Enhancement of polymer electrolyte fuel cell tolerance to CO by
Transcription
Enhancement of polymer electrolyte fuel cell tolerance to CO by
Enhancement of polymer electrolyte fuel cell tolerance to CO (by combination of different mitigation methods) HFCNordic 2013, Oslo 31.10-1.11.2013 Pauli Koski, Luis Perez Martinez, Taneli Rajala, Henri Karimäki, Kaj Nikiforow, Timo Keränen and Jari Ihonen VTT Technical Research Centre of Finland 5.7.2013 Overview VTT fuel cell research briefly HyCoRA – project introduction Motivation for research with low quality hydrogen CO impurity research with single cells Measurement system with on-line gas analysis The problem of oxygen contamination CO poisoning behavior as a function of fuel utilisation Summary and conclusions 2 5.7.2013 3 VTT Technical Research Centre of Finland VTT Technical Research Centre of Finland is the biggest multitechnological applied research organisation in Northern Europe. VTT provides high-end technology solutions and innovation services. From its wide knowledge base, VTT can combine different technologies, create new innovations and a substantial range of world class technologies and applied research services thus improving its clients' competitiveness and competence. Turnover 286 M€ Personnel 2834 Finland 5.7.2013 4 VTT Fuel Cells The main research areas today are SOFC system research SOFC stack development HT-PEMFCs and methanol reforming PEMFC systems and hydrogen quality PEMFC materials and components Enzyme catalysed & printed fuel cells Electrochemical energy storage Jari Kiviaho Senior Principal Scientist [email protected] 5.7.2013 PEMFC systems and hydrogen quality The main focus is on PEMFC system development and use of low grade hydrogen in low temperature PEMFC systems. The applications are industrial working machines and utilisation of byproduct hydrogen from industry. Jari Ihonen Principal Scientist [email protected] 5 5.7.2013 Latest publication: Optimisation of air bleed 6 5.7.2013 Research with low quality hydrogen - why and how? 7 5.7.2013 8 Why to do research with low quality hydrogen? Due to purification and quality assurance, automotive grade hydrogen is several times more expensive than industry grade. Industrial hydrogen is available at refineries for peak, reserve and regulation power production, that is essential with the growing share of solar and wind power If system level mitigation measures are applied, vehicles in fleet applications (single owner) could use much less expensive hydrogen than automotive grade hydrogen (ISO 14687-2:2012), 5.7.2013 9 Operating PEMFC with low quality hydrogen - practical issues and knowledge caps Hydrogen used in PEMFC is never 100% pure, but contains inert gases as well as contaminants (CO being most important) The effect of inert gases and contaminants is dependent on (at least): The operation mode of the anode side Fuel utilisation per pass and total fuel utilisation Anode catalyst loading and type Temperature, pressure, humidity, anode stoichiometry Membrane thickness GDL gas transport properties Aging of the components Flow homogenity in the stack 5.7.2013 10 Open questions for automotive PEMFC research Which are the most relevant contaminants? What are most relevant anode Pt loadings? What should be the temperature, pressure and humidity? What is the most relevant duration of one measurement and what should be the load profile? How to simulate change of gases (H2, O2, N2) between anode and cathode in start and stop and shut-down of the automotive PEMFC? How to simulate the effect of aging, i.e. BoL performance vs. EoL? How to minimise the effects of test station, cross-contamination, GDL aging? 5.7.2013 11 The main issue (?): Operation of anode side (stoichiometric vs. dead-end) In automotive PEMFC system fuel efficiency is over 99% (in lab 50-80%) Inert gases as well as impurities can enrich over 100 times in anode loop In laboratory test station enrichment is 2-5 times In automotive PEMFC system N2 is enriched up to levels of 10-30 %, even higher Utilisation per pass (Uf) can be as low as 15-20% (vs. test station typical 80%) Linear flow velocities are different in labs and in automotive PEFC 5.7.2013 Measuring CO contamination with exhaust gas analysis 12 13 5.7.2013 Setup for anode CO contaminant testing with on-line exhaust gas analysis C A CO N2 • Dynamic contaminant mixing to hydrogen feed allows wide range of COconcentrations to be used Humid gas feeds Diluted CO injection μMFC • Water is removed from the exhaust sample gas with a membrane drier GC H2O sensor Fuel Cell • Analysis of the exhaust gas with a gas chromatograph (GC) Membrane dryer • System built around Greenlight G60 test station Back