Over the last years, the expansion of urban areas has motivated the development of technologies such as green roofs, which in turn increases the amount of green areas in urban settlements and a series of other positive outcomes. Some of the additional benefits related to green roofs are runoff reduction [1
], mitigation of heat island effects [2
], sound and thermal insulation [3
], creation of new habitats and increments in biodiversity [4
], and an improvement in air quality [5
]. These benefits explain the substantial expansion of green roofs in Western Europe, North America, and in some countries in the southern hemisphere, such as Australia and Chile [6
In countries where climatic conditions are different to those of the northern hemisphere, mainly in rainfall and temperature, the construction and maintenance of these green areas should consider water availability during dry seasons. Even though the use of drought-tolerant plants, such as Sedum
species, can reduce the amount of water needed for irrigation, there is still an excessive use of this resource in these systems. For example, a previous study, conducted in a semiarid climate during the dry season, demonstrated that an extensive green roof planted with Sedum
could require up to 60 mm d−1
of water, which is approximately ten times larger than the evapotranspiration of reference [8
]. For this reason, in semi-arid climates an efficient use of water is essential. To achieve this goal, continuous monitoring of water content in green roofs through water sensors provides a sustainable use of this critical resource. However, according to our experience monitoring green roofs [8
], each of the typically-used sensors could cost from US$
400 to US$
1800; making this application less attractive for commercial and residential green roofs.
Microbial fuel cells (MFCs) are devices that allow bacteria to generate energy from organic and inorganic compounds. MFCs have been mainly studied for energy generation or power supply to sensors in remote locations, where battery replacement is not feasible [10
]. In addition, these systems have been studied as an alternative for remediation [11
], as biosensors of biological oxygen demand [12
], chrome concentration [14
], copper and cadmium concentration [15
] and for corrosion monitoring [16
Currently, these applications have been mainly tested under water-saturated conditions such as those proper of sediments, where water allows both the maintenance of anoxic conditions in the anode and the reduction of the internal resistance of the system [17
], facilitating ionic mobility between electrodes. However, under unsaturated conditions, as is the case of many soil environments, MFCs have been less studied due to the high internal resistance as result of the poor conductivity the soil and the ionic mobility that these conditions impose. Thus, in these systems, power generation depends on water content, by decreasing the water content in soil, cell voltage and current decreases [18
]. In spite of this, soil ecosystems have the potential to operate MFCs using plants in association with soil microbes (plant microbial fuel cells—PMFCs), taking advantage of the large soil biodiversity and the continuous supply of organic compounds by plant roots [20
]. Therefore, considering the relationship between water content and current, a PMFC could be implemented to monitor the water content in constructed areas, such as green roofs, expanding the range of applications of MFCs.
Therefore, this study aims to evaluate the performance of PMFCs systems installed in a laboratory scale green roof under non-controlled environmental conditions. Performance was evaluated in terms of current generation and its relation with the water content of the soil and solar radiation.
2. Materials and Methods
Plant Microbial Fuel Cell (PMFC) Construction and Operation
PMFCs were evaluated at a bench-laboratory and on a pilot scale in a green roof setting. The bench-laboratory experiments were performed prior to the pilot scale experiments under controlled temperature conditions (20 ± 2 °C) during 360 days. In these experiments, eight reactors consisting of plastic pots (450 mL) were filled with a plant growth substrate of a mineral matrix of 67% sand, 20% silt, 13% clay and 2.9% organic matter. Seven reactors had one seedling of each Sedum
species (S. kamschaticum
, S. rupestre
, S. album
, S. hybridum
, S. spurium
, S. sexangulare
and S. reflexum
) and one was left without plants as a control. These reactors were irrigated every week until the water drained from the bottom of the pot. Each PMFC consisted of two carbon-based electrodes: one buried at 7 cm made of circular fiber felt (92 mm diameter and 6 mm thickness); and a second electrode placed on at the surface of the substrate made of activated carbon granules (57 cm3
of volume) and graphite rod (6 mm diameter and 3 cm length) as connector. Titanium wires were used to connect the two electrodes to an external resistance of 1 kΩ. Reactors were inoculated with mud from a pond used as the water supply for irrigation located in the Campus San Joaquin at Pontificia Universidad Católica de Chile (PUC) in Santiago de Chile (33°29′ S, 70°36′ W) [9
The pilot-scale experiments were conducted in the Laboratorio de Infraestructura Vegetal (LIVE), located in the Campus San Joaquín. LIVE consists of 17 modules (4 m2
) mostly covered by Sedum
species where the thickness (5–20 cm) substrate and four drainages systems commonly used in green roofs were tested [8
]. To evaluate the relationship between soil water content and cell voltage, three identical modules with the same substrate depth (10 cm) and drainage system planted with the seven Sedum
species previously described, were selected to set-up triplicate PMFC reactors (PMFC 1, PMFC 2 and PMFC 3) at 20–30 cm distance from the sprinkler used for irrigation.
The reactors in LIVE were operated for 315 days. In the first 200 days, irrigation was applied automatically by the sprinklers, supplying 18 mm day−1, 60 mm day−1 and 15 mm day−1 between days 0–26; 27–136 and 137–200, respectively. After this time, sprinkler irrigation was suspended and water was supplied by natural rainfall.
