Final Report Summary - INSIDE (In-situ Diagnostics in Water Electrolyzers)
Executive Summary:
The INSIDE project is focused on the integration of diagnostic tools in electrolysis of water for the monitoring of performance and early detection of malfunction. Various electrolyser technologies are taken into account:
• Alkaline electrolysis with liquid electrolyte as a very established and robust technology
• Electrolysis with proton exchange membranes as very dynamic technology
• Electrolysis with anion exchange membranes as new and emerging technology that is combining aspects of both other technologies.
The diagnostics tools are based on technology with printed circuit boards, which has been successfully developed and used in fuel cell research.
The main objectives of the project INSIDE were:
• Design and construction of in-situ diagnostic tools for three electrolyser types (PEMWE, AWE, AEMWE)
• Integration of the diagnostic tools into industrially relevant electrolyser types
• Test of the diagnostic tools in stack or short stack configuration
• Definition of common test protocols for performance testing and proof of the protocols
• Identification of failures and degradation, and correlation to microscopic findings from spectroscopic and microscopic investigations
• Evaluated protocols for accelerated durability tests
Two prototypes for the PEMWE technology were developed and tested. They are both using an original graphitic bipolar plate that was modified and directly attached. In particular, the second, enhanced prototype is embedded between graphitic bipolar plates, and thus mimicking a regular BPP.
The works were hindered in the beginning by the unavailability of industrial PEMWE hardware with the option to work with the corresponding bipolar plates.
The design process for alkaline electrolysis had to face the chemistry of this technology. As the 30% KOH lye at elevated temperatures is very corrosive to many materials, the direct contact to copper and most duroplast polymers has to be strictly avoided. Although the standard material for printed circuit boards can be surprisingly stable, when it´s plain face is exposed to KOH lye, cut edges are quickly decomposed – lye also has the property to wet laminated boards internally and thus creep into a printed circuit board and corrode the circuitry there. After several fruitless attempts, a modular design was developed, where sensor modules are embedded in a rigid frame of PEEK, a stable polymer, but otherwise useless for building PCBs. An additional aspect of this design is the modularity itself, which allows for an easy upscaling and thus the application in large systems. The system was produced and “dry tested”, but could not be evaluated in electrolysis mode within the project duration.
The design for the AEMWE technology was quite straightforward, with the exemption of the need to modify parts of the attached electrolyser cell, too. The porous metallic gas / lye diffusion sheet contacting membrane/electrode and bipolar plate introduces a signal cross-talk to the lateral electrical conductivity, which could be overcome by the additional segmentation of this contact sheet.
As overarching activity, the data acquisition software has been reprogrammed for a more flexible adaptation to various requirements.
The activities were accompanied by ex-site investigations on pristine and aged and, where necessary, on model materials from the electrolysers. Experimental tools for in situ studies of electrochemical systems with Near-Ambient Pressure X-Ray Photoelectron Spectroscopy (NAP-XPS) have been successfully developed and approved using model systems. A correlation between the spectromicroscopically shown deposition of a Nickel contamination in the membrane of a PEMWE cell and the observable local deactivation of this electrolyser cell could be established.
Common testing protocols were formulated based on a literature survey and on experience from previous activities with fuel cells. They were shared with the JRC to support their harmonisation activity on electrolyser testing. The evaluation and definition of accelerated stress tests was not possible.
Two workshops were organised during the project duration, and the f-cell award o 2018 (2nd place) was grated to the beneficiaries directly involved for the development of the modular segmented bipolar plate for alkaline electrolysis
Project Context and Objectives:
Electrolysis of water is an important technology for the production of high purity hydrogen needed for numerous chemical processing applications. In the future, however, hydrogen is forecast to become important as an energy carrier that is used for storage of excess electrical energy produced by renewable sources such as solar, or wind. In fact, current energy policy requires the replacement of energy sources based on fossil energy carriers by renewable energy carriers. Reasons are the increasing depletion of sources of coal and oil, the need of significant reduction of emission of greenhouse gases (e.g. CO2), and the intention to reduce economic (and political) dependency and risks due to imports of fossil energy carriers. Moreover, nuclear power is becoming less acceptable to the general public, e.g. in Germany. However, there is a substantial discrepancy between the temporal availability of renewable energy (e.g. differences between day/night or summer/winter especially for solar) and the actual needs. Therefore, the opportunity to temporarily store excess energy is of great interest.
Consequently, there is an increasing interest in development, improvement and tailoring of electrolysers for the efficient production of hydrogen that can be used in combination with fuel cells in a wide range of applications such as mobile, automotive, aerospace, and stationary. For all these purposes the electrolysers need to be controllable, reliable, and sustainable. Due to the stable political situation Europe is a predestined place to pursue this ambitious goal. In addition, there is a drive to reduce the cost of hydrogen. The cost of producing hydrogen is closely linked to the price of electricity used to power the electrolyser.
According to Turner1 the improvement of electrolyser efficiency is only one way to lower hydrogen cost; major improvements are additionally needed in electrolyser lifetime, maintenance requirements and capital cost of the systems. From the thermodynamic point of view H2 cost can be also produced by performing electrolysis at elevated temperatures in order to minimise the electrical energy needed.
On the market there are two prominent types of low temperature electrolysers, namely alkaline electrolysers , and proton exchange membrane (PEM) electrolysers. Additionally, there is a more recent and less known type of anion-exchange membrane (AEM) electrolysers . The general occurring processes are the formation of hydrogen at the cathode and the formation of oxygen at the anode and the transport of ions between the electrodes which are separated by a diaphragm or the mentioned ion exchange membranes. The specific electrode reactions and the type of ions responsible for the internal charge transport depend on the electrolyte (alkaline or acidic). Each electrolyser technology exhibits specific properties with specific obstacles to most efficient operation and long lifetimes.
PEM electrolysers operate under acidic conditions using a proton exchange polymer electrolyte membrane (e.g. Nafion®) for the transport of protons produced from dissociation of water. In such a case only expensive noble-metals, typically Pt, are suitable catalysts, a serious drawback of this approach. Yet, PEM electrolysis benefits from the possibility of producing pressurised (200 bar) H2 (ready for transport) with little amount of extra energy.
The use of non-noble metals as catalyst (e.g. Ni) is the main advantage of alkaline electrolysers; the produced OH- ions migrate through an alkaline electrolyte (e.g. KOH) aqeous solution. The cathode and the anode side are separated by a dedicated diaphragm. Particular difficulties that need to be overcome to increase the performance are: resistances due to bubbles formation at the electrodes, reaction activation energy, and ionic transfer.
AEM electrolysers use solid alkaline ion exchange membranes as electrolyte and therefore allow the utilisation of non-noble metal catalysts, such as in liquid alkaline electrolysis. However, the ionic conduction losses of AEM are significantly higher and the membrane stability is significantly lower than in state-of-the-art PEMs. Thus, the improvement of anion exchange membranes and ionomers is a major challenge with respect to this technology.
In order to successfully manage all the possible improvements of these technologies it is necessary to monitor the performance and operation failures in-situ as well as to develop test protocols to make the long-term operation predictable. The development of such an in-situ monitoring tool and validation of test protocols are the main objectives of the R&D project INSIDE
It pursues the development of diagnostics tools for three independent technologies for water electrolysis (PEM based, alkaline, and AEM based) with individual properties and individual challenges. This development will provide operational data from inside the electrolyser systems. Conventional current-voltage characteristics, impedances and materials analytics before and after operation are not able to provide the same type and quality of data as an in-situ tool for on-line diagnostics.
This technology monitors the local current densities, which are directly linked to the local electrochemical turnover. The principle has been successfully used in the research on polymer electrolyte fuel cells for materials development, for optimisation of operation condition, and for prevention of undesired events, such as local flooding of parts of a fuel cell, which can be directly visualised during operation.
The strategic aim is to use these diagnostics tools for online monitoring with the possibility for online adaptation of operational parameters, and for the prevention of hazardous operation modes while optimising the overall performance. At the end of this project, there will be a public document with summary recommendations for the operation of present electrolysers and for the targeted development of future electrolysers – but this will not be the end of the usability of these new diagnostics tools.
In particular, online monitoring is a key technology for the application of water electrolysers in a highly fluctuating electricity market in Europe as grid stabilising highly dynamic systems, which does not dump power, but transforms excess electricity into hydrogen, a storable, transportable power source. Moreover, electrolysis of water represents a decentralised technology to provide local hydrogen refuelling stations for e.g. electromobility with the fuel cell car.
Project Results:
1. Requirements and Specifications (Work Package 2)
Proton exchange membrane water electrolysis
- The diagnostic tool covers the entire area of 76 cm².
- Connectors are in outer areas; their arrangement on the PCB has to fit into the electrolyser housing, ideally. Alternatively, the tubing and the main electrical connectors require a reconfiguration.
- The tool has 88 segments of square shape, with exception of the rim segments.
- 21 temperature sensors are covering the cell area with constant spacing.
- The PCB material itself is standard FR4 grade, as it was used for PEMFC applications.
- The PCB will replace the graphitic bipolar plate, resembling the flow structure only on the cathode side and connect to the endplate at the opposite side in first generation. In the second generation, it will replace the graphite composite bipolar plate between cells in the middle of a stack and thus resemble both cathode and anode side flow structure. Also, the thickness will be similar, of >4 mm, which is considerably thicker than conventional PCBs for PEMFC applications.
- Current bearing segments are coated with gold, the thickness is higher than for conventional PEMFC application and the 1st AEMWE prototype.
- Two 34972A type data loggers should be used
Alkaline water electrolysis
- The diagnostic tool covers the entire area with a diameter of 500 mm.
- Temperature sensors are arranged in a dense symmetric 8-pointed star pattern with intermediate positions.
- The entire PCB area, which is in not metal coated, has to be protected with a polymer coating that is stable in the alkaline ambience; in particular, the edges of internal openings are protected.
- The tool requires preconditioning in 25% KOH before going into operation.
- Either six parallel data loggers (preferred) or one larger data logger (if parallel read-out not feasible) will be used.
- Housing with ATEX zone 1 (or higher) certification will cover the data acquisition unit(s) to reduce the distance to the electrolyser stack.
Anion exchange membrane water electrolysis
- The diagnostic tool covers the entire circular area of 64 cm².
- Connectors are in outer areas.
- The tool has 59 segments, most of rectangular shape, with exception of the rim segments. Regions close the edges have more segments per area than central segments.
- 16 temperature sensors are covering the cell area with almost constant spacing.
- The current bearing segments with copper core conductors inside the PCB are completely covered with gold, similarly o PEMFC diagnostics boards.
- The tool requires preconditioning in 1% KOH to remove PCB (productions traces) before going into operation.
- Two data loggers, with a total of 80 channels should be used
2. Design of the in-situ diagnostic tools (Work Package 3)
2.1. Design and construction of the Segmented Cell for PEM electrolysers (M3-M36)
The design of the diagnostic tool for proton exchange membrane based electrolysis was planned as the technology route with least expected difficulties. The difficulties for development, however, turned out to be of organisational nature.
After the withdrawal of ITM Power Ltd., Sheffield, UK and H-TEC Systems, Lübeck, Germany from the project consortium before the start of the project, attempts were made to purchase the necessary hardware for the PEMWE route from
- ITM Power, Sheffield, UK
- H-TEC Systems, Lübeck, Germany
- Hydrogenics, Gladbeck, Germany
- Greenlight Innovation, Canada
- IRD Fuel Cells, Odense, Denmark (now named EWII Fuel Cells)
After several negotiations with all possible suppliers, an electrolyser system of the type “Elze E1050” was purchased from EWII Fuel Cells of Denmark and delivered in 2017. The cells of this system are circular, and of an area of 76.5 cm². The cell is asymmetrical, with a carbon based GDL (cathode) and titanium felt (anode).It had been decided to design a segmented PCB that connects endplate and cell cathode GDL in the 1st generation. It replaces the graphite composite bipolar plate, but does not resemble the anode side flow field (see Figure 2), which would add complexity and thickness to the prototype.
First Prototype for industry hardware
The prototype as described in the second periodic report was manufactured and tested in summer 2017. The design was carried out as a PCB with standard FR4 materials and a graphitic top plate with the cathode flow field of the regular BPP, in order to warranty the conductivity between the PCB and the graphitic plate a conductive glue was used. The rear face is contacting the regular metal contact plate of the EWII electrolyser stack with an active area of 76.6 cm2. (The bipolar plate of the stack consists of a graphitic and a metallic part.) The segmentation follows the rectangular shape of the flow plate with 44 segments, (Figure 2) fitting 2 segments into each of the full rectangles. In order to measure the temperature distributions 21 temperature sensors were homogeneously distributed
The first integration and evaluation was carried out in a small cell with original stack hardware (end plates, bipolar plates, MEAs...) out of the system. This was to facilitate fault finding and assembly, for the space in the commercial system is limited (see Figure 3) The segmented diagnostics PCB, however, was designed to fit in, regarding its area and cable connectors. The PCB itself is embedded within two modified graphitic bipolar plates, and in consequence, the stack assembly follows the procedure identical to that without PCB.
The diagnostic board was operating, but points of optimisation could be identified:
• The internal contact between the graphitic component and the PCB component was carried out with a conductive silicone based material with some elasticity. The homogeneity of this contacting could not be realised homogeneously enough to ensure a homogeneous electrical contact. The result was an insufficiently high internal resistance – at some spots. In addition this construction principle prevented a perfect sealing, and in consequence the pressure stability of this stack was lacking.
• One segment was also not operating in this prototype
• The local resolution capability and the signal cross talk between segments were meeting the expectations.
Enhanced Prototype for industrial hardware
The second prototype was designed and manufactured with a hardening conductive glue for the internal connection to overcome the resistance and inhomogeneity problems. The PCB had to be laid out new, with some modifications of the circuitry. The resulting PCB is more stable and fits better into the stack. Segmentation, number and placing of the temperature sensors are unchanged, and the board is embedded between two modified graphitic plates. The interface to the neighbouring cell is completely identical to that of a regular graphitic bipolar plate, and the anodic flow field plate described earlier does not need to be mimicked in this way. The diagnostic plate can assume principally any position in a stack.
The data acquisition also did require modifications. However, the software was upgraded in the meantime to be compatible to either a Keysight 34980A system or two Keysight 34972 systems operating in parallel.
2.2. Design and construction of the Segmented Cell for alkaline electrolysers (M3-M36)
The design of the diagnostic tool for the involved alkaline water electrolysis holds the biggest challenges amongst the three electrolysis technologies in this project. Reasons are the chemical ambience, where the system has to subsist, a heavy 30% KOH lye under elevated temperatures of 65°C, the requirement to seal this lye with a rubber frame on a smooth surface of the board, and the sheer size of the required system with a diameter of 500 mm (active area), which required a compromise between segment size and total number of segments. Finally, the data acquisition has be spatially close to the stack to avoid a weakening of the parallel analogue signals.
