The startup of microbial fuel cells (MFCs) is known to be prone to failure or result in erratic performance impeding the research. The aim of this study was to advise a quick launch strategy for laboratory-scale MFCs that ensures steady operation performance in a short period of time. Different startup strategies were investigated and compared with membraneless single chamber MFCs. A direct surface-to-surface biofilm transfer (BFT) in an operating MFC proved to be the most efficient method. It provided steady power densities of 163 ± 13 mWm−2 4 days after inoculation compared to 58 ± 15 mWm−2 after 30 days following a conventional inoculation approach. The in situ BFT eliminates the need for microbial acclimation during startup and reduces performance fluctuations caused by shifts in microbial biodiversity. Anaerobic pretreatment of the substrate and addition of suspended enzymes from an operating MFC into the new MFC proved to have a beneficial effect on startup and subsequent operation. Polarization methods were applied to characterize the startup phase and the steady state operation in terms of power densities, internal resistance and power overshoot during biofilm maturation. Applying this method a well-working MFC can be multiplied into an array of identically performing MFCs.
Microbial fuel cells (MFCs) are a promising yet challenging new technology. MFCs utilize microorganisms immobilized on an electrode as biocatalyst to generate electricity by metabolizing biodegradable organic compounds (Logan 2008). Industrial or municipal wastewater treatment is considered one of the most promising fields of application for MFCs, utilizing wastewater as fuel (Logan 2008; Liu et al. 2011; Oliveira et al. 2013; X. Zhang et al. 2013).
The performance of an MFC is dependent on the bioelectrocatalytic activity of the biofilm, which is predetermined during startup (Oliveira et al. 2013). Startup times can range from 2 to 103 days (Pant et al. 2010; X. Zhang et al. 2013); even failure of startup has been reported (Hays et al. 2011).
An MFC's development of power densities can be associated with microbial growth stages: lag phase ‘I’, log phase ‘II’, stationary phase ‘III’. The formation of an electroactive biofilm from primary inoculants is preceded by microbial acclimation (lag phase). Before generating current, the microorganisms have to adapt to the conditions inside an MFC as those are different from their natural environment; e.g. pH may be different. The substrate may require synthesis of selective enzymes and most importantly the anode needs to be colonized and utilized as electron acceptor. This acclimation and adaption process is accompanied by a shift in biodiversity. This shift may continue on, into and throughout steady state operation (Zhao et al. 2012; Higgins et al. 2013). Acclimation can be accelerated by providing conditions that resemble the indigenous environment of the inoculant bacteria. However, these conditions do not necessarily favour the development of electroactive properties of the microbial community. Consequently, there are two competing goals: avoiding long acclimation and forcing anode respiration, which inescapably goes along with acclimation and changes in biodiversity (Rismani-Yazdi et al. 2011).
Different approaches to improve MFC startup have been investigated, including: poised anode potentials (F. Zhang et al. 2013); chemical amendments to wastewater used as substrate (Liu et al. 2011); cell configuration and temperature (X. Zhang et al. 2013); chemical or physical treatment of the anode surface (Flexer et al. 2013); effects of substrate on startup (Patil et al. 2011). Inoculating an MFC with pre-acclimated cultures from another MFC is consistently reported to result in improved startup and performance. This is often done by using effluent of another MFC (Liu et al. 2011; Zhang et al. 2011) or by scraping the biofilm off a populated anode and applying it to a fresh anode (Kim et al. 2005). Power densities of these secondary biofilms (as they are derived from primary biofilms) can be several times that of the primary biofilms (Kim et al. 2005; Liu et al. 2008; Baudler et al. 2014).
Liu et al. (2008) investigated the improvement of anodic bioactivity by consecutive selection. The authors submersed a fresh anode and a populated anode in a sterile substrate solution and poised the anodes at 0.2 V vs Ag/AgCl. The disposal (detachment and attachment) of microbes from the populated anode to the fresh anode produced superior electroactivity in the secondary biofilm. The lag phase in which no microbial based currents could be observed could be reduced to 24 hours; the stationary phase was reached after 140 hours compared to 400 hours after inoculation with wastewater. The resulting current densities of the secondary biofilm were twice as high as of the primary biofilm. Third and fourth generation biofilms, however, did not display any further improvements compared to the secondary biofilms.
The main limiting factor in MFC startup is the microorganisms’ ability to acclimate to work as an electroactive community. This process can be furthered or hindered by the conditions during startup; e.g. the properties of the substrate medium, anode material and surface, the choice of inoculant, and the anode potential. Secondary biofilms start up faster and result in superior MFC performance. The microorganisms are already acclimated to work as an electroactive community. The startup limitations of primary inoculants do not apply to secondary biofilms. Previously reported startup methods for secondary biofilms either require an electrochemical workstation (potentiostat) for potential control or a primary anode has to be damaged or sacrificed to inoculate a fresh anode (scraping off the biofilm). The most commonly used method to establish secondary biofilms is the usage of effluent from another MFC. However, the biofilm needs to be established from suspended microbes and the cathode is inoculated as well in separator-less MFCs.
