Skip to main content
European Commission logo print header

ENhanced DURability materials for Advanced stacks of New solid oxide fuel CElls

Final Report Summary - ENDURANCE (ENhanced DURability materials for Advanced stacks of New solid oxide fuel CElls)

Executive Summary:
Solid oxide fuel cells represent a well consolidated device making use of hydrogen or hydrogen reach fuels to produce power and heat with very high efficiency with extremely low or zero emissions of greenhouse gases. In case of usage of biogas a neutral carbon footprint beside the high efficiency of combined heat and power (chp) system shall be considered an excellent and win win compromise. SOFCs are operating in stacks increasing the achievable power and allowing a handy management of gases, heat and power in a more complex plant where compressors, controllers and other needed components are introduced to get a complete operative system. The state of the art materials involved in an SOFC stack are quite consolidated: anode supported cells with Ni base cermet as anode, Yttria Stabilized Zirconia (YSZ) as electrolyte, Gadolinia Doped Ceria (GDC) as diffusion barrier layer, lanthanum strontium cobaltite ferrite (LSCF) as cathode, ferritic stainless steel (FSS) as structural materials and as interconnects (in this case coated with a Cr volatilization hindering product), and inert highly resistive sealants, most probably as ceramic-glass. Industrial players in the field of stack manufacturing and commercialization have proprietary solutions to solve the most known issues as Cr poisoning of the cathode (e.g. usage of a Cr getter at the interface cathode/interconnects in addition to the coating), gas leakage (e.g. layered sealants and built in delta probes), reduced ohmic resistance (e.g. usage of electrode/interconnect contacting layers). The ENDURANCE project was prepared in this context and frame with the purpose to face the still unresolved issues through innovations and improvements at all levels in order to reach and possibly outmatch the durability and reliability target demanded by the market (e.g. 50k hours of operation with a maximum of 10% of total degradation). A fully European consortium composed by 8 research institutes and 4 industrial partners worked during three years to contribute to reach to previously described goals. Starting from state of art commercial stacks operated for several thousands hours fundamental information about the most sensitive degradation modes and mechanisms were addressed. With the support of a wide literature research the experimental data were integrated in an evolved version of FMEA becoming one of the public use deliverables called Degradation Modes and Effects analysis where all degradation processes are taken into account risk ranked according to their probability (statistical frequency of the event) and harmfulness. Starting from this base thermo-mechanical, electrochemical and physical models were developed or, when already existing, refined in order to have an improved level of events and performances predictability. The main achievements in this field allowed the consortium to have and statistically prove on specific samples and excellent matching between modeled and real phenomena. On the front of materials improvements the activities focused on strategies meant to increase the overall durability by addressing the most critical issues recognized by DMEA: sealant, anode nickel coarsening, diffusion barrier layer between cathode and electrolyte. Two out of ten new glass-ceramic sealants were developed validated. One of them was tested in short stacks proving an excellent resistance to thermal cycles over the long period. A strategic solution using YSZ as interlayer between FSS and sealant was successfully verified to increase the adhesion and reduce the metal ions diffusion from the metal substrate into the sealant. Several manufacturing solutions were brilliantly applied to densify the GDC in order to improve the performances of the diffusion barrier layer (i.e. GDC impregnation, MgO as sintering aid, applcation via Pulsed Laser Deposition). The cells resulting from this step were mounted in short and real stacks, the latter operated in reformed fuel to match real operating conditions. A strategy of partial oxidation of the anode followed by its complete reduction has proven to be a promising counter action to reduce the risk of Ni coarsening, identified as one of the main source of anode performances degradation.
Moreover, a brand new method of interpretation of the electrochemical data applicable to both cells and complete stacks was introduced: the Differential Resistance Analysis (DRA). Such tool applied on real data and on model generated data demonstrated the possibility to detect nearly invisible deviations from the original electrochemical performances technically impossible to be observed on short tests by the usual tools. This allows to verify in very precise and sensitive way the evolution of the cell performances even though close or below the target of 0.1% of degradation rate per 1000 hours.

Project Context and Objectives:
Project context:
In the actual social and economical situation developed in a constantly growing public green and planet friendly policy the usage of clean and renewable power sources is a naturally leading trend. The continuous growth of the demand is far below the power potentially available from renewable and clean sources however not the theoretical but the real available energy on demand is the topic to face and this means dealing with safe and long lasting energy conversion, storage, distribution and, in one word, management. In this context Fuel cells represent a realistic and very promising solution. In the wide family of fuel cells those using solid oxide as electrolyte represent the most efficient in terms of energy conversion reaching more than 95% when used in combined heat and power mode. The Solid Oxide Fuel Cells (SOFC) have the advantage to be fuel flexible which means able to operate with pure hydrogen and with hydrogen issued from a reforming process of natural, bio- or syn-gas. In this way a combination of renewable power sources in a smart grid is providing a safe energy storage utility using water electrolysis or CO2 sequestration in the Power-to-Gas (P2G) chain considering the fuel as the most efficient, economic and durable way to store energy. The reverse of the chain Gas-to-Power makes energy available on demand wherever needed with neutral or negative carbon footprint when the carbon dioxide generated is reintroduced into the P2G process.
The advances in the last 20 years in the field of SOFC leaded to performances perfectly suitable for nearly all applications leading to well established technical solutions in terms of materials and of plants. In spite of this, the public acceptance of fuel cells is still limited by the low general knowledge about it and by a lack of awareness of their potential. The most stringent issue in the commercialization of such power sources is however more cost-related being classical and pollutant solutions considered cheaper. Reliability, durability and manufacturing cost effectiveness are thus among the most important aspects to play with to make SOFC based power plant competitive and attractive for the market.
Cost efficiency means to keep the manufacturing process at a reasonable cost, to invest on common and not expensive raw materials and to grant a long lasting life with a low risk of failure before the natural end of operativeness. The life expectation usually indicated for an SOFC stack is in the order of 50'000 hours with the perspective to move beyond this threshold. Taking into account that 10% of efficiency loss all along the operation time of the stack a degradation rate of 0.2%/1000h is considered acceptable. This means to have components and technical solutions able to meet such requests with a very low failure risk. To predict events is a risky but necessary in order to reach the goal. As long as degradation and eventually failure are related to technical solutions it should be possible to validate them over the long period (at least 3 years, if not 5) and to be prepared to find solutions in case of need.

Project objectives
The ENDURANCE project targeted some of these issues trying to face and solve them by gathering together renown scientists and industrial manufacturers with the purpose to combine knowledge on real stacks manufacturing and usage with expertise and experience from laboratories.
The main goal of the project was to start from a state of art stack with standard performances in order to set a work plan leading to improved stability, suitable counter actions and statistically validated durability.
To reach those goals the following points were introduced in the design of work:
1) degradation modes awareness and effects analysis
Collecting literature results and comparing them with data from post-mortem investigation on samples from operated stacks is meant to define in the most realistic way the criticisms existing in this system. The number of components with related materials, the interaction within those elements and the operating conditions (i.e. temperature, dual atmosphere, polarization, sources of pollutants) represent a great deal for the correct operation of the stack itself. To separate the degradation mode according to the minimum number of elements of single components (e.g. the anode, the cathode + the metallic interconnect, the electrolyte + the diffusion barrier layer, and the sealant + the metallic frame) and then to combine them in order to understand how they affect each other represented the main strategy to reach an higher level of understanding of the phenomena naturally or artificially (e.g. pollutants carried by the gases) contributing to the overall performances degradation of the stack. With this step a ranking of phenomena harmfulness based on frequency, probability, risk (consequences) can be compiled and used as a guide for the further steps of the project.
2) testing single phenomena and their combination in a stack
Once the degradation modes are identified and better understood it is possible to prove the validity of the interpretation and the efficacy of the solutions by working on specifically designed samples for near Real Life (nRL) experimental sessions. The samples are also inspired by models refined using the previously mentioned investigation activity. The same models may be verified and further improved by collecting the results issued by the new experimental sessions.
3) refining models to get closer to reality so that they can predict performances and events
A number of well designed models existed well before the redaction of the project proposal and at least three partners of the consortium were the authors of the source codes. Their activity in the project was at the same time a powerful engine to drive the project, an inspiration for further achievements in terms of phenomena - performances correlations, an excellent interpretation tools for the electrochemical and post-mortem results. These tools are potentially very powerful and may save a lot of time when becoming trustworthy predictive models able to indicate potential area of failures and to quantify the risk that the deviation from the expected behavior represents. Combining investigation on operated stacks, microsamples and models the consortium is tracking back the problem to find the real origin which may be elsewhere than where the failure was observed. This is directly related to the concept of Early Warning Output Signals (EWOS) introduced by the project and result in: the improvements applied on specific components, the creation and application of counterstrategies, and the search for a very sensitive and easy to apply degradation detection tool developed in the project.
4) EWOS and counterstrategies
To find out how to reveal at its earliest stages a source of problems is a fundamental strategy for a safe management of the stack. This means to act on time for the activation of counterstrategies or of counter actions and, in the worst of cases, to stop the operation of the stack before to reach a more dramatic level of failure. The possibility to integrate EWOS directly in the stack is one of the main purposes of the project and this is addressed by identifying the main issues (see point 1) and then the measurable output related to them. This results to be a very tricky task because of the impossibility to introduce probes and sensors in a real stack without affecting the production cost in a negative way. The best solution, acting directly on stacks is to make use of the normal outputs (i.e. Current, Voltage, Temperature, exhausted fuel and oxidant gases composition) and to work with them. On the other hand, the design and manufacturing of micro-samples replicating parts of the stack (e.g. the sealant in contact with metal, cell in the metal frame) operated in nRL with a duly designed test rig is a valid solution to control performances, materials compatibility and response to counter-actions in a very efficient and cost effective way.
5) Improvements and Statistical validation
Beside the models enhancement able to predict and then to project the performances beyond the real testing time, some improvements are needed at the components level in order to increment their efficacy and their reliability under operation. The main goal is to find out among the single elements and materials those needing to be adjusted and those clearly having to be substituted.
Industrial partners working with raw materials strongly contribute on this subject by supplying new materials, new manufacturing methods and critically discussing solutions.
A final process where micro samples, cells, and stacks are tested in simplified (pure hydrogen as fuel), complex (reformed gas as source of hydrogen), and critical (thermal and load cycles) modes is planned. This process is meant to demonstrate and statistically validate the previously achieved results.