The first attempts to the diagnostic tool had been unsuccessful, after preliminarily promising tests showed to be misleading – the seeming chemical stability of the PCB material had been falsified as machined/cut parts of the intended PCB material were exposed to the alkaline ambience.A redesign from scratch was done in this reporting period.
Design strategies that had been considered and discarded during this process were:
- PCB with protective coatings (at the edges), e.g. using epoxy based pottant materials, as common for protection of electronics
- PCBs embedded in polymer resin casts
- PCBs made out of other base materials than the common FR4 type, e.g. Teflon™
Some chemically promising concepts had to be discarded because of the lacking mechanical stability.
The final solution, which has been carried out to the end, comprises of a frame of Polyetheretherketone (PEEK), which is well known and tested as chemically stable in alkaline conditions. While the construction of a PCB completely made out of PEEK as base material is not possible, the integration of an array of sensors in a PEEK frame was so. Each segment consists of a Nickel contact segment (as can easily be seen in (Figure 6) and each underlying resistor bridge module (Figure 8) that is inserted in the PEEK frame and directly mounted to the current collecting Nickel back plate (not visible). Each module is connected via a small circuitry bridge. The interior of the segmented board is sealed against the alkaline ambience. There are no parts of circuitry or other PCB elements in contact with the KOH lye. The manifolds are completely carried out in PEEK. Such, the active areas of the electrolyser cells are in contact with the segmented Nickel, while the manifolds are in contact with PEEK.
Another beneficial aspect of this design is the scalability. The construction of a larger board would require the production of a larger amount of sensor modules, indisputably, and a correspondingly larger frame, but parts of the works would be possible as serial production. The connection of the segments requires an individual circuitry, but does not require additional electronic components. The wire bonding of the modules to the circuitry, however, would require automated technology. The used resistors are more sensitive and scaled to be able to compensate for a larger lateral cross-conductivity of e.g. electrodes than common in fuel cells, and the used temperature sensors are a miniaturised development.
The total number of data recording channels is 268. Data recording is carried out with a Keysight 34980A data logger, equipped with four scanner modules of 70 channels each, totalling to 280 channels. A total of eight pole plug connectors is used for the voltage signals and the temperatures of the ~1800 cm² cell.
The prototype was „dry tested“ for principal electrical and sensorical functionality by measuring the temperature signal and the reaction to local heating (by a hot fan), and by applying a sample current to each segment and recording the voltage response.
The size of the active area and the resulting high number of segments in the diagnostics bipolar plate unsurprisingly require a large number of data cables. Each signal is transduced in the form of a (low) voltage, and thus the loss of signal quality can be considerable, if the cables transporting hundreds of signals are too long. In consequence, the data acquisition unit has to be placed close to the electrolyser itself.
The safety certifications of the testing site, however, do not allow equipment that is not following ATEX specifications. Safety housing for the data acquisition had to be designed and constructed accordingly.
The modular design has the risk of internal leakage, if the individual modules are not thoroughly sealed, as the first tests reveal. A procedure to test for this property is advised before integration into a stack. A design update to mitigate for this issue is the use of a glue, preferably based on butyl rubber, for the final montage of the modules instead of a O ring seal. This was however necessary for the prototype evaluation. The operational test could not be carried out in the project´s duration. It is however intended nonetheless by the beneficiaries.
The modular concept was awarded with the f-cell award 2018 (2nd place).
2.3. Design and construction of the Segmented Cell for AEM electrolysers (M3-M24)
The design of the diagnostic tool for the AEMWE route is mostly dominated by geometrical aspects. The relatively mild alkaline chemical conditions did not show problematic in the preliminary tests. A challenge to the material is the self-pressurising capability of this technology, which leads to a differential pressure between anode and cathode of up to 30 bar.
The design was finalised after the following technical meetings between HELIO and DLR:
- Stuttgart, Germany, August 2015 (combined with a general meeting between DLR and Heliocentris representatives with the main issue of outlining the formal proceedings necessary after the Bankruptcy of Acta S.p.A.)
- Esslingen, Germany, November 2015 (linked to progress meeting)
- Videoconference, December 2015
- Videoconference, March 2016
The design resembles a regular bipolar plate with additional outside areas for the electrical connectors. Its active area follows the circular shape of the 64 cm² cell. The design of the segmentation pattern has been iteratively adapted to the expected needs, regarding a compromise between resolution, addressable segment area and the focus on specific areas of interest in the cell design. The number of active segments of various sizes for current density monitoring and temperature sensors requires two Keysight, type 34972 data logging units. The following aspects were observed at the first prototype:
- The initially designed PCB thickness induced some stability problems; also the stability against a differential pressure is expected relatively poor. The use of another frame design that had been used for earlier developments allows a higher thickness for the 2nd prototype.
- The contact resistance between current collectors and PCB was observed relatively high – the contact to carbon based GDLs as in PEM fuel cells is more easily established. This may be mitigated for with a higher thickness of the gold coating.
- There is a very high signal cross-talk, that requires a segmentation of the porous contact sheet, too.
The metallic foam, however, could not be segmented with mechanical methods. Milling and machining tests led to smeared out, but still attached material. In particular the experiments to only thin instead of separating the foam showed, that most of the material is compressed by this procedure and thus remains in the sheet. A more successful approach was made by wire spark eroding the sheet. A complete set of segments could be cut out of a regular sheet and glued to a segment printed circuit board. This approach was applicable, but still lacking. The height of the attached foam segments could not be maintained equally This led to an inhomogeneous compression of the foam during assembly and an inhomogeneous contact resistance and local measurement artefacts.
As a final approach, the contact sheet was laser cut, while maintaining uncut corners. This allows to mount the entire contact sheet in one single piece and to keep the distances between the segments constant.Microscopic investigations show, that there are molten remains of the lasered material, but only at the direct interface, while the pore structure beneath is unharmed
3. Polymer Electrolyte Membrane Electrolysers (Work Package 4)
After the earlier works in the work package were mainly on laboratory scale electrolyser cels and the related small segmented PCBs, the work reported in this period is on industrial electrolyser hardware.
The withdrawal of ITM Power Ltd., Sheffield, UK and H-TEC Systems, Lübeck, Germany from the project consortium before the start of the project led to the strategy for the works in the PEMWE route was to purchase a system and modify it according to the project targets. The companies contacted for this purpose are:
- ITM Power, Sheffield, UK
- H-TEC Systems, Lübeck, Germany
- Hydrogenics, Gladbeck, Germany
- Greenlight Innovation, Canada
- IRD Fuel Cells, Odense, Denmark (now named EWII Fuel Cells)
An electrolyser setup, Elze E1050 could be purchased from EWII Fuel Cells. It takes up 5.5 kW, producing 1.0 normal m³/h with a heat dissipation of about 1300 W. It has the capability to self-pressurise to 50 bars. The active area is 76.5 cm², with a circular shape. The required water purity is defined by a conductivity of <2.0 µS/cm (ASTM standard type II). The nominal maximum water temperature at the inlet is 70°C. The nominal lifetime is give as >2,500 hours
The Elze E1050 system is designed with the sole purpose of hydrogen supply, to small or medium customers, such as small living quarters or specific industry with small demand. As a consequence, the possibilities for controlling operational properties are limited. In order to follow the goals as far as possible, and to enable the system to operate under the required experimental conditions at a later point, when the laboratory safety confinements may have been solved, the following upgrades were necessary.
- Water supply bypass: The first attempt to operate a short stack with its reduced flow-through capacity was to establish a bypass that would allow the total water flow rate, monitored by the system, to remain in the safe range. It was observed, that the flow resistance of a bypass was too low. The main flow would be travelling through the bypass, and the short stack would be undersupplied. This behaviour could also be seen by means of the segmented bipolar plate. And the lack of wetting was obvious upon disassembly.
- Pressure booster pump and flow rate meter: As the EWII stack, which is designed for as-of-present unmatched self-pressurising capability up to 50 bar has a very dense structure and requires a pressure in the water supply to establish a sufficient flow. A pressure booster pump was inserted into the water flow circuit directly before the short stack. This pump increases the pressure to 3.5 bar, and can establish a flow through the stack, the control possibilities, however, are limited.
- Controllable pressurising dosing pump: As the booster pump was considered only an improvised solution, and as variable flow rates through the stack had been identified as interesting parameter, dosing pumps that can establish pressure and control flow were purchased.
- Additional heating: The Elze E1050 system does not support preheating of the water, as the full stack operation leads to balanced heating of the entire water circuit. For a short stack, the produced heat is not sufficient, and as experiments with elevated temperatures as stressor were planned, two controlled heating circuits, for heating water supply and stack individually, were added.
- Programmable controllable power supply that allows ramps and quick fluctuations, as the use of fluctuating renewable energies is envisioned for electrolysers in Europe and in operating and testing protocols, and as the present power supply lacks the relevant control properties.
Due to the lacking control possibilities of the Elze E1050 system for operating testing protocols, only some individual stress tests were performed.
As long-term experiments were not possible because of the limitations of the backup laboratory, the main focus was on elevated temperatures and operation with insufficient water supply. The experiments were carried out in a three cell short stack.
Applied stressors:
- Water temperature of 70°C (limited by the possibilities of the pumping system(s)), und not surpassing the nominal limit.
- The stack flow rate was limited by adjusting the bypass valve. Measured flow rate ~0.5 L/min. The insufficient water supply could be monitored by the locally reduced activity of the electrolyser via segmented bipolar plate.
The first prototype was tested in separate test bench, but the second prototype was integrated into the Elze E1050 system directly, as the small test bench that was used for the earlier works was not available and the application in industrial environment was one of the specific goals for the use of segmented bipolar plates. Artificial water starvation was used to evaluate general function, signal cross talk between segments and other issues. The signal cross talk was found acceptable in this range, as it can be estimated to maximally influencing the next neighbouring segment, but none further.
The 2nd prototype was used for monitoring operation under unusual conditions. In particular, the scenario of water starvations was used.
As the electrolyser is equipped several safety mechanisms, monitoring current, voltage and flow to be in a certain boundary, the application of unusual conditions was limited. nThe following scenario shows a startup of the electrolyser, with a very closely following water starvation.
As the electrolyser starts, and the water supply is sufficient to the low consumption under low load, the current density distribution evolves relatively homogeneously. After approximately one minute (Image #3), a slowly evolving hot spot at segment D2 can be seen, while the edge areas are slowly reducing their activity. This continues, while in the outlet area (bottom of the images) the activity stays for a longer while, until the water is entirely consumed in image #7. Beginning at image #6 and strikingly visible at image #7, the hotspot evolves and focuses the activity and thus, the total load of the electrolyser on coarsely one third of the area, until briefly afterwards, the voltage collapses.
The evolving water starvation can be visible by eye with some experience, and can be utilised by a software which monitors outlying values and activates countermeasures at an early stage.
The following critical conditions could be monitored
- Poisoning of the catalyst / membrane can be observed as a slow, steadily regional deactivation of a cell with a clear gradient of the current density. This behaviour was reported in the works on laboratory size PEMWE in the second reporting period.
- Even short intermediate water starvation can lead to local hot spots of the current density. The behaviour is very obvious using a tool for monitoring current density distribution, but not via voltage monitoring, where some fluctuations can have other backgrounds as well.
4. Alkaline Electrolysers (Work Package 5)
Test rig modification
During the early stages, further modifications to the test rig and test centre at NEL Notodden was performed to have full control of testing of 10 cell test stack. It was during the previous period found that it was necessary with some additional modifications to the balance of plant in order to have sufficient capability to perform transient tests of the stack. Parts that have been modified in the last period includes:
• Cooling system for lower cooling rates at low load operation
• Installation of gas analysers with higher sensitivity and better performance at low gas flow rates
• Optimised pressure and level control system.
During operation of cell stacks it has been identified that leakage of electrolyte can occur due to insufficient compression of the stack frames or misalignment during stack assembly. This is also more frequently observed during dynamic operation of the stack, or during testing under conditions outside normal operating boundaries (e.g. accelerated stress tests). Previously it has been difficult and/or hazardous to detect these leaks as it has been necessary for the operator to physically enter the test building to check for leaks. In order to improve safety and leak monitoring capability, a set of monitoring cameras have been installed in the test centre and LCD monitors mounted in the control room. This allows continuous monitoring of leaks without the need of a physical human presence in the hazardous area surrounding the test set up. An image showing an overview of the control room with the monitoring LCD screens are shown in Figure 19.
10 cell stacks which is ready for incorporation of the current distribution measurement tools being developed in the project have been constructed and tested for establishment of baseline performance numbers for comparison of stacks with the measurement tool installed. The stacks are based on NEL’s standard components, including NEL Haflex diaphragm. The electrodes used are commercially coated Ni-mesh electrodes. Images of the electrodes, frames and diaphragms are shown in Figure 20.
To prepare for testing of the 10 cell stack with the integrated segmented cell, it has been tested
in a wide range of operating conditions, including lye concentration, temperature and current loads. As part of this testing, a universal structured characterization approach for stack testing has been developed, based on a series of testing blocks which can be combined to form a complete testing campaign. The details of these testing procedures will be available in deliverable D5.1. A graphical representation of a testing campaign utilizing these blocks is given in Figure 21.
In this campaign, the testing consists of a stack conditioning/activation block, a stack characterization block and a long-term measurement block. A brief description of each block is given below:
• Conditioning step: Start-up of stack and pressurization. Investigation of stack integrity (visual check for leaks, etc.) initial gas purity, voltage and pressure drop.
• Stack characterization: Recurring unit for characterization of performance of stack. Includes polarization curves at several temperatures and pressures as well as online gas analysis of hydrogen and oxygen.
• Long term measurement: Recurring unit which can consist of one of several operating schemes: Steady state operation, batch operation (on/off), dynamic operation (solar, wind) or AST protocols (rapid shut down/start up, elevated temperature or current density).
The conditioning step is only run once per experiment while the recurring units are assembled into a series of blocks which can be repeated until the final length of the test is reached, or another end of test condition is met (normally a maximum impurity level or stack/cell voltage limit).
The NEL pressurized electrolyser stack is constructed with a rubber frame which is compressible and seals the internal compartments from the surroundings by an application of an external compression force. This force can in some operation conditions be too high for the rubber frames without causing some mechanical degradation. In normal commercial operation conditions, this phenomenon is avoided by a combination of proper control of the stack and the stack compression system. However, during stack testing in a development project like INSIDE, especially with shorter stacks, this mechanical degradation has been identified as a process which must be better controlled to avoid the influence on the measurements of the performance of other parts of the cell (e.g electrode activity and stability, dynamic operation and overall internal resistance of the stack).