pressure regulation To ventilation H2 flow meter 5.7.2013 Impurity injection control 0.05 sccm MFC Impurity 100 ppm of CO • Gas blend with known CO concentration, balanced with N2 or H2 • Precision mass flow controller (0.05-10 sccm) for impurity blend • Impurity is injected just before the cell, minimising the time delay and adsorption to surfaces and dissolution to water 14 To ventilation 200 sccm H2 3-way solenoid valve 25 ppb of CO • A 3-way valve is used to flush air traces from impurity feed line before impurity injection • The 3-way valve also allows precise timing of the injection (as long as volume between valve and H2 feed is minimised) 15 5.7.2013 Description of GC and exhaust gas preprocessing • Low cost (< 35 k€) Agilent 6890N equipped with methaniser and flame ionisation detector (FID) • With the current method, detection limit is around 100 ppb with sample analysis duration of 4.2 min • GC does not tolerate water, therefore water is removed from the exhaust sample flow with a membrane dryer Valve configuration CO CO2 • Membrane dryer has high selectivity to water and low permeability for other gases, including CO and CO2 Sample with 10 ppm of CO & CO2 5.7.2013 Measuring CO/CO2 –balance and the problem of the oxygen contamination 16 5.7.2013 17 Measuring CO/CO2 -balance – typical problems Ideally CO → CO2 conversion is the only source of CO2 In reality, quite a few error sources exist: • Fuel impurities (CO & CO2) • Membrane cross-over (CO2) • Carbon corrosion (CO2) • Humidifier water (CO2) • Air bleed impurities (CO & CO2) Some trace CO & CO2 also dissolved in liquid water Most of these can be eliminated or taken into account and subtracted from the final values In addition, “oxygen contamination” (i.e. unintentional oxygen bleed) is an issue. For example, oxygen content is about 15 mg/dm3 (0.46 mmol/dm3) in water leaving municipal water treatment plant in Helsinki 5.7.2013 18 Unintentional oxygen bleed Some test station use water bleed to keep humidifiers clean 1 dm3/h water bleed can introduce several ‰ of O2 in H2, when typical 25 cm2 test cell is used. For 1 kW stack the O2 concentration can still be tens of ppm. Water of anode humidifier can be almost saturated with oxygen, when experiments are started. This can cause high (several ‰) levels of O2 in H2 in the beginning of the experiments. Humidifier refill water (which is evaporated to anode gas), can keep the O2 concentration in H2 in level between sub ppm level and few ppm. 5.7.2013 The same conditions in all experiments and clean (COfree) initial surface High CO level (6.1 ppm) and 0.3 mgPt/cm2 Cell voltage / V Example of oxygen contamination The results show drastic effect of oxygen contamination due to initial oxygen in humidifier water CO / ppm 65°C, 25 cm2 cell, 0.6 Acm-2 Time / h 19 5.7.2013 20 Mitigation of unintentional oxygen bleed Degas anode humidifiers (nitrogen purge at high temperature) Inactivate automatic water bleed, if included in the station Inactivate automatic water refill (if possible) Measure oxygen level in humidified hydrogen By-pass anode humidifier and use anode gas recirculation for the humidification of the anode gas This solves the problem completely 5.7.2013 The effect of fuel utilization on the CO contamination dynamics 21 5.7.2013 22 What is the importance of fuel utilisation per pass? Critical questions for CO poisoning dynamics in automotive applications: Is contamination more dependent on CO concentration (in ppm) in hydrogen or in total molar rate of CO (in mol/s)? When CO can be detected at the anode exit? How are these issues affected by the catalyst type, loading and flow field structure? 5.7.2013 23 Experimental set-up for CO tolerance measurements Membrane drier – minimal CO dissolving in water Measurement of CO and CO2 at the anode exit F, mass flow controller; B: bubbler, T: thermocouple, WT: water trap, D, dryer; R, rotameter; RH, relative humidity probe; P, diaphragm pump and NV, needle valve. 5.7.2013 24 Summary of experimental conditions and materials Automotive conditions (353 K, 50% RH), but low pressure (1 bar at exit) Low Pt loading anodes (0.05 mgcm-2) for 25 cm2 cell with single channel CO levels simulate failures in ISO 14687-2:2012 quality control Fuel utilisation from 25% to 70% (automotive vs. labotarory) Air utilisation relatively low (40%), but reasonable when low pressure chosen 5.7.2013 25 Results of three ”standard measurements” with 1 ppm CO and µf = 70 % • CO slips through immediately • Enrichment would start in automotive PEMFC systems, since hydrogen is recirculated. • CO concentration reaches initial 1 ppm after 30 minutes (30 mV voltage drop) • Time scale for 50 mV voltage drop is about 