Supporting Information for

Transcription

1 Supporting Information for Toward a Practical Solar-Driven CO 2 Flow Cell Electrolyzer: Design and Optimization Gowri M. Sriramagiri,, Nuha Ahmed,, Wesley Luc, Kevin D. Dobson, Steven S. Hegedus,, Feng Jiao* Institute of Energy Conversion, University of Delaware, 451 Wyoming Rd., Newark, DE 19716, USA. Dept. of Electrical and Computer Engineering, Evans Hall, University of Delaware, Newark, DE 19716, USA. Dept. of Chemical and Biomolecular Engineering, Colburn Lab, University of Delaware, Newark, DE 19716, USA. * Corresponding author. Address: jiao@udel.edu Number of pages: 7 Number of tables: 1 Number of figures: 3 S1

2 SUPPORTING INFORMATION 6-cell configuration. The modeled parameters of the 6-cell configuration and the results are shown in Figure S1. Figure S1(a) shows all I-V curves of all the modeled configurations, with varying illuminated areas. Figure S1(b) shows the resulting plot of operating voltage and current density as cell area changes and Figure S1(c) shows the modeled SFEs and the known FE (V). From this figure, it is clear that the FE (V) curve is what dictates the SFE of a configuration with more cells than optimum, in our case, 6 or more. It can be seen that, as the output voltages of the solar array are higher in this configuration, the electrolyzer curve now falls on the constant-current region of the module I-V curve, well to the left of the array maxiumum power point. This results in JOP remaining near-constant with varying illuminated area, as seen in Figure S1(b). The same can be said for an array with 7 or more cells, which would have even higher VOC (~3.5 V and greater), with the electrolyzer curve falling even further to the left, to lower voltages, of the maximum power points. This means that, in these conditions, only FE, a function of voltage, now affects device SFE, which can be observed as the overlap of the FE and SFE curves in Figure S1(c), when plotted on different linear scales. It can hence be concluded that in configurations with excess voltage, e.g. 6 or more cells in our case, SFE is directly proportional to FE. It can also be seen that SFE values in this configuration never exceed those obtained with a 5-cell array under the same conditions because with the increasing number of cells in the array, the maximum power point of the PV array moves to larger voltages and further from the electrolyzer operating or intersection point. Potential for High SFE s with Multijunction Tandem Solar Cells Under Concentration. A water electrolysis system having the highest-reported SFE of 3%, was recently demonstrated using a InGaP/GaAs/GaInNAsSb triple-junction solar cell under 42x concentration to drive two polymer electrolyte membrane electrolyzers in series. i Using the modeling procedure demonstrated in this work, the I-V parameters of the resulting system, i.e., the multijunction tandem solar cell under concentration from reference i, driving the flow-cell electrolyzer described in this work, were modeled for a few cell areas as given in Table S1. Calculating the resulting JOP in each case, and incorporating the FE (V) of our electrolyzer, the optimum solar cell area for this combination was determined to be 1.4 cm 2, where the FE peaks to 78% at 2.7 V, giving an SFE of 14.2%. Further improvement of SFE in beyond that predicted here should be possible through independent development of the various components of the electrochemical cells, viz., electrolytes, electrode technologies, ion-separation membranes etc. However, note that the high efficiency III-V concentrator system is significantly more expensive than the commercially available Si cells used here. S2

3 Current ( ma) V OP (V) J OP (ma/cm²) FE (%) SFE (%) Current (ma) a) Voltage (V) MPPT's Intersection Pt.s Electrolyzer I-V Curve 2 cm² 6 cm² 1 cm² 14 cm² 18 cm² 21 cm² 3 cm² 31 cm² 34 cm² 38 cm² 4cm² 5 cm² b) c) Vop Jop Solar Cell Area (cm²) FE 21. cm 2, 5.7% Modeled SFE Solar Cell Area (cm²) Figure S1. (a) Modeled IV curves for a 6-cell module configuration with CO2 electrolyzer linear voltammagram, and calculated (b) VOP and IOP and (c) SFE and FE with cell area for 6 cell PV array cm² cm² cm² cm² Time (hrs.) Figure S2. Measured IOP vs. time for the configurations listed in Table 2 (main text) with different PV illumination areas in 5-cell configuration S3

