A plant-like battery: a biodegradable power source ecodesigned for precision agriculture

The natural environment has always been a source of inspiration for the research community. Nature has evolved over thousands of years to create the most complex living systems, with the ability to leverage inner and outside energetic interactions in the most efficient way. This work presents a flow battery profoundly inspired by nature, which mimics the fluid transport in plants to generate electric power. The battery was ecodesigned to meet a life cycle for precision agriculture (PA) applications; from raw material selection to disposability considerations, the battery is conceived to minimize its environmental impact while meeting PA power requirements. The paper-based fluidic system relies on evaporation as the main pumping force to pull the reactants through a pair of porous carbon electrodes where the electrochemical reaction takes place. This naturally occurring transpiration effect enables to significantly expand the operational lifespan of the battery, overcoming the time-limitation of current capillary-based power sources. Most relevant parameters affecting the battery performance, such as evaporation flow and redox species degradation, are thoroughly studied to carry out device optimization. Flow rates and power outputs comparable to those of capillary-based power sources are achieved. The prototype practicality has been demonstrated by powering a wireless plant-caring device. Standardized biodegradability and phytotoxicity assessments show that the battery is harmless to the environment at the end of its operational lifetime. Placing sustainability as the main driver leads to the generation of a disruptive battery concept that aims to address societal needs within the planetary environmental boundaries.


Derivation of the evaporation rate equation
The evaporation rate equation for a liquid in evaporation was derived using the continuity equation at the interface of the two phases, where the process is occurring. For a control volume within the evaporation interface is given by: where Q is the evaporation rate [g·s -1 ], S is the surface of the control volume [m 2 ], is the gaseous phase density [g·m -3 ], ⃗ is the particles velocity at S [m·s -1 ], and ̂ is the normal vector of the surface S [adim].
While evaporating, there's only flow on the top face of the cylinder, of area A, leaving the integral as follows: Since mass transport from the interface surface until bulk region is not driven neither by migration nor convection (assuming forced convection from an external air-flow is null, given laboratory conditions), only diffusion is left as the main cause. Thus, diffusion flow J from a substance in dissolution is given by: where is the diffusion coefficient of the solute [m 2 ·s], and ∆ is the concentration gradient of the solute between two points in space Applying the diffusion flow equation to this situation, the concentration gradient between two opposite spots of the control volume will be determined by the humidity of the air at each point: liquid-gas interface is saturated at 100% of humidity (Hsat), while at bulk region there will be the room humidity (Hr). This way, the following expression is obtained: Since the room humidity can be expressed in terms of the relative humidity (in percentage) times the saturation humidity, the previous expression rewrites as follows: where D is the diffusion coefficient of the water vapour [m 2 ·s] (dependent on T at P=cnst.), is the saturation humidity [g·m -3 ] (dependent on T at P=cnst.), % is the relative humidity [adim] (in percentage), and ∆ is the distance between the two studied points.
Since J is a flow of mass over area and time [ 2 · ], when it's multiplied by a certain area (scalar product ⃗ ·̂) the result is the amount of matter going through that area per unit of time, being equivalent to the situation given at the top part of the control cylinder (in which case the vectors are simplified to a one-dimensional expression, since both the diffusion flow and the surface are parallel).
By the continuity equation, it's known that this amount of matter is the same that is being evaporated. Thus, the following relation is obtained: Meaning so that the evaporation flow rate [g · s -1 ] will be determined by: For the purpose of this work D has been approximated to 25.6 ·10 -6 m 2 ·s -1 , its value at T = 20ºC. and Δx is set as 10 -6 m to normalize the orders of magnitude of experimental and theoretical flow rates.
Since the FlowER battery is driven by evaporation, its performance is directly dependant on the surrounding ambient conditions. For this reason, a hygrometric chart has been constructed as a tool to predict the battery performance under different conditions. The chart, shown in Fig. S3, represents the effect of the two most influential physical parameters (the temperature and the relative humidity) on our variables of interest (the theoretical flow rate and the output power). The temperature range considered for this study was 5-35 ºC, while the relative humidity included the full range from 0% to 100%. Temperatures below 5 ºC were discarded since they would compromise the flow of the active solution, due to water solvent freezing. The theoretical flow rate was computed by the model developed in Section 2.4. and the theoretical power density peak can be calculated by adapting Faraday's law as follows: Where n is the number of electrons per mole of reactive species, c0 stands for the bulk species concentration, Q is the theoretical flow rate, F is the Faraday constant and E is the theoretical Nernst cell voltage. Finally, a correction factor of 0.251 was applied to take into account all the losses affecting the cell performance. The correction factor is computed as the ratio between the power density peak delivered by the battery at time 0h (shown in Fig. 4a) and the theoretical power density peak calculated with eq. S1 for the corresponding working flow rate (Qevap= 239 µL/h ).
As depicted in Fig. S3, the FlowER behaviour can be predicted in a wide range of temperature and relative humidity conditions. The highlighted areas stand for laboratory conditions recorded over the course of this work (orange) and greenhouse optimal conditions (green). 2 Recalling that the FlowER battery has been conceived to power precision agriculture applications it is interesting to analyse the battery performance within a greenhouse environment. For instance, the theoretical power output for the worst-case scenario (HR% = 80% and T = 18 ºC) would be 2.2 mW·cm -2 , which translates to 0.6 mW·cm -2 after applying the efficiency factor. Conversely, in the warmest (T = 25ºC) and less humid (HR% = 70%) greenhouse environment, the battery would yield a corrected power density of 1.25 mW ·cm -2 . Figure S3. Hygrometric chart to predict the FlowER battery performance under different ambient conditions. A correction factor has been applied to take into account battery efficiency. Highlighted areas represent the windows of different ambient conditions: orange area stands for laboratory conditions recorded over the course of this work; green area covers optimal greenhouses conditions.

Species storage: Oxygenation state
To optimize the storage of the redox species, two experiments were set-up: deoxygenation with N2 (g) and covering of the reactants surface with a hydrophobic material (in this case, sunflower oil). Both were designed to avoid oxygen molecules from reaching the redox species, thus keeping the anodic components in the original state and, a priori, showing no difference with the cathodic ones. Here is presented the first one, as the second one is already presented in the main text. Figure S4. Current density peaks of the battery redox couple for various volumes and oxygenation states (redox species concentrations: 100mM). Hydroquinone Sulfonic Acid (H2BQS) shows a bigger difference between its samples than p-Benzoquinone, as predicted.

Characterization of the Flower Care™ monitoring unit
To get insights into the functioning of the commercial Flower Care™ unit and get an estimate of its power budget, we inserted a 10-Ω resistor (1/4W, 10% tolerance) between 3V coin cell battery (CR2032 format) and the device terminal connectors. We then measured the voltage drop across the resistor using an oscilloscope probe (P2220, Tektronix). Using such a shunt resistor is a simple common reported solution 1 . We selected a 10-Ω value because it is small enough so that it does not affect the existing circuitry; but it is large enough so that we could measure a voltage that with enough precision. Figure S5. Start-up sequence measured when the Flower Care™ is turned on. Figure S6. Advertising mode: this mode runs continuously in background, it helps the cell phone to detect the sensor during scanning for connection (the sensing is in sleep mode between these peaks). Right: advertising interval between each advertising event is found to be 1 s. Left: magnified view of one peak. Figure S7. Connection mode. Left: connection peaks/events occur every 45 ms. This mode stays active as long as the gardener tries to update data on the dedicated app. Right: enlarged view of one peak.