50 minutes • 50% of CO (1.87*10-10 mol/s) is slipping through the cell at this point Performance of the fuel cell for µf = 70 % and 1 ppm CO feed (Experiment #1); a) potential drop history, b) CO concentration at the anode outlet as a function of time and c) estimated molar flow rate of CO at the anode outlet as a function of time. The CO is injected at t = 0 min. 5.7.2013 26 Results of constant CO concentration experiment • When CO molar flow is increased, the response in the open anode cell operation is almost the same. • 50 mV cut-off voltage drop reached in 50, 55 and 60 minutes. • When CO molar flow is increased, much more of the CO is slipped through the cell. • In the real system with recirculation, the response should be different, as CO would accumulate in the system! Performance of the fuel cell for µf = 70 % (experiment #1), µf = 40 % (#2) and µf = 25 % (#3) and 1 ppm CO feed: a) potential drop history, b) CO concentration at the anode outlet as a function of time and c) estimated molar flow rate of CO at the anode outlet as a function of time. The Y axis on the right of Figure 4c is the corresponding molar flow rate of CO at the anode inlet. The CO is injected at t = 0 min. 5.7.2013 27 Results of constant CO molar rate experiment • When H2 molar flow is increased (and CO in ppm decreased), the poisoning rate is decreased. • 50 mV cut-off voltage drop reached in 30, 70 and 145 minutes for 1, 0.57 and 0.36 ppm. • When H2 molar flow is increased, more of the CO is slipped through the cell. • Enrichment of CO (in ppm) is most clear for the lowest H2 molar flow, but lowest amount (in mol/s) slipped through. Performance of the fuel cell for µf = 70 % (experiment #4), µf = 40 % (#5) and µf = 25 % (#6) a) potential drop history, b) CO concentration at the anode outlet as a function of time; and c) estimated molar flow rate of CO at the anode outlet as a function of time. 5.7.2013 28 What is the importance of fuel utilisation per pass? Is contamination more dependent on CO concentration (in ppm) in hydrogen or in total molar rate of CO (in mol/s) YES, but estimation of the effect is difficult using open anode data and single cells When CO can be detected at the anode exit? Immediately after it has been fed in the cell. However, this slip rate is dependent on number of parameters. How are these issues affected by the catalyst type, loading and flow field structure? A very good question, which must be studied for determining the right CO level in ISO 14687-2:2012 29 5.7.2013 Test station equipped to mimic an automotive system Moving to PEFC system level Miniature automotive system Passive humidification through cathode Dead end with recirculation Custom purge triggering routines Enrichment and control of nitrogen in anode loop Impurity injection with constant flow rate Flow rate measurement using H2/RH% sensor Both open anode and recirculation modes Enrichment and control of nitrogen in anode loop Stoichiometric modes in dead-end operation for impurity feeding Flow rate measurement using H2/RH% sensor Experience from real PEFC systems essential Humid gas feeds C CO N2 • • • • • • • • • • • Diluted CO injection A μMFC SV P H2 FM SV H2 flow meter Load HC H,T HC H,T PR PEMFC stack H,T Air H,T F H2 sensor FM MH P FM Fuel Cell T T P DF Tap water H2O sensor H,T FM Recirculation pump Membrane dryer FM Back pressure regulation SV HE To ventilation H2 flow meter GC 5.7.2013 30 Moving to PEFC system level • Inert gases affect the dynamics of mass transport (instability) • How does CO impact depend on this and what is the combined effect? 2 ppm CO → Next step is to construct a 1-2 kW system with on-line gas analysis suited for benchmarking CO mitigation methods and operation strategies for low quality hydrogen 5.7.2013 31 Recommendations for automotive PEMFC research 1 Exit/recirculation loop gas analysis should always be applied Measurements with recirculation of anode gas should be preferred over open anode measurements If open anode measurements are used, degassing of anode humidifier water (CO2, O2 removal) and control of MQ water is necessary In anode gas recirculation system this is avoided Quality of used hydrogen and air should be frequently controlled Cell level measurements should be verified with stack level measurements with relevant fuel utilisation per pass Catalyst active area should be measured BoL and EoL 5.7.2013 32 Recommendations for automotive PEMFC research 2 Measurements should be done with sufficiently wide temperature, pressure and humidity range The effect of anode catalyst loading should be studied whenever possible The aging (thinning) of the membrane should be taken into account Increased gas permeability