Project Results:
The first part of the Project was focused on the study of stack degradation mechanisms at different scale starting by investigation of operated stacks for several thousand hours. Numerous interfaces present in the system were investigated being their evolution potential source of failures and introduction of power losses. The raw materials used for the stack production were also characterized to evaluate the influence of pollutant on the stack lifetime introduced in the manufacturing steps.
With the support of a wide literature research, experimental data from electrochemical analysis and post-operational characterization were integrated in an evolved version of FMEA becoming one of the public usage deliverables called Degradation Modes and Effects Analysis (DMEA). In such document, all the degradation processes are taken into account and risk ranked according to their probability (statistical frequency of the event) and harmfulness.
Starting from this base, thermo-mechanical, electrochemical and physical models were developed or, when already existing, refined in order to have an improved level of events and performances predictability. Several strategies were developed and integrated in stacks to mitigate their degradation rate and reduce the risk of failures. They consisted in the introduction of improved glass-ceramic sealants with high stability at the operating conditions, low interaction with the adjoining materials and excellent resistance to thermal cycles. Additionally, YSZ as interlayer between FSS and sealant was successfully verified to increase the adhesion and reduce the metal ions diffusion from the metal substrate into the sealant.
Several manufacturing solutions were brilliantly applied to obtain denser GDC based cathode/electrolyte diffusion barrier layer (BL) in order to improve the performances of the diffusion barrier layer (i.e. GDC impregnation, MgO as sintering aid, application via Pulsed Laser Deposition). The cells resulting from this step were mounted in short and real stacks, the latter operated in reformed fuel to match real operating conditions. A strategy of partial oxidation of the anode followed by its complete reduction has proven to be a promising counter action to reduce the risk of Ni coarsening, identified as one of the main sources of anode performances degradation.
Moreover, a brand new method of interpretation of the electrochemical data applicable to both cells and complete stacks was introduced: the Differential Resistance Analysis (DRA). Such tool applied on real data and on model generated data, demonstrated the possibility to detect nearly invisible deviations from the original electrochemical performances technically impossible to be observed on short tests and by usual tools. This allows to verify in a very precise and sensitive way the evolution of the cell performances even though close or below the target of 0.1% of degradation rate per 1000 hours.
FMEA
The document Degradation Modes and Effect Analysis (DMEA) generated during the first part of the Endurance project was actualized at the light of the experimental results from studied samples. A critical analysis of information coming from the characterization of long term operated cells and the evidences arisen from micro level experiments was carried out. The complete table arisen from these studies is reported in D 7.1 with specific explanations of the reasons for every correction to the previous version. The degree of importance of those more critical degradation mechanisms have been emphasized, as well as those in which their relative importance has been changed in comparison with the preceding studies.
Hereafter a discussion about the most important conclusions deriving from the DMEA is presented.
Anode
Anode deactivation is one of the main issues leading to decrease the performance of fuel cells after long operating times. Several causes can lead to such phenomena: Ni coarsening, carbon deposition, anode poisoning, three phase boundaries (TPB) losses and micro cracks, etc. In the present project, a wide statistical investigation has been made on real samples and through refined models. The dynamic and kinetic of nickel phase mean diameter increase vs. time during operation was quantified and predicted in function of the working parameters. In addition, nano-holotomographic reconstructions have been also performed on Ni-YSZ electrodes to reveal this degradation phenomenon. From the 3D analysis, it has been found that the current density is not enhancing the nickel coarsening. It is confirmed that time and temperature are strongly affecting the nickel phase evolution reaching a more precise level or prediction by the models which is substantial to foresee the performances in a stack on the very long lasting time. A substantial increase in the Ni particle diameter at 850°C was documented. However, such a high impact of temperature could indicate a more complicated mechanism with a potential role of the ceramic backbone to limit or even inhibit the Ni coarsening at low temperature. In order to better understand the nickel coarsening mechanism, the evolution of specific surface areas of nickel with gas and YSZ have been analyzed. The metal-ceramic interface does not change significantly with time whereas the surface in contact with the gas phase is strongly affected by the Ni coarsening. This statement indicates that Ni phases coarsening in the cermet mainly result from the rearrangement of particles in contact with the gas phase while the others seem to be strongly attached to the YSZ backbone. The precise analysis of the tomographic reconstructions has allowed highlighting the crucial role of the YSZ network on the Ni coarsening in the cermet. The decrease of TPBs density is pronounced at 850°C whereas it is limited at lower temperature (750ºC). Therefore, the decrease of the TPBs density appears to be clearly linked to the Ni coarsening upon operation. Temperature appears to be a major controlling parameter. The loss in active TPBs induces a deterioration of the cell performances. In order to quantify the contribution of this mechanism to the cell degradation rate, several experiments in our in-house multi-scale modelling framework has been implemented. For the test performed in a fuel cell at i=0.5 A.cm-2 and T=850°C, it has been found with the model that the decrease of the anode exchange current density is about 10% after 1000h. The contribution of nickel coarsening can thus be estimated from the loss of performance obtained with the modelling approach. The nickel coarsening is explaining a part of the measured degradation. After 1000 h, the cell voltage is decreasing of 1.1 mV according to the simulation. This value corresponds to about 32% of the total degradation rate obtained with the experiment.
YSZ anodic back bone
Additionally, a degradation mechanism associated to the YSZ stability at the anode has been found to take place in operation. Co-existence of monoclinic and tetragonal phases in YSZ of anode is heterogeneously detected in both fresh and aged cells. In cell operation, the transformation from tetragonal to monoclinic phase is accelerated due to the generation of water in the anodic reaction. Although the formation of monoclinic phase at the anode has not produced a remarkable increase of cell internal resistance in the present project. Despite the possibility that such dual microstructure of the YSZ backbone is generated along time the major risk is related to the mechanical stability. The backbone has mainly a mechanical role while Ni network has to grant the electrical conductivity, an increase of volume due to the martensitic transformation of part of the tetragonal phase into the monoclinic one may produce a catastrophic failure in operation leading to internal cracks. This explains a result found by three dimensional characterization of the anode and, due to its high impact on the performances, was described in a model further integrated in the predictive function of the stack. Poisoning is also a matter of concern, due to the multiple possible sources and its cumulative impact on the catalytic activity of Ni-YSZ at very low contaminant concentration, thus decreasing strongly the cell performance. The most relevant anode poisoning is mainly associated to impurities based on sulphur, because practical fuels contain minor constituents as impurities. Anode poisoning due to the presence of other impurities like SiO2, Na2O, B2O3 or Al2O3 that are initially present in the raw anode materials are known to segregate at surfaces/interfaces during cell operation. External contamination during cell processing and operating by using glass seal can also appear.
Cathode
LSCF cathode de-mixing has also been investigated. For this purpose, the phase reactivity with the electrolyte and more particularly the formation of ZrSrO3 has been characterized before and after the durability experiments. Synchrotron X-ray μ-fluorescence, SEM and TEM analyses have been performed in the region of the barrier layer. The chemical profiles revealed only a slight enrichment of Sr and, in a less extent of Co, for the SOFC aged sample. The μ-fluorescence characterizations were found to be fully consistent with the high resolution SEM and EDX (Energy-Dispersive X-ray detector) analyses. The observations have revealed the formation upon operation of a secondary Sr-rich phase (which corresponds to the SrZrO3 crystalline structure identified by X-ray μ-diffraction). This secondary phase is found to precipitate in the open porosity of the barrier layer at the interface with the electrolyte. However, the amount of this secondary phase is very restricted. This observation was confirmed by the quantification of the SrZrO3 by image analysis performed on all the tested samples. Even after 9000 h in SOFC mode, the quantity of zirconates is quite limited. As a conclusion for Endurance Project, it is claimed that, in our conditions, LSCF destabilization does not significantly participate to the degradation of cell performances during fuel cell operation. Additionally, the modelling approach has also been used to interpret the role of the cell operating mode on the LSCF destabilization mechanism. The cell polarization curves and the local quantities within the O2 electrode have been computed in the experimental conditions of cell ageing. Under cathode current of i=0.5 A.cm-2 the amount of vacancies in the perovskite is increased, which would explain why the formation of SrZrO3 phase is not pronounced in SOFC operation. Deactivation of cathode due to the presence of contaminants like chromium, mainly coming from interconnect steel, is probably one of the main cathode poisoning agent. The chromium deposition and poisoning is a complex process, due to many interrelated factors, such as operating temperature, O2 partial pressure, air flow, direct or indirect contact with interconnect, applied current load, properties of electrolyte, and composition and nature of electrode materials in particular, affect the amount, location and accumulation of chromium deposition. Several measures (e.g. usage of CoMn spinel oxide as interconnect coatings and a proprietary chromium getter material as cathode contact layer) have minimized the chromium poisoning, which has substantially reduced the importance of these phenomena for the current project compared with previous ones. A more important accent is put in the role of eventual SiO2 contaminants coming from the seal. However, it is interesting to highlight that local decomposition of LSCF at the surface into secondary phases: Sr–La–O, Co–Fe–O, La2O3, SrO, CoxOy, which have been significantly increased with the aging time, and sulphur contamination from an unknown source was observed.
Diffusion barrier layer at cathode/electrolyte interface
The cathode/electrolyte interface plays a significant role on the overall SOFC performance and therefore, it has attracted ongoing research efforts. The interaction between iron-based perovskite cathodes and YSZ electrolyte forms poorly conducting secondary phases, such as SrZrO3, which degrade the cell performance. The influence of both the thickness and microstructure of the GDC interlayer has been also considered, as previous works observed a systematic variation of the electrochemical. In the GDC screen-printed interlayers, the formation of SrZrO3 occurs during the cell fabrication, mainly depending on the method deposition of interlayer and the sintering temperatures of GDC barrier layer and cathode. It has been attributed to the transport mechanisms of Sr (from cathode) and Zr (from electrolyte) via gas phase and GDC surfaces (porous and grain boundaries), respectively. SrZrO3 clusters have been placed at the GDC surfaces of barrier layer, mainly at the porous close to electrolyte/barrier-layer interface. In operation at long-term, no variation of the SrZrO3 amount was observed, but some of SrZrO3 clusters of aged cells at very long-term seems to be larger than those of fresh cells, which could be due to their coarsening. Insight on the phenomena involved in the Sr, Y, Ce and Gd diffusion has been very valuable for the fabrication of better performing future generations of cells. Since the secondary phases (SrZrO3) of GDC barrier fabricated by conventional techniques (screen-printing), which require high sintering temperatures (>1200ºC) and present high porosities, a counter strategy consisted in the analysis of dense GDC barrier layers at low temperature. For this purpose, the PLD technique has been employed to prepare dense barrier layers. Voids and delamination between electrolyte and barrier layer have been formed in the cells without annealing in the dense barrier layer, due to the cations diffusion occurred during cell testing. In the PLD GDC interlayer optimally annealed after deposition and before cathode sintering, only the cells operated at high current densities presented nano-porosites at the interlayer/electrolyte interface. On the other hand, the degradation mechanism associated to the YSZ stability in operation has been also found at the anode/electrolyte interphase. Co-existence of monoclinic and tetragonal phases in YSZ at the anode-electrolyte interphase is detected in aged cells. This transformation is accelerated in operation due to the formation of water in the anodic reaction. Therefore, it should take into account, as it may increase the probability of anode/electrolyte delamination. For this reason, special attention has been paid to the possible presence of delaminations between electrodes and electrolyte. However, the degree of optimization of the cell technology involved in the Endurance project forecast a reduced risk of such phenomena to come into scene.
Interconnects and metal frames
All metallic components of the stack (i.e. frames, spacer, interconnect plates and shaped foils) were made in a ferritic stainless steel commercial grade (i.e. K41X). This material was not under investigation for improvements in this project. However being the oxidation of interconnects in contact with the cathode a potential source of performance losses due to the electronic resistance of the thermal grown oxide and to the cathode chromium poisoning, the spinel oxide coating used to hinder Cr volatile compounds formation was studied. The samples exhibited a thin chromia scale at the alloy/coating interface plus spinel phases as a reaction layer between the chromia and the manganese cobaltite coating. The oxidized samples also exhibited MnCr2O4 spinel at the chromia/alloy interface. The formed layers showed changes in thickness, morphology and composition, attributed to changes in sample preparation and degradation experiment. The current state of the art steel components in the Endurance stacks are coated with a Mn-Co spinel oxide protective layer applied by screen printing in the areas exposed to air only. No coatings were applied on those surfaces in contact with the fuel.
Sealant
The sealant have multiple roles: joining the cell (electrolyte at the cathode side) with the metal plate acting as support and as interconnect at the same time, keeping the cathode compartment separated from the anode area in order to avoid fuel / oxidant mixing, and separating the metal frames in order to avoid short circuits between the active layers. This is why mechanical resistance, creep resistance, high electrical resistivity, low reactivity (with metal, electrolyte and gases), and gas tightness are all mandatory characteristics of sealants. In the SoA stacks manufactured by the industrial partner a compliant glass-ceramic sealant is used. The formulation insure all the previously mentioned characteristics however it is known from literature and was observed on real stack samples after long lasting operation that the sealant is affected by a number of alterations potentially dangerous for its stability. The most important ones among them are: devitrification (or increased crystallinity), ionic diffusion from the steel (i.e. Mn, Cr, Fe) with formation of brittle compounds (e.g. barium chromate), increased micro porosity due to devitrification and elements volatilization. Moreover thermal gradients and thermal cycles strongly stress the sealant increasing the risk of failure proportionally with the evolution of its microstructure or of its chemistry. In addition it was confirmed by a number of experiments in this project that the polarization of the metal plates during the normal functioning of the stack is strongly affecting the evolution of the sealant microstructure as well as the diffusion rate of cations at the glass/metal interface.
Improvement of cell materials
Based on the characterization of samples issued by operated SoA stacks and on data from literature a list of degradation modes was created (deliverable D7.1) with related risk ranking where both the probability and the harmfulness of a specific event are taken into account. This gave rise to the definition of a strategy for the improvement: of the cell by the densification of the diffusion barrier layer between the cathode and the electrolyte, and of the sealant by introducing a barrier layer (YSZ) between the metal and the sealant and by proposing a new glass composition showing better performances. According to the results specific samples were designed in order to better understand the phenomena under investigation and to statistically validate the improvements. A complete list of samples is collected in the database Book of Samples (aka BoS) accessible by all members of the consortium in the internal website.