4 Table S1. SFE s calculated for a PV-EC designed using the flow cell electrolyzer discussed in this work with high efficiency multijunction tandem cells under 42x concentration, as reported in reference 29 Area (cm²) Voltage (V) Resistive loss included, theoretical Current (ma) J'OP (ma/cm²) JOP (ma/cm²) FE (%) SFE (%) EXPERIMENTAL METHODS CO 2 electrolyzer. The electrolyzer employed in this work is a sandwich-type flow-cell reactor, schematic shown in Figure S3, comprising a large-area-25 cm 2 -nanoporous silver (np-ag) cathode and an iridiumcoated catalyst membrane (Ir-CCM) anode as described in reference ii. The np-ag catalyst was synthesized using a modified de-alloying procedure. iii In brief, a 4 x 5 x 5 cm 3 Ag-Al ingot (Al/Ag = 8:2 atom %), made by vacuum induction processing, was purchased from Sophisticated Alloys Ltd. and was then cut in to 25 cm 2 Ag-Al sheets with a thickness of 5 m using electrical discharge machining. The precursor sheets were then annealed at 819 K for 24 hrs., then quenched in a water/ice bath. Subsequently, the precursor sheets were treated in dilute hydrochloric acid to leach out Al to form the monolithic np-ag catalyst. The as-leached sheets were immediately rinsed several times in de-ionized water and then dried in a vacuum oven overnight. The Ir-CCM anode was constructed following a previously reported procedure ii. In short, a catalyst slurry was prepared by sonicating a mixture of commercial iridium black nanoparticles (surface area m 2 per gram, Premetek Co.), Nafion solution (5 wt%, Dupont), DI water, and isopropanol with a weight ratio of catalyst to dry Nafion ionomer of 4:1. The slurry was then hand sprayed onto Nafion XL (Dupont) that has been sandwiched between two self-adhesive Mylar laminates (Dupont) with a 25 cm 2 window. After a catalyst loading of 1 mg cm -2 was achieved, the as-sprayed Ir-CCM was hot pressed at 135 o C and 2 MPa for 1 min. The CO2 electrolyzer was a two-electrode two-compartment electrochemical flow cell which was constructed out of stainless steel and then plated with a 2 m gold layer to prevent corrosion under operating conditions. The np-ag catalyst was used as the cathode for CO2 reduction to CO, while the Ir- CCM was used to both separate the two compartments, as well as the catalyst for the anode to facilitate water oxidation. A CO2-saturated.5 M NaHCO3 ( %, Sigma Aldrich) aqueous electrolyte solution was fed and recirculated at a flowrate of 15 ml min -1 through the cathode compartment as shown in Figure S4

5 1. A gear pump (EW , Cole-Parmer) was used to drive the electrolyte from a reservoir, through a flow meter (1XLX3, Brooks), then through an in-line gas diffuser, and then finally through the cathode compartment. CO2 gas (Grade 5, Keen) was fed through the porous ceramic (Refractron) in excess at 25 ml min -1 to ensure saturation. The electrolyte was then fed from the electrolyzer to a gas/liquid separator, which was constructed out of a stainless-steel knockout drum for gas phase product quantification. Low solubility gases such as CO and H2 were separated while the electrolyte was recycled back to the reservoir for continuous operation. Gas products from the gas/liquid separator were fed into a 1 ml sample loop of a gas chromatograph (Shimadzu, GC- 21) equipped with PLOT Mol Sieve 5A and Q-bond PLOT columns to confirm and separate the CO and H2 products. Argon (99.999%) was used as the carrier gas and a thermal conductivity detector was used for product quantification. No electrolyte was used for the anode compartment, which prevented water from crossing over from the anode to cathode compartment that could dilute the catholyte and degrade performance. The full cell is operated between 2.4 to 3. V in different configurations, including all the voltage losses within the device due to the internal resistance, transport and kinetic limitations. For constant potential experiments, an Autolab PGSTAT128N potentiostat with a 1-amp booster was used. Figure S3. Schematic of sandwich-type electrolyzer design used in this work Photovoltaic power source. Commercially available, ~2% efficiency, SunPower Maxeon C6 TM halfarea crystalline silicon photovoltaic (PV) cells were chosen for the power source. Rated at 4 ma/cm 2 under standard sunlight (1 mw/cm 2 ), these interdigitated back contact (IBC) commercial silicon solar cells have the highest current density (JSC) and efficiency of those available on the market for terrestrial deployment. This single junction solar cell technology has all contacts processed on the rear of the device, maximizing the current generated by eliminating shading-associated optical losses on the front. The I-V curves of solar cells used in the array were measured under standard test conditions, 1 mw/cm 2 illumination at 25 o C and the measured I-V s of the individual cells were used to model the I-V performance of the array with 5 S5