increase internal air bleed Measurement durations should be similar to typical use of automotive PEMFC systems Exchange of gases should be comparable to automotive PEMFC systems Measurement during shut-down, if system test bench is used HyCoRA Hydrogen Contaminant Risk Assessment NON-CONFIDENTIAL VERSION Jari Ihonen 1.11.2013 VTT Technical Research Centre of Finland 5.7.2013 34 Project summary for HyCoRA In FCH JU 2013 call a VTT-lead consortium submitted a project proposal to the topic SP1-JTI-FCH.2013.1.5 (Fuel quality assurance for hydrogen refuelling stations). The proposal passed all the thresholds in the expert evaluation HyCoRA was invited to negotiation meetings, which are ongoing In order to minimize the time delay HyCoRA consortium is already now informing OEMs for the possibility of joining the advisory board. Due to negotiations, there can be some changes in the project plan presented in this document. 5.7.2013 35 Project consortium and budget Overall budget is about 4.2 M€ with maximum of 2.16 M€ FCH JU contribution Research partners are: VTT (coordinator &WP4 leader) CEA (WP1 leader) SINTEF (WP2&WP3 leader) JRC (WP5 leader) Industry partners are: Powercell Sweden AB Protea Ltd Since there are no automotive OEMs or hydrogen refuelling station producers/operators in the consortium, the advisory board is open for all companies in these categories. 5.7.2013 36 SP1-JTI-FCH.2013.1.5: Fuel quality assurance for hydrogen refuelling stations Overall project objective / Scope of Work The overall objective is to reduce cost of hydrogen fuel quality assurance (QA) for (Hydrogen Refuelling Station) HRSs Expected Outcome Cheaper and more reliable fuel quality assurance procedures and instrumentation for HRSs 5.7.2013 37 The main objectives for Topic 1.5 Establishing a simplified and diversified set of requirements for hydrogen fuel quality depending on fuel feedstock and production technologies (biogas, reforming, electrolysis, by-product etc.) Simplifying fuel quality control by enhance knowledge of correlations between gas impurity concentrations based on extensive in field measurements at HRS fuel nozzle Assessing ways to reduce the number of analysis methods required for complete QA 5.7.2013 38 HyCoRA strategy– Risk Assessment In HyCoRA project quantitative risk assessment is applied to optimise the needs for the QA with existing (draft) ISO standard Risk assessment provides information for the required frequency and accuracy for the gas analysis at the nozzle and/or in production Risk assessment gives the right focus for development of the new analytical methodology for the gas analysis Risk assessment requires information at least from: a) the real susceptibility for various poisonous species specifically for automotive applications – automotive FC system data needed b) probabilities for QA failure in hydrogen production site and/or at HRS – data for gas analysis methods needed c) concentration correlations between contaminant species in fuel - impurity concentrations at production sites and HRS nozzle needed 5.7.2013 39 Acknowledged sources and co-authors for literature review and experimental results Articles Nikiforow, K., Karimäki, H., Keränen, T.M., Ihonen, J. Optimization study of purge cycle in proton exchange membrane fuel cell system (2013) Journal of Power Sources, 238, pp. 336-344. Karimäki, H., Pérez, L.C., Nikiforow, K., Keränen, T.M., Viitakangas, J., Ihonen, J. The use of online hydrogen sensor for studying inert gas effects and nitrogen crossover in PEMFC system (2011) International Journal of Hydrogen Energy, 36 (16), pp. 10179-10187. Luis C. Pérez, Pauli Koski, Jari Ihonen, José M. Sousa, Adélio Mendes Effect of fuel utilization on the carbon monoxide poisoning dynamics of PEMFC, Submitted to the Journal of Power Sources Thesis Rajala, Taneli Enhancement of polymer electrolyte fuel cell tolerance to CO by combination of different mitigation methods, Master’s thesis, University of Helsinki (2013) Nikiforow, Kaj Optimization of polymer electrolyte membrane fuel cell systems – Applied study of hydrogen recirculation, Master’s thesis, Aalto University Project reports: Literature reports in FCH JU HyQ project (grant agreement n° 256773) Jari Ihonen (VTT) and Pierre-André Jacques (CEA) 5.7.2013 40 Acknowledgements for funding The work leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/20072013) under grant agreement n° 256773. This research has been conducted under the “Fuel Cell 20072013” technology program of Tekes, the Finnish Funding Agency for Technology and Innovation. The authors would also like to acknowledge their TopDrive and DuraDemo project partners. 5.7.2013 VTT creates business from technology 41