GDC Barrier Layer
Despite the very good ionic conductivity of GDC and its good barrier behaviour against the inter-diffusion of Zr, La and Sr, the design of this layer is known to significantly affect the electrochemical performances and the durability of the cell.
The main parameter affecting the efficiency of the DBL is the density. The occurrence of pores trough the DBL, indeed, reduces the barrier effect allowing the direct diffusion of ions (especially Sr-species) from the cathode toward the electrolyte. Very high sintering temperatures are normally required in order to obtain a dense DBL. However, it is well known that the high temperature promotes the miscibility between the YSZ and GDC phases, leading to a poor oxygen conducting YSZ-GDC solid solution and increasing specific resistance of the cell. For this reason, temperatures lower than 1300°C are normally considered when the GDC is sintered on the YSZ electrolyte.
Several solution found from literature were used to obtain cells with improved DBL (e.g. addition of MgO, ZnO) to be directly validated in real stacks operating under reformed fuel. The R&D process was in this case bypassed in order to be more time efficient and to dedicate further resources to alternative routes like the usage of several sintering aids (e.g. Cu oxide, Li oxide, GDC precursors) added by infiltration and by opting for a less industrial but definitely more efficient deposition method as the Pulsed Laser Deposition.

By using highly active sintering aids as well as impregnation methods to further increase the density of screen printed GDC pellets it was shown that Li, Zn as well as Cu (3-5 wt.%) improved the density of GDC compared to undoped pellets. Among these sintering aids, Cu was most effective, which even allowed for reducing the sintering temperature which helps mitigate Gd diffusion into the electrolyte layer. Unfortunately, the positive effect was not kept in screen printed layers applied on the SOFC half cells, as the already sintered substrate constrained densification of the screen printed GDC layer. The infiltration method, on the other hand, was successfully applied for screen printed layers. The best results were obtained by applying four consecutive impregnation steps of Gd/Ce-nitrate solution with intermediate nitrate burnout at 450°C, which altogether resulted in a 20 wt% weight increase of the GDC layer. Adding Cu to the nitrate solution created additional bubble-shaped porosity, and was therefore rejected. Figure 1 shows the results from electrochemical tests performed at 700°C, which shows that the improved density results lower ohmic losses as well as polarization losses.
The infiltrated GDC layer leads to a substantial improvement of the cell performance as compared to the commercial one, being the power density at 0.7 V increased from 0.59 W•cm-2 up to 0.87 W•cm-2 at 700 °C, which corresponds to a performance gain of about 40 % just by changing the GDC interlayer. Such an improvement in the cell performance supports the claim that obtaining a denser GDC interlayer by infiltration is a highly promising method. However, being this a preliminary study this option was not considered for statistical validation in the final stacks but was kept for future improvements beyond the project end.
Two types of PLD barrier layers were tested, i.e. with pre-annealing at 1150oC to further consolidate the microstructure, and 2) without pre-annealing before cathode deposition. The electrochemical tests, which were carried out at 0.5 Acm-2 and, in one case only, at 1Acm-2, and 750°C and shown in Figure 0.10 revealed that pre-annealed barrier layers achieve a higher power density than those without pre-annealing, as well as possess a lower degradation rate (figures 2 - 4). Electrochemical impedance spectroscopy measurements provided further information about the ohmic losses, which were found to be considerably lower for the annealed barrier layers, and showing the positive effect of annealing in suppressing the ohmic losses across the barrier layer.
The i-V curves acquired right after reduction (i.e. ageing time 0h) and after 1000 hours of operations are here after presented in figures 3 and 4. The cells improved with the PLD barrier are compared with the classical cells with screen printed barrier. The annealed PLD samples show a better performance than the reference cell at the beginning of the experiment confirming the improvement related to the diffusion barrier quality. It clearly results however that an annealing is needed in order to obtain a fully and better performing barrier.

According to the results obtained on single cells and the validation on the short the best solutions are the following:
• GDC doped with ZnO: displays improved performances with respect to the SoA and very low degradation behavior.
• GDC applied by PLD: displays the higher performances with respect to all the other options considered in the project. The degradation is also very low and a slight long term activation is observed on short stacks.
• GDC doped with MgO: displays slight lower performances with respect to SoA. The degradation behaviour still needs to be confirmed.