In order to achieve a better understanding and ultimately to be able to control and avoid this mechanical degradation, a study of the mechanical properties of the frame at different temperatures and compression levels has been conducted and the results of this study has been used to construct a operational zones for the stack which will be followed during stack testing which involves protocols with frequent start-up/shut downs.
The task was completed by running a characterization of a full 10 cell stack. As an example of this characterization, a polarization curve is shown in Figure 23 below. The difference between the performance of the 10 cell stack tested and a commercial 100 cell stack is mainly due to different electrode catalytic coatings, but also due to temperature regulation variations in the 10 cell stack which is not as prominent in a 100 cell stack
A full size alkaline cell will be assembled & tested according to the developed protocol in Task 5.1 and exposed for accelerated degradation conditions. It is planned that stack component degradation will be analysed by academic partners within work package 7. In order to prepare for this testing, identification of the main degradation stressors has been performed and a proposed AST protocol has been constructed.
Start up / Shut down with resulting depolarisation and current reversal is one of the main degradation phenomena, mainly due to the dissolution and precipitation of Ni and Ni oxides as shown in the Pourbaix diagram for Ni – KOH system in 80 °C.
To promote this degradation phenomenon which can cause degradation of both electrodes, current collectors and diaphragm, the following cycling protocol will be used;
● Start/Stop cycling @ 10 barg, 75 °C
● Cycling interval 5minutes (5 minutes shut down, 5 minutes start up)
● Low lye circulation during off periods (to keep temperature high, but leave produced gas in the system,
● no nitrogen purge
● 2000 cycles x 3, with periodic performance evaluation every 168 hours.
The AST protocol will be run until a detectable reduction in performance is observed, the stack will be disassembled and post mortem analyses will be performed by NEL and other partners.
The segmented bipolar plate was integrated at the first position in a short stack. The stack was filled with water for preliminary test, and a leak was detected. It was in the vicinity of the connectors, and thus an internal leak at the segmented bipolar plate itself was to be expected. After disassembly and inspection, two possible faults were identified:
- A slight mismatch in the manifold openings of the segmented bipolar plate could have led to an insufficient sealing. Then, liquid could invade between the Nickel plate and the PCB/PEEK component (see Figure 36, red arrows) and leave the stack at the borders, close to the connectors, where no sealing are present between Nickel plate and PCB.
- A sealing defect in the one (or more) of the Nickel faces of the segments (see Figure 34) could allow liquid to intrude into the segmented Bipolar plate via the active area of the next cell. (see Figure 36, orange arrows)
- As the first scenario appeared more likely and second scenario would require he larger effort, including opening more than 500 screws by hand to disassemble the entire segmented bipolar plate and searching for defective O-rings, it was decided to approach the first scenario.
- The most promising way was to fill the manifolds with a polyester resin and drill out the manifolds afterwards. Experience with such a procedure was present at the beneficiary and could be done with shipping the equipment back.
- After fulfilling this procedure, the tests were unsuccessful, as still water was leaking from the edges of the PCB component in the stack, so the segmented bipolar plate was shipped back to Stuttgart.
- The problem could not be solved within the project timeframe.
5. Anion Exchange Membrane Electrolysers (Work Package 6)
After the preliminary test on 10 cm2 cell active area performed in order to better understand the system behaviour related to variations of operating conditions (temperature, H2 pressure, electrolyte concentrations, current densities), and the preliminary tests on 64 cm2 reported in Periodic Report 1, additional test were performed on higher active area cell.
Tests were performed with the following equipment: the cell was supplied by a TDK-Lamba power supply with current limit to 50 A. The electrolyte circuit is composed by a tank connected to the cell via John Guest tubes and fittings, and a chiller FALC SBF7 to control the temperature. The temperature was measured at the electrolyte inlet. Electrolyte inlet flow rate was 5,91 L/min except for electrolyte flow rate study, and it was controlled with an Iwaki Direct Drive Pump Model RD-40E24-HR1V.
For PC recording the current was manually controlled (one current step every 10 minutes) and the corresponding potentials were measured. The test conditions of the Polarization Curve (PC) were the following:
• T = 35-70 °C
• P= 1,013 - 40 bar
• Current= 0-40 Amp
• IN cell electrolyte flow = 5,91 L/min
• Potential ramp in 64 cm2 cell
The preliminary tests shows results in agreement with the ones obtained for 10 cm2 active area cells, i.e. an influence of temperature increase on cell performances (Figure 1), while at standard operating conditions pressure and electrolyte concentration have no influence on performances. Also different electrolyte inlet flow rates have poor influence on cell performances.
5.1. Test of AEM cells and stacks under accelerated conditions (M17-M28)
The goal of the second study was the study of electrolytic cell performances using 1 wt % Buffer (K2CO3/KHCO3) solution and 1 wt % KOH solution as electrolyte and try to correlate the results with the possible cell lifetime and corrosion. Tests were performed by means of an Arbin BT2000-L testing system provided with 18 measure channels; in order to obtain higher current densities channels can be connected together, to reach the maximum current density of 1 A/cm2. Each channel is controlled and operated via the MITS Pro software: steady state conditions and preconditioning time, current ramp and related cell voltage limit, current steps and dwell time for PC recordings are controlled by the software, and output parameters as cell potential and temperature are also automatically filed. Temperatures are recorded via K-type thermocouples at the electrolyte inlet.
The electrolyte is heated by an Ascon M5 temperature controller and supplied to cells with a SANSO PDH-E054 IT9-CE pump with an inlet flow rate of 9,5 L/min. Two cells with buffer and two cell with KOH were used, one operating in continuous and one operating with start/stop cycles for each electrolyte type. At the beginning of each test a PC was recorded in order to obtain “zero time” parameters. The start-up procedure for all the cells was the following:
1. Current ramp starting from I = 0 A with voltage limited to 2.0 V, H2 pressure = 1,013 bar
2. Preconditioning for about three hours at 45 - 48 °C, pH2 = 1,013 bar, I = 4.32 A.
After the preconditioning one cell for each electrolyte type was operated in continuous at T = 45 – 48 °C, pH2 = 34 bar, I = 4.32 A (corresponding to a current density of 0.45 A/cm2, while the other two cells were operated with the following start/stop cycles:
i. ON ≥ 60 min, T 45 - 48 °C, pH2 = 34 bar, I = 4.32 A;
ii. OFF ≥ 30 min, pump ON, T= 45 – 48 °C, pH2 = 34 bar.
3. In continuous durability tests a big loss in performance was observed for the cell with 1% KOH solution in comparison with 1% buffer solution with a degradation rate of 0.2 and 0.06 mV/h respectively, the KOH showed a degradation 34 times bigger than buffer solution. By the other hand, the decrease in performances can be justified by the decrease of KOH concentration in the solution due to CO2 absorption form air.
4. Figure 5 and Figure 6 show ON/OFF cycle performance on cell with both electrolyte solutions.
5. It is possible to observe a performance difference between 1 wt % buffer and 1 wt % KOH solutions: performances decrease with time in both the cells, but the one using buffer solution as electrolyte showed a faster potential increase with time.
Results showed that the decrease of performances, especially using KOH, is quiet slow, so start/stop cycles are not suitable to achieve a degradation useful for failure mechanisms studies.
5.2. Integration of Segmented Cell into AEM electrolyser (M24-M28)
The PCB was assembled in a 64 cm2 active area single cell (CS471_161108), in order to evaluate PCB resistance to cell assembling pressures and to verify possible electrolyte and gas leakages. At the beginning tests were planned, in agreement with DLR, on a small stack. The cell configuration was the following:
• End plate;
• Insulator plate provided with a bipolar with a loop for connection to power supply;
• # 2 active cells separated by the PCB, with standard MEAs;
• End plate with connection to the power supply.
After the positioning of the endplate, the cell was pressed at 200 bar by means of a manual press, and the screws was tightened with a torque wrench applying a 20 Nm torque. During that phase no signs of PCB damage was observed.
The cell, provided with proper supports (see Figure 7), was then connected to the electrolyte circuit in order to test possible leakages. The electrolyte circuit is composed by a tank connected to the cell via John Guest tubes and fittings, and a chiller FALC SBF7 to maintain the temperature at the operation temperature of 45°C. Electrolyte flow is assured by a SANSO PDH-E054 IT9-CE pump with an inlet flow rate of 9,5 L/min.
Major electrolyte leakages were detected few seconds after the system start-up in the contact regions between the PCB and the active cells frames. In these conditions no gas leakage test was possible.mElectrolyte leakages are mainly due to the rough PCB surface: the traces of electrical connections don’t allow the o-ring to ensure the proper sealing. Different strategies were evaluated to solve the leakage: use of frames of different material (less stiff), use of different sealing (e. g. silicon gasket instead of o-rings), use of a glue to enhance the sealing, in order to avoid leakages without completely modify the frame design. This solution is in fact too much time consuming and it doesn’t agree with project duration. The strategy was to try a glue coupled with the use of a less stiff material frame, because the use of additional gaskets was not suitable for the purpose due to frame configuration.
At a first attempt it was decided to use a silicon glue (LOXEAL 59∙30) with a temperature range of use from -60°C to 300°C. Before the assembly of the cell, the material was used to glue together a frame with a dummy PCB to test the possibility to easily separate the electronic board and the frame, and to test material resistance to the glue. The silicon was then allowed to dry for one night (the time needed is about 8 ours). After drying the frame and the board can be easily separated with the help of a spatula, while the silicon applied on the surface can be detached by hands, with no damage for both the components. The glue was deposited around the o-rings at the electrolyte outlet and at the electrolyte inlet/outlet, and on the external boundaries of the frame, and the cell was assembled again as described above and leaved under pressure overnight to allow the silicon to dry and ensure the sealing.
The cell was then tested for leakages with the system described above, starting with the electrolyte at room temperature. After 30 minutes of electrolyte circulation no leakage was detected, and the temperature was raised to 45°C. Even at operating temperature no leakage appeared, anyway the circulation was continued for about 4 hours, before the gas leakage test described below.
Hydrogen is fed into the stack by a gas line connected to a gas cylinder. The pressure in the stack can be monitored by a digital pressure gauge and controlled in order to be set to the desired value. A piping connects the electrolyte tank to a bubble flow meter used to measure the amount of hydrogen permeated in the liquid. The procedure to detect hydrogen leakage is the following:
1. The cell is pressurized by hydrogen at the desired pressure. After 10 minutes from pressurization an operator checks external hydrogen leaks by using a gas sniffer. All frames are carefully checked over their circumference. If no leaks are present the test continues to the following phase.
2. The hydrogen flow permeating into the liquid lines is measured and recorded by means of a bubble flow meter. The maximum allowable leak for each cell is related to the hydrogen permeability from the membranes for each pressure value. If the measured flow are below the previously mentioned value the leak test is passed.
3. Finally, the cell is carefully depressurized.
In order to avoid possible damages to the PCB due to the pressure it was decided to increase the pressure with step of 5 bar, repeating the procedure described above for each step. The cell was then pressurized to 5 bar. No external leakage was detected by means of the sniffer. The measured hydrogen flow was matching with the one due to membrane permeation in these conditions, thus no hydrogen leakage was detected in the electrolyte flow.
The pressure was then raised to 10 bar. As the pressure started to increase bubbles started to be observed at the electrolyte outlet. Anyway, the cell was tested for external leakage, and again no leakage was detected. The measured hydrogen flow showed that at 10 bar the most part of the gas fed to the cell enter the electrolyte circuit instead of the gas outlet, as expected by the observation of bubbles in the tank.
The hydrogen flow to the cell was stopped, the cell was depressurized and the electrolyte circulation was stopped. The massive hydrogen leakage in the electrolyte must be avoided in order to prevent H2/O2 mixing in the tank when the current will be applied to the cell. Anyway, in order to better clarify the cell behavior respect to gas leakage, it was necessary to test the cell in operating conditions but at p = 1 bar, in order to allow the production of both hydrogen and oxygen. The electrolyte circuit used was the same used for the electrolyte leakage test, but the bubble flow meter was connected directly to the hydrogen outlet.
The current was supplied with a TDK-Lambda GEN 100-50-3P400 power supply connected to the bipolar (positive) and to the end plate (negative). Operation conditions for the tests were the following:
• Electrolyte temperature = 45 °C
• Electrolyte flow rate = 9, 5 L/min
• Hydrogen pressure = 1,013 bar
• Current = 32 A
• Voltage limit = 2 V per cell (4 V)
The operating current of 32 A was not reached, even increasing the voltage limit: this evidence is usually related to a poor contact between MEA’s components. This can be due to different hardness of the PCB respect to standard bipolar plate used to assemble the stack. The problem can be also worsen by the thickness of the PCB, which is about 0,1 mm thinner than the standard BP.
Measured H2 flow rate was compared with the flow rate calculated on the basis of Faraday’s law for the maximum current reached by the stack, taking into account also H2 loss due to the measured membrane permeability.
The measured H2 flow rate was always higher than expected from theory: differences are great also taking into consideration that the test was performed at room temperature and errors due to the use of a bubble flow meter. This result showed that the measured flow is not only due to the hydrogen produced by the electrolysis process, but also to oxygen, and it is perfectly in compliance with the observation made during the leakage test described above. Also this problem can be due, or worsened, by the poor contact between MEA components, or between the GDLs and the PCB, but it can also be caused by damages occurred in the frames or in the PCB.
In order to have a better understanding the stack was disassembled and all the hardware components were carefully observed.
As an effect of the pressure applied during stack assembly the silicon glue tend to accumulate near the o-ring borders and also on electrolyte channels.
In Figure 3 the hydrogen outlet is also shown: it appears plugged by the silicon glue.
The presence of the silicon glue in electrolyte channels doesn’t prevent electrolyte circulation, but it can affect electrolyte diffusion in the active area of the cells. As far as concerns the plugging of H2 outlet, it can be responsible for the massive internal leakage observed during tests. During next assemblies different spreading methods for the glue will be applied, and possibly the silicon quantity will be reduced in order to obtain a good sealing without the risk of electrolyte and hydrogen channels plugging.
Cracks were also observed in the frame of cell #1: the material used for the frame (Noryl) is less stiff but also less resistant to several assembly/disassembly operations. Also cracks in frame surfaces can obviously affect hydrogen leakage. Noryl frames will be anyway used was to assemble the stack devoted to project development, because the design allows to house also PCB thicker than 1 mm. Increasing the PCB thickness can be a strategy to enhance the board resistance to stack pressure, thus increasing cell sealing and facilitating the contact between MEA’s components. In future test frames will be replaced at every stack disassembly in order to prevent damages. Furthermore, after stack assembly the glue will be allowed to dry at ambient pressure, given that the pressure applied overnight possibly enhance the PCB bending and frame rupture.