6 and 6 cells connected in series. The cells were connected in series using commercial tabbing methods. The number of cells in series was determined by analysis described in the main text. Their output was measured individually and after tabbing. A small increase in series resistance (RS) which occurred due to the tabbing interconnection was characterized and addressed in the model. In each experimental configuration, the array was setup for the desired cell area using the shadow masks and connected to the electrolyzer via a Keithley 244TM Source Meter Unit in 2-wire mode as an ammeter in series. Voltage readings were taken every ten minutes at the electrolyzer, every fifteen minutes at the PV module terminal, and the operating current (IOP) was logged at 1 minute intervals. Gas products from the gas/liquid separator were fed every 15 minutes into a 1 ml sample loop of a gas chromatograph (Shimadzu, GC- 21) equipped with PLOT Mol Sieve 5A and Q-bond PLOT columns to confirm and separate the CO and H2 products. Solar Simulator Construction for Illumination Source. The cells were characterized individually under a calibrated AM1.5G class AAA simulator made by OAI, that illuminated up to 15 x15 cm 2 area. However, to achieve larger area illumination of the array, a home-made solar simulator, incorporating eight 15 W GE halogen light bulbs, was constructed. All the lamps are connected to a rheostat to control the input power and, hence, the light intensity output. Four lamps were attached to a rigid rail that could be raised to a desired height to adjust the light intensity variation and uniformity. The light intensity calibration is achieved by tuning the input power to the system to obtain the required short circuit current (ISC) output of the array based on the simulator value of ISC. For a module made with 5 cells of known I-V characteristics, the power to the four lamps was adjusted to achieve an array ISC equal to the lowest ISC of the individual cells in series. The bulbs rapidly heat the cells, so a cooling system comprising of a fan and a heat-sink under the stage was employed. However, there was still significant radiative heating, which heated the cells to around 4 o C, beyond the standard solar cell test conditions of 25 o C. This resulted in a voltage loss of 13 mv for the array which is a closer approximation to outdoor operation conditions, where the cells would typically operate at around 5 o C. Since ISC is relatively insensitive to temperature, no correction was made to the ISC used for lamp calibration. System Operation. Voltage can be adjusted in quanta of VMP (the voltage at maximum power) of cells in series while current can be adjusted by reducing the illuminated cell area. Experiments were designed to verify the model results and their dependence on cell area in a 5-cell configuration. This variation in illuminated solar cell area was achieved by using stainless steel adjustable shadow masks on each cell. Using masks was easier and more reversible than cutting cells into smaller areas. It is noted that using masks for area reduction did decrease measured cell efficiency due to an increasing ratio of dark to illuminated surface with small-area illumination, leading to a relative increase in dark current, hence lower VOC and fill-factor. For this reason, lower efficiencies for masked small cell areas are expected, compared S6

7 to those predicted from modeling. However, this loss would not occur in a dedicated PV module design for EC applications. REFERENCES i J. Jia, L. C. Seitz, J. D. Benck, Y. Huo, Y. Chen, J. W. Ng, T. Bilir, J. S. Harris and T. F. Jaramillo, (216) Solar water splitting by photovoltaic-electrolysis with a solar-to-hydrogen efficiency over 3%. Nat. Commun. 7, (DOI:1.138/ncomms13237). ii Luc, W.; Rosen, J.; Jiao, F. An Ir-based anode for a practical CO2 electrolyzer. Catalysis Today 216. iii Lu, Q., Rosen, J., Zhou, Y., Hutchings, G. S., Kimmel, Y. C., Chen, J. G., & Jiao, F. (214) A selective and efficient electrocatalyst for carbon dioxide reduction. Nature Communications 5:3242. S7