The best performing solutions were introduced in stacks for the validation of their behaviour at operating conditions. It is worth to mention that the PLD barrier layer, being fully dense, were considered as reference for the best performances. Preliminary tests on button cells allowed to determine the impact of annealing treatments on the barrier layer stability.
The cells including the thin-film (PLD) barrier layer applied displayed significantly higher performances with respect to all the other materials prepared by conventional technologies (e.g. screen printing). The use of ZnO as sintering aid in GDC improves the performances with respect to the reference cells, while the addition of Li2O or MgO leads to lower performances. It is worth noticing that the cluster having the same barrier layer composition, such as GDC-ZnO and GDC-Li2O, despite their different position in the stack tower, display the same performances, demonstrating a good homogeneity of temperature and gas distribution in the stack.
The degradation of the stack (figure 5) was calculated as the percentage of voltage loss, at 0.4 A/cm2, normalized on 1000 h operation. In order to have reliable data, only the values of the voltages after maintenance (>3200 h) were used. The initial and final values of the voltages were taken as the average on 48 h at the two extremes of the range.
Some clusters showed a slight increase of the performances (negative degradation, figure 6, especially for GDC-PLD, the reference cells and one of the clusters including GDC-MgO. A slight increase of the performances was also observed on one of the clusters including GDC-Zn. On the basis of these results, although the best electrochemical performances were obtained with GDC-PLD and GDC-Zn, the selected materials with respect to the degradation behaviour are GDC-PLD and GDC-MgO (excluding the reference).

Sealing and YSZ barrier layer
Ten samples of glass compliant sealants were tested according to the procedure developed and described in the Handbook of tests procedures and protocols (D 4.1). Additionally, the use of Yttria-stabilized zirconia (YSZ) has been proposed and tested as barrier layer preventing metal ions diffusion from the steel substrate toward the glass and enhancing the compatibility between the glass and the metal frame. YSZ layers have been applied by means of plasma spraying on as-rolled FSS substrates after sand blasting. Glasses were applied on steel substrates with and without YSZ coating and tested at 780°C in static air for 100 hours. Based on such results the two best performing glass compositions (GS-A and GS-I) in terms of compatibility with the substrate and stability at the operating conditions over time were further tested for four consecutive thermal cycles of 250 hours in order to investigate the effect of such conditions on the microstructures and interfaces. Both compositions resulted well adherent to the metal substrate for the whole test duration and a layer constituted by Cr, Mn, Fe was detected at the steel/glass interface. Despite the increase in crystallinity and the formation of new phases, no negative consequence or defect formation were observed up to 1000 hours. The behavior of the selected glasses confirmed the observation made on the preliminary tests and allowed to propose such compositions as improved materials for stacking.
YSZ layers were porous and irregular in their thickness, however, the porosity decreased as effect of the thermal aging. Additionally, the effectiveness of the YSZ layer as diffusion barrier was confirmed detecting a confined chromium-rich oxide at the steel/YSZ interface without further diffusion of substrate elements through the coating. In conclusion GS-A and GS-I can be indicated as valid compositions for sealing planar stainless steel based SOFCs considering their mechanical compatibility with the steel, stability at high temperature and limited interaction with the substrate. The application of YSZ coating enhances the sealing adhesion to the substrate and constitutes a further protection of the glass confining the substrate elements diffusion.
Before to substitute the state of art sealing with the new formulation further tests were made in order to check out the influence of thermal cycling on glass-ceramic sealants under more robust conditions using a testing furnace was purchased which enables constant cooling rates over a broad temperature range. Thermal cycling between 300 and 800°C for several hundred times showed that, in comparison to samples kept at a constant temperature of 800°C, the cycling (in particular the structure and crystal/glass ratio) had no negative impact. It is thus expected that by keeping the chemical interaction with neighbouring layers low, long lifetime of the glass exceeding those expected by the state of art one can be achieved.
Statistical Validation Experiments
Three stacks were delivered by SolidPower with various combinations of state-of-the-art and improved cells and glass seal. Two stacks were tested under thermal cycling conditions. The introduction of the improved glass seal had an effect on the durability of the stack and a lower degradation for this type was found. The stack used for load cycling had no degradation in power density after 100 cycles but there was a decrease in OCV in all cells.
A short stack, Stack #751, was mounted using improved glass seal and state-of-the-art cells in order to check the stability under operation and the resilience to thermal cycling considered among the most critical conditions for stacking materials. According to quality assurance test from SolidPower the #751 stack was operated at an OCV of 4.81 V and a power of 106 W at 84.9 % FU. After heat up at DLR the stack had an OCV of 4.79 V and a power of 106.4 W at 84.9 % FU which is almost identical to the QC measurement from SolidPower. In total 16 thermal cycles were performed during the operating time of 580 h. The results are expressed in figure 7 while figure 8 refers to a comparison between the improved stack and a stat of the art one (stack #710) operated in identical conditions. The OCVs of the cells in stack #751 were more uniform compared to stack #710. The gap between best and worst cell was only 7 mV compared to 43 mV. This can be attributed to the improved glass seal in the stack. The initial power density of the stack was 314 mW/cm² at 80% FU. After the 1st thermal cycle the power density increased slightly to 317 mW/cm² and then stayed the same for the next 15 thermal cycles. Therefore the degradation of the stack was 0 % for the 16 thermal cycles. The first stack #710 had a better initial power density of 323 mW/cm² but exhibited a higher degradation rate.
A 6 cells short-stack from SolidPower was operated for 4000 h at nominal operating conditions (0.4 A/cm2, 80% FU) with diluted hydrogen. The overall degradation was approximately 0.5%/1000h. The stack showed an initial rapid degradation during the first 100 h, which correspondeds to a loss of OCV probably due to a deterioration of the sealing. This initial degradation was followed by a slow activation up to 1700 h of operation. EIS measurements showed that this was mainly due to the improvement of polarization processes occurring respectively around 200 Hz and 2 kHz, both of them attributed to the anode. This improvement probably stems from changes in the anode microstructure with time (e.g. increase of the porosity). After 1700 h, the short-stack performances start to degrade mainly due to an increase of the ohmic resistance.
Load cycling
The stack #716 was used for load cycling. This test was performed with the same gas flow of 1.15 slpm H2 and 0.77 slpm N2 on the anode side and 21.3 slpm air on the cathode side. Operating temperature was 750°C. First a polarization curve was recorded. Then the current was increased at 10 A for 30 minutes and then changed to 32 A at a rate of 0.5 A/min where it was kept for another 30 minutes before it was decreased again to 10 A at the same rate like before. It was intended to perform 100 such load cycles with the stack. At the start the stack had a power density of 304 mW/cm² at 80% FU and an OCV of 4.74 V, see Figures 9 and 10. The stack power at 84 % FU was 98 W. SolidPower recorded an OCV of 4.84 V and a power of 105 W at 84.1 % FU which is higher than the values at DLR. At DLR cell 4 of the stack had a rather low OCV of 1151 mV which was about 50 mV less than the other 3 cells which are at 1200 mV and it also has the lowest power density in the stack. In total 100 load cycles were performed over the course of 1250 h. Since the test bench used was not equipped with an automatic mode the test had to be performed via manual input. The stack was kept on the lower boundary of the load profile over the night and weekends. The power density at 80% FU after the test was 308 mW/cm² which is a slight improvement compared to the start, see figure 11. There was a change in the OCV of the stack from 4.74 V at the beginning to 4.68 V at the end. The biggest OCV loss was recorded for cells 1 and 3 with 29 mV and 25 mV, respectively. But no cell showed an abnormal behaviour at the last polarization curve. The curves are still overlapping (figure 12)

