plansoec PROJECT REPORT MAY 2011 R&D and commercialization roadmap for SOEC electrolysis R&D of SOEC stacks with improved durability

Transcription

1 plansoec R&D and commercialization roadmap for SOEC electrolysis R&D of SOEC stacks with improved durability PROJECT REPORT MAY 2011 Support program: ForskEL2010 Energinet.dk Project acronym: plansoec Project no.: Participants: Topsoe Fuel Cell A/S H2 Logic A/S RISØ DTU Fuel Cells and Solid State Chemistry Division Page 1 of 65

2 Table of content Introduction... 3 PART 1 - R&D and commercialization roadmap for SOEC electrolysis Background & purpose Roadmap methodology & process Electrolysis state-of-the-art & theoretical potential Alkaline electrolysis (AEC) Proton Exchange membrane eletrolysis (PEM) SOEC Potential benefits of SOEC Market requirements for electrolysers Onsite industrial (1A) Central industrial (1B) Onsite CHP (2A) Onsite off-grid balancing (2B) Central grid balancing (2C) Onsite hydrogen fuel (3A) Central hydrogen fuel (3B) Central synthetic fuel (4A) De-central synthetic fuel (4B) SOEC R&D & commercialisation roadmap formulation PART 2 - R&D of SOEC stacks with improved durability Summary Corrosion resistance of interconnect alloys Reducing atmosphere Oxidizing atmosphere Cell and stack element testing Overview of the tests Initial cell performance Durability under co-electrolysis (cell tests) Durability under steam electrolysis Cell tests Stack element tests Stack testing N N Stack modeling Appendix A. List of project milestones (Work packages 6-7) B. List of internal reports Page 2 of 65

3 Introduction This report provides results of the plansoec project supported by Energinet.dk ForskEL 2010 program. Overall the project has been divided into two parts each with related work packages: PART 1: Formulation of a R&D and commercialization roadmap for SOEC electrolysis o WP2 Study specification o WP3 SOEC state-of-the-art and R&D need analysis o WP4 Market requirement and outlook analysis o WP5 R&D & commercialisation roadmap formulation PART 2: Conducting R&D of SOEC stacks with improved durability o WP6 Durability of SOEC cell and stack components o WP7 SOEC stack testing and modelling This reporting is therefore also divided into two corresponding parts, with different authors and contributors from the project participants. Part 1 authors and contributors: Andreas Richter Topsøe Fuel Cells A/S Claus Friis Pedersen Topsoe Fuel Cells A/S 1 Mogens Mogensen RISØ DTU - Fuel Cells and Solid State Chemistry Division Søren Højgaard Jensen RISØ DTU - Fuel Cells and Solid State Chemistry Division Mikael Sloth H2 Logic A/S Part 2 authors and contributors Ming Chen RISØ DTU - Fuel Cells and Solid State Chemistry Division Jens Ulrik Nielsen Topsøe Fuel Cell A/S Topsoe Fuel Cells A/S has acted as overall administrative project manager, with main effort contribution to part 1, acting as a developer and potential future manufacturer of SOEC technology. H2 Logic A/S has contributed to the part 1 efforts acting as a developer of hydrogen refuelling infrastructure, potentially a customer for future SOEC technology. RISØ DTU - Fuel Cells and Solid State Chemistry Division has contributed to both part 1 and part 2, as the research institute is already today active in R&D of SOEC technology. 1 Moved to -> Haldor Topsøe A/S late 2010 Page 3 of 65

4 PART 1 - R&D and commercialization roadmap for SOEC electrolysis Page 4 of 65

5 1. Background & purpose The purpose has been to develop a R&D and commercialisation roadmap for hydrogen and CO production plants based on the solid oxide electrolysis cell (SOEC) technology. Electrolysis for energy storage Production of hydrogen and carbon monoxide (CO/syngas) through use of renewable electricity for water/co 2 SOEC electrolysis may benefit the overall energy system as shown in the figure below. Hydrogen can act as storage medium for renewable energy and can be utilised for industrial gas supply, fuel for transport and supply of hydrogen to combined heat and power units (CHP) in e.g. households. Hydrogen and CO (syngas) can be used as feedstock for production of various CO 2 neutral synthetic hydrocarbon fuels such as methane (synthetic natural gas SNG), dimethylether (DME) or clean diesel. The SOEC production plants, the CHP units as well as the fuel cell electric vehicles (hydrogen &/or synthetic fuelled) may also provide various balancing services to the power grid. Formulating a roadmap to guide future R&D SOEC technology is still on an early R&D stage but years of extensive R&D within SOFC technology provides a strong platform for an accelerated commercialisation. However, in order to guide the future SOEC R&D activities towards reaching commercial market requirements a detailed roadmap is necessary. An overall strategy for R&D of various electrolysis technologies in Denmark already exists 2, formulated in the Hydrogen Production working group in the Danish Hydrogen and Fuel Cell Partnership. The SOEC roadmap developed as part of the plansoec project supplements the overall strategy, by conducting an updated analysis of state-of-the-art. Also plansoec provides a detailed analysis of requirements for different market applications for SOEC, which enables formulation of precise and detailed R&D targets. 2 Danish Hydrogen & Fuel Cell Partnership Electrolysis hydrogen production R&D strategy Page 5 of 65

6 2. Roadmap methodology & process The process and methodology for the roadmap formulation is outlined in the figure below: First step was to specify the content of the roadmap and the relevant data to be gathered, analysed and developed. This also included developing the data structures for the analysis in order to ensure that all necessary parameters for a complete roadmap was analysed and developed. Next step were to analyse state-of-the-art (SoA) for various electrolysis technologies as well as the future theoretical potential for the technologies. SoA for SOEC should of course be the starting point for the roadmap and the theoretical potential enabled a later comparison with the market requirements, in order to evaluate if the technology holds the potential to actually meet the requirements. Similar analyses were made for alkaline and PEM electrolysis in order to provide a benchmarking basis. In parallel with the SoA analysis, market requirements for a number of various potential market segments for SOEC technology were analysed. This involved financial calculation of business cases for SOEC use and which technical and economical targets that the technology must meet in order to provide a feasible case. The SoA and market targets was then summarized into a roadmap, showing detailed R&D steps and targets in order to move the SOEC technology from SoA onwards reaching of the various market requirements. Page 6 of 65

7 3. Electrolysis state-of-the-art & theoretical potential 3.1 Alkaline electrolysis (AEC) Alkaline electrolysers (AEC) represent a very mature technology that is the current standard for largescale electrolysis. The anode and cathode materials in these systems may be made of nickel and nickelplated steel, respectively. A good performance of both electrodes is obtained using Raney nickel, which is a nano-porous Ni formed by leaching either Zn or Al from Ni-Zn or Ni-Al-alloys. Electrocatalysts are often added to the electrodes. The electrocatalyst can also be noble metals like platinum, rhodium or iridium, but a large selection of non-noble catalysts is also available such as Co 3O 4 in the oxygen electrode and MoS 2 in the hydrogen electrode. The electrolyte in these systems is a highly caustic KOH aqueous solution. The electrodes and the evolved gasses are separated by a diaphragm with very fine pores, It was traditionally made of asbestos, but today it may be made of NiO or other materials stable in KOH. Key advantages of this technology include 1) its relative simplicity, 2) its maturity, and 3) its durability. Key disadvantages are 1) the difficulty of formation and growth of gas bubbles in highly caustic electrolyte, which make it necessary to have large containers into which the KOH electrolyte can be pumped from each of the electrodes and then release the gas bubbles over time, 2) its difficulty to produce hydrogen at high pressures at which the separation of the gas bubbles from the electrolyte is increasingly difficult, and 3) a large overvoltage for oxygen evolution at the relative low operation temperature, typically below 100 C. The difficulty with producing high pressure hydrogen for storage results in the added need for an external compressor, which adds cost and complexity to the system. The electrode reactions in an AEC are: Cathode: 4H 2O(l) + 4e - 2H 2(g) + 4OH - (aq) Anode: 4OH(aq) O 2(g) + 2H 2O(l) + 4e - 2H 2O(l) 2H 2(g) + O 2(g) (1) Most commercial AECs are operated at ºC. Increasing the AEC operation temperature above 200 C may significantly increase the performance and the electricity-to-hydrogen efficiency. A known obstacle for operating at elevated temperature is the lower stability of the materials. Candidates for hightemperature cell- and separator-materials with improved stability are currently investigated at Risø DTU. Page 7 of 65

8 State-of-the-art figures and theoretical potentials for AEC is listed below. State-of-the-art AEC + theoretical potential Parameter Data for Data for theoretical potential state-of-the-art A Capacity range (Nm3/h) System price /Nm3/h Actual market price: At mass production: System electrical beginning of life (kwh/nm3) & Power consumption increase per year due to cell degradation 2-4% 2-4% (@ 24x365 operation) Outlet pressure 32 bar Up to 100 bar Cell/stack lifetime (hours) System lifetime (years) A) Data depending on system capacity and outlet pressure Conditions for potential Increased market for AEC Increased electrode kinetics and elevated operation temperature Pressurized operation (100 bar) at elevated temperature (200 C) The current SOA AEC systems (> 2MW system) cost is 600 /kw which is equal to a system capacity price of 3000 /Nm3/h with an electric consumption of the system of 5 kw/nm3. Given a reduction in the electrical consumption from 5 kw/nm3 to 4 kw/nm3, the theoretical potential is estimated to 2400 /Nm3/h. The electric consumption of SOA AEC cells is kwh/nm3. The electric consumption of the system is somewhat higher due to losses in e.g. diaphragm, current collectors, wires and AC-DC inverters. The electric consumption of AEC systems is expected to decrease from 5.2 kwh/nm3 to 3.9 kwh/nm3, if the cell-temperature in AEC systems can be increased to app. 200 C and the pressure increased to app. 100 bar. The alkaline technology is currently used commercially almost exclusively to produce hydrogen for industrial purposes that require very pure hydrogen MW plants have been constructed in connection with chemical fertilizer factories. Few kw AEC systems could be economically attractive for small CHP plants and refueling of light vehicles. New stack design for small systems may have the potential to reduce the electrolysis stack price of 60-70%. In Denmark the development of AEC is especially concentrated around HIRC, Risø DTU and GreenHydrogen.dk. All figures not referred, are based on Partnerskabet Brint og Brændselsceller i Danmark (2009). Elektrolyse i Danmark, Strategi for F, U & D , Energinet.dk and Jens Oluf Jensen et al. (2008). Pre-investigation of water electrolysis, Energinet.dk. 3 Market intelligence 4 Danish Hydrogen & Fuel Cell Partnership Roadmap Alkaline MW-Electrolyser 5 J. O. Jensen et al. Pre-Investigation of Water Electrolysis, Energinet.dk (2008) 6 A. Mortensgaard. Elektrolyse i Danmark, Strategi for F, U & D , Energinet.dk (2009) Page 8 of 65

9 3.2 Proton Exchange membrane eletrolysis (PEM) Polymer electrolysers also know as proton exchange membrane electrolysers (PEMEC) are built around a proton conductive polymer electrolyte. The membrane consists of a solid fluoropolymer which has been chemically altered to contain sulphonic acid groups, SO 3H, which easily release their protons and thus is an ion exchange resin. The electrodes are typically made of a support with catalysts made of IrO 2 and Pt in the anode and cathode respectively. The key advantages of PEMEC are a high production rates at low temperature and compact design. The solid electrolyte allows for operation at high pressure. The disadvantage of the PEMEC is the high cost of the electrolyte and catalyst. The electrode reactions in a PEMEC are: Anode : 2H 2O 4H + + O 2 + 4e - Cathode : 4H + + 4e - 2H 2 Cell : 2H 2O 2H 2 + O 2 (2) PEMECs are usually constructed using the filter press-type design. They do not require electrolyte circulation because the electrolyte is a solid ion-exchange resin. The electrodes are either embedded in the surface of the membrane or pressed closely against the two opposing faces of the sheet of resin material. A ribbed or corrugated solid separator plate is interposed between cells, providing electric continuity between one cell and the next while separating the hydrogen from the oxygen in adjacent cells. This type of cell is usually cooled by circulating water through the cavity between the metal separator and the electrode plate. Hydrogen or oxygen evolved into this cavity is swept out by the coolant stream and is separated from the water outside the cell. PEMECs have fast response time and start-up/shut-down characteristics. Hydrogen generation starts immediately at ambient conditions. State-of-the-art figures and theoretical potentials for PEMEC is listed below. State-of-the-art PEM + theoretical potential Parameter Data for Data for theoretical state-of-the-art potential Capacity range (Nm3/h) System price /Nm3/h System electrical beginning of life (kwh/nm3) 6-6, Power consumption increase per year due to cell degradation (@ 2-4% 2-4% 24x365 operation) Outlet pressure 30 bar 200 bar Cell/stack lifetime (hours) System lifetime (years) 3 years 10+ Conditions for potential Larger implementation of PE- MEC Increased electrode kinetics and elevated operation temperature Pressurized operation (100 bar) at elevated temperature (200 C) All figures not referred are based on Partnerskabet Brint og Brændselsceller i Danmark (2009). Elektrolyse i Danmark, Strategi for F, U & D , Energinet.dk and Jens Oluf Jensen et al. (2008). Pre-investigation of water electrolysis, Energinet.dk. 7 Danish Hydrogen & Fuel Cell Partnership Road Map for LT PEM Elektrolyse Page 9 of 65

10 Existing PEMEC demonstrates already a part of the potential of the PEM technology in the form of compact, simple and secure installations producing pressurized hydrogen. There is a development potential that may lead to an overall reduction in manufacturing costs and increase the efficiency. Lower material costs can be achieved using new electrolyte membranes and new non-noble metal-containing catalyst materials. Such components are under development for low temperature PEMEC. Increased efficiency is expected when the operating temperature is increased, although it is not expected in the short term that high temperature PEMEC (200 C) will be operated at thermoneutral potential as is possible with SOEC. Because the excess heat is produced at temperatures between 150 and 200 C, it may be used for steam production or used for district heating, with a very high overall efficiency. In Denmark the development of PEM electrolysis concentrated around DTU Chemistry, DTU Physics and IRD. 3.3 SOEC An SOEC is in principle the same cell as a solid oxide fuel cell (SOFC). The SOEC consists of two porous electrodes on either side of a thin oxygen-ion conducting electrolyte. The electrolyte is usually made of yttria-stabilized zirconia (YSZ). The cathode and anode can be made of Nickel/YSZ and YSZ/LSM (Lanthanum Strontium Manganate). Other materials like CGO (Gadolinia doped Ceria) and STN (niobium doped strontium titanate) is likely to be used in future SOECs with improved performance. The preferred operation temperature is in the range of C. A key advantage of SOECs is the capability of a high production rate at a high efficiency. Other advantages are low material costs and a possibility for co-electrolysis of H 2O and CO 2. A disadvantage of SOECs is the demand for material stability at high temperature. The electrode reactions in a SOEC are: Cathode : 2H 2O 2H 2 + 2O e - Anode : 2O e - O 2 Cell : 2H 2O 2H 2 + O 2 (3) Various designs of SOECs have been produced and tested, but the planar design is widely acknowledged as the optimum since it offers the shortest current paths and thus the lowest internal resistance. Modern SOECs are electrode supported to allow a minimization of the electrolyte thickness and thereby reduce the internal resistance of the cell. The support layer is made of ferritic stainless steel or Ni/YSZ and typically produced by tape-casting. The steel support is more mechanical robust than the Ni/YSZ support which allows a reduction of the thickness of the support layer. The cathode and the electrolyte are spray painted on the support tape. Half-cells are stamped and subsequently sintered at high temperature. The anode is spray painted on the sintered half-cell and the full cell is sintered again. Finally current collectors are applied to both sides of the cell. In order to produce a usable voltage, the SOECs are serially connected by means of interconnect layers. These are typically made of ferritic stainless steel when the operation temperature is below 800 ºC. Page 10 of 65

