportunistic utilisation of the widest diversity of substrate sources is one that will be able to operate in a plethora of environments including terrestrial, underground, underwater and (provided that organic matter exists) space. In the cases where the available organic matter is waste, such robots can also have a benevolent role in the environment, for example, organic pollution clean-up.

The way into the future appears to lie with smaller and higher numbers of MFCs connected as stacks. The latest model currently undergoing intense testing is of 1.4 mL total volume (0.7 mL per half-cell; see Figure 8 A–C). It is envisaged that this system will be more suitable for developing stacks of relatively small footprint and mass payload (Figure 8 D) and, therefore, ideal for mobile robotics. It is the authors’ intention to continue pursuing this direction in an effort to better understand the MFC technology and capitalise on its merits.

Figure 8. A–C) miniature-sized MFC with 1.4 mL total volume with continuous flow design. D) stack of 15 spherical MFCs undergoing early tests.

Acknowledgements The work has been funded over the years by UK Engineering & Physical Sciences Research Council (EPSRC), Grants EP/I004653/1, EP/H019480/1, EP/D027403/1 and GR/S80448/01 as well as by the EU FP-6 IP IST 027819 ICEA.

Summary For any large number of continuously fed units in stacks, there is an energy cost associated with the fluid flow of input and output. Although a gravity gradient may be helpful, there still are essential peripherals that are needed for the maintenance of such a stack. A true self-sustainable system must produce enough energy to run these peripheral maintenance modules.[16] This is not a trivial challenge, but unless it is met, such a system will never be a net energy provider. The engineering challenges faced in the development of the robotic applications are similar to those that will be met in the development of the MFC technology for any practical application. The mobility of the robotic agents increases the level of complexity and could perhaps be considered unnecessary in scenarios where the fuel can be supplied continuously (as opposed to be collected). There is an enormous effort from the MFC scientific community to develop MFC stationary systems for wastewater treatment applications, and the recent discovery that urine can be directly utilised in MFCs[19] potentially opens up possibilities for energy recovery perhaps never thought of before. ChemSusChem 2012, 5, 1020 – 1026

Outlook

Keywords: fuel cells · microbes · robots · stacks · symbiosis

[1] B. Webb in Proceedings of the Towards Intelligent Mobile Robots Conference (Eds.: U. Nehmzow, C. Melhuish), TIMR, Bristol, 1999. [2] W. G. Walter, Sci. Am. 1950, 182, 42 – 45. [3] D. McFarland, E. Spier, Robot. Auton. Syst. 1997, 20, 179 – 190. [4] B. E. Logan, B. Hamelers, R. Rozendal, U. Schrçder, J. Keller, S. Freguia, P. Aelterman, W. Verstraete, K. Rabaey, Environ. Sci. Technol. 2006, 40, 5181 – 5192. [5] N. Beecroft, F. Zhao, J. R. Varcoe, R. C. T. Slade, A. Thumser, C. AvignoneRossa, Appl. Microbiol. Biotechnol. 2012, 93, 423 – 437. [6] B. R. Ringeisen, E. Henderson, P. K. Wu, J. Pietron, R. Ray, B. Little, J. C. Biffinger, J. M. Jones-Meehan, Environ. Sci. Technol. 2006, 40, 2629 – 2634. [7] B. Cohen, J. Bacteriol. 1931, 21, 18 – 19. [8] S. Wilkinson in Proceedings of the 2000 IASTED Int. Conference on Robotics and Applications, (Ed.: M. H. Hamza), IASTED/Acta, 2000, Paper 318 – 037. [9] I. Ieropoulos, C. Melhuish, J. Greenman, in Proceedings of the 7th European Conference in Artificial Life, (Eds.: W. Banzhaf, T. Christaller, P. Dittrich, J. T. Kim, J. Ziegler), Springer, Berlin, 2003, pp. 792 – 799.

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemsuschem.org

1025

I. A. Ieropoulos et al. [10] P. Aelterman, K. Rabaey, HT. Pham, N. Boon, W. Verstraete, Environ. Sci. Technol. 2006, 40, 3388 – 3394. [11] I. Ieropoulos, J. Greenman, C. Melhuish, Int. J. Energy Res. 2008, 32, 1228 – 1240. [12] I. Ieropoulos, J. Greenman, C. Melhuish, I. Horsfield in Artificial Life XII (Eds.: H. Fellermann, M. Dçrr, M. M. Hanczyc, L. L. Laursen, S. Maurer, D. Merkle, P-A. Monnard, K. Stoy, S. Rasmussen), MIT Press, Cambridge, MA, 2010, pp. 733 – 740. [13] I. Ieropoulos, J. Greenman, C. Melhuish, I. Horsfield in Bioelectrochemical Systems: From Extracellular Electron Transfer to Biotechnological Applications (Eds.: K. Rabaey, J. Keller, P. Lens), IWA Publishing, London, 2009, pp. 409 – 420. [14] I. Kelly, O. Holland, C. Melhuish in Proceedings of the 5th International Symposium on Artificial Life and Robotic for Human Welfare and Artificial Liferobotics (Eds.: M. Sugisaka, H. Tanaka), AROB, 2000, pp. 470 – 475. [15] C. Melhuish, I. Ieropoulos, J. Greenman, I. Horsfield, Auton. Robot. 2006, 21, 187 – 198.

1026

www.chemsuschem.org

[16] J. Greenman, I. Ieropoulos, C. Melhuish in Modern Aspects of Electrochemistry No. 52: Applications of Electrochemistry and Nanotechnology in Biology and Medicine I (Ed.: N. Eliaz), Springer, Berlin, 2011, pp. 239 – 285. [17] G. Bennett, J. Lombardo, B. Rick in Space Nuclear Power Systems (Eds.: M. El-Genk, M. Hoover), Orbit Book Co., Malabar, FL, 1987, pp. 437 – 540. [18] G. Bennett, J. J. Lombardo, R. J. Hemler, J. R. Peterson in Proceedings of the 21st Intersociety Energy Conversion Engineering Conference (IECEC’86), San Diego, CA, 1986, pp. 1999 – 2011. [19] I. Ieropoulos, J. Greenman, C. Melhuish, Phys. Chem. Chem. Phys. 2012, 14, 94 – 98.

Received: April 23, 2012

 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemSusChem 2012, 5, 1020 – 1026