Management of electrochemical tests and data generation
This activity, fundamental to operate organize and follow the micro and macro samples under operation in order harmonize the data acquisition, to generated samples for the post-operation characterization where to find out chemical and physical features to correlate with the performances, and to feed the modelling process with fresh data. This activity, carried out by WP4 had the following main objectives:
• To set up and perform micro and macro experiments
• To record on line data and behavior near Real Life and cycled tests
• To define the protocol of sample management and transfer
WP4 played the fundamental role of connection between on one side WP3 and WP7 and on the other side WP5 and WP6. WP7 was closely related to WP4 for stack and microsample ageing protocols as planned for statistical validation and improvements. The studied samples gave inputs to WP6 and, as a consequence, WP2 and WP3. During the course of the project, samples with potentially improved durability emerged. The micro-level and macro-level tests are reported in this work package.
Two main achievements are worth to be taken into account: the micro-level counter measures and the introduction of Differential Resistance Analysis (DRA).
The first contemplates the previously mentioned improvement of DBL density, introduction of YSZ as inter layer between steel and sealant, the formulation and test of a new glass sealing. Such results were also positively confirmed by the statistical validation process. It is however important to mention that an investigation on the improvement of the amount of electrocatalytically active sites by obtaining smaller Ni particles in the anode by redox cycling, as better described here after, was experimented and duly described in the Handbook of test procedures and protocols (D 4.1). Its implementation is under experimentation on special micro samples and will be introduced in the stack management practice once statistically validated.
The DRA deserves a special and well detailed discussion being a brand new mathematical method to apply to the most simple output of a cell or a stack, the i-V curve. Its simplicity not needing special equipments and offering a very accurate vision of the state of health of a cell or of a stack deserve a special mention.
Improvements anode materials
The accumulated experience in redox cycling of single anodes was used for the performance of initial reduction and redox cycling on button cells, which was carried out by impedance measurements by IEES. The initial fast reduction of NiO (several minutes) causes rapid formation of water which fills the pores in the anode and increases the polarization resistance. The artificial evacuation of water by applying reverse (electrolysis) mode indicates a positive effect resulting in lower resistance, better VAC and shorter reduction time.
The oxidation process in the complete cell followed 2 approaches: (i) replacement of the reduction blend with an oxidizing one, which simulates oxidation conditions that may occur during emergency shutdown and (ii) electrochemical oxidation which simulates operation in conditions of fuel starvation when the oxidation occurs locally in the active layer close to the electrolyte/anode interface. The oxidation of the cell anode performed at 600oC by introduction of oxygen which covers the total volume of the anode, decreases the cell resistance after reduction (figure 13), i.e. the behaviour is similar to that observed for single anodes.
The positive effect of the pre-oxidation treatment at 600°C can be explained with expulsion of the NiO particles at the outer anode surface, which was registered in the performed experiments by post mortem analysis. Perceptibly, the local oxidation obtained in conditions simulating fuel starvation does not have a positive effect and an increase of the cell resistance after oxidation/reduction treatment is registered. Although the oxidation level (evaluated by the increase of the cell resistance during oxidation) is smaller in comparison to that obtained by oxidation from the gas supply, the effect of the localized oxidation is quite strong and causes cracks in the active surface which in some of the experiments caused cell failure.
The obtained results have provided new directions for improvement of the cell and stack performance by introduction of preliminary and periodic oxidation treatments for improved redox tolerance.
Differential Resistance Analysis (DRA)
Being one of the consequences of an advancement in DMEA and models predicted by the project the definition of specific EWOS a part of the effort of this WP was focused on implementation of fine methods able to amplify the signals related to the degradation processes. The Differential Resistance Analysis (DRA) by simply using volt-amperometric curves (i.e. VAS or i-V curves) successfully increased the sensitivity of the degradation analysis up to 30 times making thus possible to observe very low degradation rates (e.g. 0.1% voltage loss at constant current load per 1000 hours) in short time experiments (e.g. 100 hours) and to define the quality of the cell or of the stack right after activation.
The application of Intensive Active Tests and Analyses (IATA) provides more sensitive and multi-dimensional detection of EWOS using the multi-parametric nature of the fuel cell, since it is operating with derivatives which are more sensitive to small deviations.
IATA combines 2 types of measurements: volt-ampere curves (VAC) and impedance in well defined working points. In the second part of the project a new analysis for IATA development named Differential Resistance Analysis was introduced. It offers a new approach for analysis of VAC curves and their comparative assessment through the parameter Differential Resistance Rd.
A preliminary analysis of different VACs showed that they are sensitive to operating parameters and degradation which reflects in the change of their shape. At constant operating conditions those changes should be caused by degradation. The question is how to perform “Shape analysis”.
The innovative approach “Shape analysis” is based on the analysis of Rd as a function of the current I, named Differential Resistance Analysis (DRA). A meaningful example is presented in figures 14a and 14b. Three characteristic regions can be separated: Region I which is connected with activation losses, Region 2, which concerns the transport losses and Region 3 where the gas diffusion limitations are dominating. A characteristic point is the minimum of the differential resistance Rd,min which is an important characteristic point for the cell performance reflecting the state of health at constant operating conditions, since it is determined by the intrinsic properties of the system and not by the external conditions (for instance load current).
The development of the DRA method provides for the creation of EWOS technique for fuel cells. As far as the Differential Resistance described above is a derivative function, it can be discussed as similar to the impedance (which is also a derivative) at zero (or very low) frequency. The defined value of Rd,min in the DRA corresponds to the zero point of the Rd derivative, i.e. it can be regarded as a second derivative. As a result, the property of Rd,min to change becomes observable much earlier than that of the other FC parameters. Thus the nature of this parameter, or performance indicator, defines its properties to register early recognition of FC degradation performance and thus an adequate diagnostic.
DRA method proved to be a useful tool with increased sensibility towards degradation, since it works with the derivatives of the measured parameters, which are in principle more sensitive to small deviations. It gives an opportunity for collection of reliable data from shorter tests avoiding accelerating test conditions.
Since the DRA indicators are more sensitive, sharp deviations in their time dependence can serve as EWOS for increased degradation. Every indicator has higher selectivity in respect to different degradation source (∆U* is more sensitive to activation losses and Rd,min, to transport hindrances). In addition impedance measurements can give more precise information about the origin of the degradation (ohmic losses, electrodes polarization etc.).
The operation with derivatives, however, increases the influence of the noise. The spectral transform based on splines approximation improves the noise immunity. However, the analysis needs higher quality of the measurements and not a big number of data. For one VAC 30-40 points are sufficient, but, they should be measured after stabilization of the operating conditions (about 20-40 seconds).
An example is shown in figure 15 where i-V curves simulated by the most up to date and validated model issuing form Endurance project are presented in function of time up to 36'000 hours.
Advanced sample analyses
Most of the operated samples underwent a post-experiment deep investigation oganized at various levels:
• cell level
o Nickel coarsening dependence from temperature
o Nickel depletion in electrolysis mode
o Nickel redistribution under long lasting operation at high fuel utilization and nearby the outlet fuel manifold
o beneficial effect of partial oxidation and complete reduction on anodes
o YSZ-GDC interaction on the long lasting period
o LSCF partial decomposition on the long lasting period
o LSCF instability under air starvation
o Three phase boundaries concentration reduced under operation
• stack level
o Glass sealing evolution in function of polarization, temperature and pressure, thermal cycles
o stainless steel - Glass sealing interaction with cations diffusion from the metal substrate under operation, effect of temperature and polarization
o Spinel cobalt manganese coating on steels densification during operation with related absorption of Cr from the stainless steel.
o beneficial effect of YSZ as an interlayer between the metal frame and the glass sealing.

The methods used to characterize samples were adapted according to the goal using superficial and bulk analyses for more general usage (e.g. phases distribution at the cathode and at the anode on a 15'000 hours operated segmented cell from a stack, figures 16 and 17)
Among the huge amount of data collected in this project topic it is worth to mentioning the following results gathered from two of the last samples:
• Segmented cell operated under electrical load during 15000 hours (the second segmented cell operated at HTc which electrochemical performances were quoted in WP4). 20 segments distributed from inlet segments (1, 6, 11, 16) to outlet segments (5, 10, 15, 20) as reported in figure 18. together with the operating parameters
• Three button cells from a group of 6 operated at various temperatures, and SOFC or SOEC mode at CEA facility. Those selected for the post-mortem investigations are C2 (SOEC mode, 850°C, 50/50 H2/H2O, 2000h), C4 (SOFC mode, 850°C, 50/50 H2/H2O, 1000h) and C6 (SOFC mode, 750°C, H2 100%, 9000h). In order to have more details on the whole set of studied cells please refer to table 4.1 in WP4 chapter.
• Samples called Metal-glass-metal were studied after being tested in order to check the correlation with the failure or the evolution of the materials and of the interfaces in case of successful test.
With reference to the segmented cell operated for 15000 hours it is possible to discuss about the surface and bulk analyses reported in figures 16 and 17. Two main conclusions may be drawn whatever the considered segment: at high scale, large stripes (from upper left to lower right) depleted in cubic and enriched in tetragonal zirconium oxide phases were observed. Correspondingly, the fluorescence is lower in these stripes. At low scale, one sees the ellipsoidal footprint left by the metallic current collector. The edges appear depleted in tetragonal zirconium oxide transformed into the monoclinic phase. On the opposite it appears that the zones of higher monoclinic ZrO2 correspond to lower fluorescence and lower tetragonal YSZ. Higher cubic phases could also be involved, however when the monoclinic phase is observed it is difficult to assume the effective presence of a cubic one. As evidenced on upper Figure 17 (yellow part), images of monoclinic ZrO2 have the shape of 3-branch stars exactly located between the current collector zones. The area surrounding the current collector footprint are active during the operating time because electrons are passing through such areas. The metallic fluorescent zones are then connected with higher content of monoclinic ZrO2 (and/or cubic) and lower content of tetragonal YSZ. Indeed, the geometrical features of it, as well as the corresponding change on the gas flow distribution within the substrate and adjacent to it, and the concentration of electrons flowing from the anode to the current collector could be at the origin of local physical phenomena generating the occasionally observed Ni deposition and the zirconia phase transformation. It is worth noting that beside the representativeness of the pictures presented in Figure 17 a little difference was remarked comparing the inlet segments (i.e. numbers 1 and 6) with to outlet ones (i.e. numbers 5 and 10) which seems to have a more homogeneous aspect. The two main differences of the two couple of segments are the temperature (ca. 100°C higher at the outlet side) and the fuel composition (dry hydrogen at the inlet, wet hydrogen, with more than 80% of water vapour, at the outlet).
Working on cross sections from the same segmented cell it was possible to observe the effect of the local working parameters (i.e. temperature, fuel composition) on the evolution of the Ni (figure 18 and 19). To better understand such results it was important to carry on similar investigations on single cells operated in various conditions as shown in figure 20.
The set of pictures of figure 20 is clearly showing that: C2 has the largest Ni grain-size, the YSZ-CERMET interface lost a part of the Ni grains and some seems to be partially oxidized. C4, operated for half the time as C2 at the same temperature, the Ni grains coarsening is visible but it is lower than for C2, theYSZ-CERMET interface has still all Ni grains. C6, operated nine times longer than C4 but at lower T, shows the smallest Ni coarsened grains and a still good anode-electrolyte interface. Temperature seems then to affect more than time the Ni evolution in the cermet electrode with a visible effect already after 1000 hours and still clearly effective after 2000 hours. Beside, operating the same kind of cell for 9000 hours at 750°C (which is classically the average temperature of a real stack) has a limited effect on Ni coarsening. The cell operated in electrolysis mode has shown a strong Ni loss at the electrode-electrolyte interface that might lead to a very high degradation rate. A more detailed quantification of the Ni grain size and of the related TPBs has been made with high resolution cross section analyses as reported in the following chapter
In Figure 18, the Ni particles were highlighted in red to have a better visibility. It is worth to recall that the segmented cell is made of a common anode + electrolyte substrate supporting a segmented cathode applied by screen printing. The observed difference between the left hand picture and the right hand one is only the position in the test bench which means: a) a difference in operating temperature of ca. 100°C , and b) a difference in the fuel composition (i.e. H2/H2O ratio) due to the imposed fuel consumption. Comparing the present results with those previously described the effect of the temperature on the Ni grain coarsening is confirmed while the fuel composition might have a different effect on the Ni distribution. To investigate over this aspect on a large number of samples a specific tool was developed combining image analysis, compositional analysis and physical data. The Ni volume fraction was quantified according to the transformation of the following formula
Ni(vol%)=100* Nivol /(Nivol + YSZvol + porosityvol) (1)
Where Nivol and YSZvol corresponds to the amount of surface occupied by the specific element taken into account. The Ni amount can be quantified by the chemical composition of the measure area with a higher definition compared to the image analysis, and with the contribution of the all present Ni (e.g. the partially oxidized, the Ni diffused, the Ni volatilized and then re-solidified elsewhere than a well defined grain crystal). Knowing the density of the Ni and of the YSZ and knowing that the sum of all volume fraction must necessarily be 100% it was thus possible to transform the equation (1) into the equation (2) to obtain the equivalent volume fraction of Ni. In this equation two only variables are introduced: the Ni weight percent and the porosity volume fraction (i.e. the easiest measurable image analysis parameter). The final equation is hereafter referred:
Ni% eq.= ((wtNi*densYSZ)*(100-Por))/((100-wtNi)*densNi+(wtNi*densYSZ)) (1)