11 State-of-the-art SOEC + theoretical potential Parameter Data for state-of-the-art Data for theoretical potential Conditions for potential Capacity range (Nm3/h) Demonstration of reliability System price /Nm3/h Large scale production System electrical beginning of life (kwh/nm3) Power consumption increase per year due to cell degradation (@ 24x365 operation) (3.7)* 3.2 (3.6)* Large systems 8% 2% Outlet pressure 8 bar 50 bar Cell/stack lifetime (hours) Development of electrodes with decreased gas impurity vulnerability System lifetime (years) *The electrical consumption for the SOEC is quoted without the heat for steam generation. If the heat is generated from electricity the electrical consumption is the number in brackets (see next section for further details). The manufacture cost of 5 kw SOFC modules is expected to be 200$/kW at a production rate of 500 MW/yr. Assuming SOFC- and SOEC-modules costs the same per cell-area, and that SOFC voltage is 0.7 V and SOEC voltage is 1.3 V (and SOEC and SOFC current density is the same (i.e. the overvoltage and internal resistance is the same)) the potential SOEC-module manufacture cost is 240 /(Nm 3 /h). The system price is estimated to be 2.5 to 3 times higher than the stack cost. The heat formed due to ohmic and polarization losses in cells interconnects and current collectors may be used to supply the necessary heat to the cell as electrolysis is particularly endothermic at high temperature. This enables a very high system electrical efficiency.the remaining figures are taken from Preinvestigation of water electrolysis and Elektrolyse i Danmark, Strategi for F, U & D One of the specific benefits of SOEC compared with conventional electrolysis technology is its ability to make combined H 2O and CO 2 electrolysis and thus the ability to make cheap synthetic (non-fossil) fuel to e.g. the transport sector. The electrolysis process is endothermic, which means it consumes heat. Combined with the high operating temperature this means that almost no waste heat is produced and this gives a very high efficiency - considerably higher than for low-temperature electrolysis. The high operating temperature also enable relatively cheap electrode- and electrolyte materials can be used (no precious metals). Further efficiency and improvement of the economy can be achieved by pressurized operation of the SOEC. The pressure can be achieved by evaporation of high pressure feed water (liquid) using low grade heat and is therefore expected to be inexpensive compared with other pressurization methods. The high operating temperature and high pressure makes it possible to integrate further catalysis of the synthesis gas to synthetic fuel in one system. It is shown that state-of-the-art SOFC cells are reversible and that they provide a good starting point for developing SOEC cells. For a Danish venture in the area, it is an advantage that development can build on the great competences already found in Denmark on the SOFC field. Here, in the last 20 years there has been a large activity in basic R & D as well as in the development of cheap methods of production and module development and one of the few existing production facilities are placed in Denmark. This R & D effort - in line with the technology is approaching the commercial market - has increased significantly within the latest 5 years. In 2010 the activity was app. 200 man years / year (in Denmark) divided on both research institutions (especially Risø DTU) and private companies (especially Topsoe Fuel Cell A/S). 8 J.H.J.S. Thijssen. The Impact of Scale-Up and Production Volume on SOFC Manufaturing Costs, NETL (2007) 9 A. Mortensgaard. Elektrolyse i Danmark, Strategi for F, U & D , Energinet.dk (2009) 10 J. O. Jensen et al. Pre-Investigation of Water Electrolysis, Energinet.dk (2008) Page 11 of 65

12 3.4 Potential benefits of SOEC Because the charge carrier in the SOEC electrolyte is O --, SOECs can be used for co-electrolysis of H 2O and CO 2. Co-electrolysis is relevant for production of CO 2-neutral synthetic fuel. The charge carrier in the PE- MEC and AEC electrolyte is H + and OH - respectively, which means that these cell types are at present more difficult to use for co-electrolysis. Moreover, due to the low operating temperature, AECs and PEMECs are very susceptible to CO. The electric energy requirement for the electrolysis process decreases with increasing temperature and this also makes SOEC s advantageous compared with AEC s and PEMEC s. Steam electrolysis (i.e. SOEC electrolysis) requires app. 3.1 kwh/nm3 H 2whereas water electrolysis requires app. 3.5 App. kwh/nm3 H 2. The difference in energy requirement (0.4 kwh/nm3) is the energy needed for steam generation. Low grade heat for steam generation is normally cheaper than electricity. For this reason, SOEC operation may be cheaper than AEC or PEMEC operation. In the table State-of-the-art SOEC + theoretical potential the low grade heat for steam generation is not included. In short, the thermodynamic and the kinetic of all three electrolysis technologies may benefit from increased pressure and temperature operation. However, serious stability issues of the electrodes and the electrolyte must be addressed before high pressure and temperature AECs and PEMECs can be widely commercialized. The graph compares i-v curves measured with an AEC, a PEMEC and a SOEC. As seen, there is a great difference in the cell voltage between the different cells. The major difference between AEC and PEMEC is the difference in the inner resistance (the slope of the i-v curves). Unfortunately the low internal resistance of PEMECs is not free: The active material in the PEMEC electrodes is typically platinum! The low inner resistance in the SOECs is due to the high operating temperature. The electrodes in SOECs are made of Nickel and low-precious materials. The difference in the starting voltage between PEMEC and SOEC is also due to the high operating temperature of the SOEC. The efficiency is given as 1.5 V divided with the cell voltage. This corresponds to the energy contained in the produced hydrogen (HHV) divided with the electricity consumption of the cell. If the cell is operated below 1.5 V, the energy must be supplied as heat instead. Hence, at low current density it is possible to operate both AECs and PEMECs with efficiencies close to 100%. However, such an operation is not commercially optimized since the hydrogen production rate is proportional to the current density. In other words, both AECs and PEMECs will be operated with efficiencies significantly lower than 100%. In contrast to this, the SOEC can be operated with a high current density and a high efficiency at the same time. Page 12 of 65

13 4. Market requirements for electrolysers R&D targets for SOEC technology depend on the conditions in each of the potential market segments. In total 9 various market segments have been identified and analyzed with respect formulating market driven targets and requirements for electrolysis technology. The table below shows the different segments and the author of the individual analysis. Market description (status & outlook) Market Author 1A Onsite Industrial <2000 kg/day for onsite use H2 Logic 1B Central Industrial >2000 kg/day for onsite/distribution Topsoe 2A Onsite CHP Households/buildings All 2B Onsite off-grid balancing Remote areas with small power grid H2 Logic 2C Central grid balancing Central areas with large power grid Topsoe 3A Onsite hydrogen fuel Production consumption onsite H2 Logic 3B Central hydrogen fuel Central production for distribution H2 Logic 4A Central synthetic fuel Central production for distribution Topsoe 4B De-cetral synthetic fuel De-central production for e.g. fuel and biogas upgrade Topsoe Definitions Onsite: hydrogen is produced and consumed onsite Central: hydrogen is produced onsite in large quantities and either used at the site or distributed as bulk/commodity to various use 4.1 Onsite industrial (1A) Worldwide 50 million tons of hydrogen is commercially used as a compound/feedstock in various industrial processes due to its unique physical features, among other the following industries: Power plant generator cooling Materials processing and Food industry Semiconductors and electronics Petroleum and chemical industry A large share of industrial hydrogen is consumed in the petroleum and chemical industry e.g. for use in oil refineries, these markets are covered in part 1B Central Industrial due to the much larger capacities and generally lower price levels required. Hydrogen for the various others markets (than petroleum and chemical industry) is mainly produced on large central plants based on reforming of natural gas or byproduct hydrogen from chemical processes and then distributed to the site of consumption, either by truck (gaseous or liquid) or via pipeline. Only a small part is produced onsite through either reforming or electrolysis and only in areas where either low cost electricity is available or where cost of trucked-in hydrogen is high due to long distance to nearest central production facility. Improved electrolysis efficiency e.g. through use of SOEC technology can help expand the market opportunities to areas with lower trucked-in hydrogen prices and higher electricity prices. The value proposition for onsite produced hydrogen in opposition to delivered hydrogen is the potential for a lower hydrogen cost and a higher purity of the hydrogen. A higher purity of the hydrogen can have a Page 13 of 65

14 positive effect on the industrial production process in which the hydrogen is used. E.g. in power plants where hydrogen is used for cooling a higher gas purity may enable an increase in the generation efficiency and in powered metallurgy processes the quality of the manufactured metal units may be improved. Delivered hydrogen from a gas company is typically in the purity range of 99,95% (N3) whereas onsite produced hydrogen easily can reach as high as 99,9995% (N5). Trucked-in hydrogen can also be provided at similar purity however costs increase drastically with the purity due to several extra quality checks required. Conservatively however the added value of a higher purity is not taken into account as the need for this only account to a fraction of the total annual industrial gas amount. Prices for delivered/trucked-in hydrogen vary significantly from region to region. The main parameter is the distance to a nearby central production facility and the operation condition for the plant (cost of fuel and taxes). In Europe main hydrogen central production facilities are placed in Germany and the Netherlands thus hydrogen prices here are generally lower depending on the distance to the plants. Denmark in general is a middle priced country with prices higher than in Germany whereas e.g. Norway is a high price country with prices higher than in Denmark. Please note that the price difference is only indicative and may vary much depending on actual region and consumption level. For onsite electrolysis hydrogen production the electricity cost is the main parameter. Despite of generally high electricity prices in Denmark this does however not account to electrolysis hydrogen production as this is exempted for the major electricity taxes. Overall Denmark therefore represents an electricity price slightly above average in Europe when used for electrolysis. The medium price level of industrial gas in Denmark and the slightly higher electricity cost therefore makes Denmark a well covering and average reference country for calculating cost targets for SOEC. Below are shown indicative prices of trucked-in hydrogen in Denmark (based on market intelligence). The related electrolyser size in terms of Nm3/h capacity is shown, when operating the plant 300 days a year 15 hours a day (51% utilization). As hydrogen most often is not consumed 24 hours a day at a site the hydrogen production plant cannot operate all day, as cost of storing hydrogen generally will impact economics negatively (outlet pressure of electrolyser is too low at present and cost of compressors too high to make storage feasible). Consumption kg/year (at end-user site) Hydrogen price /kg (including transport & bulk/trailer rental) Electrolyser capacity Nm3/hour (operating 15 hours/day 300 days/year) 400 kg/year 31 /kg 1 Nm3/h kg/year 23 /kg 5 Nm3/h kg/year 19 /kg 20 Nm3/h kg/year 14 /kg 50 Nm3/h kg/year 11 /kg 150 Nm3/h kg/year 10 /kg 400 Nm3/h As can be seen the hydrogen price is greatly affected by the quantity. In the lower consumption range transportation and rental costs takes up a high share, whereas at higher quantity the main share is the base hydrogen cost. Please note that prices are indicative only, as these depend very much to the region and specific deals negotiated between the customer and gas supplier. All prices are based on gaseous trucked-in hydrogen as liquid hydrogen typically is only available at few locations in selected regions. The varying hydrogen price results in different target prices for the SOEC technology depending on the size in terms of capacity. The higher hydrogen price at small capacities and thus higher acceptable CAPEX fits well with the anticipated SOEC technology evolvement starting at small capacities and high prices. Page 14 of 65

15 Below is shown an economical calculation for onsite SOEC to match the competing trucked-in H2 price at different consumptions and with a 3 year payback of CAPEX including financial costs (interest rate) during the payback period. The calculation results in the key technical targets for SOEC to be competitive. As can be seen from the table, early introduction of SOEC plants with a small capacity can happen at high CAPEX and low efficiency levels. As technology evolves and prices are reduced and plant size can be increased, use of SOEC can gradually be introduced to larger consumption applications. A higher utilization of plant than the anticipated 51% will greatly impact the economics, however whether this is possible depends on the exact site. If onsite production is to match the lowest cost of trucked-in hydrogen (~ 6/kg) the CAPEX target would be less than per Nm3/hour capacity. Hydrogen price from year 4 and onwards (after full coverage of CAPEX) will enable great savings compared to trucked-in hydrogen. Despite significant accumulated savings over a 10 years period, industrial customers tend to require a 3 year payback if they are to invest in onsite production instead of trucked-in. Market driven targets for SOEC in Onsite Industrial market Data/Target parameter Data/Targets Electrolyser capacity (Nm3/h) Annual H2 consumption (Nm3) Operation hours (15 hours days/year 51%) Competing H2 price ( /kg) Total electrolyser CAPEX ( ) Payback time (years) Annual financial expenses (% interest rate of CAPEX during payback period) 5% 5% 5% 5% 5% 5% Electrolyser consumption (kwh/nm3- beginning of life) 4 3,95 3,9 3,85 3,8 3,75 Degradation/year (2,5% - included in calculation) 2,5% 2,5% 2,5% 2,5% 2,5% 2,5% Service cost/year OPEX (% of CAPEX) 7,5% 6,5% 5,5% 4,5% 3,5% 2,5% Electrolyser system target price ( /Nm3/h) Power price incl. tax (DKK/kWh) 0,85 0,83 0,81 0,79 0,77 0,75 Stack life time before refurbishment (1.000 x hours) Refurbishment cost (% of CAPEX) Minimum outlet pressure (bar) Electrical consumption, electricity price and OPEX percentage of CAPEX is reduced as plant size increases. Power prices are including all applicable Danish taxes after deducting tax exemptions. Anticipated base electricity market price excluding taxes is 0,4-0,5 DKK/kWh). Use of excess heat and value hereof is not taken into account. Additional opportunity: On-site Carbon Mono-Oxide electrolysis SOEC is the only known electrolysis technology which has demonstrated efficient electrolysis of CO 2 to CO. Investigations at Risø (and elsewhere) have shown that electrolysis of CO 2 is almost as efficient as electrolysis of H 20 and that degradation rates are very low for CO 2 electrolysis 11. Consequently, the cost of producing 1 Nm 3 of CO by SOEC electrolysis will be comparable to the cost of producing 1 Nm 3 of H 2 by SOEC electrolysis. The reason to consider CO independently is that the market prices of CO are about 6 times higher than the market prices of H 2. Furthermore, there are significant health and safety issues involved in the storage of CO, which might add further to the practical cost. It is therefore a potentially attractive business to use SOEC to make on-site CO on demand from stored CO 2. Using the same assumptions as in the table above, it is found that with CO-prices six times higher (per Nm 3 ) than hydrogen prices, the acceptable CO 2 electrolyser system targets prices ( /Nm 3 /h) are from 7 to 10 times higher for CO-production than for hydrogenproduction. 11 Sune Ebbesen, Performance and Durability of Solid Oxide Electrolysis Cells, Capture and conversion of CO2 into sustainable hydrocarbon fuels, Roskilde April Page 15 of 65

16 4.2 Central industrial (1B) Most industrial hydrogen used today is produced at large central plants based on natural gas reforming. The produced hydrogen is either consumed at the production site (e.g. large oil refineries) or distributed as industrial gas to end-user sites where the consumption do not make natural gas reforming feasible. Central hydrogen production from natural gas is a very cost effective solution enabling a production price as low as 2/kg before distribution cost (price may vary from production site and region). For central electrolysis production to provide the same low cost of 2/kg, would require a CAPEX on per Nm3/hour and 3,6 kwh/nm 3 and a low power price of 0,034/kWh. (see section 4.7) Electrolysis technology in general has its present strength in small scale onsite applications whereas natural gas reforming is well suited for large scale and central locations. This also means that moving electrolysis from onsite (small scale) to central (large scale) may not yield that many benefits, e.g. in terms of improved energy efficiency or cost reductions, as is the case for natural gas reforming. This correlation may change over the years as the electrolysis technology matures. Page 16 of 65