The goal of this study was to develop a startup strategy that quickly establishes steady state conditions. An important issue was the reduction of microbial acclimation, as this is the most error-prone step in startup. Several strategies were investigated and evaluated; the biofilm transfer (BFT) method described hereafter clearly provided the best results.
MFC setup and operation
All experiments were carried out in tubular shaped membraneless single chamber MFCs with direct air direct methanol fuel cell (DMFC) cathodes (12.56 cm2, 0.5 mg cm−2 Pt, Elcomax, Germany). The electrode spacing was 12 cm; the netvolume 190 ml; the anodes were comprised of two layers of carbon felt (Sigracell®, KFD 2.5 EA, SGL Group, Germany, 12.56 cm2) fixed together via conductive two component epoxy adhesive (EC 261C, Polytec PT, Germany). A stainless steel contact wire (X5CrNi18-10) was fixed in the bonding area between the two carbon felt layers. The adhesive had to harden at 150 °C for 2 hours. This process may have led to some degree of unintended heat treatment of the anodes. The MFCs were operated in batch mode with synthetic growth medium as substrate. The chemical oxygen demand (COD) of the substrate solution was 3,500 mg l−1 (HACH LANGE cuvette tests) where NaAc·3H2O (3.88 g l−1) and glucose (1.64 g l−1) each provided half the COD. Nutrient salts (50 mg l−1 K2HPO4, 0.30 g l−1 (NH4)2HPO4, 56 mg l−1 Na2HPO4, 34 mgl−1 Na2SO4, 10 mg l−1 FeSO4·7H2O, 1.30 g l−1 Na3PO4·12H2O, 0.80 g l−1 CH4N2O), vitamins and trace elements were added according to Vogl et al. (2016). The resulting conductivity of the substrate solution was 3.58 mS cm−1 (HACH Sension 156). The substrate was replaced when the COD fell below 500 mg l−1. The MFCs were unstirred and operated under ambient room temperature varying from 21.0 to 23.5 °C.
Measurements and analysis
The cell voltages were monitored and logged via a multi-channel data logger ALMEMO 5690-2M09 (Ahlborn, Germany). In order to determine the stage of microbial acclimation and biofilm formation polarization curves were measured from resistance (R) = 20 kΩ to R = 100 Ω. The IV-characterization via varying external resistances is a commonly used method to describe MFCs. Quasi-stationary conditions should be achieved for every R before the cell voltage is taken. Otherwise too high (decreasing order of R) or too low (increasing order of R) voltages and power densities are measured. In order to achieve quasi-stationary conditions the MFCs were put in open circuit (OC) mode 2 hours prior to the measurements. Each resistance was applied for 15 minutes. The power curve was calculated using Ohm's law where power in watts (P) = V2/R where V is voltage in volts and R is resistance in ohms (Ω), and plotted over the current density. All power and current densities are normalized to the projected surface area of the equally sized electrodes (12.56 cm2, one side). The trend of power densities during startup gave information about the development of the exoelectrogenic biofilm.
A conventional inoculation with anaerobic sludge (ANS) served as reference for the BFT. ANS was favoured because of the well-known interspecies electron transfer accompanying the anaerobic digestion of biomass. It originated from a laboratory-scale reactor for the digestion of microalgae biomass. The sludge was diluted (2 g l−1) and aerated for 2 h to inhibit methanogens. During inoculation the MFCs were short circuited to the cathode in order to provide the highest possible anode potential and thus the maximum theoretical energy gain for the microorganisms. It has been reported that higher anode potentials favour biomass production during startup (Liu et al. 2008; Rismani-Yazdi et al. 2011; Carmona-Martínez et al. 2013). After 48 hours the MFCs were emptied, the cathodes rinsed to remove residual sludge, and the substrate added into the MFCs. The MFCs remained short circuited. The progress of MFC startup was monitored by power and polarization curve measurements. Once constant power densities were achieved, a fixed external load of 1 kΩ was applied.