where the term "Por" stands for the porosity percentage volume fraction, wtNi is the amount of NI measured by EDXS and expressed in Ni wt.%, densNi and densYSZ are the density of Ni and of YSZ respectively.
Supported by the results acquired on the C4 and C6 samples the amount of YSZ is considered stable or only minimally affected by the operating period, therefore the results are expressed as the ratio Ni wt.% / YSZ wt% (i.e. Y wt.% + Zr wt.% + O wt.%) also known as Ni relative content and as Ni equivalent volume fraction. The Ni/YSZ ratio is supposed to be homogeneous all along the Cermet volume. A line profile acquired from the YSZ-Cermet interface to the edge of the electrode should be flat showing only a difference at the very beginning of the profile and the same is supposed to happen at the Ni equivalent volume fraction. Any deviation from linearity should be explained as a diffusive phenomenon or a redistribution of the elements. In case of lack of coherence between the two profiles (i.e. Nieq and Ni/YSZ vs. distance from the electrolyte) a redistribution phenomenon instead a coarsening diffusive process.
The low amount of Ni found at the interface with the electrolyte is related to the manufacturing process and does not represent an issue. The increase of Ni in the first layers (still in the active volume of the anode) is worth to be mentioned as this is localized in the inlet segment (number 6) while it is displaced and widened in the outlet segment. For segment 6 the relative amount of Ni remains quite constant from distance 5 μm to distance 150 μm with a small variation nearby the edge. There is a visible divergence in the Ni equivalent curve which may be related to a different density of distribution of Ni grain in correspondence to the edge. Beside this, the segment 10 shows a strong alteration of the Ni distribution in the core of the cermet with values comparable with the inlet segment only starting from distance 100μm. The final part of the electrode is similar between inlet and outlet suggesting that the main alteration in Ni distribution is mainly localized in the half of the cermet closer to the active zone (i.e. to the interface with the electrolyte). For a deeper discussion on Ni moving across the anode a more generic vision is needed. With this purpose the segment numbers and the "from the electrolyte to the edge of the anode" curves are combined to draw two contour graphs (Figure 19) giving an interesting overview on the Ni distribution after 15'000 hours of operation.
Both contour graphs show that Ni is unevenly distributed with an higher concentration in the the zone between 2 and 5 μm in segment 6 and 7, a beginning of redistribution beyond 7 μm of distance with related Ni depletion below this distance in the remaining segments. The graphs suggest a redistribution of Ni far from the active zone to the central lateral volume coherently with the fuel stream direction. Such change was unnoticed by the simple observation of the cross section and seems to be coherent with the change in fuel stream chemical composition along the active zone aside the anode-electrolyte interface from the inlet to the outlet manifolds. The remote part of the anode is less affected. Few hypotheses may be proposed to open the discussion:
1. The Ni grain coarsening is a kinetic phenomenon mainly related to the local operating temperature
2. The Nichel is not only moving through solid state diffusion (as in the coarsening case) but also by cycles of evaporation and deposition related to the temperature and the H2/H2O ratio in the fuel stream
3. The Ni changes in content between the segments at the inlet manifold and those at the outlet manifold suggest a faster performance degradation nearby the outlet due to the cermet alteration.
It is here supposed that the formation of Ni volatile compounds (e.g. Ni hydroxide) entering into the fuel stream is favored in the active area of segments from 7 to 10, especially because of higher temperature and H2/H2O ratio. The volatile Ni oxidized compound are then reduced in the adjacent zones where they meet a fuel richer in hydrogen. To explore how close to reality is such hypothesis a high definition investigation has been planned on the same fragments at the synchrotron facility of Grenoble to be carried out on July 2017. The results will be shared within the consortium. Diffusion, coarsening and, if confirmed, Ni volatilization are natural phenomena not related to malfunctioning cells. To be aware of such phenomena means might be an interesting step forward which will lead to suitable strategies to increase the cell durability.