17 4.3 Onsite CHP (2A) Efforts are being pursued in various demonstration projects nationally and internationally to use electrolysers for onsite hydrogen production in e.g. households combined with onsite storage and later conversion in fuel cells to electricity and heat (Combined Heat and Power - CHP). The idea or rational is that households can serve as decentralized balancing plants for fluctuating renewable energy, whilst also providing heat in areas without access to the central district heating grid. The market addressed is therefore to some extent similar to that of natural gas based CPH, except that areas without access to the natural gas grid also are potential ones. For further on the market potential in CPH see the Danish national SOFC strategy 12. For a CPH system each household is to pay back the system through own use of the produced heat and electricity. Also the economical profit to pay back the system is to be achieved in the price difference between the electricity consumed for the electrolyser and the given market electricity price when using the electricity from the fuel cell as well as the savings in costs for heating. Converting hydrogen to electricity & heat should only be done when the electricity price in the grid is high and hydrogen production only when electricity price in the grid is low. Therefore the profit to pay back the CHP lies in the electricity price gab, plus the potential savings in heat costs when using excess heat from electrolyser and fuel cell. To quantify the electricity price gap today Nordpool prices from 2007 and 2010 has been used. To quantify the possible electricity price gap in the future electricity price projections onwards 2050 has been used 13. The various electricity prices have been plotted to the graph shown to the right To calculate the potential profit of this scheme it is for simplicity assumed that ideal operation is possible, i.e. that electricity can be purchased at the lowest possible prices and sold again at the highest possible prices. This is shown schematically in the graph to the right for Nordpool prices in It is seen that the size of the potential electricity price gab depends critically on how many hours this is achievable and on the electricity prices a given year. DKK/Mwh hrs 1000 hrs purchase Sale In order to establish some guideline targets, the simple assumptions is made that hydrogen production is conducted 50% of the year and fuel cell conversion in the remainder 50%. Concerning actual price gab two scenarios has been set up with a price gab of respectively 0,6 DKK/kWh and 0,4 DKK/kWh. The gabs are based on a low electricity price of respectively 0,1 and 0,2 DKK/kWh and a high price on 0,6 and 0,7 DKK/kWh. All prices are market price excluding all applicable taxes. These are simple and also very optimistic assumptions, however they have been formulated to show what price gabs that a CHP system with 12 "Dansk strategi for udvikling af SOFC-brændselsceller " page 7-10, Partnerskabet Brint & Brændselsceller SOFCstrategigruppen 13 Klimakommisionen 2010, Modelanalyser af indpasning af fluktuerende elproduktion i energisystemetfremtidsbillede for Danmark i 2050 Page 17 of 65

18 onsite hydrogen production would require. Further analysis could be done to determine scenarios for both price gabs and number of power price allocation on hours per year. However such analysis is beyond the scope of this study. Target market for CHP is households that have no access to district heating. Today heat for such households are provided from various sources e.g. natural gas, biomass or electricity. As share of renewable electricity in grid are to increase in future it is assumed that households using electricity for heating will also increase (either direct or through use of heat pumps). A simple assumption is therefore made that a household using a CHP system would alternately have used electricity for heating, thus the value of excess heat from electrolyser and fuel cell is valued at the given electricity price in the grid. Excess heat from the electrolyser is valued at the present low electricity price and excess heat from the fuel cell at the present high electricity price. Possible storage of heat is not taken into account. Storage of hydrogen is to be done at >35 bar without compression in order to reduce cost of the CHP system, thus this should also be the outlet pressure of the electrolyser (target). Targets for the electrolyser and fuel cell electrical efficiencies have been set as the maximum realistic achievable targets. A payback time of 5 years has been assumed as this is assumed the longest time acceptable for ordinary consumers. Given the above assumptions a calculation on operation of a CHP system in a normal sizes Danish household has been calculated, as outlined in the table on the following page. The calculations result in a target price per kw electrolyser and a theoretical hydrogen price (to enable comparison with other electrolyser markets) as well as various other targets. As can be seen from the table the electrolyser when used in a CHP system is to reach prices per kw electrolyser system on between DKK and a target hydrogen price on DKK/kg. The calculation also results in a theoretical service cost that may be too low, even though the percentage of CAPEX is on level with the assumption for other markets. Higher service cost will impact the economics. Based on above calculations the targets for electrolysers in the CHP market is summarized in the table below. Market driven targets for SOEC in CHP market Target parameter Target Note Hydrogen sales price 2,2-2,5 /kg Turnover of heat & electric divided by annual H2 production Electrolyser system target price ( /Nm3/h) Annual service cost System life time 3% of CAPEX > hours / 10 years 40% of total CHP cost Stack life time before refurbishment > hours / 5 years Electrolyser operating 50% of a year System electrical consumption System production capacity 3,75 kwh/nm3 ~0,42 Nm3/hour Outlet pressure >35 bar To avoid cost for compression into storage Page 18 of 65

19 Onsite CHP household calculation case Assumptions Data Note Household annual electricity consumption kwh Corresponds to an average Danish household Fuel cell annual electricity production kwh Operating 50% of year Fuel cell system size 0,68 kw Capable of producing kwh in 50% of a year Fuel cell system electrical efficiency (target) 55% A 80% of heat loss recoverable for household Hydrogen production needed annually for fuel cell Nm3 To produce kwh at 55% efficiency Electrolyser capacity 0,42 Nm3/h Capable of producing Nm3 in 50% of year Electrolyser system electrical efficiency (target) 80% A 50% of heat loss recoverable for household System electrical consumption (kwh/nm3) 3,75 Corresponds to 80% efficiency Electrolyser system size 1,56 kw To provide 0,42 Nm3/h at 80% efficiency Annual service cost CHP system (target) 3% Of CPH system CAPEX Electrolyser share of CHP system CAPEX 40% Remainder CAPEX to cover fuel cell & heat system Results economic calculation Electrolyser operation Price gab 0,6 B Price gab 0,4 C Note Annual electricity consumed for hydrogen production kwh At the given electrolyser efficiency Cost of electricity consumed 682 DKK DKK At the given low electricity price Annual heat recovered from electrolyser 682 kwh 50% of loss recovered Value of heat recovered from electrolyser 68 DKK 136 DKK At the given Low electricity price Total annual cost electrolyser operation 614 DKK DKK Fuel cell operation Annual electricity produced by fuel cell kwh 50% of household annual consumption Value of electricity produced DKK DKK At the given High electricity price Annual heat recovered from fuel cell kwh 80% of loss recovered Value of heat recovered from fuel cell 756 DKK 648 DKK At the given High electricity price Total annual profit from fuel cell operation 2.856DKK DKK Annual service cost CHP system 292 DKK 159 DKK 3% of CHP system CAPEX Total annual profit created DKK DKK Profit minus costs/expenses 5 years accumulated profit / target CHP CAPEX DKK DKK CHP CAPEX recoverable after 5 years Electrolyser target CAPEX DKK DKK 40% of CHP CAPEX Electrolyser system target price (DKK/Nm3/h) DKK DKK CAPEX divided by electrolyser size Target price hydrogen 18,5 DKK/kg 16,35 DKK/kg Theoretical H2 price for CHP turnover Minimum lifetime (operating only 50% of a year) hours For electrolyser to operate total 5 years A) Cell degradation over operation period not taken into account B) Low electricity market price is 0,1 DKK/kWh and high is 0,7 DKK/kWh resulting price gab is 0,6 DKK/kWh C) Low electricity market price is 0,2 DKK/kWh and high is 0,6 DKK/kWh resulting price gab is 0,4 DKK/kWh Page 19 of 65

20 4.4 Onsite off-grid balancing (2B) Several remote regions worldwide have off-grid power systems typically based on diesel power and heat generation, that are not connected to a larger power grid system. Diesel electricity generally results in significant higher prices than in large power grid systems with central power plants. Also the more remote region the higher diesel costs due to longer distribution distances for the fuel. Due to generally increasing costs of diesel fuel, several remote regions are investing in renewable energy sources like wind turbines. However due to the wind fluctuations the share of wind power in the off-grid is limited and diesel generators are still needed to provide balancing production, thus increasing the average electricity price in the grid. Besides wind, this fluctuation issue may also account to other renewable energy sources like hydro or solar power, depending on the regional conditions. The value proposition for producing hydrogen in off-grid systems is similar to the market for Micro CHP (2A) as well as central grid balancing (2C); namely to produce hydrogen when plentiful of renewable energy is available, and convert it to electricity and heat when there is lack of energy. The higher costs of diesel electricity compared to prices in large power grids creates potential for a much larger difference in low and high power prices. This is illustrated in the picture to the left, showing an indicative power price window for respectively Denmark and Greenland. Energy supply in Greenland is very distributed on few large cities and several smaller settlements. In some of the larger cities hydro power is available, whereas in many settlements diesel electricity and power generation is only available option. As the price range window is the main parameter affecting feasibility of hydrogen energy storage, off-grid systems could be an earlier market for grid-balancing than large power grid systems. Also the smaller scale of off-grid systems fits well with the anticipated technological starting level for SOEC at lower capacity sizes. Calculating the economical feasibility of SOEC as well as the related technical targets differs greatly from region to region due to differences in diesel costs, renewable energy share, type and costs. It is beyond the scope of this project to conduct analysis of several remote off-grid applications, as this will not have any impact on Danish energy political goals (only few off-grid systems exists on Danish islands). Instead a theoretical calculation example is conducted showing what different electricity inlet prices for SOEC hydrogen production may enable in outlet electricity price when the produced hydrogen is used in a SOFC fuel cell. To evaluate wetter the achieved electricity price intervals has a market potential, these are compared to the electricity price intervals in Greenland, which has a large share of diesel power generation. The calculation is done for a balancing plant with a 1MW electrical capacity from the SOFC fuel cell. The SOEC electrolyser is half the size, in terms of production capacity compared to the fuel cell hydrogen consumption. The fuel cell operates 4 months a year and the electrolyser 8 months a year. Storage is sized for 1 month fuel cell operation or 2 months hydrogen production. Page 20 of 65

21 To enable a feasible case SOEC is to reach the long term potential in terms of a low energy consumption (3,6 kwh/nm3) and a low cost of 2.000/Nm3/h). The reason for selecting SOFC is the anticipation that this technology is available when SOEC reaches is full potential. SOFC is expected to have the same CAPEX cost close to a conventional gas engine for power plant and an efficiency of 60% 14. Excess heat from both the electrolyser and fuel cell is not taken into account as this is also available from diesel generators. Three different inlet electricity prices for the hydrogen production are used: A theoretical low electricity price of 0,03/kWh reflecting low cost hydro power 0,06/kWh the electricity price from a wind turbine when operating on full commercial terms A theoretical high electricity price of 0,11 kwh - reflecting Danish market price including taxes Based on the above parameters the resulting outlet power price from the SOFC fuel cell has been calculated, with a simple payback of the CAPEX in 10 years, as outlined in the table below. Off-grid SOEC/SOFC balancing plant calculation case Parameter Hydrogen production - inlet power price ( /kwh different scenarios) Fuel cell - outlet power 10 year simple payback of CAPEX ( /kwh) Low power price Commercial wind power High power price 0,03 0,06* 0,11 0,29 0,34 0,45 Total plant CAPEX (SOEC, SOFC and H2 Storage) Annual indirect OPEX (% of SOEC and SOFC CAPEX) 3,65% Fuel cell hydrogen consumption (Nm3/kWh) 555 Fuel cell efficiency (@ 3kWh/Nm3 hydrogen) 60% Electrolysis energy consumption (kwh/nm3) 3,6 SOFC fuel cell CAPEX ( /kw) SOEC electrolyser CAPEX ( /Nm3/h) H2 Storage CAPEX ( /kg H2) 105 * Average electricity price from a commercial wind turbine operation & investment case. Source: EMD International A/S "Vindmøllers Økonomi" Feb Other assumptions SOFC fuel cell system size of 1MW electricity production Electrolyser half size of fuel cell (on a hydrogen production/consumption ratio) Fuel cell operates 4 month a year (24 utilization) Electrolyser operates 8 months a year (24 utilization) Hydrogen storage dimensioned for 1 month fuel cell operation or 2 months hydrogen production Hydrogen storage at ~20bar in steel vessels (type 1) directly from electrolyser without use of compression SOFC fuel cell CAPEX expected to be close to conventional power gas engine costs today Value of excess heat from production and fuel cell not taken into account, as this is also available from diesel generators The resulting outlet prices from the fuel cell can then be compared with the average price levels in different off-grid power systems worldwide. As an example a comparison is made with prices in Greenland, where prices vary significantly depending on each settlement and whether there is access to hydro power. The table shows the highest and lowest electricity production cost in Greenland, as well as the average cost across 70 different settlements in Greenland. As can be seen the price range is quite significant and the average production cost on 0,65/kWh is substantially higher than the outlet price from the fuel cell on 0,34/kWh when based on electricity from wind power. 14 Danish Energy Agency & Energinet.dk "Technology Data for Energy Plants" June 2010 Page 21 of 65

22 4.5 Central grid balancing (2C) Electrolysis can be used for central grid balancing by producing hydrogen when excess electricity is available at low prices and use this hydrogen to produce power later when electricity is needed in the grid and prices are high. This power can be produced either by a gas engine of fuel cells. This is shown schematically in the figure below. To calculate the potential profit of this scheme, it is assumed that ideal storage operation is possible as described in section 2a, where electricity is always purchased at the lowest possible price and sold at the highest. The conversion efficiencies used for the calculation are 2020 values 15 as shown next to the figure below. The electricity prices used are Nordpool prices from 2007 and 2010 and estimated Nordpool prices in are the same as shown in section 2A. The heatprice for all three years are assumed to be 0,3 kr/kwh 17 Electricity Efficiency SOEC 100% Gas Engine 42% Electricity SOEC H2 Storage Gas Engine 43% Heat District Heating SOFC 55% Electricity 35% Heat The operating income of storing and reusing hydrogen (not including depreciation of capital expenses, maintenance or interest rates) are shown in the figures below for both gas engines and SOFC used for power generation Storage 2010 Storage 2007 Storage Storage 2010 Storage 2007 Storage k /year/mw(soec) k /year/mw(soec) Operating income for electricity storage based on a gas engine Operating income for electricity storage based on SOFC 15 The Danish Energy Agency, 2010, Technology Data for Energy Plants 16 Klimakommisionen 2010, Modelanalyser af indpasning af fluktuerende elproduktion i energisystemetfremtidsbillede for Danmark i Natural gas unit replacement price according to H2 Logic Page 22 of 65

23 What is noticeable from these results is that: Even in a year with high electricity prices like 2010, there is in principle money to be made form storing electricity The high electrical efficiency of SOFC is quite important for the operating income To calculate the possible profitability, the operating income should be compared to the operating costs consisting of depreciation, maintenance costs and a desired interest rate. Here the following standard prices are used together with an interest rate of 10% for SOEC and SOFC due to the fact that this technology is still quite un-proven, while 5% is used for gas turbines. Data from 18 SOEC SOFC Gas turbine Gas storage Investment k /kwe 0,57 1,5 1,5 0,13 19 Yearly operating and maintenance costs k /MW Technical lifetime Years Desired interest 10% 10% 5% 5% Yearly cost k /MW 80,4 137,9 49,9 13 By adding the SOEC and storage yearly costs with the cost of gas-turbine/sofc and comparing this with the operating incomes in the figures above, it is found that energy storage based on SOEC is NOT a straight forward profitable business. This result is in line with previously published analysis 20 To compare these results with the calculated profit of existing electricity production, the operating income of a gas engine operating on natural gas (see figure 2.C.3) is also considered using the same assumptions. In this case electricity and district heating is produced from natural gas (at 2.5 kr/nm 3 ) when electricity prices are attractive. Electricity Trad 2010 Trad Gas Engine District Heating Yearly income k /MW Trad Natural Gas Oper at i ng h our s Reference configuration for natural gas based power production Calculated operating income for natural gas based power production 18 The Danish Energy Agency, 2010, Technology Data for Energy Plants 19 Assuming 48 hours storage, an SOEC efficiency of 3,5 kwh/nm3 and a storage investment of 105 /kg 20 Claus Krog Ekman and Søren Højgaard Jensen, Prospects for large scale electricity storage in Denmark, Energy Conversion and Management 51, P , 2010 Page 23 of 65