Cell voltage was measured once per week during the first 280 days of operation using a multimeter UT55 (Uni-Trend Technology Limited, Dongguan, China). After that time, and for detailed measurements of the green roof PMFC reactors, data were collected every 10 min using data logger EM50 (Decagon®, Pullman, WA, USA). Power and current densities were normalized using the projected area of the buried carbon felt (circular electrode of 0.0067 m2). Polarization and power density curves were made, varying external resistance from 50 Ω to 80 kΩ every 45 min (chosen based on the period of time observed necessary to reach steady-state measurements in the bench-laboratory experiments).
Environmental parameters such as solar radiation, temperature, relative humidity and precipitation were measured every 5 min by a photosynthetically active radiation (PAR) sensor, air temperature sensor (ECT) and high-resolution rain gauge (ECRN-100) (Decagon®, Pullman, WA, USA), respectively. Substrate water content, temperature and electrical conductivity were measured every 5 min by a GS3 sensor (Decagon®, Pullman, WA, USA).
The PMFCs performance showed differences in relation to previous studies in two aspects, that is, start-up time and radiation. The start-up period was shorter than those reported under laboratory conditions, where the start-up time was observed in more than 100 days [9
]. This difference in the start-up may be due to the fact that the PMFCs tested in this study were installed on a green roof built about a year before the experiment, which could lead to a greater number of roots and therefore a higher organic matter content available for microorganisms, allowing the fast establishment of electroactive microorganisms (EAMs) in anodes. The solar radiation during the testing period was greater than that previously used in the laboratory for PMFC experiments, which does not exceed 596 ± 161 μmol m−2
]. However, a relationship between solar/light radiation and current was not observed in contrast to previous PMFC studies conducted with wetland plants [24
]. There seems to be a slight increase in current during the night, which could be related to variations in the moisture content allowing an increase of ion mobility through the substrate or affecting the release of exudates from the roots. Previous studies had reported that the exudation of carbohydrates increases with the soil water content [26
]. Crassulacean acid metabolism of Sedum
might also lead to exudation of organic compounds in darkness periods when carbon is fixed. Hence, the plant metabolism can affect the exudation response to different factors, for example light intensity [27
]. However, to the best of our knowledge, in the literature, there is no conclusive evidence related to this process. Further detailed research is needed to understand the root exudation of CAM species.
The maximum power density observed in PMFCs in LIVE (114.6 μW m−2
) showed values lower than those reported for PMFCs using wetland plants (6–222 mW m−2
], that is expected considering no waterlogging condition. However, they are greater than those obtained under laboratory conditions using each of the seven species of Sedum
, where S. hybridum
achieved the highest power density (92 μW m−2
) followed by S. rupestre
(15.5 μW m−2
]. Similar substrate temperatures suggested that this parameter is not responsible for the observed differences in the power density curve, but there seems to be a relationship between two variables: power and water content, due to a lower value being observed for the reactor with the lowest power density (PMFC 2), while the PMFCs with the highest and the second highest power density had higher moisture values. However, PMFC 3, with the highest current density, did not have a significantly larger amount of water in the substrate, suggesting that there is an additional factor that could influence power generation. In laboratory experiments, differences in the anodic community were observed, where the PMFC with the highest power density showed a higher abundance of bacteria of the family Micrococcaceae
, suggesting the importance of bacterial community structure for performance [9
]. Dunaj et al. proposed that soil characteristics and bacterial community structure are key factors in the performance of soil-based MFC [29
]. Further research is necessary to study the effect of these parameters on power production in PMFC systems settled in actual natural or engineered environments.
A similar behavior between current density and water content was seen, where an increase in the current and water content is observed as a result of irrigation. This is expected because the water allows better cation transport between electrodes and anoxic conditions in the anode [18
]. Additionally, high correlation coefficient values were obtained, demonstrating a correlation between both the two variables. This suggests that PMFCs could be used as moisture biosensor, mainly in semiarid climate where water is a critical resource and efficient management is needed.
Soil moisture is an important parameter because it not only affects plant growth and development, but also microbial activity and biogeochemical cycling of nitrogen and carbon [30
]. For this reason, water content measurements are needed to keep a suitable water supply. The use of soil moisture sensors provides a reliable alternative for improving the water use efficiency, as water is applied in response to soil moisture variation instead of using a regular schedule. However, the commercial moisture sensors are relatively expensive for domestic or commercial applications in large natural and engineered environments. Thus, PMFC emerges as a new alternative to provide low-cost devices that could be designed covering large surface areas using granular carbon-based materials, without a negative impact on the soil/substrate microbiome [9
]. Similar to conventional resistance type moisture sensors, which rely on variations in resistance to the current as a function of soil water content [31
], the PMFC can measure the current generated, which changes in function of the soil resistance that relies on soil moisture. Although further studies and calibration steps are needed to design an optimized PMFC reactor, this type of biosensor could be a self-sustainable device, including a potential additional advantage in relation to conventional moisture sensors.