Selected from the campaign of experiments performed on glass sealants (GS) the results gathered by the usage of YSZ as diffusion barrier between GS and metal substrate are hereafter shortly presented. The best performing new composition is here described in terms of microstructural features with and without YSZ in the state "as cured" (0h), mid-term aged (500h) and fully aged (1000h). Figure 21 shows a set of pictures with related profile line of elements distribution where only the main elements of the steel are taken into account (i.e. Fe, Cr, Mn). The bulk of the glass was characterized by the presence of homogeneously distributed polygonal crystals. After ageing for 500 h, some irregularities, as well as the formation of a dark interlayer and the growth of bright crystals with polygonal shape were observed between the glass and the metal. After 1000 h testing including four thermal cycles, the glass was still well adherent to the substrate and no further significant evolution of the glass was observed. EDX line profile further confirmed the formation of a Cr-Mn spinel phase on top of the metal during basically the first 750 h of operation, and the efficient role of the YSZ layer to prevent chemical interactions between the metal and the glass. Additionally, it was found that the wetting and, consequently, the adhesion of the glass on the YSZ coating/metal is improved compared to the non-coated metal.
Predictive modelling
The most enhanced predictive modelling issued by the ENDURANCE project are collected in the deliverable D6.1: series of thermo-electrochemical and mechanical models that were developed for improving the understanding and prediction of SOFC stack degradation. The model formulation and implementation were driven by experimental results (WP4, 5 and 7) with the goal of providing new capabilities for degradation modelling. Therefore, formulations based on physical principles were favoured to capture the regimes in degradation patterns, as a function of operation history. The model scales span from the characteristic size of the material phases in the electrodes to that of the stack and both thermo-electrochemical and thermo-mechanical aspects were investigated. The developed frameworks are integrated, i.e. continuum micro-models developed for phenomena analyses were implemented in full details up to the stack scale.
The study of electrochemical and mechanical degradation required analyses at different scales. The above mentioned deliverable is therefore organized from the smallest size (features in the heterogeneous electrodes) to the stack scale. The focus in the thermo-electrochemical analyses was placed on selected phenomena: Ni coarsening (Sections 1.3 4.1.2 4.2.1 and 4.2.2) and cathode destabilization/zirconate formation (Section 1.4 4.1.3 4.2.1 and 4.2.2) at interfaces, which were studied by time-lapse analyses and/or time-dependent simulations. The identification of processes in electrochemical impedance spectroscopy measurements (EIS) and understanding of reaction pathways are presented in Sections 2 and 4.1.1. Electrode fracture was investigated at different scales in Sections 1.1 1.2 3 and 4.3.4. Component irreversible deformation and geometrical imperfections were implemented in the stack thermo-mechanical simulations for understanding their effect on the reliability (Sections 3 and 4.3).
The dynamic and effects of Ni coarsening were studied using 3-D imaging (x-ray nanotomography and FIB-SEM/EDX serial sectioning, figure 22). Under SOFC stack conditions, the rate of growth of the Ni phase is initially higher and then decreases. Time-lapse degradation analyses with 3-D discrete element simulations of the polarization resistance showed good agreement with the degradation measured by DRT on segmented-cells (figure 23). The evolution of the phase diameter and effective TPB length was analyzed using a standard power-law model, which results from the physically-based modelling of the sintering of two metallic particles. The value of the fitted exponent indicates a surface diffusion mechanism for the coarsening with a constraining effect from the YSZ phase, with dependence on the temperature. The computed effect on the thermo-elastic properties was mild. New measurement concepts and metrics were also developed for further insights into the reasons for the degradation and identifying potential risks when extrapolating for long-term predictions. Within the time window investigated in the project, the degradation is due mainly to the decrease in the number of TPB sites, but not altered accessibility by the transport of products/reactant (i.e. quality), even though the microstructure is not uniformly utilized.
A continuum model for LSCF-based electrode was developed to investigate in details the reaction pathways (figure 24). Four reactions are taking place in the oxygen electrode that can be split in two parallel reactive pathways: the bulk and surface paths. After combined calibration under anodic and cathodic polarization, the ratio of the two paths were investigated as a function of operation and microstructural parameters. The charge transfer at the TPBs, which can extend in the whole thickness of the composite layer, was found to dominate the electrode response in electrolysis and fuel cell modes. This information is of high relevance for providing guidelines for electrode improvements and understanding the effects of alterations observed in post-test analyses. The model has also been used to interpret the role of the cell operating mode on the LSCF destabilization mechanism. The computed distribution of vacancies within the electrode is strongly dependent on the electrode polarization. Under cathodic (anodic) current, the amount of vacancies in the perovskite is increased (decreased). Therefore, the deviation from stoichiometry tends to zero under anodic current, which drives the demixion and the precipitation of SrO. This would explain why x-ray and 2-D electron microscopy showed a formation of SrZrO3 phase much more pronounced in SOEC mode than in SOFC mode. The detrimental effect of the SrZrO3 inclusions on the conductivity of the YSZ/GDC interface was quantified by 3-D time lapse finite-element transport analyses using FIB-SEM/EDX dataset from WP5. The quantitative morphological measurements of SrZrO3 detected at the interface suggested a mild degradation during SOFC operation, though much less pronounced than in SOEC mode. The conductivity degradation due to the sole presence of the SrZrO3 inclusions predicted by finite-element simulations remained also mild (linear relative increase reaching 4% after 4700 h).
The governing equations of the continuum composite electrode models were implemented in 1-D, 2-D and 3-D cell and stack models developed by the partners. The 1-D framework for EIS simulations first enabled a finer understanding of the processes detected by EIS measurements on segmented-cells, compared to experimental sensitivity analysis. In total, 7 contributions could be identified with the help of the model: (P1) oxygen adsorption ( f<1 Hz), (P2) gas conversion on the anode side (f=3.5 Hz), (P3) charge transfer in the cathode (f=20 Hz) , (P4) gas-phase diffusion in the anode (f=120 Hz), (P5) charge transfer in the anode (f=1000 Hz) , (P6) anion transfer at the LSCF/GDC interface (f=2000 Hz) and (P7) anion transfer at the GDC/YSZ interface (f=8000 Hz). This is a significant improvement over the understanding available at the start of the project. The detection of variations in the contributions to the degradation that cause alone a decrease of the cell voltage in the range of 1% can be practically expected, using the full (or selected) frequency response (figure 25).
The continuum composite electrode models developed for analysis at the micro-scale were also coupled to CFD for thermo-electrochemical simulations of the SOLIDpower stack design (figure 26). The developed degradation laws described above were complemented with semi-empirical relationships for the degradation of the YSZ electrolyte and anode scaffold from the literature, and MIC oxidation model. The degradation measured in short-stack testing could be captured. The capability of segmented-cell tests to access local information, i.e. evolution of the spatial distribution, allowed an in-depth examination of the model predictions. The anode degradation model based on percolation theory could predict the overall decrease in performance, but differences in the spatial evolution of the degradation were observed, showcasing potential improvements for the future.
The computed temperature profiles were imported into a stack thermo-mechanical contact model. The focus was placed on the effects of stack operation history and component imperfections. Experimental setups together with a model-based parameter estimation method were used to measure the primary and secondary creep properties and strength for implementation in the stack model (figure 27). Simulations were performed with combinations of MICs with different pre-deformation to better describe the reality. The effect of pre-deformation on the uniformity of contact pressure is significant already after assembly and clear differences are observed in the evolution upon testing. The inlet of the active area is the most critical zone, which is qualitatively in line with the spatial distribution of the ohmic loss monitored during segmented-cell test. The cell probability of failure during combined steady-state operation and thermal cycling was computed, based on the 4-point strength measurements. Because of the uncertainty in the measured Weibull parameters, the computed values are provided as indicators for comparative studies, rather than threshold values. For the studied conditions, the detrimental effect of MIC imperfections on the computed cell probability of failure were found significant in operation, but negligible at room temperature. The computed risk of micro-cracking in the Ni-YSZ anode during thermal cycling is limited (50% to generate 72 cracks in the active area of a cell). A case study of the effect on the electrochemical performance based on non-destructive imaging of redox cycling suggests that the performance loss will be low. Direct and accurate validation of the prediction of stack mechanical failures remains challenging and may be advanced by follow-up studies. Establishing direct relationships between contact pressure history and electrical contact will be of relevance understanding further links between thermo-electrochemical and mechanical aspects.
Summarized, current understanding leads to the belief that continuous stack performance degradation with long term operation stems from the incremental addition of cumulating processes, without these necessarily leading to end-of-life (on the contrary, rather flattening out with prolonged operating time), and that accidental, sudden, or eventual stack failure is likely rather due to thermomechanical issues, provoked either by cycling or by progressive contact loss between the constituting layers.
Communication and Dissemination (WP8)
The communication and dissemination activities of the ENDURANCE project were focused on the following targets:
• to contribute to the public acceptance of fuel cells as a reliable and appealing power source by working on public awareness
• to create a continuous and mutually profitable interaction with other EU projects in the field
• to attend fairs and other public events in order to contribute to the exploitation of the results

The activities were coordinated by WP8 and involved the whole consortium in order to have the point of view of the industrial partners comforting the strategies of the researchers.
A serious game called "The Lost Colony" was prepared on a multilingual platform (UK, IT, BG, FR, ES, NL) to be distributed in schools targeting teen-agers. The purpose is to get them closer to the renewable power sources and more aware of fuel cells as an option for clean energy storage and secondary power source. The beta version was successfully tested during two public events. A number of national initiatives were also carried on un various countries in order to reach the possible largest audience.
Two inter-project international workshops were organized welcoming experts from all over Europe and creating the possibility to share results and achievements in order to reach an higher level of maturity for all.
The participation to international fairs dedicated to clean energy and fuel cells offered then the opportunity to meet institutions and people from all over the world and belonging to other domains still related with energy management or to specific applications. Narrowing the target it was possible to better understand how to shape our results and achievements in a more attractive way.

Potential Impact:
The ENDURANCE project resulted successful in matching the initially planned objectives:
1) to increase the awareness on degradation phenomena and the risk they represent for the life extension of the stack and its components
2) to compile a number of documents useful for further developments and projects: Degradation Modes and Effects Analysis; Handbook of test protocols and procedures; proceedings on Degradation Mechanisms in Solid oxide cells and systems
3) to suggest and validate materials, cells and manufacturing processes for more reliable cells and stacks (i.e. denser diffusion barrier layer, more stable and cycle resistant sealant, partial anode redox process against Ni coarsening)
4) to enhance existing models and to introduce new predictive models related to thermo-mechanical, electrochemical and physical behavior at the cell and at the stack level
5) to set up a sensitive tool (Differential Resistance Analysis) for the interpretation of the stack (or cell) output (i-V curves) in order to investigate since the early stages of operation the performance level and the degradation rate.
The above mentioned achievements contribute to increase the reliability of the SOFC stacks in terms of performances, durability and safety. The technical solutions have been implemented by the manufacturing company into real stacks actually under operation in reformed fuel in order to prove their efficiency and suitability for real applications. Such improvements were obtained without affecting the manufacturing cost which means, in terms of Cost/hours of operation, an effective positive impact on the suitability for the market.
The usage of the DRA to check the stability of the cell is an excellent quality test for the stack components and its periodical application under operation allows a constant monitoring of the stack state of health. This is one of the most advanced and simple to apply EWOS making possible to detect and counter act in due time possible deviations from normality of the stack components behavior.
The impacts of such achievements can be distinguished into two main domains:
1) Technical improvements and opening of new fields of investigation,
2) Manufacturing of reliable stacks for the market.
The first point has given rise to initiatives corresponding to the invitation of the various members of the consortium to participate to new FCH_JU and FET projects under submission. The investigation methods, the technical solutions and, in general, the scientific achievements presented in 16 papers and more than 40 participation to conferences and workshops, have started a positive trend motivated to continue and to improve the line traced during the ENDURANCE project.
The second point is obviously going to be observed in the next future where natural gas and then bio- and/or syn-gas will be used as fuel. This smooth transition is needed to get the society ready to accept fuel cell based power generator as the best alternative to the classical and pollutant fossil fuel based devices. In order to speed up this process for the sake of the environment conservation a number of activities (at the national and international level) were activated.
Hereafter a detailed list refers about both the exploitation (e.g. attending to fairs and public events) and the dissemination (e.g. publications, website, large audience initiatives) activities.
Introduction and main achievements
The communication and dissemination activities of the ENDURANCE project were focused on the following targets:
• to contribute to the public acceptance of fuel cells as a reliable and appealing power source by working on public awareness
• to create a continuous and mutually profitable interaction with other EU projects in the field
• to attend fairs and other public events in order to contribute to the exploitation of the results