24 Comparing, the calculated operating income of natural gas power generation with the CAPEX estimated previously reveals that traditional electricity production based on gas turbines can be a profitable business at the moment (year 2007 and 2010), but that this is NOT expected to be the case in year 2050 with assumptions used here. This is in-line with the remarks in Vindsporet where Energinet.dk expects that very high peak electricity prices are necessary to make electricity production in low-wind periods attractive. Based on the previous discussion, it is found that SOEC energy storage is NOT a profitable business in the near future but that it will become increasingly interesting towards 2050 and in the same period traditional electricity production based for example on natural gas will become increasingly difficult to make profitable. It could therefore be attractive to investigate if some kind of transitional scheme could be proposed which would allow a gradual transition from electricity production to electricity storage at low investment and operating costs. One such possible configuration is shown in the figure below. Electricity can be produced from hydrogen or natural gas in an SOFC and electricity can be stored using an SOEC. Combined SOFC/SOEC Investment k /kwe 1,5 Electrical efficiency 55% Yearly operating and maintenance costs k /MW Technical lifetime To reduce investment cost the SOFC and SOEC is combined into one unit which can operate in both modes using the same stacks. A best case estimate of the possible operating costs in this case is shown in table to the left. Years Operating frequency Desired interest Yearly cost k /MW 66% 5% 96,6 Comparing this with the calculated operating income shown in the following figure, it is seen that this scheme could potentially be profitable in both 2010 and 2050, assuming that: 1. Combined SOFC & SOEC operation is possible at the same costs previously assumed for SOFC only 2. Interests rates are low, corresponding to a well proven technology In order to pursue this kind of configurations in the future more research will be needed on dual-mode SOEC/SOFC operation. Electricity Storage 2010 Trad 150 SOEC H2 Storage SOFC District Heating k /year/mw 100 Natural Gas SOEC/SOFC dual-mode for storage and power generation Estimated operating income for the dual mode configuration Page 24 of 65

25 4.6 Onsite hydrogen fuel (3A) Leading car manufacturers pursue use of hydrogen as fuel in fuel cell electric vehicles, in order to enable zero emission and fossil independent propulsion of vehicles through with the same ease of use as fossil fuels in terms of fast refueling and long operation range. Hydrogen and fuel cells are therefore supplementing batteries as these cannot enable long range or fast refueling. Today hydrogen powered prototype Fuel Cell Electric Vehicles (FCEV) can achieve ~31 km/l Gasoline Equivalent (G.E. - tank-to-wheel), which is significantly longer as compared to similar gasoline combustion engine vehicles at 16,8 km/l G.E. 21. The long term potential towards 2035 is 43,4 km/l G.E 22. Taking into account energy consumption for hydrogen production and dispensing (well-to-wheel) a full sized FCEV today will consume 0,5 kwh/km with potential improvement onwards to 0,271 kwh/km. This is for a full-size SUV car with a range above 600 km and 3 min. refueling (see figure below). The data is based on the state-of-the-art figures from WP3 on efficiency improvement of hydrogen production as well as increasing in vehicle fuel efficiency (tank-to-wheel) to 0,57 kg H 2/100 km in Today small sized battery electric vehicles achieve between 0,16 kwh/km to 0,3 kwh/km 23 (depending on consumption for cabin heating) with only 150 km range and 10+ hours refueling. Increasing range and reducing refueling time to match hydrogen will result in significant higher energy consumption. Besides providing efficient, zero emission and convenient vehicle propulsion hydrogen can also help balance and integrate renewable energy, both in the energy sector as a whole as well as enabling use hereof for transport. This is also in line with the recommendations in the strategic plan 24 of the Danish Transmission System Operator Energinet.dk, where production of hydrogen is seen as an important storage medium for enabling high inputs of fluctuating wind power. Where 1,5 million battery vehicles (70% of Danish car fleet) can balance 2,5-5 hours of full wind power production, hydrogen production can store 6-8 weeks of wind power production. Such long term storage of energy is required to achieve very high wind power inputs to the grid in future. As gasoline fuel provision has matured during the last 100 years any new fuel including hydrogen will require substantial upfront investments with a long payback time. A recent international study 25 presented under the leadership of McKinsey has revealed that extra total cost of ownership for hydrogen infrastructure and fuel cell vehicles in Europe, compared to gasoline, onwards 2030 will be 100 million. From 2030 the TCO for hydrogen will be positive and roll-out can happen without losses on a commercial basis Battery Electric Vehicle testing conducted by Icelandic New Energy Page 25 of 65

26 For onsite hydrogen production and SOEC technology the hydrogen end-user price dispensed at pump determines the various technical targets to be reached. The higher price dispensed at pump, the less strict targets for the onsite production technology and vice versa. For vehicle users to prefer or select hydrogen both the vehicle cost and the fuel or operation cost per driven kilometer has to be the same. Under the anticipation that car manufacturers in future ensure the necessary competitive vehicle purchase and service price, the importance for the hydrogen sales price and thus the onsite production technology is the fuel cost per driven kilometer. Hydrogen should be priced to ensure the same cost per driven km as a gasoline vehicle in future. With an anticipated market introduction of FCEV s during the competition basis is most likely hybrid gasoline vehicles (HEV) with a fuel economy of 20 km/liter gasoline. As outlined in the figure below this approach enables selling hydrogen at a quite high price due to the high fuel efficiency in a FCEV compared to HEV. The figure shows the present average cost of gasoline today and the resulting cost per driven kilometer in a HEV (20 km/liter). With a corresponding cost per driven kilometer in a FCEV with a fuel economy of 30km/liter gasoline equivalent the hydrogen sales price dispensed at pump can be as high as 9,69/kg. From the time of market introduction of hydrogen ( ) the fuel economy of both FCEV and HEV can be expected to increase. HEV can evolve into a plug-in version (PHEV) with the first kilometers driven very energy efficient on batteries alone. However such hybridization can also be applied to the FCEV, if it proves economically feasible. Further FCEV is at an early technology stage with more efficiency improvement potential compared to the conventional combustion engine. However to ensure a conservative assumption the energy efficiency improvement of FCEV and HEV is anticipated to be the same over the years, thus the efficiency improvement can be removed from the calculation equation. Overall, despite the use as a vehicle fuel in competition with gasoline, the possible hydrogen sales price dispensed at pump on 9,69/kg is comparable to the lower target price of onsite produced industrial hydrogen gas (400Nm 3 /hour plant). Thus use of hydrogen as vehicle fuel may be an achievable market for SOEC or other onsite production technologies already in short-to-medium term. If an increase in oil price and thus gasoline price is taken into account this will only further increase the hydrogen sales prices due the fuel economy difference between HEV and FCEV. When oil price increases, this is most beneficial to the most energy efficient technology, if the cost per driven kilometer will be the same for vehicle users. However to keep assumptions conservative increase in oil price is not taken into account. Page 26 of 65

27 For onsite hydrogen production at a Hydrogen Refueling Station (HRS) to enable a sales price of 9,69/kg the following targets have been formulated: Targets for onsite hydrogen production at hydrogen refueling stations To enable 70MPa hydrogen price of < 10/kg dispensed at pump with CAPEX coverage in <5 year Target parameter Hydrogen price dispensed at pump Target <10 /kg Electrolyser capacity range (Nm 3 /h) System life time Stack life time before refurbishment > hours / 10 years > hours / 5 years Electrolyser consumption (kwh/nm3- beginning of life) 3,9 kwh/nm3 Degradation/year 2,5% Outlet pressure Electrolyser system target price (CAPEX) >50 bar Service cost/year (% of CAPEX) 3,5% Stack refurbishment cost (% of CAPEX) 25% /kw Power price is the Danish average the past 12 months (2011) + all applicable taxes = 0,1/kWh 25% VAT is included in the hydrogen price a pump CAPEX coverage is for both the hydrogen production and hydrogen refueling station. The table outlines the needed evolvement in various key targets for onsite electrolysis production in order to reduce the hydrogen price dispensed at pump at hydrogen refueling stations (HRS) to < 10/kg beyond The stated pump price onwards is the theoretical one needed if the CAPEX investment for both the production and HRS are to be paid back in less than 5 years (simple payback). The operation is based on Danish conditions where all applicable electricity and fuel taxes are paid and electricity costs based on present prices. Onsite production investment costs (CAPEX) are to be reduced from state-of-the-art to around per Nm 3 /hour capacity. The electrical system consumption per produced Nm 3 hydrogen is to be decreased to 3,9 kwh in The potential for SOEC technology is down to 3,6 kwh, however this may take decades to reach, thus the 3,9 kwh is an achievable target in medium term. Outlet pressure from the electrolyser also has to be increase from 10bar today to around 50bar. Reaching the targets for onsite production will enable a full commercial roll-out of HRS s where operation can payback the CAPEX in less than 5 years and enable sale of hydrogen at a price competitive with fossil fuels. Onwards different market phases are expected for use of hydrogen for transport, starting with demonstration and moving towards market introduction. Throughout these phases the electrolysis technology is to be developed and matured towards reaching the targets. Already today electrolysers are marketed for use in demonstration hydrogen refueling stations, despite at the present high prices of per Nm3/hour capacity. When prices are reduced to an early deployment of hydrogen refueling stations with electrolysers is possible. Page 27 of 65

28 4.7 Central hydrogen fuel (3B) Central production of hydrogen and distribution to hydrogen refueling stations, face the same challenges as the market for centrally produced industrial hydrogen gas (1B), as illustrated in the picture below: Central production of hydrogen today is based on natural gas which is very cost effective, enabling a very low outlet price on approximate 2/kg, before distribution cost (price may vary from production site and region). When adding the distribution cost, the price when delivered at the hydrogen station may be close to the price that onsite production may provide when reaching the targets (see market 3A). Central electrolysis production therefore has to compete with both a low cost natural gas reforming and onsite production. This also means that the technical targets for central electrolysis have to be stricter than for onsite, as can be seen from the figure. Onsite production can provide a compelling case with in CAPEX cost per Nm 3 /hour capacity and an energy consumption of 3,9 kwh/nm3 and a power price of 0,1/kWh (see market 3A). Central production would in comparison require and 3,6 kwh/nm 3 and a low power price of 0,034/kWh. Electrolysis technology in general has its present strength in small scale onsite applications whereas natural gas reforming is well suited for large scale and central locations. This also means that moving electrolysis from onsite (small scale) to central (large scale) may not yield that many benefits, e.g. in terms of improved energy efficiency or cost reductions, as is the case for natural gas reforming. This correlation may change over the years as the electrolysis technology matures. Also sale of excess heat and synergy with central power plant operation, e.g. balancing of wind electricity, may provide extra revenue sources for central electrolysis compared to onsite. Page 28 of 65

29 4.8 Central synthetic fuel (4A) One of the main motivations to consider using electrolysis for synthetic hydro-carbon based fuels is that these types of fuel are compatible with existing infrastructure in particular with respect to storage and transportation. A scheme which has recently been proposed by Energinet.dk for electrolysis assisted central synthetic hydro-carbon based fuel production is shown in the figure below. Possible scheme for centralised synthetic fuel production Here excess electricity is used for electrolysis and the produced hydrogen is stored in large caverns like Stenlille. Gasified biomass together with the electrolysed hydrogen is used for a range of applications depending on demand. These applications include production of heat and power, production of synthetic transportation fuels and production of synthetic natural gas. An economical analysis of the production of synthetic transportation fuels based on such a centralised synthesis gas production has recently been performed in the EUDP project Green Synfuels. Here the cost of methanol production based on biomass gasification and additional hydrogen from electrolysis has been evaluated and these results will be used for the following discussion. Biomass H 2 +CO+CO 2 H2O Methanol synthesis H2 Gasification Electrolysis CH3OH Methanol The basic process flow for the methanol production is shown simplified in the figure to the left tons of wood per day is gasified. For methanol production the produced syngas contains a surplus of CO2, which can either be removed or additional hydrogen can be added. If the CO 2 is removed 523 tons/day of methanol is produced. If on the other hand 200 tons/day of electrolysed hydrogen is added, the daily methanol output will be 1053 tons Page 29 of 65

30 instead. Oxygen is needed for the gasification process and this is supplied by the electrolysis process. As no storage is considered is not possible to turn the electrolysers completely of, but they can operate between 33% (enough oxygen is produced for the gasification) and 100% (enough hydrogen is produced to convert all CO2 to methanol). In the Green Synfuels report the cost of three scenarios are calculated. Cost is expressed as the equivalent oil price which would make this scheme competitive with methanol produced from fossil fuels. Here it is found that: Using gasification only (no hydrogen added) or using gasification and alcaline electrolysers an equivalent oil price of roughly 150 USD/barrel will make the synthetic methanol production competitive. Using gasification and SOEC electrolysers, an equivalent oil price of only 120 USD/barrel is needed to make the synthetic methanol production competitive. Such cost projections depends critically on a number of assumptions, which can be found in 26. The most important SOEC assumptions are capital costs of 0.55 k /kw and an effective lifetime of the SOEC system of 10 years. This lifetime is based on the assumption that the SOEC stack (1/3 the cost) needs to be replaced every 5 years, whereas the other system components (2/3 the cost) will last for 20 years. The distribution of the different capital and operating costs of the SOEC case is shown in the figure to the left. Here it is interesting to see that supply of wood is the largest cost and that the cost of gasification is considerable larger than the cost of SOEC and methanol synthesis. In total the gasified biomass accounts for 59% of the cost. As the biomass gasification technologies matures in the future further cost reductions could be anticipated Another aspect to consider is of course the use of gas storage, which would allow the electrolysers to operate more cost effectively and to increase the output of the system as hydrogen for converting all CO 2 to methanol would be available all the time. The figure below shows a calculation of the cost structure of methanol production with and without gas storage. As cavern storage is assumed the cost of gas storage has been neglected. This is probably fair for hydrogen storage, whereas it is not obviously true for the oxygen storage needed for operating the gasifier in periods without electrolysis. The calculation is based on electrolyser operation in 50% of the operating time and use of stored gases in the other 50% of the time. The electrolyser capacity of course needs to be doubled compared to the previous cases and the average methanol output is increased to 1053 tons per operating day compared to previously an average of 876 tons. 26 Appendix 1 and 2 of Figur 1, %20publikationer/2010/forudsaetninger_for_samfundsoekonomiske_analyser_paa_energiomraade.pdf Page 30 of 65

31 4 3,5 3 kr/kg Methanol 2,5 2 1,5 Other Electricity SOEC Methanol Synthesis Biomass gasification Wood 1 0,5 0 No storage Storage Calculated cost structure for SOEC assisted methanol production with and without gas storage It is found that with these assumptions the cost of methanol is about 10% lower with storage than without. In which case break even is achieved at a crude oil price around 110 USD/barrel. Price projections from Energistyrelsen 27 expects crude oil prices to reach110 USD/Barrel around With the average crude oil price for 2011 being above 100 USD/barrel 28 this analysis shows that profitable central synthetic fuel production based on renewable sources may indeed be realistic in a not too distant future. The previosus calculations shows that with 0,55 k /kw and 10 years of effective lifetime, the SOEC CAPEX only consists 7-14% of the methanol price. It could therefore be argued that higher SOEC prices might be acceptable for this application. For the roadmap discussions of WP5, it is assumed that SOEC prices between 1 and 0,5 k /kw are needed for the introduction of central synthetic fuels. However, one potential issue for the near term introduction of SOEC for central synthesis gas production is that relatively large electrolyser capacities are required. In the example above, the 200 tons hydrogen per day corresponds to an SOEC capacity of 140 MW without storage and 280 MW with storage. With the estimated early installed electrolysis capacity being of the order of 200 MW, it is evident that it will take some years before SOEC systems of this scale can be installed. 27 Figur 1, %20publikationer/2010/forudsaetninger_for_samfundsoekonomiske_analyser_paa_energiomraade.pdf 28 Page 31 of 65