The placement of the anodes on top of each other limited diffusion into and from the contact area. The resulting ion build-up in the contact area and the spreading of the biofilm reduced the contact resistance. This decrease in contact resistance provides a way of monitoring the BFT progress. If the anodes are kept in contact too long, the contact resistance and therefore the potential drop between the anodes will increase again. This is likely due to nutrient depletion and subsequently reduced biofilm activity, leading to an increase in biofilm resistance. 24 hours was found to be sufficient for the microorganisms to migrate onto the secondary anode and still prevent nutrient depletion in the contact area (increase in potential drop). After the transfer, the secondary anode was placed in an MFC of its own (secondary BFT-MFC); the primary anode remained in the primary MFC. The startup process was monitored via IV measurements; the MFC remained short circuited until stable power densities were obtained.
BFT only requires an operating MFC to establish a secondary biofilm. No electrochemical workstation is necessary, no anode must be damaged to inoculate another, and a fresh anode can be inoculated without inoculating the cathode at the same time. Thus, BFT can overcome the main disadvantages of previously reported startup strategies for secondary biofilms.
RESULTS AND DISCUSSION
Biofilm transfer compared to conventional inoculation
The MFCs inoculated via BFT displayed rapidly increasing power densities. No microbial lag phase ‘I’ could be observed after the inoculated anodes were put in MFCs of their own (‘BFT’ in Figure 2). The BFT-MFCs reached stationary power densities within 8 days compared to ∼30 days after conventional inoculation. The power densities remained stable until the end of the experiment (chronologically the black curve would start at day 60) at 124 ± 15 mWm−2 with one significant drop after the addition of fresh substrate. This shows how sensitive young and immature biofilms may react to changes in environmental conditions.
An experimental repetition of the first experimental series was carried out (data not shown). In this repetition the BFT-MFCs did show a lag phase similar to the ANS-MFCs, as a redox shock was suspected to have caused the lag phase after the anodes were put in MFCs of their own and fresh substrate was added. To avoid a lag phase it is crucial that the conditions inside the secondary MFC are similar, at best identical, to the MFC the BFT took place in. Observing this critical aspect resulted in further improvement of startup behaviour (see the following section).
Second experimental series
In order to ensure similar environmental conditions in the primary and secondary MFCs the following methodological changes were applied in this experimental repetition. (1) The substrate medium was stored in a sealed bottle at room temperature for 48 hours to reduce suspended electron acceptors and to lower the redox potential of the substrate medium to below the anode's potential. COD loss was negligible; the redox potential could be reduced by 471 ± 47 mV. Anaerobic pretreatment of the substrate medium was reported to improve MFC startup (Rodrigo et al. 2009). (2) Effluent from the respective primary MFC (10%) was added into the secondary MFC at the first substrate addition. By this, suspended enzymes and mediators present in the primary MFC were introduced into the secondary BFT-MFC. (3) The BFT was carried out well within a batch cycle (7 days), not at its beginning or end. To maintain identical experimental conditions, these three changes were applied to both BFT and ANS startups alike.
The methodological improvements also affected the ANS-MFCs, which reached stationary operation after 9 days (Figure 3(c)) compared to 30 days in the first series (Figure 2). Their power densities kept increasing in the stationary phase ‘III’ reaching the power densities of the BFT-MFCs after 16 days. However, secondary biofilms (BFT) are consistently reported to outperform primary biofilms (ANS) (Kim et al. 2005; Liu et al. 2008; Baudler et al. 2014). It stands to reason that the effluent from a primary MFC, which was added to the first substrate addition, introduced pre-acclimated suspended exoelectrogens into the ANS-MFCs. These might have enriched the anode biofilm, slowly attributing secondary biofilm properties to the primary biofilm. The gradual OCP build-up of the ANS-MFCs (Figure 3(b)) shows that the anode was only slowly being utilized as microbial electron acceptor by the suspended microbes.
Power curve development and power overshoot during startup
During the IV measurement, the external resistance is consecutively decreased. Thereby, the cell voltage drops and the current increases. In case of power overshoot the MFC's current increases beyond its actual short circuit current (SCC) and then doubles back to the SCC (Figure 4). There are several papers dealing with this issue. The reasons are not fully understood (Zhu et al. 2013). Insufficient biofilm maturation is linked to power overshoot (Winfield et al. 2011); however, there is no consent on the subject. Watson & Logan (2011) reported that biofilm enrichment does not prevent power overshoot. Hong et al. (2011) found that power overshoot can be eliminated by adapting an MFC to high currents using low external loads; whereas high loads lead to power overshoot. Zhu et al. (2013) confirm that a high anode potential (e.g. low external load) prevents power overshoot. A lack of anodic capacitance was also reported to be the cause of power overshoot (Peng et al. 2013). Anodic capacitance, the transient charge storage within the anode biofilm, is affected by both operation conditions and biofilm maturation. Therefore, a lack of capacitance and power overshoot may be related effects of the same root cause, e.g. acclimation to high external loads. In our study, power overshoot occurred during the first days after primary inoculation and BFT. The overshoot occurred while the MFCs were still short circuited and the anode potential was high. This observation points towards insufficient biofilm maturation as cause of power overshoot, rather than an acclimation to low anode potentials. Power overshoot ceased within 2 to 3 days after the external load (1 kΩ) was applied. The external load was about twice as high as the internal resistances of the MFCs, which were determined from the slope of the IV-curves as 505 ± 57 Ω and 524 ± 126 Ω for the BFT-MFCs and the ANS-MFCs, respectively. The ceasing of the power overshoot after the application of the high external loads is inconsistent with the findings of Hong et al. (2011). However, the high internal resistance of our MFCs resulted in high ohmic losses limiting MFC current. During IV measurements, the doubling back point where current demand exceeds the biofilm's ability to provide current may not have been reached any more after day 9. In all experiments, power overshoot was more pronounced at MFCs inoculated via BFT than at the reference MFCs inoculated with ANS. To what extent biofilm maturation and/or the applied external resistance were responsible for the ceasing of the power overshoot was not determined.