The activities were coordinated by WP8 and involved the whole consortium in order to have the point of view of the industrial partners comforting the strategies of the researchers.
A serious game called "The Lost Colony" was prepared on a multilingual platform (UK, IT, BG, FR, ES, NL) to be distributed in schools targeting teen-agers. The purpose is to get them closer to the renewable power sources and more aware of fuel cells as an option for clean energy storage and secondary power source. The beta version was successfully tested during two public events. A number of national initiatives were also carried on un various countries in order to reach the possible largest audience.
Two inter-project international workshops were organized welcoming experts from all over Europe and creating the possibility to share results and achievements in order to reach an higher level of maturity for all.
The participation to international fairs dedicated to clean energy and fuel cells offered then the opportunity to meet institutions and people from all over the world and belonging to other domains still related with energy management or to specific applications. Narrowing the target it was possible to better understand how to shape our results and achievements in a more attractive way.
General view
This specific WP was constructed on two pillars:
• Internal knowledge management and communication within the consortium aiming to gather, organize and safely store consortium results, achievements and virtual products (testing procedures, protocols etc., experimental results) and to ensure friendly access to the accumulated intellectual resources. The main tool for its implementation was the Internal part of the project web site - a virtual “hub” for project information and pooling of partner resources. Thus it ensured vast dissemination and solid management of the knowledge, expertise, and experience acquired during the project. The Internal web site was continuously updated during the duration of the project. Some of its components (Handbook of Experiments, part of Knowledge pool) were implemented in the final (external) version of the web site.
• External targeted (Industry, Scientific Community, General Public) dissemination/exploitation of the project results and achievements shortening the "lab-to-public" pathway applying new and original tools. For increase of the dissemination impact specific deliverables (described below) for dissemination after the termination of the project were planned and realized, including free access final version of the project web site which will operate at least 3 years more.
The implementation of WP 8 targets was distributed in three Tasks (Consortium Communication; Exploitation; Dissemination) covering the whole project duration (m1–m36).
Communication and knowledge management
The internal web site was the main tool for building the communication backbone needed for the project implementation (Figure 1). It was designed and served as a Web Library & Data Bank where all project-related information was accumulated. The FileMaker family of software products was used by Partner 6 (IEES) to create three Databases:
• Book of Samples (BoS): it ensured the Coordinator and every Partner access to all the research and logistic information related to the project at any time and from anywhere. The BoS has catalogued each sample from the moment of its creation, following its shipment, recording all testing and characterization results, allowing ENDURANCE team members to view the status of all samples or track a specific one. The BoS concept was thoroughly discussed during the negotiation period and its first version was introduced during the Kick off meeting. The final operating (beta) version went live in m 3. During the project implementation 92 samples have been monitored, staring from their preparation and passing through all the stages of their testing and characterization in different laboratories (including data collection and analysis), till their final storage. Detailed information about BoS has been presented in Milestone MS2 “Setting up of BoS, HoE”.
• Hanbook of Experiments (HoE): the initial version was transformed into D 4.1 titled "Handbook of Tests Procedures and Protocols". This database collected and stored testing procedures and protocols that have already been implemented in tasks relevant to ENDURANCE's as well as such that have been developed during its implementation. The HoE, (developed by Partner 6) assigned each testing protocol a unique ID according to a unique serial number and the protocol type:
o Existing harmonized tests: Four harmonized testing procedures for SOFC (Test Modules) developed under the Research & Training Network (RTN) FCTESTNET (Fuel Cells Testing & Standardization NETwork) and published by JRS have been uploaded.
o Existing experimental procedures: (from publications and other projects), about 30 procedures have been uploaded (some of them from previous and running projects as REAL SOFC, DESIGN, ADEL, SCOTAS, STACKTEST, SOCTESQA);
o ENDURANCE experimental procedures for: Testing (7); Characterization (9) and Manufacturing (2), developed during the project implementation. More detailed information about HoE operating during the project implementation are given in Milestone MS2 “Setting up of BoS, HoE”.
Currently in the field of PEM fuel cells for automotive applications an important JRC-led crosscutting activity ("Harmonization of Test Protocols") is running between several FCH JU projects. Recently similar initiative has been initiated for stationary applications where the protocols for PEMFCs are in more advanced stage. In the field of SOFC additional efforts are needed for accumulation of an information source as a base for the development of harmonized procedures. Accumulated previous experience is also needed for accelerating tests where non-reversible degradation caused by the selected performance conditions should be avoided. One of ENDURANCE’s objectives in respect to exploitation and dissemination was to accumulate existing information and to store developed project procedures for testing and characterization of SOFC. Thus for the FCH community an open access digital deliverable “Handbook of testing procedures and protocols” (D.4.1) was recorded on dissemination disks and uploaded in the Final version of the web site. This approach of ENDURANCE may be multiplied in other projects and thus step by step Data bank can be structured and uploaded (for instance in JRC, or FCH JU web sites).
• Knowledge Pool (KP): this data base was introduced in the beginning of the project for pooling the intellectual resources and enhancement the ENDURANCE team knowledge. It stored publications and materials relevant to the project tasks and goals. The KP database development and implementation was not initially planned; however, the need to gather and store state of the art knowledge and background on ENDURANCE-related topics (papers, references, research) arose during the initial stages of the project. Thus, the KP was developed by Partner 6 as a virtual "pool" of resources both fed and used by all project members. More than 150 documents were uploaded. Some of them which concern testing procedures were included in D 4.1.
In addition to the three data base useful and necessary for the project management information was internally stored and disseminated: Meeting documents, presentations (including the presentations of the 2 Workshops organized in the frames of the project) etc.
• Publication Intent was very useful tool for implementation of the GA requirements for publication, according to which all partners should have been notified of any intent of project results publication within a reasonable time frame. Partners stated their intent for publication by filling out a form, which automatically notified the team members by email as well as stored all publication intents for future reference. ENDURANCE project highly recommends the “Publication Intent” tool for the management of other projects.
Exploitation and Dissemination
The Exploitation and Dissemination activities were structured in parallel, based on the development of common tools.
The main Exploitation and Dissemination goals could be summarized as:
• Engage – get input/feedback from the community;
• Promote – ‘sell’ ENDURANCE outputs and results;
• Inform – educate the community;
• Raise awareness – let others know about ENDURANCE and Hydrogen Technologies.
The development of the exploitation plan and strategy were closely related to the introduction of forward and backward communication strategy with stakeholders and research groups. The project is finalized with 23 published papers and 47 participations on scientific events which are listed in the Final version of the web site. It is expected that the number of papers will further increase. Part of the final deliverables was designed in accordance with this strategy (HoE, Workshop proceedings, Serious game, Final External Web site):
• External web site: One of the strong exploitation and dissemination tools developed in the beginning of the project was the external web site, which has been structured to ensure illustrative presentation of project implementation, advancements and other useful for the stakeholders materials (Figure 2). In addition to the description of ENDURANCE philosophy, objectives, challenges, it was ensuring current information about the progress in the context of improvement of fuel cell performance through project achievements. Separate information sections were directed towards the main project target groups - scientific, industrial and public.
Other tools for establishment of the exploitation and dissemination plan have been also applied: direct contacts with the stakeholders, scientific society and potential customers via planned and dedicated participation on fairs, (Hannover Messe, Sofia EXPO, f-cell Stuttgart), conferences, organization of Workshops in collaboration with other FCH JU projects, other dissemination events (Green Radio in Genoa; Festival of Science, Children’s scientific happenings, FCH JU Information days in Bulgaria, Smart Specialization in the domain of Fuel Cells and Hydrogen technology) (Figs. 3, 4). Those events have been also introduced in the INFO CENTER of the Web site.
• Final version of the External web site: www.DURABLEPOWER.eu

In every project the strongest and valuable achievements are realized in the end of the project and described in papers, reported on conferences, or patented after its termination. Due to the experience of the partners in previous projects, a new approach for increasing the impact of the project achievements has been proposed. One of the project deliverables was the Final version of ENDURANCE web site which, after an appropriate re-design, will continue to operate at least 3 years after the termination of the project (Figure 5). The most valuable information about the project and its activities and results is included. Three of the final deliverables were prepared and uploaded in a form applicable for further dissemination and exploitation of project achievements and for long term increase of the public awareness:

o “Handbook of Testing Procedures and Protocols” (D.4.1.): it includes collection of test procedures and protocols – gathered, developed and/or applied within ENDURANCE. The deliverable is described in Section 1 of this report. It was prepared due to collaboration between WP4 and WP8. CDs were also recorded for dissemination on scientific events.

o Proceedings of the Workshop “Degradation Mechanisms in Solid Oxide Cells and Systems" (D.8.1.)
Following the recommendations of FCH JU for increase of the networking between projects working is similar areas, ENDURANCE team was the initiator of the planned Workshop “Degradation Mechanisms in Solid Oxide Cells and Systems" which was co-organized together with four FCH JU projects DIAMOND, SOPHIA, SOCTESQA and Eco and held in Barcelona, Spain (February 17, 2017). It gathered high-level researchers and industrial representatives involved in the development and market introduction of solid oxide cell systems.
The aim of the workshop was to ensure a platform for professional discussions and exchange of expertise in one of the most challenging issues in the commercialization of SOFC: degradation and durability. Different aspects covering cell components, seals and interconnectors, as well as stacks and systems were discussed. Diverse approaches were presented: from modeling and prediction of degradation, through experimental confirmation and observations to diagnostic tools for early detection and mitigation strategies. The Workshop Proceedings (about 350 pages) include an abstract and full presentation of the talks. There is an open access via the Final version of ENDURANCE web site, as well as CDs for dissemination during participation of ENDURANCE members on different scientific events.

o Serious Game (D.8.2 and D. 8.3)
The dissemination towards the public has more general goal – to increase the public awareness of the fuel cells and hydrogen importance for sustainable and secure energy via the project achievements. Young people - pupils, students are an important targeted group. In this direction a special deliverable product has been planned with longer term impact – a clip and a video game. After the start of the project they were united in a more sophisticated product – the so called Serious Game, named “The Lost Colony” (Figure 6). Its first version was successfully promoted on the Festival of Science in Genoa.
The goal of the Serious Game is to transfer knowledge related to fuel cells. Following the spiript of the game, the players jump in the year 2115 when the Earth’s energy sources are nearly depleted; therefore, humans have decided to colonize Mars where energy is produced using fuel cells - devices that use Hydrogen and Oxygen. To explore the Martian colony the gamers have to complete a series of didactic mini-games present in the 3D environment. At the end they should have learned the basics on how fuel cells and electrolyzers work and can be used.
The game in its advanced version may use the following languages: English, Italian, Spanish, Dutch and Bulgarian. It can be downloaded from the final project web site. CDs are recorded for further dissemination.
Higher levels of the Serious Game can be further developed where the players can design original application solutions. Competitions can be organized in different EU countries and regions, starting with the FCH JU members. “The Lost Colony” can be further developed in FCH JU cross-cutting activity and transformed in attractive dissemination tool with long term impact.

List of Websites:
www.durablepower.eu