32 4.9 De-central synthetic fuel (4B) An energy source which is attracting increasing attention at the moment is biogas and the potentially wide range of applications it may obtain by upgrading it to pipeline quality. By increasing the capacity of biogas production in Denmark and by upgrading the biogas to pipeline quality, it is possible at the same time to reduce CO 2 emissions and dependence on foreign energy sources. An important feature of upgraded biogas is that biogas is typically produced at relatively small plants ( Nm 3 /h) to reduce the transportation cost of the often energy thin biomass used for biogas production. Consequently, biogas can complement large centralised biomass facilities as described in section 4A and make use of bio resources which can not economically be transported to centralised facilities. Furthermore, the use of small plants will also make it simpler to introduce SOEC based electrolysis compared to the larger central plants discussed in section 4A. A number of countries e.g. Sweden and Germany are starting to upgrade biogas to pipeline quality by removing the 30-40% of CO 2 in the raw biogas to obtain >97% CH 4 which is necessary to match existing pipeline requirements for heat value and Wobbe index. This CO 2 wash/removal typically cost 29 around 1.1 kr/nm 3 CH 4. As an alternative to removing the CO 2 from the biogas, it has recently been proposed to use electrolysis to convert CO 2 to methane, through either: CO 2 + 4H 2 = CH 4 + 2H 2O or CO + 3H 2 = CH 4 + H 2O CH4+CO2 Biogas Svovl rens Methanation H2 CH4 SNG H2O Electrolysis This process requires some additional gas cleaning (sulphur removal), electrolysis of H 2O or CO 2 and catalytic methanation as indicated in the figure above. The present Danish biogas production (2008) is around 4 PJ/year and the potential is estimated to be 40 PJ. The electrolysis based upgrade scheme will increase the total amount of gas produced and assuming 40 PJ biogas produced and an average CO 2 concentration of 35% (volume) the electrolysis upgrade scheme would increase the energy content the total gas produced to above 60 PJ. Electrolysis based upgrade of biogas could therefore contribute significantly to some of the key targets of the anticipated future energy structure as described in Vindsporet, where Energinet.dk expects that: 20 PJ of natural gas will be produced by electrolysis. The estimates above shows that this could potentially all be from electrolysis based biogas upgrade Assuming that 14 kwh (4 Nm 3 of H 3.5 kwh/nm 3 ) of electricity is needed to produce 10 kwh gas (1 Nm 3 of CH 4), then biogas upgrade could contribute with more than 50% of the total anticipated electrolyser power consumption of 50 PJ expected necessary in year Energistyrelsen.. Page 32 of 65

33 In the following, some simple estimates of the possible profitability of electrolysis based biogas will be shown and compared with CO2 removal upgrade as the reference case. Much more elaborate studies will be published elsewhere as part of two on-goin projects. Independent of the biogas upgrade technology, subsidies are at the moment needed to make biogas competitive with fossil natural gas. The possible subsidies for upgrading Danish biogas to pipeline quality has not yet been decided. It is however expected 30 that government support of the order of 2.7 kr/nm 3 CH 4 will be provided. For the following economical calculations it is assumed that biogas can be provided at 3.8 kr/nm 3 (CH 4) and that CO 2 wash costs 1,1 kr/nm 3 (CH 4). Estimated cost of different traditional biogas applications include upgrade by CO2 removal from Energistyrelsen, Anvendelse af biogasressourcerne og gasstrategi herfor, Maj 2010 It is evident from the figure above that in this case upgrade of biogas is not necessarily a very profitable business with an estimated profit of (2,5 + 2,7) (3,8 + 1,1) kr/nm 3 = 0,3 kr/nm 3. This is however not the purpose of the following discussion, here the question is, can Electrolysis upgrade compete with CO 2 wash? For the electrolysis based biogas upgrade profitability calculation, the same CAPEX and OPEX assumptions are used as in section 2C. Two electricity prices are used in the calculations, one is the present prices which are found as an average of actual Nordpool prices from , the other is the estimated prices in Furthermore, the analysis also compares the profitability of an upgrade scheme based on SOEC with a scheme based on Alcaline electrolysors. For the alkaline electrolysors the following parameters 31 are used,: Cost Interest rates Maintenance Lifetime Efficiency Alcaline Elecrolyser 0,6 /kw 5% 5% 20 years 4,2 kwh/nm 3 In the simplest operating mode of the system, a constant flow of upgraded biogas is produced. When electricity prices are low, the electrolysers operate with an excess capacity and produce hydrogen for biogas upgrade and for hydrogen storage. In periods with high electricity prices, biogas is upgraded with stored hydrogen. 30 Energistyrelsen, Anvendelse af biogasressourcerne og gasstrategi herfor, Maj J.M. Zolezzi et al, Large scale hydrogen production from wind energy in the Magallanes area for consumption in the central zone of Chile, Journal of Power Sources 195 pp , 2010 Page 33 of 65

34 The figure below shows a calculation of biogas profit versus electrolyser operating time for four different cases. In these cases no value is ascribed to the heat produced. Here it is seen that With present electricity prices, the potential profit of electrolysis based upgrade is comparable to that of CO 2 wash and the optimum time operating time of the electrolysers is fairly long (>60%) Everything else equal, the electrolysis based upgrade becomes less attractive in year 2050 than it is today. This is because average electricity prices increase, whereas gas prices are unchanged in these calculations. The profit difference between SOEC and alkaline electrolysers are marginal in this case. This is because the better efficiency of SOEC is compensated by the longer lifetime (20 vs 5 years) of Alcaline electrolysers Upgrade profit (kr/nm3 CH4) 0,5 SOEC SOEC ,25 Alcaline Alcaline % 20% 40% 60% 80% 100% -0,25-0,5 Electrolysis frequency Profit kr/nm3 upgraded biogas as function of the operating time of the electrolysors This result may not seem very promising for electrolysis based upgrade, however there are good reasons to consider the electrolysis approach. It is hardly fair not to ascribe any value to the heat produced. The majority of this heat comes (at least in the SOEC case) as high temperature (>200 C) steam and water, which could be used for district heating. Assuming that half of the produced heat can be sold as district heating at a price of 0.3 kr/kwh, this adds 0,6 kr/nm 3 biogas to the electrolysis case 32. SOEC is still an immature technology and significant lifetime improvements can be expected in the future. Assuming a lifetime of 20 years (instead of 5) in year 2050 adds 0,35 kr/nm 3 biogas to the profit in the SOEC case. In the future gas engines are expected to provide the electricity needed during periods of low windpower production. It could therefore be argued that in these periods it does not make much sense to convert biogas to SNG at one place and at the same time convert SNG to power at another place. In these periods, it would be more effective to convert the biogas directly into power. Here SOEC has the potential of operating also in SOFC mode and hence (with some system modifications) to be able to convert biogas to power when relevant. Neglecting the additional investment cost of dual SOEC/SOFC operation, this dual mode operation adds 0,7 kr/nm 3 biogas to the SOEC case in year The figure below shows the different cost and profit contributors for a 2050 SOEC scenario assuming a 40% SOEC and a 40% SOFC operating time and 20% operating time based on stored hydrogen. It is fur- 32 At the maximum profit mode of 40% operating time Page 34 of 65

35 thermore assumed that 50% of the heat is used for district heating and that the SOEC/SOFC lifetime is 20 years. 6,00 kr/nm3 Biogas 5,00 4,00 3,00 2,00 Profit Støtte NG price Heat value Biogas IR+Maint Depreciation Electricity 1,00 0,00 Gas price Gas value Power Price Power Value Average Profit Best case SOEC/SOFC based biogas upgrade in Biogas is upgraded in 60% (with electrolysis in 40%) of the time and is used directly for power generation in 40% of the time This indicates the potential of the electrolysis based biogas upgrade in particular in the SOEC case, where lifetime improvements and dual mode operation are considered realistic 5-10 years into the future. To estimate the SOEC price (k /kw) which makes SOEC relevant for introduction in early biogas upgrade systems, it is assumed that a profit of at least 0,3 kr/nm 3 CH 4 is needed, similar to the profit of the competing CO 2 wash. 50% of the heat is used at 0,3 kr/kwh. Furthermore, an effective SOEC lifetime of 10 years is assumed as in section 4A - Balance of plant components has a lifetime of 20 years and stacks can be replaced every 5 year. Average Nordpool prices from are use as the cost of electricity. The most profitable operation in this case is actually obtained with the electrolysers operating 80% of the time, however this is hardly advantageous for the overall energy system. It is therefore assumed that the electrolyser operating time is 50%. In this case an SOEC introduction price of 1,1 k /kw is found to provide the desired profit of 0.3 kr/nm3 CH4. The cost and profit elements in this case are shown in the figure above. Page 35 of 65

36 5. SOEC R&D & commercialisation roadmap formulation Based on the analysis of state-of-the-art and market requirements, a roadmap for the SOEC market entry and R&D efforts is proposed here. The market entry roadmap is based on a comparison between the market requirements found in section 4 and the future technical development state estimated by Topsoe Fuel Cells (TOFC). Two of the most important technical development parameters to consider here are price and capacity. The figure bellows shows the SOEC system unit cost and capacity anticipated by TOFC from 2012 to The capacity here is the maximum capacity of a SOEC system unit, expected to contain several stacks. Larger capacities for a given application can be obtained by coupling several units together into a single system. The prices are based on simple extrapolation from previously published TOFC SOFC cost projections 33. For a given SOFC system price it is assumed that the stack makes up 33% of the cost and the balance of plant (BoP) accounts for the rest. Due to the higher current density and cell voltage of SOEC compared to SOFC, the cost per kw for a SOEC stack is assumed to be 25% of the cost per kw of a SOFC stack. The SOEC BoP is expected to be similar but somewhat simpler than the SOFC BoP. SOEC BoPs are simpler because neither fuel reforming nor off-gas burners are required and the airflow is much smaller. Here, it is estimated that the SOEC BoP cost/kw are 2/3 of the SOFC BoP cost/kw. 18 SOEC unit sales price (k /Nm3/h) SOEC BoP SOEC Stack 2012 (1 Nm3/h) 2015 (10 Nm3/h) 2017 (75 Nm3/h) 2020 (300 Nm3/h) SOEC Unit release year and capacity With these simple assumptions, it is found that the SOEC system price is expected to drop from /Nm3/h in 2012 to /Nm3/h in At the same time the (maximum) capacity per SOEC system unit is expected to increase from 1 Nm3/h to 300 Nm3/h. It is also noticeable, that the BoP price is expected to dominate the SOEC system price. 33 EUDP applications , 2010 and Green Natural Gas, 2011 Page 36 of 65

37 In the figure below, the maximum market entry SOEC prices found in section 4 are plotted against the minimum capacity needed for a specific application. By comparing the two previous figures, it is possible to propose early ( ), medium term ( ) and long-term (2020+) markets. This is shown schematically in the figure below, where the estimated SOEC prices are shown with bars. To make the details of the graph easier to see, the CO2 on-site industrial graph has been removed. From the figure below it is found that: Early markets are on-site industrial electrolysis, in particular CO 2-electrolysis, but H 2O electrolysis could also be attractive. Medium term markets are on-site hydrogen fuel, off-grid balancing and de-central syngas (biogas upgrade) Other markets are expected to be mature after See larger version of the graph by the end of section 5. Page 37 of 65

38 The order of markets entry for different electrolysis applications found here are in good agreement with the order shown in the Danish elecrolysis strategi 34. The main difference is that this analysis expects the introduction of electrolysis for biofuel (biogas) and hydrogen transportation fuel before the introduction of electrolysis for µchp. Based on this the market entry analysis some key aspects of a R&D roadmap proposal becomes fairly straight forward. First of all it is noticed that the commercial sale of SOEC systems could potentially start almost as soon as BoP components are developed and reliable systems are demonstrated. This indicates that high priority should be given to: BoP component development and system demonstration. An important aspect here is the ability to flexibly and reliably couple several system units into a single operating system matching the capacity needed for a given application. Development and demonstration of reliable and durable SOEC cells, stacks and systems. Requirements on low degradation and long lifetime requirements makes it desirable to investigate electrolysis at operating temperatures below 800 C as this may significantly improve corrosion related degradation of BoP components and stack interconnects. Demonstration and development of pressurised stacks and systems. All early and medium term markets found above require output gas pressures between 10 and 50 bar, in particular for use of hydrogen as transportation fuel In section 4 it was found that dual SOEC/SOFC mode operation could have significant influence on the profitability of mid-term applications like bio-gas upgrade and on-site off-grid balancing. It is therefore recommended that this aspect is investigated before Biogas will contain quite substantial amounts of sulphur and to simplify the sulphur removal of biogas upgrade systems, it is also proposed to dedicate R&D resources to sulphur tolerant SOEC cells before Furthermore, pure oxygen for biomass gasification will be desired for centralised syngas/synthetic fuel production. Assuming that pilot tests of synthetic fuel production are desirable before 2020, it is proposed that SOEC systems which can provide pure oxygen are developed before The issues here are mainly expected to be corrosion related and the developments steeps could include qualification of corrosion resistant interconnects based on advanced coatings and novel steel types. Finally, it should be stressed that the cost and capacity projections shown for SOEC systems are based on the assumption that the SOFC development and product roll-out continues as expected by TOFC. The R&D and demonstration funding necessary to make the SOFC market introduction progress as planned is therefore essential to the market introduction of SOEC systems. A SOEC R&D roadmap proposal reflecting this discussion is shown on the next page. This roadmap is in good agreement with table 6.3 of Elektrolyse I Danmark, with the main differences that the targets for the current density are more conservative in this report, and that this report also stresses the importance of reducing corrosion related degradation. 34 Figur 3.6, Elektrolyse i Danmark, Strategi for F,U & D , Partnerskabet for Brint og brændselsceller, August 2009 Page 38 of 65

39 Page 39 of 65

40 Page 40 of 65

41 PART 2 - R&D of SOEC stacks with improved durability Page 41 of 65

42 0. Summary The objectives of the work packages 6-7 in this project were multiple: 1) To investigate durability of solid oxide cells (SOCs) and stack components under industrially relevant ( harsh ) electrolysis operating conditions; 2) to investigate performance of standard TOFC (Topsoe Fuel Cell A/S) SOC stacks (based on stateof-the-art solid oxide cells) under mild electrolysis operating conditions ( 0.75 A/cm 2 ); 3) to further develop SOEC stack computer models available at Risø DTU and TOFC. Accordingly four lines of work were carried out in the here reported project: - Investigation of corrosion resistance of interconnect alloys. - Cell and stack element testing. - SOEC stack testing. - SOEC stack modeling. Good progress has been made on several of the activities. Below the outcome and most important achievements are listed for the four lines of the work: Corrosion resistance of interconnect alloys In the present project, superior interconnect alloys identified in the project ForskEL (Crofer22APU, Crofer22H, E-Brite, AL29-4) and shaped ICs (interconnects) employed in state-of-the-art SOC stacks, with and without protective coatings, were selected for study. The corrosion resistance was evaluated at 850 o C in both reducing and oxidizing atmospheres, simulating environments in SOEC cathode and anode compartments, respectively: In state-of-the-art SOC stacks, shaped ICs based on Crofer22APU with a protective coating on the oxygen side, are employed. In the present project, a corrosion resistance K p = 5-15 X g 2 /cm 4 s at 850 o C was demonstrated on uncoated Crofer22APU in reducing atmospheres with high steam content, simulating the atmosphere in the SOEC cathode compartment. Crofer22H has slightly better corrosion resistance as compared to Crofer22APU. Meanwhile Crofer22H fulfills other requirements as well as Crofer22APU and is relatively cheaper. It is therefore desirable to employ Crofer22H in SOEC stacks. In oxidizing atmosphere, coated Crofer22APU has a corrosion resistance K p = 1-4 X g 2 /cm 4 s at 850 o C. To allow stable operation for 2 years, it requires an oxide scale thickness of < 10 µm or K p < 5 X g 2 /cm 4 s. As clarified via the oxidation tests in the present project, corrosion of uncoated Crofer22APU in reducing atmosphere with high steam content is more severe than that of coated Crofer22APU in oxidizing atmosphere. This is further supported by the results obtained via stack element testing carried out in the present project. It can therefore be concluded that uncoated Crofer22APU does not have sufficient corrosion resistance for SOEC applications at 850 o C or above for either oxygen or fuel side. This calls for protective coatings on both sides. Alternatively the operation temperature must be reduced. Page 42 of 65

