Vapor-fed bio-hybrid fuel cell

Background Concentration and purification of ethanol and other biofuels from fermentations are energy-intensive processes, with amplified costs at smaller scales. To circumvent the need for these processes, and to potentially reduce transportation costs as well, we have previously investigated bio-hybrid fuel cells (FCs), in which a fermentation and FC are closely coupled. However, long-term operation requires strictly preventing the fermentation and FC from harming each other. We introduce here the concept of the vapor-fed bio-hybrid FC as a means of continuously extracting power from ongoing fermentations at ambient conditions. By bubbling a carrier gas (N2) through a yeast fermentation and then through a direct ethanol FC, we protect the FC anode from the catalyst poisons in the fermentation (which are non-volatile), and also protect the yeast from harmful FC products (notably acetic acid) and from build-up of ethanol. Results Since vapor-fed direct ethanol FCs at ambient conditions have never been systematically characterized (in contrast to vapor-fed direct methanol FCs), we first assess the effects on output power and conversion efficiency of ethanol concentration, vapor flow rate, and FC voltage. The results fit a continuous stirred-tank reactor model. Over a wide range of ethanol partial pressures (2–8 mmHg), power densities are comparable to those for liquid-fed direct ethanol FCs at the same temperature, with power densities >2 mW/cm2 obtained. We then demonstrate the continuous operation of a vapor-fed bio-hybrid FC with fermentation for 5 months, with no indication of performance degradation due to poisoning (of either the FC or the fermentation). It is further shown that the system is stable, recovering quickly from disturbances or from interruptions in maintenance. Conclusions The vapor-fed bio-hybrid FC enables extraction of power from dilute bio-ethanol streams without costly concentration and purification steps. The concept should be scalable to both large and small operations and should be generalizable to other biofuels and waste-to-energy systems. Electronic supplementary material The online version of this article (doi:10.1186/s13068-017-0755-7) contains supplementary material, which is available to authorized users.

). An incubator is used to precisely control the temperature. A cold trap is optionally attached after the fuel cell (FC) to collect the FC products for gravimetric and FTIR analysis.

Vapor-Liquid Equilibrium
For ideal solutions, where all chemical components have similar properties, it is possible to determine the partial pressure exerted by a certain component in the vapor phase given the fraction of that component in the liquid phase. The relationship between the partial pressure and liquid composition is given by Raoult's Law: [1] = = (S1) where yi is the mole fraction of component i in the vapor phase, P is the total pressure, Pi is the partial pressure of component i, xi is the fraction of i in the liquid phase, and Pi sat is the vapor pressure of pure i at the given temperature.
In strongly nonideal systems such as ethanol-water mixtures, Raoult's Law fails to predict the partial pressure of the more dilute component (ethanol in the case of fermentation). The activity coefficient term corrects for solution nonideality, and in general, the relationship between the mole fraction of a volatile component in liquid solution and the partial pressure of that component in the vapor phase is given by a modified Raoult's Law: [1] The activity coefficient depends strongly on both liquid-phase composition and temperature. Wilson's equation [1] is an empirical model that predicts the activity coefficients of each component and, applied to the ethanol-water system at room temperature, yields the curve shown in Figure S2, known as an XY diagram. Based on the predictions of Wilson's equation, it is possible to determine the partial pressures of ethanol and water at a given liquid composition and temperature. Further, since the flow rate of carrier gas is controlled in the experimental setup, it is possible to calculate the expected mass flow rates of ethanol and water vapor, which we confirmed using FTIR spectroscopy and gravimetric analysis. The data and standard deviations are shown in Figure S2.

CSTR Model
The continuous stirred-tank reactor (CSTR) is a chemical reactor model in which there is a continuous inlet and outlet flow. In the ideal version of this model, the contents of the reactor are assumed to be well-mixed; that is, there are no temperature gradients or concentration gradients within the reactor, so the reaction occurs uniformly throughout the reaction volume. This also implies that the concentration of components in the outlet stream is identical to their concentrations in the reactor. This model is well suited to homogenous reactions in gaseous or liquid phase, and is useful in approximating some heterogeneous reactions that involve a solid phase.
From the diagram in Figure S3, we can write a mole balance for each chemical species. For chemical species A, at steady state, where N indicates a molar rate. The left-hand terms are (left to right): inlet flow, outlet flow, and chemical generation (which will be negative if A is being destroyed). The right-hand side is the accumulation of A, which is zero at steady state.
For the direct ethanol FC, it is only necessary to consider the concentration of ethanol since the partial pressure of water remains nearly constant for each of the compositions tested.
In principle the generation term can have a complicated dependence on the concentrations of the various species in the reactor. However, Figure 2 shows that the reaction is pseudo-first order in ethanol partial pressure at low concentrations ( < 1.7 mmHg). If we make this assumption, then the reaction rate (in mmHg/s) can be written as: In the case of ethanol, we substitute the product of the reactor volume and the reaction rate for the generation term. Note that this new term must be converted to a molar rate using the ideal gas law.
It is convenient to rewrite the molar flow rate terms ( ) in terms of the flow rates and partial pressures as follows: where is the partial pressure and is the volumetric gas flow rate. By substituting equations S4 and S6 into S5 and rearranging, the outlet ethanol partial pressure can be obtained in terms of known quantities ( , , , ): Using Faraday's constant with 4 electrons per molecule of ethanol oxidized, an equation for the fuel cell current density ( ) can also be obtained: Equation S8 is plotted on Figure 4, using = 2.68 cm 2 , k = 36 min -1 , V = 1.75 mL (the volume of the fuel reservoir) and , = 1.7 mmHg. Finally, the conversion as a function of flow rate can also be obtained by substituting into its definition and rearranging: in plotted in Figure S4 using the same constants as in Figure 4. and fed into a FC (with area of 2.68 cm 2 ). The FC current density (at 200 mV) was measured at intervals of 1 s. The hourly and daily averages are plotted as blue and red lines, respectively. A leak in the vapor delivery was found and fixed on day 118. Figure S7 shows the concentrations of the major components of the control (non-bubbled) fermentation vs. time, for about 70 days (rather than 90 days, as for the bubbled fermentation). As with Figure 7 for bubbling, the ethanol is plotted vs. the right-hand abscissa, and the acetic acid is x10. The control fermentation, as expected, quickly slows and stops due to yeast being unable to metabolize in high concentrations of ethanol. One obvious observation is that the glucose concentration is quite low in the 3, 7, 10, and 15 day samples. Together with the discussion of Figure  7, this again seems to indicate severe glucose limiting of the fermentation. The glycerol concentration initially plummets, then stays low (~0.5%) the rest of the time. The acetic acid concentration never exceeds 0.18%, and falls off over a long time (probably evaporating). These behaviors are different from the bubbling data. Between 23 days and 70 days, the glucose concentration doesn't increase, because we stopped feeding the non-bubbling fermentation after 23 days. (With no feeding, there was still 1.5% glucose at 70 days.) Lastly, the Y+P concentration is very steady, another notable difference from the bubbling fermentation, indicating limited yeast reproduction. We are following up to see if these apparent changes in the yeast's phenotype reflect changes in genotype.  ----------------------------------------