All reference MFCs, inoculated with ANS, displayed a characteristic microbial lag phase ‘I’. During this lag phase metabolic activity and growth is inhibited, e.g. by a lack of substrate specific enzymes. Electroactivity in the community is low since there is no metabolic electron liberation from substrate oxidation, yet. The utilization of the anode as electron acceptor must prevail over alternative metabolic pathways (OCP development, Figure 3), making the MFC vulnerable to undesired developments or even startup failure during the lag phase. In all ANS-MFCs, acclimation (lag phase ‘I’) was up to twice as long as the log phase ‘II’. In the stationary phase ‘III’ the ANS-MFCs still displayed signs of microbial changes, e.g. the steady increase in power densities in ‘III’, Figure 3(c).
The BFT completely eliminated the error-prone lag phase ‘I’ in BFT-MFC startup and resulted in stable and steady power densities after half the startup time of the ANS-MFCs (Figures 2 and 3). No changes in MFC performance were observed in the stationary phase ‘III’. Erratic MFC behaviour, often appearing if flat anodes are used (F. Zhang et al. 2013), can be avoided. All changes in the microbial community structure that follow a primary inoculation had already occurred in the primary (precursor) MFC prior to BFT. The microbial community is as a result of all differentiations in biodiversity and selection processes that had occurred during the time of operation of the primary MFC (selection and enrichment of exoelectrogens). The advanced microbial socialization (syntrophy, symbiosis, interspecies electron transfer) and possibly even parts of the functional morphology of the biofilm matrix (ecological niches) are transferred along with the biofilm.
Any bacterial attachment and biofilm formation on the anode is preceded by the formation of a conditioning film, an adhesion of molecules from the medium matrix onto the anode surface. The conditioning film can have significant impact on biofilm development (Hwang et al. 2013). Since the BFT is carried out inside an operating MFC, the conditioning film consists only of chemical species inherent to an MFC. These are unlikely to hinder but rather favour the formation of an electroactive biofilm.
A freshly transferred community lacks the protection of a matured biofilm; it is vulnerable to environmental changes (Figure 2, no. 2). To avoid a lag phase after BFT, the conditions inside the secondary MFC (BFT-MFC) must resemble the conditions of the primary (precursor) MFC. The properties of the first substrate addition to a BFT-MFC should be adjusted accordingly (pH, conductivity, redox potential, temperature, COD, etc.).
BFT can also serve to enrich exoelectrogens from mixed culture biofilms (e.g. from wastewater) by transferring them onto a new anode (microevolution and selection) as an alternative to the method proposed by Liu et al. (2008). The BFT is easier to apply, does not require a potentiostat for potential control, and the secondary biofilm does not need to be established from planktonic microorganisms (no lag phase). However, only one anode can be inoculated per MFC at a time and BFT is not applicable to all anode geometries. The method of Liu et al., in theory, allows for the inoculation of multiple anodes simultaneously using only one populated anode.
The study provides an easy-to-apply startup strategy for laboratory-scale MFCs. It is meant as a guide to establishing defined conditions for reproducible experiments. A total of 12 MFCs were put into operation via BFT. With the exception of the previously stated test series in which the MFCs did display a lag phase, all MFCs started up without complications. The surface-to-surface transfer of an electroactive biofilm ensures well working electroactive microbial communities from primary inoculants. A microbial lag phase ‘I’ as part of startup can be avoided entirely. This results in shorter startup times and improved operation performance, thus providing the basis for reproducible and comparable research with well-performing electroactive biofilms.
The authors would like to acknowledge the ‘Bavarian State Ministry of Sciences, Research and the Arts’ for funding of the research.