43 Cell and stack element testing Risø DTU has previously shown that state of the art SOFC technology works well also for electrolysis as long as current densities are modest. At single cell level and 850 o C, up to close to 1 A/cm 2 is achievable with modest degradation. In the present project, 9 cell tests and 3 stack element tests were carried out at Risø DTU, with a total test period of exceeding 5000 hours. The purpose was to investigate performance and durability of solid oxide cells under industrially relevant ( harsh ) electrolysis conditions (current density 1 A/cm 2 ) and to identify degradation mechanisms. The tests were carried out on three types of solid oxide cells, 2.0G, 2.1G and 2.5G, which differ mainly in the oxygen electrode. Both the 2.0G and 2.1G cells have a LSM/YSZ composite oxygen electrode, prepared by Production method A and Production method B respectively (LSM: (La,Sr)MnO 3, YSZ: (Y,Zr)O 2). In the 2.5G cells, the LSM/YSZ electrode is replaced by a LSCF/CGO electrode with a CGO barrier layer at the interface between the electrolyte and the oxygen electrode (LSCF: (La,Sr)(Co,Fe)O 3, CGO: (CeGd)O 2). The testing program covered both coelectrolysis and steam electrolysis. Under co-electrolysis, the 2.1G cells showed a degradation rate of mv/1000h at -1 A/cm 2 and >800 mv/1000h at -1.5 A/cm 2. The 2.5G cells showed better durability with a degradation rate of 400 mv/1000h at -1.5 A/cm 2. Under steam-electrolysis, the 2.1G cells showed a degradation rate of 463 mv/1000h at 850 o C, A/cm 2 (determined via stack element testing). The cell degradation under high electrolysis current is related to both the Ni/YSZ and LSM/YSZ electrodes. Loss of percolation in the electrodes, Ni coarsening and re-distribution, diffusion of Y/Zr into Ni particles, formation of cracks in the electrolyte layer, and detachment of the LSM/YSZ electrode were identified as possible degradation mechanisms. To allow stable electrolysis operation at -1.5 A/cm 2, both better oxygen and fuel electrodes are required. A stack element test utilizing Ag-based braze as seal for the fuel side was carried out at 850 o C, -0.5 A/cm 2 for 900 hours (steam electrolysis). Post-mortem analyses showed that after 900 hours testing at 850 o C, Ag-based braze still adhered very well to both the interconnect and the cell. The braze did not show any mechanical deformation. The results proved feasibility of applying Agbased braze as seal for SOEC stacks. Further development work shall focus on optimizing production and sealing procedure. Stack testing Risø DTU has previously demonstrated an electrolysis stack test at ~850 o C and -0.5 A/cm 2 with no degradation. As a step forward, electrolysis stack tests at A/cm 2 were carried out in the present project. During the project period, two identical stacks (N-063 and N-126) were produced at TOFC and tested at Risø DTU. The stacks contained three types of solid oxide cells with a footprint of 12X12 cm 2 : 2.0G, 2.1G and 2.5G. The stack N-063 was tested at 800 o C, A/cm 2 for steam electrolysis. The test was carried out smoothly for 170 hours and hereafter terminated due to a failure in steam supply, which unfortunately destroyed the stack. The stack N-126 was tested at 800 o C for co-electrolysis of steam and carbon dioxide. The test was carried out at -0.5 A/cm 2 for 70 hours and A/cm 2 for 1000 hours, combined with detailed electrochemical and post-mortem characterizations. The following conclusions can be made based on the results obtained on N-126: Page 43 of 65

44 o With respect to the initial electrolysis cell performance at the stack level, the three types of solid oxide cells follow the order of 2.5G>2.1G>2.0G. This is consistent with the results obtained via single cell testing. o The 2.0G cells showed no degradation, but actually activation under co-electrolysis at 800 oc and A/cm 2 for 1000 hours. Post-mortem analyses revealed only minor microstructural changes in the tested cells. o The 2.1G and 2.5G cells showed inferior durability as compared to 2.0G. Whether this is a true picture of cell durability of different cell types is yet too early to be concluded. To clarify this, different types of cells with equally good microstructure and identical active layers shall be employed in future stack testing. In the present project, we have demonstrated that state of the art SOFC technology works well for electrolysis at 800 o C and A/cm 2 with negligible degradation. Stack modeling In the present project, TOFC s SOFC/SOEC stack model was updated with the most recent cell and stack test results obtained within this project. The focus was to update the cell ASR (area specific resistance) expression. The new ASR expression for 2.1G and 2.5G cells was then implemented into TOFC s stack model. The model calculated results were compared with the measured results from Stack N-126, showing good agreement at the beginning of the electrolysis period. With respect to further improvement work on SOEC stack modeling, the following shall be considered: (1) Including effects of CO, CO 2, or CH 4 in the ASR expression; (2) Minimizing temperature change during the measurements dedicated for model validation. In the formulation of the project, nine milestones for the expected achievements were formulated. These are attached to this report as Appendix A. Out of these milestones, only one milestone - M6.3 has not been met in this project. The results characterized by all the other milestones have been presented in the previous biannual reports and in the result section of the present report. With respect to the milestone M6.3, the planned activities on oxidation modeling were stopped with the unexpected leave of an employee allocated to this task. Considering the staff difficulties and time limitation, it was decided to reallocate the project resources set aside for this task to other tasks in this project. The milestone M6.3 is still considered relevant and is recommended to be addressed in future projects. The results obtained in the project are further documented in the next four sections and in a number of internal reports. An overview of these reports is presented in Appendix B. The objectives of the work packages 6-7 in this project were multiple: 1) To investigate durability of solid oxide cells (SOCs) and stack components under industrially relevant ( harsh ) electrolysis operating conditions; 2) to investigate performance of standard TOFC (Topsoe Fuel Cell A/S) SOC stacks (based on stateof-the-art solid oxide cells) under mild electrolysis operating conditions ( 0.75 A/cm 2 ); 3) to further develop SOEC stack computer models available at Risø DTU and TOFC. Accordingly four lines of work were carried out in the here reported project: - Investigation of corrosion resistance of interconnect alloys. - Cell and stack element testing. - SOEC stack testing. - SOEC stack modeling. Page 44 of 65

45 1. Corrosion resistance of interconnect alloys The purpose of this task was to evaluate corrosion resistance of interconnect alloys in SOEC stack-relevant environments. Superior interconnect alloys identified in the project ForskEL and shaped ICs (interconnects) employed in state-of-the-art SOC stacks, with and without protective coatings, were selected for the study. The corrosion resistance was evaluated at 850 o C in both reducing and oxidizing atmospheres, simulating environments in SOEC cathode and anode compartments, respectively. This task was carried out by Risø DTU Reducing atmosphere In state-of-the-art SOC stacks, shaped ICs based on Crofer22APU with a protective coating on the oxygen side, are employed. In the project ForskEL , Crofer22APU, Crofer22H, E-Brite, AL29-4 showed superior corrosion resistance in the oxidation tests of non-coated alloys. In the present project, metal sheets of these four alloys and uncoated shaped ICs were selected for oxidation tests in reducing atmosphere. The oxidation tests were carried out at 850 o C in H 2/Ar or H 2/N 2 with 50% steam (4.5% H 2, 45.5% Ar or N 2, 50% steam), simulating the atmosphere in an SOEC cathode compartment. The corrosion kinetics was monitored as average mass increase versus time. For most of the investigated alloys, the mass increase followed the parabolic rate law. The calculated oxidation rate constants (based on the mass increase) are plotted in Figure 1. The oxidation rate constants of Crofer22APU metal sheet and of IC show only a small difference, indicating that the effect of shaping on the corrosion kinetics of Crofer22APU is minor. This is further supported by SEM observations on the oxide scale of Crofer22APU metal sheet and of IC (not shown in this report). With respect to the metal alloy sheets, Crofer22H shows the least oxidation among the four alloys. Besides, most of the alloys showed different oxidation kinetics in H 2/Ar with 50% steam as compared to H 2/N 2 with 50% steam, which may be ascribed to the different testing environments, i.e. difference in oxygen partial pressure, Cr evaporation rate etc. This however needs further investigation. The composition and the microstructure of the oxide scale were investigated by XRD, SEM &EDS. Formation of (Mn,Cr) 3O 4 spinel on top of a Cr 2O 3 layer was observed for all the four alloys. As shown in Figure 2, Crofer22APU has an oxide scale of 4-5 µm in thickness after being at 850 o C in H 2/N 2 with 50% steam for 940 hours, while the oxide scale thickness of Crofer22H is uneven. The white phases in the metal matrix of Crofer22H and at the metal-oxide interface were confirmed by EDS as Laves phase, (Fe,Cr,Si) 2(Nb,Si) 1. Precipitation of the Laves phases happens mostly at the grain boundaries, which can retard the cation outward diffusion and therefore slow down the oxidation process. Kp, g 2 cm -4 S H 2 /Ar - 50% H 2 O H 2 /N 2-50% H 2 O 0 IC Crofer22APUCrofer22H Alloy AL29-4 E-Brite Figure 1 Oxidation rate constants of uncoated alloy sheets and uncoated ICs at 850 o C in H2/Ar or H2/N2 with 50% steam (4.5% H2, 45.5% Ar or N2, 50% steam). Page 45 of 65

46 Metal Oxide scale Figure 2 SEM back-scatter images on the cross-sections of uncoated alloy sheets oxidized at 850 o C in H2/N2 with 50% steam (4.5% H2, 45.5% N2, 50% steam) for 940 hours, left Crofer22APU and right Crofer22H. Crofer22APU is used as interconnect alloy in state-of-the-art SOC stacks. In the present project, a corrosion resistance K p = 5-15 X g 2 /cm 4 s and an oxide scale thickness of 4-5 µm after 940 hours at 850 o C were demonstrated on uncoated Crofer22APU in reducing atmospheres with high steam content, simulating the atmosphere in the SOEC cathode compartment. To allow stable operation at 850 o C for 2 years, it requires a scale thickness of < 10 µm or K p < 5 X g 2 /cm 4 s. Based on scale thickness or oxidation rate constant, we could conclude that uncoated Crofer22APU does not have sufficient corrosion resistance for SOEC applications at 850 o C or above for either oxygen or fuel side. This is further supported by the results obtained via stack element testing in the present project (Section 2.4 in this report). To improve the corrosion resistance, it calls for protective coatings on both the oxygen side and the fuel side, or a lowered operating temperature for SOEC stacks, or alloys with better corrosion resistance and at the same time fulfilling other requirements as ICs. Crofer22H has slightly better corrosion resistance as compared to Crofer22APU. Meanwhile Crofer22H fulfills other requirements as well as Crofer22APU and is relatively cheaper. It is therefore desirable to employ Crofer22H in SOEC stacks, though protective coatings are still needed for operations at 850 o C. Page 46 of 65

47 1.2. Oxidizing atmosphere Shaped ICs based on Crofer22APU, with protective coatings on the oxygen side, are used in the state-ofthe-art SOC stacks. In the present project, we also studied the effect of shaping on the corrosion resistance of coated Crofer22APU. The corrosion kinetics of flat metal sheet and shaped IC was studied at 850 o C in compressed lab air with 1% H 2O or in pure oxygen for periods up to 2000 hours. Both the metal sheet and the IC were coated with a protective coating consisting of Co 3O 4 and a perovskite. 5 Kp, g 2 cm -4 S Metal sheet (air with 1% H 2 O) Metal sheet (oxygen) 0 IC (air with 1% H 2 O) IC (oxygen) Figure 3 Oxidation rate constants of coated Crof22APU metal sheets or ICs at 850 o C in compressed lab air with 1% H2O or in pure oxygen. As shown in Figure 3, coated Crofer22APU has a corrosion resistance K p = 1-4 X g 2 /cm 4 s at 850 o C in oxidizing atmosphere. The shaped ICs have a smaller K p than the metal sheets. This was however not reflected in the SEM observations. The oxide scale of Crofer22APU metal sheet is quite similar to that of IC, indicating similar oxidation kinetics. Another difference was found between air and oxygen-treated samples. For both metal sheet and shaped ICs, the derived K p value in air with 1% H 2O is only half of that in oxygen. SEM observations on the cross-sections of Crofer22APU metal sheets oxidized in air with 1% H 2O and oxidized in oxygen confirm that, in both samples, the dense oxide scale consists of an inner layer of Cr 2O 3 (2-3 µm) and an outer layer of (Co,Cr,Mn) 3O 4 (7-10 µm). On top of the (Co,Cr,Mn) 3O 4 layer is the porous perovskite layer, which is part of the coating. EDS analyses indicate that the amount of Cr and Fe in the oxide scale and in the porous coating (i.e oxidized Cr and Fe) is comparable for the air and oxygentreated samples. The difference in weight gain might be caused by different oxygen uptake in both the oxide scale and the coating and different evaporation rate. Besides, it is worth mentioning that the thickness of the Cr 2O 3 layer here is thinner (2-3 µm) than the one (4-5 µm) in uncoated Crofer22APU oxidized in reducing atmosphere with high steam content. As clarified via the oxidation tests in the present project, corrosion of uncoated Crofer22APU in reducing atmosphere with high steam content is more severe than that of coated Crofer22APU in oxidizing atmosphere. This is further supported by the results obtained via stack element testing carried out in the present project (see Section 2.4 in this report). It can therefore be concluded that uncoated Crofer22APU does not have sufficient corrosion resistance for SOEC applications at 850 o C or above for either oxygen or fuel side. This calls for protective coatings on both sides. Alternatively the operation temperature must be reduced. Page 47 of 65

48 In the present project, the following milestones were set up in the project plan: M6.2 Corrosion resistance of coated superior interconnect alloys (identified in the project ForskEL or obtained from a third party) investigated in an SOEC stack-relevant environment (high steam content (>50%), CO 2-containing atmosphere, etc.). M6.3 A model able to account for the oxidation behaviour of a Fe-Cr-Mn complex alloy completed and verified with experimental oxidation data. The milestone M6.2 is considered fulfilled. With respect to the milestone M6.3, the planned activities on oxidation modeling were stopped with the unexpected leave of an employee allocated to this task. Considering the staff difficulties and time limitation, it was decided to reallocate the project resources set aside for this task to other tasks in this project. The milestone M6.3 was therefore not fulfilled within this project. Page 48 of 65