In order to enhance contacts between MEA’s components new GDLs with different thickness and hardness will be developed.
As far as the electrolyte and hydrogen leakages on the stack assembled with PCB were solved by using and additional sealing system as described in the Periodic report related to Period 2, operation tests on stacks assembled with PCB were performed in order to understand possible influence of the diagnostic tools on stack performances and if it is possible to operate the stacks in company standard operating conditions. The PCB was assembled in different 3-cells stacks with a cell active area of 64 cm2 and in a 19-cells stack (see Figure 1) with the same active area. All the hardware used in stacks is the standard for stack assembly. The test bench used for the tests on 3-cells stacks was the following:
• A tank with a capacity of 4 L equipped with a Pt1000 resistance for electrolyte heating and a visual electrolyte level control system.
• A SANSO PD-E51-IT6 pump
• John Guest tubes and fittings to connect the tank to the pump and to the stack
• A Cotek AK-650-24 to supply both the pump, the temperature controller and the Pt1000.
• a TDK-Lambda GEN 100-50-3P400 power supply to supply the stack while two different devices were used for heating depending on the test:
• A chiller FALC SB7 connected to the pump/stack with John Guest tubes and fittings.
• A WEST N6500 controller used for durability tests (the chiller water reservoir is not enough to provide heating during the whole week end without refilling).
The test bench used for the 19-cells stack consisted of a stack module equipped with a control module (see Figure 2), i. e. an EL250 system modified for INSIDE test. The main modifications were the following:
• The cooling fans control was modified in order to allow the stack temperature variation up to 55°C;
• The tank was directly connected to a chiller FALC SB 7 in order to allow the stack heating up to 55 °C and fast cooling;
• The “low stack pressure” alarm threshold was modified in order to decrease the pressure down to 10 bar.
Unfortunately major hydrogen leakages occurred during the tests performed on 19-cells stack operated at standard pressure, as well as in the 3-cells stacks, so the tests on the 19-cells stack were stopped waiting for the prototype 1.1 while tests on small stacks were performed at reduced pressure, 21 bar instead of standard 30 bar, given that the pressure is one of the parameters that shoes less influence in stack performances, in order to avoid leakages and possible damages to the PCB.
Tests in operating standard conditions (electrolyte temperature = 45 °C, electrolyte flow rate = 9, 5 L/min, hydrogen pressure = 21 bar and current = 32 A) were performed on a 3-cell stack without the PCB and on different 3-cells stack assembled with the PCB, in order to understand possible influence of the diagnostic tool on stack performances. Results in Figure 1 shows that the PCB assembled in the stack doesn’t introduce deviations from the standard stack behavior, and also that the assembly of the system integrated with PCB is reproducible, a result not necessarily predictable given that the additional sealing system is hand worked.
Results of tests on stacks assembled with the PCB (discussed in details in the next paragraphs) showed that the resolution of the obtainable data is not satisfactory, due to the lateral conductivity of the contact sheet between electrochemically active layer and bipolar plate, which lead to a prototype #1.1 with an integrated segment GDL. For technical reasons due to GDL cutting and GDL manufacturing tools, it was possible to integrate the segmented GDL only on the cathode side, so tests were performed with the PCB assembled on the cathode side in order to facilitate the comparison between the different prototypes.
5.3. Test of the Segmented Cell under normal and accelerated conditions (M27-M46)
Tests on the integration of a PCB in AEM stack according to the protocols described in D8.1 were performed, with minor changes from the protocols, on a 3-cells stack (AES565_180215) assembled with the PCB I (prototype 1) on the cathode side. The cathode side was chosen because the new generation PCB with integrated segmented GDL will be available only with GDL suitable for cathode side of the cell.
Respect to the test protocol foreseen in D8.1 the dynamical behavior of industrially hardened fats cycle was skipped because the stack load profile is not defined, while the dynamical behavior solar cycle was not performed because the power supply must be operated manually, so it is not possible for the moment to perform tests with a timeline longer than the personnel working hours. For the same reason the polarization curves were performed taking into account only the mandatory load values.
Again, for the same reason, during the test on operating parameters sensitivity the stack was put in off conditions (I = 0 A, p = 1,013 bar, no electrolyte circulation) overnight and switched on and preconditioned every morning.
As far as concerns the polarization curves, in order to obtain the stack pressurization it is needed to have hydrogen production and it is difficult to maintain the stack at the operative pressure when the production is stopped (i. e. at I = 0 A).
This is due to the use of a simple relief valve in the pressure gauge and also to the particular configuration of the stack: even if the 3 M tape can avoid electrolyte and external H2 leakages, still small internal H2 leakages are present, even if in the range of acceptance limits fixed internally for quality control purposes. During all the tests anyway the measured H2 flows were compared with the nominal flows calculated using Faraday’s Law in order to check the possible increase of internal leakages. For this reason the polarization curves were performed on reverse respect to the protocol, i. e. starting from 100% of maximum current density and going down to 0% before to raise again to 100%.
The polarization curves with the mandatory points listed in table 3 of D8.1 (in the following text called “long PC”) were performed after stack preconditioning, after every test on operating parameters sensitivity and before and after the durability and the ageing tests. For the test on operating parameters sensitivity a shorter polarization curve (in the following text called “short PC”) with points corresponding to 100%, 75%, 50%, 25% of maximum current density were recorded for each value of the parameter under study.
For every step of all polarization curves the step time was 6 minutes, as suggested in D8.1: 4 minutes of stabilization time and 2 of analysis time: only the data recorded in the last 2 minutes were considered in calculation of the parameters mean values.
The pressure was measured with a pressure gauge, so it was impossible to acquire the data number requested for standard deviation calculation. The same is valid for H2 flow, measured manually with a bubble flow meter, especially for low currents, because flows are so low that is impossible to have more than one measure in 2 minutes.
The stack was assembled using insulator plates on both endplates and with special bipolar plates on cells # 1 and 3 connected to the power supply: these bipolar plates have four loops distributed evenly on the circumference in order to distribute as homogeneously as possible the current supplied on all the bipolar surface. These bipolar plates in principle are useful to prevent damages to both the cell and PCB due to uneven current distribution.
The peculiar stack configuration with the PCB substituting a BP doesn’t allow the measure of single cells potentials, so only the total stack voltage was measured, via the PCB, during tests,
The stack was pressurized using a pressure gauge equipped with a Hamlet relief valve with pressure range from 3.4 to 24 bar and the flow was measured using a bubble flow meter.
The electrolyte circulation system and other hardware used for tests are described in the previous paragraph
In order to prevent stack damages during the test a safety potential limit was fixed on the power supply: the limit is the same used to protect the stack during the current start up ramp, and it is 2,3 V per cell, corresponding in this case at 6,9 V. The power supply stop to supply current when the voltage limit is reached.
The stack preconditioning is composed of a first heating step of one hour with the stack switched off and with electrolyte circulation on, in order to stabilize the stack temperature to the standard operative temperature of 45°C. When the temperature is stable the stack hydrogen production is started performing a current ramp up to 32 A with the potential limit described above. When the stack was operating at 32 A, it was again checked for external H2 leakages using a sensor (this checked was repeated daily for all the tests duration), and for internal leakages by measuring the flow with the bubble flowmeter, then the pressure was raised up to 21 bar. The stack potential was then allowed to stabilize for at least one hour before starting the test. After stack preconditioning a long PC was recorded. The first tests performed were the one related to operating parameters sensitivity, needed as stack characterization tests. The stack was preconditioned as described above before the test on each parameter.
After the recording of the first PC, the electrolyte temperature was lowered in order to have a stack temperature of 25°C, taking into account the stack self-heating due to the current. When the temperature and the other operating parameters were stabilized, a short PC was measured. The temperature was then raised to 35°C, 45°C, 55°C allowing the operation parameters to stabilize before the short PC. A long PC was performed again after the stack temperature was decreased to 45°C.
Given that the curves for ascending and descending power show no hysteresis, polarization curves comparison for every operation parameter considered are shown for ascending current only, in order to simplify the graphical rendering.
Polarization curves recorded at different temperatures show, as expected, a significant increase of performances increasing the temperature. The operation temperature of 45°C was chosen in order to find a better compromise between stack performances and durability, especially respect to membrane degradation.
The next test performed was the one on pressure variation sensitivity. Operation pressure for the current technology is 35 bar, but the peculiar configuration of the stack suggested to operate at lower pressure, 21 bar, in order to prevent damages to the PCB or hydrogen leakages. The pressure in the stack is checked and maintained using a pressure gauge connected to the hydrogen outlet and equipped with a relief valve and a two-way valve to pressurize and de-pressurise the stack.
Short PC was measured for each pressure value, 21, 14 and 7 bar, and a long PC was performed at the end of the pressure variation test after the stack was put again in standard operation conditions and allow to stabilize. Other operation parameters are standard.
The pressure, as shown in Figure 3, doesn’t affect stack performances. Higher pressure are preferred in order to reduce hydrogen humidity.
Tests on variation of electrolyte concentration was performed at standard concentration, i. e. 1 Wt % KOH, and at 0,5 Wt % concentration.
Similar results were obtained with a 2-cell stack (AES561_180108) assembled with PCB III (prototype 1.1). Unfortunately with prototype 1.1 it was not possible to complete the tests in accelerated conditions due to a failure in a PCB segment after about 172 hours of continuous operation in standard conditions, while with the stack AES565 the tests were continued as described in detail below before the same kind of failure happened.
Stack AES565 was tested for:
• about 240 hours durability test in standard operating conditions
• about 190 hours durability test at maximum operating current (I = 47 A)
• about 37 hours start/stop test as described in D8.1.
Figure 8 shows a comparison in PC recorded for stack AES565 after different test step: after stack preconditioning, i. e. with the stack operated for few hours (Ustack = 5,240 V), after the characterization tests (Ustack = 5,215 V) and after the durability test at high current (Ustack = 5,370 V). The test was supposed to be stopped when the potential limit of 6,9 V was reached, but degradation was to slow to allow the potential to reach the limit, so it was decided to switch to the start/stop test, including also a temperature shock, in order to try to accelerate the degradation. The manual operations on the power supply didn’t allow to perform cycling overnight and during the week end, the temperature controller used instead of the chiller didn’t allow the temperature to remain stable at 45°C overnight and during the week-end, thus affecting also the stack potential. Unfortunately the unstable potential value didn’t allow an evaluation of stack degradation, and the PCB failure during test didn’t allow the recording of a PC.
5.4. Identification of electrochemical behaviour under critical operating conditions (M29-M48)
The most critical condition for the stack is to operate at standard current and temperature without electrolyte flow, which is also acting as a coolant: without electrolyte flow stack temperature increases causing at the same time membrane drying, which can result in serious damage to the membrane itself. For this reason standard electrolysers are equipped with a safety control that switch off the current when a critical voltage (usually 2 V per cell) is reached. A starvation test was performed in AES561 stack (2-cells stack equipped with PCB III (prototype 1.1) on cathode side. For this test the potential limit was set higher, up to 2,3 V per cell, in order to preserve (if possible) the stack for further investigations. With the power supply used for the test the current is not abruptly switched off, but it is lowered by the power supply in order to allow the potential to decrease below the limit.
The stack was allowed to stabilize at standard operating conditions (in this shorter test the pressure was set at 30 bar) for about 3 hours before the starvation test, that was performed by switching off the electrolyte pump leaving the stack at standard pressure and current. The stack potential was not strongly affected by the lack of electrolyte until all the water remaining in the cells (basically trapped in GDLs) was evaporated. After about 45 minutes the potential started to raise up very quickly together with the temperature, until the potential limit of 4,6 V was reached and the power supply started to decrease the current. A graph showing stack voltage and temperature behavior during the starvation test is shown in Figure 10, while in Figure 11 is shown a current density map recorded before the start of the starvation test compared with one at the end, i. e. when the current started to decrease.
When the current started to decrease the pump was switched on again, and after few minutes, when the stack hydration was completed, the stack was operating again with the same parameters recorded after preconditioning (32 A, 30 bar, 44,9°C, V = 3,51 V after preconditioning vs 32 A, 30 bar, 44,4°C, V = 3,52 V after the pump restarting).
6. Investigation of Failure Mechanisms (Work Package 7)
Ru-, Ir, and mixed Ir-Ru anode test materials were successfully prepared and integrated in the MEAs of a PEM electrolyzer
Anode materials for a PEM electrolyser were characterized ex situ before and after water electrolysis with XRD, TEM, STEM, XPS
The experimental tools for in situ studies of electrochemical systems with Near-Ambient Pressure X-Ray Photoelectron Spectroscopy (NAP-XPS) have been successfully developed and approved using model systems
Application of NAP-XPS to the anode of a PEM electrolyser allowed us to unveil the degradation mechanism during electrolysis and suggest the stabilization mechanism
Preliminary corrosion tests of raw FR4 material for PCB production were carried out under AE conditions
PEM-based MEAs comprising various unsupported and supported Ru-, Ir, and mixed Ir-Ru anode test materials were prepared by the DLR partner and investigated at the beginning of life and after ageing by CNRS-UdS and UAES partners using various ex situ (XRD, SEM, EDX,TEM, STEM, XPS, AFM, Kelvin probe force microscopy –KPFM, etc.) and quasi in situ approaches (conducting AFM, NAP-XPS, soft X-ray Absorption Spectroscopy);
Application of NAP-XPS and soft X-ray absorption to various types of Ir-based anodes of a PEM electrolyser allowed us to unveil the anion-type (OII-/OI-) red-ox oxygen evolution reaction mechanism and propose a tentative origin of higher stability of Ir- relative to Ru-based anodes, the latter operating through cation red-ox (as proven during the 1st reporting period);
Degradation/restructuring of ionomer in the PEM-based MEAs after (during) operation has been evidenced with material-sensitive, conductive AFM measurements performed by UAES partner and NAP-XPS measurements performed by CNRS-UdS partner;
Ionomer loss in the electrodes after operation has been detected by UAES partner using material-sensitive, conductive AFM.
UAES partner set up a fluid cell measurements technique to monitor the material/component degradation of membranes, electrodes or half-MEAs, and tested this approach on model proton-exchange membranes/MEAs;
UAES partner investigated thin films of anion-exchange ionomer with conductive AFM tapping mode (TUNA) measurements and discovered swelling and increasing conductivity of the film caused by the current flow.
The AWE conditions require the protection of any copper surface and the protection of cut, drilled or machined PCB edges against the invasion KOH lye. The PEMWE and AEMWE conditions did not show problematic corrosion.
Nickel contaminations in the membranes of operated PEMWE could be locally correlated to the reduced activity of this cell.