49 2. Cell and stack element testing Risø DTU has previously shown that state of the art SOFC technology works well also for electrolysis as long as current densities are modest (less than 0.5 A/cm 2 at stack level, T ~800 o C. At single cell level and higher temperature up to close to 1 A/cm 2 is achievable with modest degradation). The purpose of Task 2 in the present project was to investigate performance and durability of solid oxide cells under industrially relevant ( harsh ) electrolysis conditions (current density 1 A/cm 2 ) and to identify degradation mechanisms Overview of the tests In the current project, 9 cell tests and 3 stack element tests were carried out at Risø DTU, with a total test period of exceeding 5000 hours. The tests were carried out on three types of solid oxide cells, 2.0G, 2.1G and 2.5G. The 2.0G and 2.1G cells have a 300 µm thick porous Ni/YSZ support layer, a µm thick Ni/YSZ cermet fuel electrode, a µm thick YSZ electrolyte, and a µm thick LSM/YSZ composite oxygen electrode. The oxygen electrode in 2.0G cells is prepared by Production method A whereas in 2.1G cells by Production method B. In the 2.5G cells, the LSM/YSZ electrode is replaced by a LSCF/CGO electrode with a CGO barrier layer at the interface between the electrolyte and the oxygen electrode. A proprietary glass (AS) was used as seal for all the cell tests and most of the stack element tests. The detailed testing conditions are summarized in Table 1. The testing program covered both co-electrolysis and steam electrolysis. Table 1 Testing conditions for the cell and stack element tests Test number Test type Cell type Sealing Inlet gas cleaning Testing condition Temp., Fuel gas oc 11test48 Cell 2.1G AS 1 Yes 800 CO2/H2O/H2 (45/45/10) 14test109 Cell 2.1G AS Yes 800 CO2/H2O/H2 (45/45/10) 14test113 Cell 2.1G AS Yes 850 CO2/H2O/H2 (45/45/10) 14test111 Cell 2.1G AS Yes 850 CO2/H2O/H2 (45/45/10) 14test110 2 Cell 2.1G AS Yes 11test49 Cell 2.5G AS Yes 800 CO2/H2O/H2 (45/45/10) 14test112 2 Cell 2.5G AS Yes 15test8 Cell 2.5G AS Yes 850 H2/H2O (50/50) 15test9 Cell 2.1G AS Yes 800 H2/H2O (50/50) Oxidant gas Current density/reactant utilization/electrolysis period, A/cm 2 /%/hour O2-1/62%/900 O2-1.5/62%/300 O2-1/62%/660 O2-1.5/62%/250 O2-1.5/62%/400 O2-1.25/67%/500 O2-1.5/80%/250 15test4 Stack 2.0G Ag+AS 3 Yes 850 H2/H2O O2-0.5/27%/900 element (50/50) 15test6 Stack 2.1G AS Yes 850 H2/H2O O2-1.5/80%/100 element (50/50) 11test51 Stack 2.1G AS Yes 850 H2/H2O O2-1.5/80%/890 element (50/50) 1 Proprietary glass. 2 No long-term electrolysis testing. 3 Ag-based braze as seal for the fuel side and glass as seal for the oxygen side. Page 49 of 65

50 2.2. Initial cell performance Figure 4 presents the measured polarization (iv) curves at three temperatures, with H 2O/H 2 (50/50) or CO 2/H 2O/H 2 (45/45/10) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSM/YSZ (2.1G, 14test111) or the LSCF/CGO (2.5G, 14test112) oxygen electrode. The derived ASR (area specific resistance) values are plotted in Figure 5. The 2.5G cells exhibit lower ASR than the 2.1G cells at 800 and 750 o C, for both co-electrolysis and steam electrolysis Voltage, mv % H 2 O-50% H 2 2.1G 850 o C 800 o C 750 o C 2.5G 850 o C 800 o C 750 o C Current density, A/cm Voltage, mv % CO 2-45% H 2 O-10% H 2 2.1G 850 o C 800 o C 750 o C 2.5G 850 o C 800 o C 750 o C Current density, A/cm 2 Figure 4 Performance of the 2.1G and 2.5G cells during steam electrolysis or co-electrolysis of steam and carbon dioxide. ASR, Ω cm G 50% H 2 O-50% H 2 in SOFC mode 50% H 2 O-50% H 2 in SOEC mode 45% CO 2-45% H 2 O-10% H 2 in SOEC mode 2.5G 50% H 2 O-50% H 2 in SOFC mode 50% H 2 O-50% H 2 in SOEC mode 45% CO 2-45% H 2 O-10% H 2 in SOEC mode Temperature, o C Figure 5 Representative ASR (area specific resistance) of the 2.1G and 2.5G cells. ASR was calculated from the measured iv curves with fuel utilization corrected, i.e. ASREL = (Cell voltage at A/cm 2 OCV)/Current density (0.25 A/cm 2 ). Page 50 of 65

51 2.3. Durability under co-electrolysis (cell tests) In the present project, 5 cells tests were carried out to investigate the durability of solid oxide cells under co-electrolysis, with four tests on 2.1G and one test on 2.5G. The cell voltage degradation with time is presented in Figure 6. At -1 A/cm 2, the 2.1G cells exhibited a degradation rate of mv/1000hours for the entire durability test period and a degradation rate of 200 mv/1000hours (at 800 o C) for the final/steady stage. This is comparable with the degradation of 2.0G cells reported in the project ForskEL : 200 mv/1000hours for the final/steady stage at 850 o C and -1 A/cm 2. Increasing the current density to -1.5 A/cm 2 for the 2.1G cells lead to increased degradation with a degradation rate of exceeding 800 mv/1000hours. The 2.5G cells showed lower degradation rate, only 400 mv/1000hours at 800 o C and -1.5 A/cm Cell voltage, mv Time, hour 2.1G, 800 o C, -1 A/cm 2 2.1G, 800 o C, -1.5 A/cm 2 2.1G, 850 o C, -1 A/cm 2 2.1G, 850 o C, -1.5 A/cm 2 2.5G, 800 o C, -1.5 A/cm 2 Degradation rate, mv/1000h G 800 o C -1 A/cm 2 Degradation rate for the entire durability test period Final/steady degradation rate 2.1G 800 o C -1.5 A/cm 2 2.1G 850 o C -1 A/cm 2 2.1G 850 o C -1.5 A/cm 2 2.5G, 800 o C -1.5 A/cm 2 Test type Figure 6 Cell voltage degradation (left) and degradation rate (right) for tests operated with CO2/H2O/H2 (45/45/10) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSM/YSZ (2.1G) or the LSCF/CGO (2.5G) oxygen electrode. Figure 7 presents the ohmic (R s) and electrode polarization (R p) resistance degradation with time during the co-electrolysis test period and Figure 8 plots the degradation rates. For 2.1G cells, the R s degradation is predominantly influenced by current density and to a less extent influenced by temperature. The R s degradation rate was increased significantly by increasing current density from -1 A/cm 2 to -1.5 A/cm 2 and decreased slightly by increasing temperature from 800 o C to 850 o C. The electrode polarization resistance behaved differently. The R p degradation rate decreased with either increasing current density or decreasing temperature. At 800 o C and -1.5 A/cm 2, the 2.5G cells showed less R s degradation than the 2.1G cells and showed R p activation. Page 51 of 65

52 R s, Ω cm G, 800 o C, -1 A/cm 2 2.1G, 800 o C, -1.5 A/cm G, 850 o C, -1 A/cm 2 2.1G, 850 o C, -1.5 A/cm 2 2.5G, 800 o C, -1.5 A/cm R p, Ω cm G, 800 o C, -1 A/cm 2 2.1G, 800 o C, -1.5 A/cm G, 850 o C, -1 A/cm 2 2.1G, 850 o C, -1.5 A/cm 2 2.5G, 800 o C, -1.5 A/cm Time, hour Time, hour Figure 7 Ohmic (left) and electrode polarization (right) resistance degradation with time. Degradation rate, Ω cm 2 /1000h R s 2.1G 800 o C 2.1G 800 o C 2.1G 850 C 2.1G 850 o C 2.5G, A/cm A/cm 2-1 A/cm A/cm A/cm 2 o C Degradation rate, Ω cm 2 /1000h R p 2.1G 800 o C 2.1G 800 o C 2.1G 850 C 2.1G 850 o C 2.5G, A/cm A/cm 2-1 A/cm A/cm A/cm 2 o C Figure 8 Degradation rate for the ohmic and electrode polarization resistance during the entire durability test period. The electrode degradation mechanisms were investigated by detailed impedance & post-mortem analyses. Figure 9 presents DRT (distribution of relaxation times) plots, derived from measured impedance spectra during the electrolysis test period. For 2.1G cells at 800 o C and -1 A/cm 2, the R p degradation was associated with two processes occurring in the LSM/YSZ electrode: the high-frequency process (charge transfer) and the medium-frequency process (oxygen dissociation/recombination). At 800 o C and -1.5 A/cm 2, the R p degradation is associated with the high-frequency process (charge transfer) in the LSM/YSZ electrode and the medium frequency process in the Ni/YSZ fuel electrode. This is further supported by post-mortem results. Figure 10 presents percolation images for a reference cell and a tested cell. The white particles represent percolating Ni. Loss of Ni percolation and growth of Ni particles in the active Ni/YSZ electrode and support are easily seen in the tested cell. Besides, cracks in the YSZ electrolyte layer (close to the LSM side, not shown in this report) and the interface of Ni/YSZ particles becoming irregular with porosity and/or secondary phases are also clearly visible (Figure 11). EDS analyses on the Ni/YSZ cathode revealed diffusion of Y/Zr into Ni particles in the 2.1G cell tested at 800 o C and -1.5 A/cm 2. The observed phenomena may be caused by either chemical or structural changes (or both) in a sub-micro/nano-meter scale, which requires more advanced tools such as TEM to clarify. Page 52 of 65

53 Figure 9 DRT plots of the cell tests with CO2/H2O/H2 (45/45/10) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSM/YSZ oxygen electrode, left operated at -1 A/cm 2 (11test48), right operated at -1.5 A/cm 2 (14test109). Ni/YSZ YSZ Figure 10 InLens images of the cell cross-section at the fuel inlet, left 14test110 (reference), right 14test109 (operated at -1.5 A/cm 2 ). YSZ Ni-YSZ Figure 11 SEM back-scatter image of the cell cross-section at the fuel inlet. The cell was tested at 800 o C and -1.5 A/cm 2. Page 53 of 65

54 2.4. Durability under steam electrolysis Cell tests In the present project, two cell tests (15test8 and 15test9) were carried out with purposes of both validating a newly built cell testing rig and investigating cell durability under steam electrolysis. Both tests showed poor initial performance, which was most likely caused by non-optimal contacting. The test 15test8 was carried out on a 2.5G cell at 850 o C, A/cm 2, 67% reactant utilization and the cell showed a degradation rate of 241 mv/1000h for the entire durability test period. The test 15test9 was carried out on a 2.1G cell and at 800 o C, -1.5 A/cm 2, 80% reactant utilization and the cell showed a cell degradation rate of 633 mv/1000h for the entire durability test period test8 Cell voltage Cell temperature test9 Cell voltage Cell temperature Cell voltage, mv mV/1000h V= 276mV/1000h 400 mv/1000h for h 276 mv/1000h for h 241 mv/1000h for the entire period Time, hour Temperature, o C Cell voltage, mv V= 633 mv/1000h for the entire period Time, hour Figure 12 Cell voltage and temperature for tests operated with H2O/H2 (50/50) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSCF/CGO (left) or the LSM/YSZ (right) oxygen electrode, left - 15test8, 2.5G, 850 o C, A/cm 2, 67% reactant utilization; right 15test9, 2.1G, 800 o C, -1.5 A/cm 2, 80% reactant utilization Temperature, o C Stack element tests In the present project, three stack elements were carried out, all tested for steam electrolysis durability. The stack element design developed in the project ForskEL was employed, in which etched Crofer22APU plates were used as interconnects. The interconnect at the oxygen electrode (LSM/YSZ) side was coated with a protective coating consisting of Co 3O 4 and a perovskite on both sides. The interconnect at the fuel electrode (Ni/YSZ) side was coated with NiO on the non-etched side (in contact with the current collector plate). For testing conditions see Table Test with Ag braze as seal In the test 15test4, Ag-based braze was used as seal for the fuel side. By replacing glass with Ag-based braze, the problem related to Si emission from the glass on the fuel side should be eliminated, which would otherwise poison the Ni/YSZ electrode. For the oxygen side, glass was still used, as it has no detrimental effect there (due to low P(H 2O)). The test was carried out at 850 o C, -0.5 A/cm 2, 27% reactant utilization, with H 2O/H 2 (50/50) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSM/YSZ oxygen electrode. Figure 13 plots the stack element voltage degradation with time in comparison with 11test29. The stack element test 11test29 was carried out (in the project ForskEL ) under same conditions, except that glass seal was used for both fuel and oxygen sides. Page 54 of 65

55 Voltage, mv test29 - Glass seal 15test4 - Ag braze seal Time, hour Figure 13 Voltage of the stack element for the tests operated with H2O/H2 (50/50) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSM/YSZ oxygen electrode at 850 o C, -0.5 A/cm 2, 27% reactant utilization. 11test29: glass as seal; 15test4: Ag braze as seal for the fuel side and glass as seal for the oxygen side. 11test29 was carried out in the project ForskEL The initial performance of 15test4 was a bit inferior to that of 11test29, which was most likely caused by non-optimal contacting. 15test4 degraded 230 mv for the first 250 hours and activated for the remaining test period, with the performance getting closer to that of 11test29. Post-mortem analyses showed that after 900 hours testing at 850 o C, Ag-based braze still adhered very well to both the interconnect and the cell. The braze did not show any mechanical deformation. These results proved feasibility of applying Agbase braze as seal for SOEC stacks. Further development work shall focus on optimizing production and sealing procedure. Metal spacer Ag braze Interconnect Figure 14 SEM image on the cross-section of the interconnect-ag braze interface at the fuel inlet for the test 15test4. Page 55 of 65

56 Tests with glass as seal Two stack element tests with AS glass on both the fuel side and the oxygen side were carried out. 15test6 showed poor performance and the results are therefore not presented here. The stack element test 11test51 was carried out on a 2.1G cell at 850 o C, -1.5 A/cm 2, 80% reactant utilization with H 2O/H 2 (50/50) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSM/YSZ oxygen electrode. The cell showed a degradation rate of 463 mv/1000h for the entire durability test period and a degradation rate of only 313 mv/1000h for the final/steady stage. In the project ForskEL , a degradation rate of mv/1000h for the final/steady stage was observed on 2.1G cells tested at 850 o C, A/cm 2, 67% reactant utilization. This clearly shows accelerated degradation with increasing current density. Post-mortem analyses on 11test51 detected the following microstructural changes: Loss of percolation in the Ni/YSZ and LSM/YSZ electrodes. Ni coarsening and re-distribution. Cracks in the YSZ electrolyte layer (close to the LSM side) and detachment of the LSM/YSZ electrode. The fuel-side IC (uncoated) exhibited more severe corrosion than the oxygen-side IC (coated) (Figure 16). Voltage, mv V(cell voltage) = Cell Stack element Cathode (Ni-YSZ) IC Anode (LSM-YSZ) IC 463 mv/1000h (entire test period) 313 mv/1000h (final/steady) V(cathode IC) = -0.4 mv/1000h (entire test period) V(anode IC) = 12.3 mv/1000h (entire test period) Time, hour Figure 15 Cell and stack element voltages for the stack element test 11test51 operated with H2O/H2 (50/50) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the LSM/YSZ oxygen electrode at 850 o C, -1.5 A/cm 2, 80% reactant utilization. Figure 16 SEM back-scatter images on the cross-sections of the interconnects at the center of the stack element (at the bottom of the gas channel), left oxygen side IC, right fuel side IC. Page 56 of 65

57 Based on the cell and stack element tests carried out in the present project, we conclude the following: Under co-electrolysis, the 2.1G cells showed a degradation rate of mv/1000h at -1 A/cm 2 and >800 mv/1000h at -1.5 A/cm 2. The 2.5G cells showed better durability with a degradation rate of 400 mv/1000h at -1.5 A/cm 2. Under steam-electrolysis, the 2.1G cells showed a degradation rate of 463 mv/1000h at 850 o C, A/cm 2 (determined via stack element testing). The cell degradation under high electrolysis current is related to both the Ni/YSZ and LSM/YSZ electrodes. Loss of percolation in the electrodes, Ni coarsening and re-distribution, diffusion of Y/Zr into Ni particles, formation of cracks in the electrolyte layer, and detachment of the LSM/YSZ electrode were identified as possible degradation mechanisms. To allow stable electrolysis operation at -1.5 A/cm 2, both better oxygen and fuel electrodes are required. In the present project, the following milestones were set up in the project plan: M6.1 Electrolysis cell tests on state-of-the-art SOFC cells at current densities of 1-2 A/cm 2 with various CO 2/H 2O ratios carried out and combined with detailed post mortem & electrochemical characterizations. Major degradation mechanisms at the cell level under various harsh electrolysis conditions reported. M6.4 A stack element test sealed with Ag-based composite braze and carried out over a period of more than 500 hours and post test characterizations reported. M6.5 Stack element tests with coated interconnects at a current density of 1.5 A/cm 2 carried out and combined with detailed post mortem & electrochemical characterizations. Major degradation mechanisms at the stack level under harsh electrolysis conditions reported. All the three milestones are considered fulfilled. Page 57 of 65

58 3. Stack testing The purpose of this task was to investigate performance of standard TOFC (Topsoe Fuel Cell A/S) SOC stacks (based on state-of-the-art solid oxide cells) under mild electrolysis operating conditions ( 0.75 A/cm 2 ). Risø DTU has previously demonstrated an electrolysis stack test at ~850 o C and -0.5 A/cm 2 with no degradation. As a step forward, electrolysis stack tests at A/cm 2 were carried out in the present project. During the project period, two identical stacks (N-063 and N-126) were produced at TOFC and tested at Risø DTU. The stacks contained three types of solid oxide cells with a footprint of 12X12 cm 2 : 2.0G (Cell 2/5/8), 2.1G (Cell 1/4/7/10) and 2.5G (Cell 3/6/9). Cell 1 is the cell at the bottom of the stack, which in comparison with the other cells in the stack, normally shows inferior performance due to non-optimal contacting and is therefore not taken into consideration N-063 Before the electrolysis testing, the stack N-063 was first characterized under various conditions. Figure 17 presents the cell voltages during iv measurements in both SOFC and SOEC modes. Higher cell voltage corresponds to lower ASR (area specific resistance) in SOFC mode but higher ASR in SOEC mode. The cells in N-063 follow the order of 2.5G > 2.1G 2.0G with respect to the initial performance at o C in both SOFC and SOEC modes. After initial characterizations, the stack was tested at 800 o C, A/cm 2, with 434 L/h H 2O/H 2 (90/10) supplied to the Ni/YSZ fuel electrode and 60 L/h oxygen to the oxygen electrode. The test was carried out smoothly for 170 hours and hereafter terminated due to a failure in steam supply, which unfortunately destroyed the stack. Therefore no post-mortem work was performed on the stack N-063. Figure 18 presents the average cell voltages at the start and the end of the steam electrolysis period. The 2.5G cells showed the best performance (lowest voltage) both at the beginning and at the end of the electrolysis period. The test period is however too short to enable a sound evaluation of the cell durability under steam electrolysis at the stack level SOFC: 760 o C, 20A, H 2 +N 2 /Air 1.15 SOEC: 800 o C, -20A, H 2 +H 2 O/O G 2.5G 2.1G G Voltage, V Voltage, V G 2.5G 1.05 Cell 03 Cell 06 Cell 09 Average Cell 01 Cell 04 Cell 07 Cell 10 Average 0.78 Cell 02 Cell 05 Cell 08 Average Cell 03 Cell 06 Cell 09 Average Cell 01 Cell 04 Cell 07 Cell 10 Average Cell 02 Cell 05 Cell 08 Average Figure 17 Initial cell performances for the stack N-063. Left: measured cell voltages in SOFC mode at 760 o C, 20A, with 150 L/h H2 and 100 L/h N2 supplied to the Ni/YSZ fuel electrode and 960 L/h air to the oxygen electrode; Right: measured cell voltages in SOEC mode at 800 o C, -20 A, with 360 L/h H2O/H2 (50/50) supplied to the Ni/YSZ fuel electrode and L/h oxygen to the oxygen electrode. Page 58 of 65

59 1.4 Start End 1.3 Voltage, V G 2.5G 2.1G Cell type Figure 18 Average cell voltages at the start and the end of the electrolysis period. The stack N-063 was operated at 800 o C, A/cm 2 for 170 hours with H2O/H2 (90/10) supplied to the Ni/YSZ fuel electrode and oxygen supplied to the oxygen electrode N-126 Figure 19 presents initial cell performance for the second stack N-126. The cells follow the order of 2.5G > 2.1G > 2.0G with respect to the initial performance at o C in both SOFC and SOEC modes SOFC: 760 o C, 20A, H 2 +N 2 /Air 1.13 SOEC: 800 o C, -20A, H 2 +H 2 O/O G 2.1G G Voltage, V G Voltage, V G 2.1G Cell 02 Cell 05 Cell 08 Average Cell 03 Cell 06 Cell 09 Average Cell 01 Cell 04 Cell 07 Cell 10 Average 1.03 Cell 02 Cell 05 Cell 08 Average Cell 03 Cell 06 Cell 09 Average Cell 01 Cell 04 Cell 07 Cell 10 Average Figure 19 Initial cell performance for the stack N-126. Left: measured cell voltages in SOFC mode at 760 o C, 20A, with 150 L/h H2 and 100 L/h N2 supplied to the Ni/YSZ fuel electrode and 960 L/h air to the oxygen electrode; Right: measured cell voltages in SOEC mode at 800 o C, -20 A, with 540 L/h CO2/H2O/H2 (45/45/10) supplied to the Ni/YSZ fuel electrode and L/h oxygen to the oxygen electrode. After initial characterizations, the stack was tested at 800 o C, -0.5 A/cm 2 for 70 hours and afterwards A/cm 2 for 1000 hours with 540 L/h CO 2/H 2O/H 2 (45/45/10) supplied to the Ni/YSZ fuel electrode and 60 L/h oxygen to the oxygen electrode. The cell voltage evolution with time is presented in Figure 20. Figure 21 plots the cell voltages at the start and the end of the electrolysis period at A/cm 2 and the degradation rates. The 2.5G cells showed the best performance at the beginning of the durability test period at A/cm 2 and hereafter degraded with an average degradation rate of 96 mv/1000h. The 2.0G cells behaved in an opposite way, being the worst in the beginning and afterwards activating during the entire durability test, with an average activation rate of -73 mv/1000h. At the end of the durability test at A/cm 2, the 2.0G cells have the best performance, followed by 2.5G and hereafter 2.1G. During the durability test, the 2.1G cells behaved very strangely, which is most likely due to unstable contacting between cells and interconnects. The cells showed accelerated degradation with jumps. The average degradation rate for 2.1G is 152 mv/1000h. Page 59 of 65

60 -0.5 A/cm A/cm 2 Voltage, V Voltage, V G Cell 02 Cell 05 Cell G Cell 03 Cell 06 Cell 09 Voltage, V Time, hour 2.1G Cell 01 Cell 04 Cell 07 Cell 10 Figure 20 Cell voltage degradation with time for the stack N-063 operated at 800 o C, -0.5 A/cm 2 for 70 hours and A/cm 2 for 1000 hours with 540 L/h CO2/H2O/H2 (45/45/10) supplied to the Ni/YSZ fuel electrode and 60 L/h oxygen to the oxygen electrode. Voltage, V Start End 2.0G 2.5G 2.1G Cell type Degredation rate, mv/1000h Cell 02 Cell 05 Cell 08 Average Cell 03 Cell 06 Cell 09 Average Cell 04 Cell 07 Cell 10 Average 2.0G 2.5G 2.1G Cell type Figure 21 Left: average cell voltages at the start and the end of the electrolysis period at 800 o C, A/cm 2 for 1000 hours with 540 L/h CO2/H2O/H2 (45/45/10) supplied to the Ni/YSZ fuel electrode and 60 L/h oxygen to the oxygen electrode. Right: cell voltage degradation rate. Page 60 of 65

61 Post-mortem analyses on the tested stack N-126 revealed the following: After electrolysis testing at 800 o C for 1070 hours (70 hours at -0.5 A/cm 2 and 1000 hours at A/cm 2 ), the cells exhibited only minor microstructural changes. For all the cells, loss of Ni percolation was observed in the active Ni/YSZ electrode, within 5 µm from the Ni/YSZ YSZ interface. Further away from the interface, the Ni percolation seems to be intact. Besides, few cracks were found in the YSZ electrolyte layer, which could be caused due to electrolysis testing or SEM sample preparation. The 2.1G cells showed a large degree of inhomgeneity in the LSM/YSZ oxygen electrode, as compared to the one in 2.0G. Besides, there are also structural differences between 2.5G and 2.0G. Whether these could account for the inferior durability of 2.1G and 2.5G as compared to 2.0G is yet too early to be concluded. After electrolysis testing at 800 o C for 1070 hours (70 hours at -0.5 A/cm 2 and 1000 hours at A/cm 2 ), the IC has an oxide scale thickness of 1 µm for both the fuel-side and the oxygen-side. Based on the stack testing results, the following can be concluded here: With respect to the initial electrolysis cell performance at the stack level, the three types of solid oxide cells follow the order of 2.5G>2.1G>2.0G. This is consistent with the results obtained via single cell testing. The 2.0G cells showed no degradation, but actually activation during co-electrolysis at 800 o C and A/cm 2 for 1000 hours. Post-mortem analyses revealed only minor microstructural changes in the tested cells. The 2.1G and 2.5G cells showed inferior durability as compared to 2.0G. Whether this is a true picture of cell durability of different cell types is yet too early to be concluded. To clarify this, different types of cells with equally good microstructure and identical active layers shall be employed in future stack testing. The origin of the cell-interconnect interface instability shall also be explored in the following project ForskEL Development of SOEC cells and stacks. In the present project, the following milestones were set up in the project plan: M7.1 A standard TOFC 10-cell SOFC stack (based on state-of-the-art SOFC cells with a footprint of 12X12 cm 2 ) delivered to Risø DTU. M7.2 Another standard TOFC 10-cell SOFC stack (based on state-of-the-art SOFC cells with a footprint of 12X12 cm 2 ) delivered to Risø DTU. M7.3 Electrolysis stack tests at current densities equal to or exceeding 0.75 A/cm 2 carried out over a period of more than 1000 hours and combined with detailed post mortem & electrochemical characterizations. All the three milestones are considered fulfilled. Page 61 of 65

62 4. Stack modeling The purpose of this task was to update TOFC s SOFC/SOEC stack model with the most recent cell and stack test results obtained within this project. The focus was to update the cell ASR (area specific resistance) expression. This task was carried out by TOFC. The ASR expression is a mathematical expression of the type: ASR model = Function (T, i d, P a, P b, P c) where T is temperature, i d is current density, P a, P b, P c are partial pressure of all the gasses which are relevant for the reaction rate. In the present project, the existing ASR expressions of 2G and 2.5G were updated for both SOFC mode and SOEC mode (steam electrolysis only). The work was carried out by first evaluating if new variables are needed for the existing ASR expressions. This was done by plotting ASR old/asr fu against various variables, where ASR old is the ASR value calculated using the existing (old) ASR expression in TOFC s stack model and ASR fu is the measured value. If a systematic trend is observed, it means there is an effect which the existing ASR expression did not include. For example, Figure 22 plots ASR old/asr fu for 2G cells against current density, where ASR old was based on 2.0G and ASR fu on 2.1G. It is clearly shown that whenever the current density is too high, the existing ASR expression gives a too high ASR value. The ASR dependence on current density needs to be adjusted for 2.1G. Similar analyses on other variables showed that it is not necessary to adjust dependence of partial pressure of the individual gasses. ASRold/ASRfu Current density A/m2 Figure 22 ASRold/ASRfu for 2G cells against current density. The next step was to fit the existing ASR expression with the data obtained in the present project. This was done by optimizing various parameters in the ASR expression to minimize the sum of the square error, ((ASR fu ASR model)/ ASR model) 2, where ASR model is the value calculated using the updated/improved ASR expression and ASR fu is the measured value. As shown in Table 2, the sum of the square error was clearly reduced after the optimization. A plot of ASR model/asr fu against current density was then made and no overall trend was found. This proves that the new ASR expression is able to capture the dependence of measured ASR on current density. Similar analysis proves that the new ASR expression is also able to capture the dependence of measured ASR on temperature. Page 62 of 65

63 Table 2 Sum of square error ((ASRfu ASRmodel)/ASRmodel) 2 by the old model and the new model developed in the present project. Old (SOFC) New (SOFC) Old (SOEC) New (SOEC) Sum of square error for 2G Sum of square error for 2.5G In order to check if the new ASR expression is able to capture the overall variation of the measured ASR values with all variables, a plot of ASR model against ASR fu was made (Figure 23). If the new expression is able to capture all systematic variations, this plot should be just a straight line through the point (0, 0) with a slope of 1 and this is exactly shown in Figure ASRfit ASRfu Figure 23 ASRmodel against ASRfu for 2G cells. The new ASR expression for 2.1G and 2.5G cells was then implemented into TOFC s SOFC/SOEC stack model. The model calculated results were compared with the measured results from Stack N-126, showing good agreement at the beginning of the electrolysis period. With respect to further improvement work on SOEC stack modeling, the following shall be considered: (1) Including effects of CO, CO 2, or CH 4 in the ASR expression; (2) Minimizing temperature change during the measurements dedicated for model validation. In the present project, the following milestone was set up in the project plan: M7.4 Validation of SOEC stack model versus experiments carried out and capability of the model to reflect measured cell and stack performance demonstrated. The milestone M7.4 is considered fulfilled. Page 63 of 65

64 5. Appendix A. List of project milestones (Work packages 6-7) M6.1 Electrolysis cell tests on state-of-the-art SOFC cells at current densities of 1-2 A/cm 2 with various CO 2/H 2O ratios carried out and combined with detailed post mortem & electrochemical characterizations. Major degradation mechanisms at the cell level under various harsh electrolysis conditions reported. M6.2 Corrosion resistance of coated superior interconnect alloys (identified in the project ForskEL or obtained from a third party) investigated in an SOEC stack-relevant environment (high steam content (>50%), CO 2-containing atmosphere, etc.). M6.3 A model able to account for the oxidation behaviour of a Fe-Cr-Mn complex alloy completed and verified with experimental oxidation data. M6.4 A stack element test sealed with Ag-based composite braze and carried out over a period of more than 500 hours and post test characterizations reported. M6.5 Stack element tests with coated interconnects at a current density of 1.5 A/cm 2 carried out and combined with detailed post mortem & electrochemical characterizations. Major degradation mechanisms at the stack level under harsh electrolysis conditions reported. M7.1 A standard TOFC 10-cell SOFC stack (based on state-of-the-art SOFC cells with a footprint of 12X12 cm 2 ) delivered to Risø DTU. M7.2 Another standard TOFC 10-cell SOFC stack (based on state-of-the-art SOFC cells with a footprint of 12X12 cm 2 ) delivered to Risø DTU. M7.3 Electrolysis stack tests at current densities equal to or exceeding 0.75 A/cm 2 carried out over a period of more than 1000 hours and combined with detailed post mortem & electrochemical characterizations. M7.4 Validation of SOEC stack model versus experiments carried out and capability of the model to reflect measured cell and stack performance demonstrated. Page 64 of 65

65 B. List of internal reports. Risø DTU: BC-1312 BC-1317 Oxidation tests of ICs and IC alloys at 850 C - Part II - Post mortem (Janet J. Bentzen) BC-1318 Impedance analysis of the cell tests in plansoec project (Xiufu Sun, Per Hjalmarsson, Ming Chen). BC-1319 N-063 Vand elektrolyse test med forskellige celler i Fuelcon (Jens Høgh). BC-1320 N-126 Co elektrolyse test med forskellige celler i gammel teststand (Jens Høgh). BC-1321 Oxidation tests of ICs and IC alloys at 850 C in reducing atmospheres with high steam content (Ming Chen). BC-1322 Oxidation tests of coated ICs and IC alloys at 850 C in oxidizing atmospheres (Ming Chen). BC-1323 Post-mortem analyses on the stack N-126 tested at 800C for co-electrolysis (Ming Chen, Janet J. Bentzen, Jens Høgh). TOFC: 1. Fitting of SOEC cell test data 14test110 (2.1G cell) to the SOFC/SOEC stackmodel at TOFC (Jens Ulrik Nielsen). 2. Fitting of SOEC cell test data (11test49 2.5G cell) to the SOFC/SOEC stackmodel at TOFC (Jens Ulrik Nielsen). 3. Modelling of results from stacktesting of N-126 (Jens Ulrik Nielsen). Page 65 of 65