Ageing of industrial PEMWE samples could be attributed to ionomer distribution and mechanically induced cracks.
The stability of segmented PCB was investigated. The corrosion in heavy alkaline ambience was a triking problem, but only when cut edges oft he PCB material is in contact. In dilute alkaline ambience, this was not a problem, but Rhodium traces, presumably from the diffusion barrier below the gold coating, was found on teh segmented bippolar plate for AEMWE.
7. Test Protocols and Control Strategy (Work Package 8)
A literature study was performed what testing protocols are used in research groups over the world for alkaline, PEM and AEM electrolysers. Existing international standards for electrolysers were evaluated if they could provide descriptions of tests. In addition previous FCH-JU projects stating that they had developed their own testing procedures were contacted and asked if these procedures could be made available. However these procedures were usually not publically available. A contact with European international activities on harmonisation of electrolyser testing protocols was established, especially IEA Annex 30 and JRC electrolyser testing harmonisation activity.
The only already established international standards for electrolysers are two international standards for water electrolysers released by ISO: ISO 22734-1 Hydrogen generators using water electrolysis process — Part 1: Industrial and commercial applications and ISO 22734-2 ... Part 2: Residential applications. The tests described there are for verifying the reliability of the electrolyser system. All tests are described in a very short form and cannot be well used for the present case of collecting information from processes in the electrolyser.
Searching for some typical electrolyser application power profile two cases were identified: industrial electrolysers are operated according to the hydrogen demand following a daily/weekly profile. This applies e.g. for electrolysers used in the industrially hardened fat industry. Electrolysers fed by renewable energy have to follow the power supply profile which is quite dynamic and shows also long periods without operation. As an example the power profile of a photovoltaics field on an April day in Germany is given assuming direct use of the solar power respectively testing the maximum dynamics required for applications in this field.
The primary degradation mechanisms reported by researchers are electrodes degradation for all types of electrolysers, membrane degradation (thinning, ion exchange, loss of conductivity) for PEM and AEM, anode corrosion including diffusion layer, connection and bipolar plate corrosion and temperature variations causing leaking, corrosion and failure in BOP. High temperatures, periods of OCV, high current density and temperature variations are named as causes for degradation.
Based on results from literature and the project participicants’ own experience basic testing protocols were suggested and discussed among the partners. In line with previous FCH-JU project testing protocol activities established for PEM fuel cells and using the experience and verified wording from these projects, test modules and test procedures were defined. A test module is a basic piece of test described well in detail. A test procedure is a sequence of test modules making up a test in a practical experiment.
The basic test modules defined are
• Polarisation Curve
• Operating parameter X sensitivity (with X standing alternatively for temperature, pressure, differential pressure, liquid flow anode, liquid flow cathode, concentration)
• Steady State Endurance
• Dynamic Behaviour Industrially Hardened Fats Cycle
• Dynamical behaviour Solar Cycle
The test modules are written in such a way that they can be applied to all types of electrolysers differentiating for the different way of operation wherever necessary within the test module description.
These test modules are then combined into suggested procedures for basic characterisation for the different types of electrolysers. The purpose of this basic characterisation is to evaluate the quality of performance of the electrolyser and to collect data for internal processes (using the new current density distribution measurement) during normal operation of an electrolyser.
A literature study was performed what degradation mechanisms are especially relevant for which type of electrolyser, how these are investigated and if there are accelerated tests available.
From the literature survey it is not exactly clear what states of the electrolyser will contribute to what degradation effect. Also no public accelerated stress tests (AST) are reported. However it seems to be clear that several conditions can accelerate degradation in general: these are prolonged operation at high current/power density, voltage-controlled dynamic operation (reported for PEM electrolysers), increased impurity levels in the water (for membrane electrolysers), long periods of OCV for the stack (especially at higher temperature) and frequent temperature variations as well as higher temperature operation. Furthermore during highly dynamic operation fast fluctuations in the system might be induced by inadequate control (e.g. in relative pressure, KOH supply) that might as well induce degradation. Tests along this line are suggested for this project. For different types of electrolysers different degradation mechanisms seem to be more relevant giving somewhat different accelerated testing protocols suggested for the electrolysers.
Based on results from literature and the project participicants’ own experience accelerated testing protocols were suggested and discussed among the partners. It was decided that it is necessary to show a clear warning that these tests should not be applied to expensive equipment that should still be used after the test because the tests might be very destructive to the performance. Preferentially such tests should only be applied to single cells and small stacks.
For alkaline electrolysers the following accelerated testing protocols are suggested and written:
• Long term operation at constant maximum current (several periods of steady state endurance test at standard operating conditions for 200 h at high current density alternating with polarisation curve characterisation)
• Accelerated start-stop test (repeating 30min on, 30 min off with accelerated heating/cooling)
• Long term operation at increased temperature (several periods of steady state endurance test at maximum temperature for 200 h alternating with polarisation curve characterisation)
• OCV at increased temperature (several periods of steady state endurance test at 0 A for 200 h alternating with polarisation curve characterisation)
For PEM electrolysers the following accelerated testing protocols are suggested and written:
• Long term operation at constant maximum current
• Voltage-controlled dynamical stress test (30 seconds at voltage corresponding to a cell current of 1 A/cm2 during the initial polarisation curve; 30 seconds 0.1 V; sets of 200 cycles)
• Long term operation with increased water impurity levels (using redistilled water with 1 mg/l Ca2+, 0.1 mg/l Cu2+ and 0.1 mg/l Fe2+ added)
• OCV at increased temperature
• Accelerated start-stop test
• Long term operation at increased temperature
For AEM electrolysers the following accelerated testing protocols are suggested and written:
• Long term operation at constant maximum current
• Accelerated start-stop test
The test procedures have not yet been verified. Verification experiments will show if the tests can be operated as defined here in practical laboratory operation, if they will give reproducible results and what type of degradation mechanism is accelerated by these tests. For final evaluation of such a test a very long term operation of an electrolyser at normal conditions and extensive characterisation during operation as well as after use would be necessary to compare the accelerated degradation achieved to the regular degradation phenomena. However such a comparison cannot sufficiently be done during this project due to limited time and different focus of the project.
The combined evaluation of these protocols was not possible to this point.
As a contribution to the JRC´s activities in harmonised electrolyser testing, the consortium shared their generic testing protocols with the JRC as concrete starting point for more detailed definitions.
Potential Impact:
The project is directly aimed towards improved performance and lifetime of the state-of-the-art water electrolysis technologies available. The creation of a tool which monitors the electrochemistry in-situ and on-line will enable the manufacturers to find improved operations conditions with higher efficiencies. Moreover, hazardous operation mode can be identified and methods can be developed to anticipate and avoid critical conditions, extending the lifetime and sustainability of water electrolysers.
The changes currently going on in the European electricity market will demand increasing decentralised capacities for energy storage. The use of hydrogen as an energy carrier will not only reduce the dependence of the European countries on oil and gas exporters. It will also provide a location independent method of storing excess electricity, and in combination with fuel cell and turbine power plant it will provide even a method of redistributing the stored energy. The increasing incorporation of solar, hydroelectric, and windmill plants with their unsteady output profiles into the European electricity grid will require the ability to operate with variable load in a reliable manner. In particular, frequent load changes are a challenge to the lifetime and performance stability of electrolysers, and thus the monitoring of these systems is an existential issue for the reliability and tunability. The capability of the segmented cell tool for on-line monitoring the internal electrochemistry is the key property for the diagnostics of load changes.
The creation of the in-situ tool for on-line monitoring the electrolyser performance will improve the overall efficiency of electrolysers and will lay path for a decentralised use of electrolyser technology for the stabilisation of the European electricity grid and the provision of hydrogen as an environmentally friendly fuel. This development and its impact on the energy policy will emphasise the leading role the European Union in the world in sustainable energy technologies.
Wider societal impacts are indirect and related to the general willingness of European society and policy makers to push forward the transition from fossile to renewable energies and sustainable technologies in general. This project is adding another bit to this big puzzle.
The dissemination activities for this project are mostly research and industry related. On top of publications, conferences and workshops, the winning of a an award and the corresponding ceremony during the f-cell conference soirée was probably the most public dissemination event, and was attended attended by representatives from industry and research, and also from politics and press. Another unexpected dissemination activity, a page in the official calendar of the German Aerospace Center was unfortunately stopped by a decision of the headquarters.
The exploitation of results by the individual beneficiaries is mostly depending on the nature of their institutions/ companies:
DLR
The German Aerospace Centre as holder of patents for segmented boards for PEM fuel cell diagnostics and grantor of licenses to industrial companies, e.g. S++ Simulation Services, has a very high interest to continue this by creating patents and grating licences on similar developments in the field of electrolysis technology. It is the task of the German Aerospace Centre to advance technologies towards usability, and not to bring them onto the market itself, but enable and motivate industry to do so. In addendum to the financial flowback by licence costs, which may be ideally used for initial research, the patents are used as performance figures of department and institute. Patents also secure the high-tech position of Europe and Germany, which is another mission of DLR. Concretely, two technological advances in the field of monitoring of electrochemical converters are envisioned.
The technological advance created in this project is always used as basis for future R&D activities. As DLR is active in adapting the segmented board technology also to other electrochemical converters, each bit of experience will carry to further activities. The project INSIDE is a part of DLR´s strategy to extend and strengthen the position in the field of in-situ and in-operando analytics.
As a research institution within the Helmholtz Association, DLR has the interest of publishing and presenting created foreground. Successful high quality dissemination is internally a performance figure and externally a medium for developing collaborations and new projects, and based on that, funding possibilities. Thus, it is essential for the existence as a research institution to address the scientific public and the non-scientific public as well. Concrete research ideas are the collaboration with the University of Aarhus with the goal to correlate modelling results on PEMWE hardware with experimental in-operando data, and to evaluate operation strategies together with the University of Lappeenranta.
Both the technological basis and the scientific reputation from the created foreground will be the background for acquisition of third-party funding and thus one of the supporting pillars of the German Aerospace Centre and its future activities. Concretely, a R&D project is planned, together with Enapter S.r.l. (the beneficiary initially named Heliocentris Italy in this project), which will make a direct link from INSIDE to the public and is expected to generate a significant societal impact.
NEL
NEL will use knowledge developed during the project in its R&D activity focused on development of new extremely flexible alkaline water electrolysers with improved energy efficiency. The tool will serve the development of new generation of electrocatalytic compositions based on recent achievements of nanotechnology. The tool will be an instrument for the detection of possible failures in advanced cells in-situ instead of making several post mortem analyses and therefore will reduce time and cost of development of new products.
Utilization of the diagnostic tools will allowed to reduce development of new products in 2-3 times, by reduction of time for ex-situ testing of materials for both electrodes and diaphragms, reduction of expenses for verification of mathematical modelling of heat and flow distribution along and across of the cells. So, direct reduction of the cost is expected at the level of 0.4 million euro/year. Share of new products in the profit will be increase and generate profit increase of app. 20%.
Enapter S.r.l. (formerly Heliocentris ITaly, S.rl.l. HELIO)
Heliocentris Italy S.r.l. offers a unique technology and very innovative system devoted to UPS saving system and a variety of off-grid applications. The devices can be coupled with a variety of renewable energy sources and must be able to resist also to adverse environmental conditions such as high temperatures and/or humidity. Given that the technology is new, a further insights on lifetime and failures modes are needed in order to improve devices durability and reliability in stress conditions described above (environmental conditions but also operation with discontinuous and non-stable energy supply form renewable sources), and the diagnostic tool developed during the project is a very powerful tool to obtain the information needed to reach the goal, in addition reducing times devoted to development. Information obtained with the diagnostic tool will be also relevant to improve stack performances reducing at the same time to production and material costs, especially in view of the development of a new stack generation starting from the beginning of 2018. The new generation and the cooperation with DLR is currently going into anoher project with an expected societal impact.
CNRS / UdS
CNRS-UdS is a research institution and will participate in dissemination measures, including any scientific publications relating to foreground and its content will be made available in the public domain thus demonstrating the added-value and positive impact of the project on the European Community. CNRS-UdS has the interest of publishing and presenting results of the research and development activities in this project in international scientific journals, such as Electrochimica Acta, Electrochemical Communications, Journal of Power Sources, Journal of Solid State Electrochemistry, etc., and conferences and exhibitions, such as meetings of the Electrochemical Society (ECS), International Society of Electrochemistry (ISE), Gordon Research Conferences (where E. Savinova will act as a chair of GRC “Fuel Cells” in 2014). Technology improvements resulting from this project will be protected by patents jointly with the industrial partners.
UAES
UAES is a University of Applied Sciences and dedicated to teaching and applied research. UAES has strong interest in publishing the scientific results (considering the confidentiality issue) in peer reviewed international journals for driving the expertise in the research area of ion exchange membranes and catalysts. A successful cooperation with the DLR where nanoscopic analysis of MEAs was performed by UEAS before and after Electrolyser operation has led to several common peer reviewed papers. Further publications on degradation detected in a cell with segmented PCB are in preparation. Talks at international conferences, i.e at the ISE Spring Meeting in Buenos Aires and the Hyceltech in Porto, strengthen the scientific landscape and the institutes profile and were used to install further international cooperation. .
As a second goal, UAES is dedicated to education, also in the field of renewable energy. Since 3 semesters, a lecture including experiments has been established within a master course for students interested in this field. The project activities provide opportunities for young researchers and students. This includes an outstanding education leading to PhD degrees, scientific international exchange, and excellent job opportunities. A facebook contribution and a video of INSIDE activities dedicated to reach the younger generation via youtube have been produced and will be updated and complemented.
List of Websites:
fb.me/insideelectrolysis.eu
The original website is temporarily unavailable.
Further contact details:
Indro Biswas
+49 711 6862 603
indro.biswas@dlr.de
The INSIDE project is focused on the integration of diagnostic tools in electrolysis of water for the monitoring of performance and early detection of malfunction. Various electrolyser technologies are taken into account:
• Alkaline electrolysis with liquid electrolyte as a very established and robust technology
• Electrolysis with proton exchange membranes as very dynamic technology
• Electrolysis with anion exchange membranes as new and emerging technology that is combining aspects of both other technologies.
The diagnostics tools are based on technology with printed circuit boards, which has been successfully developed and used in fuel cell research.
The main objectives of the project INSIDE were:
• Design and construction of in-situ diagnostic tools for three electrolyser types (PEMWE, AWE, AEMWE)
• Integration of the diagnostic tools into industrially relevant electrolyser types
• Test of the diagnostic tools in stack or short stack configuration
• Definition of common test protocols for performance testing and proof of the protocols
• Identification of failures and degradation, and correlation to microscopic findings from spectroscopic and microscopic investigations
• Evaluated protocols for accelerated durability tests
Two prototypes for the PEMWE technology were developed and tested. They are both using an original graphitic bipolar plate that was modified and directly attached. In particular, the second, enhanced prototype is embedded between graphitic bipolar plates, and thus mimicking a regular BPP.
The works were hindered in the beginning by the unavailability of industrial PEMWE hardware with the option to work with the corresponding bipolar plates.
The design process for alkaline electrolysis had to face the chemistry of this technology. As the 30% KOH lye at elevated temperatures is very corrosive to many materials, the direct contact to copper and most duroplast polymers has to be strictly avoided. Although the standard material for printed circuit boards can be surprisingly stable, when it´s plain face is exposed to KOH lye, cut edges are quickly decomposed – lye also has the property to wet laminated boards internally and thus creep into a printed circuit board and corrode the circuitry there. After several fruitless attempts, a modular design was developed, where sensor modules are embedded in a rigid frame of PEEK, a stable polymer, but otherwise useless for building PCBs. An additional aspect of this design is the modularity itself, which allows for an easy upscaling and thus the application in large systems. The system was produced and “dry tested”, but could not be evaluated in electrolysis mode within the project duration.
The design for the AEMWE technology was quite straightforward, with the exemption of the need to modify parts of the attached electrolyser cell, too. The porous metallic gas / lye diffusion sheet contacting membrane/electrode and bipolar plate introduces a signal cross-talk to the lateral electrical conductivity, which could be overcome by the additional segmentation of this contact sheet.
As overarching activity, the data acquisition software has been reprogrammed for a more flexible adaptation to various requirements.
The activities were accompanied by ex-site investigations on pristine and aged and, where necessary, on model materials from the electrolysers. Experimental tools for in situ studies of electrochemical systems with Near-Ambient Pressure X-Ray Photoelectron Spectroscopy (NAP-XPS) have been successfully developed and approved using model systems. A correlation between the spectromicroscopically shown deposition of a Nickel contamination in the membrane of a PEMWE cell and the observable local deactivation of this electrolyser cell could be established.
Common testing protocols were formulated based on a literature survey and on experience from previous activities with fuel cells. They were shared with the JRC to support their harmonisation activity on electrolyser testing. The evaluation and definition of accelerated stress tests was not possible.
Two workshops were organised during the project duration, and the f-cell award o 2018 (2nd place) was grated to the beneficiaries directly involved for the development of the modular segmented bipolar plate for alkaline electrolysis
Project Context and Objectives:
Electrolysis of water is an important technology for the production of high purity hydrogen needed for numerous chemical processing applications. In the future, however, hydrogen is forecast to become important as an energy carrier that is used for storage of excess electrical energy produced by renewable sources such as solar, or wind. In fact, current energy policy requires the replacement of energy sources based on fossil energy carriers by renewable energy carriers. Reasons are the increasing depletion of sources of coal and oil, the need of significant reduction of emission of greenhouse gases (e.g. CO2), and the intention to reduce economic (and political) dependency and risks due to imports of fossil energy carriers. Moreover, nuclear power is becoming less acceptable to the general public, e.g. in Germany. However, there is a substantial discrepancy between the temporal availability of renewable energy (e.g. differences between day/night or summer/winter especially for solar) and the actual needs. Therefore, the opportunity to temporarily store excess energy is of great interest.
Consequently, there is an increasing interest in development, improvement and tailoring of electrolysers for the efficient production of hydrogen that can be used in combination with fuel cells in a wide range of applications such as mobile, automotive, aerospace, and stationary. For all these purposes the electrolysers need to be controllable, reliable, and sustainable. Due to the stable political situation Europe is a predestined place to pursue this ambitious goal. In addition, there is a drive to reduce the cost of hydrogen. The cost of producing hydrogen is closely linked to the price of electricity used to power the electrolyser.
According to Turner1 the improvement of electrolyser efficiency is only one way to lower hydrogen cost; major improvements are additionally needed in electrolyser lifetime, maintenance requirements and capital cost of the systems. From the thermodynamic point of view H2 cost can be also produced by performing electrolysis at elevated temperatures in order to minimise the electrical energy needed.
On the market there are two prominent types of low temperature electrolysers, namely alkaline electrolysers , and proton exchange membrane (PEM) electrolysers. Additionally, there is a more recent and less known type of anion-exchange membrane (AEM) electrolysers . The general occurring processes are the formation of hydrogen at the cathode and the formation of oxygen at the anode and the transport of ions between the electrodes which are separated by a diaphragm or the mentioned ion exchange membranes. The specific electrode reactions and the type of ions responsible for the internal charge transport depend on the electrolyte (alkaline or acidic). Each electrolyser technology exhibits specific properties with specific obstacles to most efficient operation and long lifetimes.
PEM electrolysers operate under acidic conditions using a proton exchange polymer electrolyte membrane (e.g. Nafion®) for the transport of protons produced from dissociation of water. In such a case only expensive noble-metals, typically Pt, are suitable catalysts, a serious drawback of this approach. Yet, PEM electrolysis benefits from the possibility of producing pressurised (200 bar) H2 (ready for transport) with little amount of extra energy.
The use of non-noble metals as catalyst (e.g. Ni) is the main advantage of alkaline electrolysers; the produced OH- ions migrate through an alkaline electrolyte (e.g. KOH) aqeous solution. The cathode and the anode side are separated by a dedicated diaphragm. Particular difficulties that need to be overcome to increase the performance are: resistances due to bubbles formation at the electrodes, reaction activation energy, and ionic transfer.
AEM electrolysers use solid alkaline ion exchange membranes as electrolyte and therefore allow the utilisation of non-noble metal catalysts, such as in liquid alkaline electrolysis. However, the ionic conduction losses of AEM are significantly higher and the membrane stability is significantly lower than in state-of-the-art PEMs. Thus, the improvement of anion exchange membranes and ionomers is a major challenge with respect to this technology.
In order to successfully manage all the possible improvements of these technologies it is necessary to monitor the performance and operation failures in-situ as well as to develop test protocols to make the long-term operation predictable. The development of such an in-situ monitoring tool and validation of test protocols are the main objectives of the R&D project INSIDE
It pursues the development of diagnostics tools for three independent technologies for water electrolysis (PEM based, alkaline, and AEM based) with individual properties and individual challenges. This development will provide operational data from inside the electrolyser systems. Conventional current-voltage characteristics, impedances and materials analytics before and after operation are not able to provide the same type and quality of data as an in-situ tool for on-line diagnostics.
This technology monitors the local current densities, which are directly linked to the local electrochemical turnover. The principle has been successfully used in the research on polymer electrolyte fuel cells for materials development, for optimisation of operation condition, and for prevention of undesired events, such as local flooding of parts of a fuel cell, which can be directly visualised during operation.
The strategic aim is to use these diagnostics tools for online monitoring with the possibility for online adaptation of operational parameters, and for the prevention of hazardous operation modes while optimising the overall performance. At the end of this project, there will be a public document with summary recommendations for the operation of present electrolysers and for the targeted development of future electrolysers – but this will not be the end of the usability of these new diagnostics tools.
In particular, online monitoring is a key technology for the application of water electrolysers in a highly fluctuating electricity market in Europe as grid stabilising highly dynamic systems, which does not dump power, but transforms excess electricity into hydrogen, a storable, transportable power source. Moreover, electrolysis of water represents a decentralised technology to provide local hydrogen refuelling stations for e.g. electromobility with the fuel cell car.
Project Results:
1. Requirements and Specifications (Work Package 2)
Proton exchange membrane water electrolysis
- The diagnostic tool covers the entire area of 76 cm².
- Connectors are in outer areas; their arrangement on the PCB has to fit into the electrolyser housing, ideally. Alternatively, the tubing and the main electrical connectors require a reconfiguration.
- The tool has 88 segments of square shape, with exception of the rim segments.
- 21 temperature sensors are covering the cell area with constant spacing.
- The PCB material itself is standard FR4 grade, as it was used for PEMFC applications.
- The PCB will replace the graphitic bipolar plate, resembling the flow structure only on the cathode side and connect to the endplate at the opposite side in first generation. In the second generation, it will replace the graphite composite bipolar plate between cells in the middle of a stack and thus resemble both cathode and anode side flow structure. Also, the thickness will be similar, of >4 mm, which is considerably thicker than conventional PCBs for PEMFC applications.
- Current bearing segments are coated with gold, the thickness is higher than for conventional PEMFC application and the 1st AEMWE prototype.
- Two 34972A type data loggers should be used
Alkaline water electrolysis
- The diagnostic tool covers the entire area with a diameter of 500 mm.
- Temperature sensors are arranged in a dense symmetric 8-pointed star pattern with intermediate positions.
- The entire PCB area, which is in not metal coated, has to be protected with a polymer coating that is stable in the alkaline ambience; in particular, the edges of internal openings are protected.
- The tool requires preconditioning in 25% KOH before going into operation.
- Either six parallel data loggers (preferred) or one larger data logger (if parallel read-out not feasible) will be used.
- Housing with ATEX zone 1 (or higher) certification will cover the data acquisition unit(s) to reduce the distance to the electrolyser stack.
Anion exchange membrane water electrolysis
- The diagnostic tool covers the entire circular area of 64 cm².
- Connectors are in outer areas.
- The tool has 59 segments, most of rectangular shape, with exception of the rim segments. Regions close the edges have more segments per area than central segments.
- 16 temperature sensors are covering the cell area with almost constant spacing.
- The current bearing segments with copper core conductors inside the PCB are completely covered with gold, similarly o PEMFC diagnostics boards.
- The tool requires preconditioning in 1% KOH to remove PCB (productions traces) before going into operation.
- Two data loggers, with a total of 80 channels should be used
2. Design of the in-situ diagnostic tools (Work Package 3)
2.1. Design and construction of the Segmented Cell for PEM electrolysers (M3-M36)
The design of the diagnostic tool for proton exchange membrane based electrolysis was planned as the technology route with least expected difficulties. The difficulties for development, however, turned out to be of organisational nature.
After the withdrawal of ITM Power Ltd., Sheffield, UK and H-TEC Systems, Lübeck, Germany from the project consortium before the start of the project, attempts were made to purchase the necessary hardware for the PEMWE route from
- ITM Power, Sheffield, UK
- H-TEC Systems, Lübeck, Germany
- Hydrogenics, Gladbeck, Germany
- Greenlight Innovation, Canada
- IRD Fuel Cells, Odense, Denmark (now named EWII Fuel Cells)
After several negotiations with all possible suppliers, an electrolyser system of the type “Elze E1050” was purchased from EWII Fuel Cells of Denmark and delivered in 2017. The cells of this system are circular, and of an area of 76.5 cm². The cell is asymmetrical, with a carbon based GDL (cathode) and titanium felt (anode).It had been decided to design a segmented PCB that connects endplate and cell cathode GDL in the 1st generation. It replaces the graphite composite bipolar plate, but does not resemble the anode side flow field (see Figure 2), which would add complexity and thickness to the prototype.
First Prototype for industry hardware
The prototype as described in the second periodic report was manufactured and tested in summer 2017. The design was carried out as a PCB with standard FR4 materials and a graphitic top plate with the cathode flow field of the regular BPP, in order to warranty the conductivity between the PCB and the graphitic plate a conductive glue was used. The rear face is contacting the regular metal contact plate of the EWII electrolyser stack with an active area of 76.6 cm2. (The bipolar plate of the stack consists of a graphitic and a metallic part.) The segmentation follows the rectangular shape of the flow plate with 44 segments, (Figure 2) fitting 2 segments into each of the full rectangles. In order to measure the temperature distributions 21 temperature sensors were homogeneously distributed
The first integration and evaluation was carried out in a small cell with original stack hardware (end plates, bipolar plates, MEAs...) out of the system. This was to facilitate fault finding and assembly, for the space in the commercial system is limited (see Figure 3) The segmented diagnostics PCB, however, was designed to fit in, regarding its area and cable connectors. The PCB itself is embedded within two modified graphitic bipolar plates, and in consequence, the stack assembly follows the procedure identical to that without PCB.
The diagnostic board was operating, but points of optimisation could be identified:
• The internal contact between the graphitic component and the PCB component was carried out with a conductive silicone based material with some elasticity. The homogeneity of this contacting could not be realised homogeneously enough to ensure a homogeneous electrical contact. The result was an insufficiently high internal resistance – at some spots. In addition this construction principle prevented a perfect sealing, and in consequence the pressure stability of this stack was lacking.
• One segment was also not operating in this prototype
• The local resolution capability and the signal cross talk between segments were meeting the expectations.
Enhanced Prototype for industrial hardware
The second prototype was designed and manufactured with a hardening conductive glue for the internal connection to overcome the resistance and inhomogeneity problems. The PCB had to be laid out new, with some modifications of the circuitry. The resulting PCB is more stable and fits better into the stack. Segmentation, number and placing of the temperature sensors are unchanged, and the board is embedded between two modified graphitic plates. The interface to the neighbouring cell is completely identical to that of a regular graphitic bipolar plate, and the anodic flow field plate described earlier does not need to be mimicked in this way. The diagnostic plate can assume principally any position in a stack.
The data acquisition also did require modifications. However, the software was upgraded in the meantime to be compatible to either a Keysight 34980A system or two Keysight 34972 systems operating in parallel.
2.2. Design and construction of the Segmented Cell for alkaline electrolysers (M3-M36)
The design of the diagnostic tool for the involved alkaline water electrolysis holds the biggest challenges amongst the three electrolysis technologies in this project. Reasons are the chemical ambience, where the system has to subsist, a heavy 30% KOH lye under elevated temperatures of 65°C, the requirement to seal this lye with a rubber frame on a smooth surface of the board, and the sheer size of the required system with a diameter of 500 mm (active area), which required a compromise between segment size and total number of segments. Finally, the data acquisition has be spatially close to the stack to avoid a weakening of the parallel analogue signals.
The first attempts to the diagnostic tool had been unsuccessful, after preliminarily promising tests showed to be misleading – the seeming chemical stability of the PCB material had been falsified as machined/cut parts of the intended PCB material were exposed to the alkaline ambience.A redesign from scratch was done in this reporting period.
Design strategies that had been considered and discarded during this process were:
- PCB with protective coatings (at the edges), e.g. using epoxy based pottant materials, as common for protection of electronics
- PCBs embedded in polymer resin casts
- PCBs made out of other base materials than the common FR4 type, e.g. Teflon™
Some chemically promising concepts had to be discarded because of the lacking mechanical stability.
The final solution, which has been carried out to the end, comprises of a frame of Polyetheretherketone (PEEK), which is well known and tested as chemically stable in alkaline conditions. While the construction of a PCB completely made out of PEEK as base material is not possible, the integration of an array of sensors in a PEEK frame was so. Each segment consists of a Nickel contact segment (as can easily be seen in (Figure 6) and each underlying resistor bridge module (Figure 8) that is inserted in the PEEK frame and directly mounted to the current collecting Nickel back plate (not visible). Each module is connected via a small circuitry bridge. The interior of the segmented board is sealed against the alkaline ambience. There are no parts of circuitry or other PCB elements in contact with the KOH lye. The manifolds are completely carried out in PEEK. Such, the active areas of the electrolyser cells are in contact with the segmented Nickel, while the manifolds are in contact with PEEK.
Another beneficial aspect of this design is the scalability. The construction of a larger board would require the production of a larger amount of sensor modules, indisputably, and a correspondingly larger frame, but parts of the works would be possible as serial production. The connection of the segments requires an individual circuitry, but does not require additional electronic components. The wire bonding of the modules to the circuitry, however, would require automated technology. The used resistors are more sensitive and scaled to be able to compensate for a larger lateral cross-conductivity of e.g. electrodes than common in fuel cells, and the used temperature sensors are a miniaturised development.
The total number of data recording channels is 268. Data recording is carried out with a Keysight 34980A data logger, equipped with four scanner modules of 70 channels each, totalling to 280 channels. A total of eight pole plug connectors is used for the voltage signals and the temperatures of the ~1800 cm² cell.
The prototype was „dry tested“ for principal electrical and sensorical functionality by measuring the temperature signal and the reaction to local heating (by a hot fan), and by applying a sample current to each segment and recording the voltage response.
The size of the active area and the resulting high number of segments in the diagnostics bipolar plate unsurprisingly require a large number of data cables. Each signal is transduced in the form of a (low) voltage, and thus the loss of signal quality can be considerable, if the cables transporting hundreds of signals are too long. In consequence, the data acquisition unit has to be placed close to the electrolyser itself.
The safety certifications of the testing site, however, do not allow equipment that is not following ATEX specifications. Safety housing for the data acquisition had to be designed and constructed accordingly.
The modular design has the risk of internal leakage, if the individual modules are not thoroughly sealed, as the first tests reveal. A procedure to test for this property is advised before integration into a stack. A design update to mitigate for this issue is the use of a glue, preferably based on butyl rubber, for the final montage of the modules instead of a O ring seal. This was however necessary for the prototype evaluation. The operational test could not be carried out in the project´s duration. It is however intended nonetheless by the beneficiaries.
The modular concept was awarded with the f-cell award 2018 (2nd place).
2.3. Design and construction of the Segmented Cell for AEM electrolysers (M3-M24)
The design of the diagnostic tool for the AEMWE route is mostly dominated by geometrical aspects. The relatively mild alkaline chemical conditions did not show problematic in the preliminary tests. A challenge to the material is the self-pressurising capability of this technology, which leads to a differential pressure between anode and cathode of up to 30 bar.
The design was finalised after the following technical meetings between HELIO and DLR:
- Stuttgart, Germany, August 2015 (combined with a general meeting between DLR and Heliocentris representatives with the main issue of outlining the formal proceedings necessary after the Bankruptcy of Acta S.p.A.)
- Esslingen, Germany, November 2015 (linked to progress meeting)
- Videoconference, December 2015
- Videoconference, March 2016
The design resembles a regular bipolar plate with additional outside areas for the electrical connectors. Its active area follows the circular shape of the 64 cm² cell. The design of the segmentation pattern has been iteratively adapted to the expected needs, regarding a compromise between resolution, addressable segment area and the focus on specific areas of interest in the cell design. The number of active segments of various sizes for current density monitoring and temperature sensors requires two Keysight, type 34972 data logging units. The following aspects were observed at the first prototype:
- The initially designed PCB thickness induced some stability problems; also the stability against a differential pressure is expected relatively poor. The use of another frame design that had been used for earlier developments allows a higher thickness for the 2nd prototype.
- The contact resistance between current collectors and PCB was observed relatively high – the contact to carbon based GDLs as in PEM fuel cells is more easily established. This may be mitigated for with a higher thickness of the gold coating.
- There is a very high signal cross-talk, that requires a segmentation of the porous contact sheet, too.
The metallic foam, however, could not be segmented with mechanical methods. Milling and machining tests led to smeared out, but still attached material. In particular the experiments to only thin instead of separating the foam showed, that most of the material is compressed by this procedure and thus remains in the sheet. A more successful approach was made by wire spark eroding the sheet. A complete set of segments could be cut out of a regular sheet and glued to a segment printed circuit board. This approach was applicable, but still lacking. The height of the attached foam segments could not be maintained equally This led to an inhomogeneous compression of the foam during assembly and an inhomogeneous contact resistance and local measurement artefacts.
As a final approach, the contact sheet was laser cut, while maintaining uncut corners. This allows to mount the entire contact sheet in one single piece and to keep the distances between the segments constant.Microscopic investigations show, that there are molten remains of the lasered material, but only at the direct interface, while the pore structure beneath is unharmed
3. Polymer Electrolyte Membrane Electrolysers (Work Package 4)
After the earlier works in the work package were mainly on laboratory scale electrolyser cels and the related small segmented PCBs, the work reported in this period is on industrial electrolyser hardware.
The withdrawal of ITM Power Ltd., Sheffield, UK and H-TEC Systems, Lübeck, Germany from the project consortium before the start of the project led to the strategy for the works in the PEMWE route was to purchase a system and modify it according to the project targets. The companies contacted for this purpose are:
- ITM Power, Sheffield, UK
- H-TEC Systems, Lübeck, Germany
- Hydrogenics, Gladbeck, Germany
- Greenlight Innovation, Canada
- IRD Fuel Cells, Odense, Denmark (now named EWII Fuel Cells)
An electrolyser setup, Elze E1050 could be purchased from EWII Fuel Cells. It takes up 5.5 kW, producing 1.0 normal m³/h with a heat dissipation of about 1300 W. It has the capability to self-pressurise to 50 bars. The active area is 76.5 cm², with a circular shape. The required water purity is defined by a conductivity of <2.0 µS/cm (ASTM standard type II). The nominal maximum water temperature at the inlet is 70°C. The nominal lifetime is give as >2,500 hours
The Elze E1050 system is designed with the sole purpose of hydrogen supply, to small or medium customers, such as small living quarters or specific industry with small demand. As a consequence, the possibilities for controlling operational properties are limited. In order to follow the goals as far as possible, and to enable the system to operate under the required experimental conditions at a later point, when the laboratory safety confinements may have been solved, the following upgrades were necessary.
- Water supply bypass: The first attempt to operate a short stack with its reduced flow-through capacity was to establish a bypass that would allow the total water flow rate, monitored by the system, to remain in the safe range. It was observed, that the flow resistance of a bypass was too low. The main flow would be travelling through the bypass, and the short stack would be undersupplied. This behaviour could also be seen by means of the segmented bipolar plate. And the lack of wetting was obvious upon disassembly.
- Pressure booster pump and flow rate meter: As the EWII stack, which is designed for as-of-present unmatched self-pressurising capability up to 50 bar has a very dense structure and requires a pressure in the water supply to establish a sufficient flow. A pressure booster pump was inserted into the water flow circuit directly before the short stack. This pump increases the pressure to 3.5 bar, and can establish a flow through the stack, the control possibilities, however, are limited.
- Controllable pressurising dosing pump: As the booster pump was considered only an improvised solution, and as variable flow rates through the stack had been identified as interesting parameter, dosing pumps that can establish pressure and control flow were purchased.
- Additional heating: The Elze E1050 system does not support preheating of the water, as the full stack operation leads to balanced heating of the entire water circuit. For a short stack, the produced heat is not sufficient, and as experiments with elevated temperatures as stressor were planned, two controlled heating circuits, for heating water supply and stack individually, were added.
- Programmable controllable power supply that allows ramps and quick fluctuations, as the use of fluctuating renewable energies is envisioned for electrolysers in Europe and in operating and testing protocols, and as the present power supply lacks the relevant control properties.
Due to the lacking control possibilities of the Elze E1050 system for operating testing protocols, only some individual stress tests were performed.
As long-term experiments were not possible because of the limitations of the backup laboratory, the main focus was on elevated temperatures and operation with insufficient water supply. The experiments were carried out in a three cell short stack.
Applied stressors:
- Water temperature of 70°C (limited by the possibilities of the pumping system(s)), und not surpassing the nominal limit.
- The stack flow rate was limited by adjusting the bypass valve. Measured flow rate ~0.5 L/min. The insufficient water supply could be monitored by the locally reduced activity of the electrolyser via segmented bipolar plate.
The first prototype was tested in separate test bench, but the second prototype was integrated into the Elze E1050 system directly, as the small test bench that was used for the earlier works was not available and the application in industrial environment was one of the specific goals for the use of segmented bipolar plates. Artificial water starvation was used to evaluate general function, signal cross talk between segments and other issues. The signal cross talk was found acceptable in this range, as it can be estimated to maximally influencing the next neighbouring segment, but none further.
The 2nd prototype was used for monitoring operation under unusual conditions. In particular, the scenario of water starvations was used.
As the electrolyser is equipped several safety mechanisms, monitoring current, voltage and flow to be in a certain boundary, the application of unusual conditions was limited. nThe following scenario shows a startup of the electrolyser, with a very closely following water starvation.
As the electrolyser starts, and the water supply is sufficient to the low consumption under low load, the current density distribution evolves relatively homogeneously. After approximately one minute (Image #3), a slowly evolving hot spot at segment D2 can be seen, while the edge areas are slowly reducing their activity. This continues, while in the outlet area (bottom of the images) the activity stays for a longer while, until the water is entirely consumed in image #7. Beginning at image #6 and strikingly visible at image #7, the hotspot evolves and focuses the activity and thus, the total load of the electrolyser on coarsely one third of the area, until briefly afterwards, the voltage collapses.
The evolving water starvation can be visible by eye with some experience, and can be utilised by a software which monitors outlying values and activates countermeasures at an early stage.
The following critical conditions could be monitored
- Poisoning of the catalyst / membrane can be observed as a slow, steadily regional deactivation of a cell with a clear gradient of the current density. This behaviour was reported in the works on laboratory size PEMWE in the second reporting period.
- Even short intermediate water starvation can lead to local hot spots of the current density. The behaviour is very obvious using a tool for monitoring current density distribution, but not via voltage monitoring, where some fluctuations can have other backgrounds as well.
4. Alkaline Electrolysers (Work Package 5)
Test rig modification
During the early stages, further modifications to the test rig and test centre at NEL Notodden was performed to have full control of testing of 10 cell test stack. It was during the previous period found that it was necessary with some additional modifications to the balance of plant in order to have sufficient capability to perform transient tests of the stack. Parts that have been modified in the last period includes:
• Cooling system for lower cooling rates at low load operation
• Installation of gas analysers with higher sensitivity and better performance at low gas flow rates
• Optimised pressure and level control system.
During operation of cell stacks it has been identified that leakage of electrolyte can occur due to insufficient compression of the stack frames or misalignment during stack assembly. This is also more frequently observed during dynamic operation of the stack, or during testing under conditions outside normal operating boundaries (e.g. accelerated stress tests). Previously it has been difficult and/or hazardous to detect these leaks as it has been necessary for the operator to physically enter the test building to check for leaks. In order to improve safety and leak monitoring capability, a set of monitoring cameras have been installed in the test centre and LCD monitors mounted in the control room. This allows continuous monitoring of leaks without the need of a physical human presence in the hazardous area surrounding the test set up. An image showing an overview of the control room with the monitoring LCD screens are shown in Figure 19.
10 cell stacks which is ready for incorporation of the current distribution measurement tools being developed in the project have been constructed and tested for establishment of baseline performance numbers for comparison of stacks with the measurement tool installed. The stacks are based on NEL’s standard components, including NEL Haflex diaphragm. The electrodes used are commercially coated Ni-mesh electrodes. Images of the electrodes, frames and diaphragms are shown in Figure 20.
To prepare for testing of the 10 cell stack with the integrated segmented cell, it has been tested
in a wide range of operating conditions, including lye concentration, temperature and current loads. As part of this testing, a universal structured characterization approach for stack testing has been developed, based on a series of testing blocks which can be combined to form a complete testing campaign. The details of these testing procedures will be available in deliverable D5.1. A graphical representation of a testing campaign utilizing these blocks is given in Figure 21.
In this campaign, the testing consists of a stack conditioning/activation block, a stack characterization block and a long-term measurement block. A brief description of each block is given below:
• Conditioning step: Start-up of stack and pressurization. Investigation of stack integrity (visual check for leaks, etc.) initial gas purity, voltage and pressure drop.
• Stack characterization: Recurring unit for characterization of performance of stack. Includes polarization curves at several temperatures and pressures as well as online gas analysis of hydrogen and oxygen.
• Long term measurement: Recurring unit which can consist of one of several operating schemes: Steady state operation, batch operation (on/off), dynamic operation (solar, wind) or AST protocols (rapid shut down/start up, elevated temperature or current density).
The conditioning step is only run once per experiment while the recurring units are assembled into a series of blocks which can be repeated until the final length of the test is reached, or another end of test condition is met (normally a maximum impurity level or stack/cell voltage limit).
The NEL pressurized electrolyser stack is constructed with a rubber frame which is compressible and seals the internal compartments from the surroundings by an application of an external compression force. This force can in some operation conditions be too high for the rubber frames without causing some mechanical degradation. In normal commercial operation conditions, this phenomenon is avoided by a combination of proper control of the stack and the stack compression system. However, during stack testing in a development project like INSIDE, especially with shorter stacks, this mechanical degradation has been identified as a process which must be better controlled to avoid the influence on the measurements of the performance of other parts of the cell (e.g electrode activity and stability, dynamic operation and overall internal resistance of the stack).
In order to achieve a better understanding and ultimately to be able to control and avoid this mechanical degradation, a study of the mechanical properties of the frame at different temperatures and compression levels has been conducted and the results of this study has been used to construct a operational zones for the stack which will be followed during stack testing which involves protocols with frequent start-up/shut downs.
The task was completed by running a characterization of a full 10 cell stack. As an example of this characterization, a polarization curve is shown in Figure 23 below. The difference between the performance of the 10 cell stack tested and a commercial 100 cell stack is mainly due to different electrode catalytic coatings, but also due to temperature regulation variations in the 10 cell stack which is not as prominent in a 100 cell stack
A full size alkaline cell will be assembled & tested according to the developed protocol in Task 5.1 and exposed for accelerated degradation conditions. It is planned that stack component degradation will be analysed by academic partners within work package 7. In order to prepare for this testing, identification of the main degradation stressors has been performed and a proposed AST protocol has been constructed.
Start up / Shut down with resulting depolarisation and current reversal is one of the main degradation phenomena, mainly due to the dissolution and precipitation of Ni and Ni oxides as shown in the Pourbaix diagram for Ni – KOH system in 80 °C.
To promote this degradation phenomenon which can cause degradation of both electrodes, current collectors and diaphragm, the following cycling protocol will be used;
● Start/Stop cycling @ 10 barg, 75 °C
● Cycling interval 5minutes (5 minutes shut down, 5 minutes start up)
● Low lye circulation during off periods (to keep temperature high, but leave produced gas in the system,
● no nitrogen purge
● 2000 cycles x 3, with periodic performance evaluation every 168 hours.
The AST protocol will be run until a detectable reduction in performance is observed, the stack will be disassembled and post mortem analyses will be performed by NEL and other partners.
The segmented bipolar plate was integrated at the first position in a short stack. The stack was filled with water for preliminary test, and a leak was detected. It was in the vicinity of the connectors, and thus an internal leak at the segmented bipolar plate itself was to be expected. After disassembly and inspection, two possible faults were identified:
- A slight mismatch in the manifold openings of the segmented bipolar plate could have led to an insufficient sealing. Then, liquid could invade between the Nickel plate and the PCB/PEEK component (see Figure 36, red arrows) and leave the stack at the borders, close to the connectors, where no sealing are present between Nickel plate and PCB.
- A sealing defect in the one (or more) of the Nickel faces of the segments (see Figure 34) could allow liquid to intrude into the segmented Bipolar plate via the active area of the next cell. (see Figure 36, orange arrows)
- As the first scenario appeared more likely and second scenario would require he larger effort, including opening more than 500 screws by hand to disassemble the entire segmented bipolar plate and searching for defective O-rings, it was decided to approach the first scenario.
- The most promising way was to fill the manifolds with a polyester resin and drill out the manifolds afterwards. Experience with such a procedure was present at the beneficiary and could be done with shipping the equipment back.
- After fulfilling this procedure, the tests were unsuccessful, as still water was leaking from the edges of the PCB component in the stack, so the segmented bipolar plate was shipped back to Stuttgart.
- The problem could not be solved within the project timeframe.
5. Anion Exchange Membrane Electrolysers (Work Package 6)
After the preliminary test on 10 cm2 cell active area performed in order to better understand the system behaviour related to variations of operating conditions (temperature, H2 pressure, electrolyte concentrations, current densities), and the preliminary tests on 64 cm2 reported in Periodic Report 1, additional test were performed on higher active area cell.
Tests were performed with the following equipment: the cell was supplied by a TDK-Lamba power supply with current limit to 50 A. The electrolyte circuit is composed by a tank connected to the cell via John Guest tubes and fittings, and a chiller FALC SBF7 to control the temperature. The temperature was measured at the electrolyte inlet. Electrolyte inlet flow rate was 5,91 L/min except for electrolyte flow rate study, and it was controlled with an Iwaki Direct Drive Pump Model RD-40E24-HR1V.
For PC recording the current was manually controlled (one current step every 10 minutes) and the corresponding potentials were measured. The test conditions of the Polarization Curve (PC) were the following:
• T = 35-70 °C
• P= 1,013 - 40 bar
• Current= 0-40 Amp
• IN cell electrolyte flow = 5,91 L/min
• Potential ramp in 64 cm2 cell
The preliminary tests shows results in agreement with the ones obtained for 10 cm2 active area cells, i.e. an influence of temperature increase on cell performances (Figure 1), while at standard operating conditions pressure and electrolyte concentration have no influence on performances. Also different electrolyte inlet flow rates have poor influence on cell performances.
5.1. Test of AEM cells and stacks under accelerated conditions (M17-M28)
The goal of the second study was the study of electrolytic cell performances using 1 wt % Buffer (K2CO3/KHCO3) solution and 1 wt % KOH solution as electrolyte and try to correlate the results with the possible cell lifetime and corrosion. Tests were performed by means of an Arbin BT2000-L testing system provided with 18 measure channels; in order to obtain higher current densities channels can be connected together, to reach the maximum current density of 1 A/cm2. Each channel is controlled and operated via the MITS Pro software: steady state conditions and preconditioning time, current ramp and related cell voltage limit, current steps and dwell time for PC recordings are controlled by the software, and output parameters as cell potential and temperature are also automatically filed. Temperatures are recorded via K-type thermocouples at the electrolyte inlet.
The electrolyte is heated by an Ascon M5 temperature controller and supplied to cells with a SANSO PDH-E054 IT9-CE pump with an inlet flow rate of 9,5 L/min. Two cells with buffer and two cell with KOH were used, one operating in continuous and one operating with start/stop cycles for each electrolyte type. At the beginning of each test a PC was recorded in order to obtain “zero time” parameters. The start-up procedure for all the cells was the following:
1. Current ramp starting from I = 0 A with voltage limited to 2.0 V, H2 pressure = 1,013 bar
2. Preconditioning for about three hours at 45 - 48 °C, pH2 = 1,013 bar, I = 4.32 A.
After the preconditioning one cell for each electrolyte type was operated in continuous at T = 45 – 48 °C, pH2 = 34 bar, I = 4.32 A (corresponding to a current density of 0.45 A/cm2, while the other two cells were operated with the following start/stop cycles:
i. ON ≥ 60 min, T 45 - 48 °C, pH2 = 34 bar, I = 4.32 A;
ii. OFF ≥ 30 min, pump ON, T= 45 – 48 °C, pH2 = 34 bar.
3. In continuous durability tests a big loss in performance was observed for the cell with 1% KOH solution in comparison with 1% buffer solution with a degradation rate of 0.2 and 0.06 mV/h respectively, the KOH showed a degradation 34 times bigger than buffer solution. By the other hand, the decrease in performances can be justified by the decrease of KOH concentration in the solution due to CO2 absorption form air.
4. Figure 5 and Figure 6 show ON/OFF cycle performance on cell with both electrolyte solutions.
5. It is possible to observe a performance difference between 1 wt % buffer and 1 wt % KOH solutions: performances decrease with time in both the cells, but the one using buffer solution as electrolyte showed a faster potential increase with time.
Results showed that the decrease of performances, especially using KOH, is quiet slow, so start/stop cycles are not suitable to achieve a degradation useful for failure mechanisms studies.
5.2. Integration of Segmented Cell into AEM electrolyser (M24-M28)
The PCB was assembled in a 64 cm2 active area single cell (CS471_161108), in order to evaluate PCB resistance to cell assembling pressures and to verify possible electrolyte and gas leakages. At the beginning tests were planned, in agreement with DLR, on a small stack. The cell configuration was the following:
• End plate;
• Insulator plate provided with a bipolar with a loop for connection to power supply;
• # 2 active cells separated by the PCB, with standard MEAs;
• End plate with connection to the power supply.
After the positioning of the endplate, the cell was pressed at 200 bar by means of a manual press, and the screws was tightened with a torque wrench applying a 20 Nm torque. During that phase no signs of PCB damage was observed.
The cell, provided with proper supports (see Figure 7), was then connected to the electrolyte circuit in order to test possible leakages. The electrolyte circuit is composed by a tank connected to the cell via John Guest tubes and fittings, and a chiller FALC SBF7 to maintain the temperature at the operation temperature of 45°C. Electrolyte flow is assured by a SANSO PDH-E054 IT9-CE pump with an inlet flow rate of 9,5 L/min.
Major electrolyte leakages were detected few seconds after the system start-up in the contact regions between the PCB and the active cells frames. In these conditions no gas leakage test was possible.mElectrolyte leakages are mainly due to the rough PCB surface: the traces of electrical connections don’t allow the o-ring to ensure the proper sealing. Different strategies were evaluated to solve the leakage: use of frames of different material (less stiff), use of different sealing (e. g. silicon gasket instead of o-rings), use of a glue to enhance the sealing, in order to avoid leakages without completely modify the frame design. This solution is in fact too much time consuming and it doesn’t agree with project duration. The strategy was to try a glue coupled with the use of a less stiff material frame, because the use of additional gaskets was not suitable for the purpose due to frame configuration.
At a first attempt it was decided to use a silicon glue (LOXEAL 59∙30) with a temperature range of use from -60°C to 300°C. Before the assembly of the cell, the material was used to glue together a frame with a dummy PCB to test the possibility to easily separate the electronic board and the frame, and to test material resistance to the glue. The silicon was then allowed to dry for one night (the time needed is about 8 ours). After drying the frame and the board can be easily separated with the help of a spatula, while the silicon applied on the surface can be detached by hands, with no damage for both the components. The glue was deposited around the o-rings at the electrolyte outlet and at the electrolyte inlet/outlet, and on the external boundaries of the frame, and the cell was assembled again as described above and leaved under pressure overnight to allow the silicon to dry and ensure the sealing.
The cell was then tested for leakages with the system described above, starting with the electrolyte at room temperature. After 30 minutes of electrolyte circulation no leakage was detected, and the temperature was raised to 45°C. Even at operating temperature no leakage appeared, anyway the circulation was continued for about 4 hours, before the gas leakage test described below.
Hydrogen is fed into the stack by a gas line connected to a gas cylinder. The pressure in the stack can be monitored by a digital pressure gauge and controlled in order to be set to the desired value. A piping connects the electrolyte tank to a bubble flow meter used to measure the amount of hydrogen permeated in the liquid. The procedure to detect hydrogen leakage is the following:
1. The cell is pressurized by hydrogen at the desired pressure. After 10 minutes from pressurization an operator checks external hydrogen leaks by using a gas sniffer. All frames are carefully checked over their circumference. If no leaks are present the test continues to the following phase.
2. The hydrogen flow permeating into the liquid lines is measured and recorded by means of a bubble flow meter. The maximum allowable leak for each cell is related to the hydrogen permeability from the membranes for each pressure value. If the measured flow are below the previously mentioned value the leak test is passed.
3. Finally, the cell is carefully depressurized.
In order to avoid possible damages to the PCB due to the pressure it was decided to increase the pressure with step of 5 bar, repeating the procedure described above for each step. The cell was then pressurized to 5 bar. No external leakage was detected by means of the sniffer. The measured hydrogen flow was matching with the one due to membrane permeation in these conditions, thus no hydrogen leakage was detected in the electrolyte flow.
The pressure was then raised to 10 bar. As the pressure started to increase bubbles started to be observed at the electrolyte outlet. Anyway, the cell was tested for external leakage, and again no leakage was detected. The measured hydrogen flow showed that at 10 bar the most part of the gas fed to the cell enter the electrolyte circuit instead of the gas outlet, as expected by the observation of bubbles in the tank.
The hydrogen flow to the cell was stopped, the cell was depressurized and the electrolyte circulation was stopped. The massive hydrogen leakage in the electrolyte must be avoided in order to prevent H2/O2 mixing in the tank when the current will be applied to the cell. Anyway, in order to better clarify the cell behavior respect to gas leakage, it was necessary to test the cell in operating conditions but at p = 1 bar, in order to allow the production of both hydrogen and oxygen. The electrolyte circuit used was the same used for the electrolyte leakage test, but the bubble flow meter was connected directly to the hydrogen outlet.
The current was supplied with a TDK-Lambda GEN 100-50-3P400 power supply connected to the bipolar (positive) and to the end plate (negative). Operation conditions for the tests were the following:
• Electrolyte temperature = 45 °C
• Electrolyte flow rate = 9, 5 L/min
• Hydrogen pressure = 1,013 bar
• Current = 32 A
• Voltage limit = 2 V per cell (4 V)
The operating current of 32 A was not reached, even increasing the voltage limit: this evidence is usually related to a poor contact between MEA’s components. This can be due to different hardness of the PCB respect to standard bipolar plate used to assemble the stack. The problem can be also worsen by the thickness of the PCB, which is about 0,1 mm thinner than the standard BP.
Measured H2 flow rate was compared with the flow rate calculated on the basis of Faraday’s law for the maximum current reached by the stack, taking into account also H2 loss due to the measured membrane permeability.
The measured H2 flow rate was always higher than expected from theory: differences are great also taking into consideration that the test was performed at room temperature and errors due to the use of a bubble flow meter. This result showed that the measured flow is not only due to the hydrogen produced by the electrolysis process, but also to oxygen, and it is perfectly in compliance with the observation made during the leakage test described above. Also this problem can be due, or worsened, by the poor contact between MEA components, or between the GDLs and the PCB, but it can also be caused by damages occurred in the frames or in the PCB.
In order to have a better understanding the stack was disassembled and all the hardware components were carefully observed.
As an effect of the pressure applied during stack assembly the silicon glue tend to accumulate near the o-ring borders and also on electrolyte channels.
In Figure 3 the hydrogen outlet is also shown: it appears plugged by the silicon glue.
The presence of the silicon glue in electrolyte channels doesn’t prevent electrolyte circulation, but it can affect electrolyte diffusion in the active area of the cells. As far as concerns the plugging of H2 outlet, it can be responsible for the massive internal leakage observed during tests. During next assemblies different spreading methods for the glue will be applied, and possibly the silicon quantity will be reduced in order to obtain a good sealing without the risk of electrolyte and hydrogen channels plugging.
Cracks were also observed in the frame of cell #1: the material used for the frame (Noryl) is less stiff but also less resistant to several assembly/disassembly operations. Also cracks in frame surfaces can obviously affect hydrogen leakage. Noryl frames will be anyway used was to assemble the stack devoted to project development, because the design allows to house also PCB thicker than 1 mm. Increasing the PCB thickness can be a strategy to enhance the board resistance to stack pressure, thus increasing cell sealing and facilitating the contact between MEA’s components. In future test frames will be replaced at every stack disassembly in order to prevent damages. Furthermore, after stack assembly the glue will be allowed to dry at ambient pressure, given that the pressure applied overnight possibly enhance the PCB bending and frame rupture.
In order to enhance contacts between MEA’s components new GDLs with different thickness and hardness will be developed.
As far as the electrolyte and hydrogen leakages on the stack assembled with PCB were solved by using and additional sealing system as described in the Periodic report related to Period 2, operation tests on stacks assembled with PCB were performed in order to understand possible influence of the diagnostic tools on stack performances and if it is possible to operate the stacks in company standard operating conditions. The PCB was assembled in different 3-cells stacks with a cell active area of 64 cm2 and in a 19-cells stack (see Figure 1) with the same active area. All the hardware used in stacks is the standard for stack assembly. The test bench used for the tests on 3-cells stacks was the following:
• A tank with a capacity of 4 L equipped with a Pt1000 resistance for electrolyte heating and a visual electrolyte level control system.
• A SANSO PD-E51-IT6 pump
• John Guest tubes and fittings to connect the tank to the pump and to the stack
• A Cotek AK-650-24 to supply both the pump, the temperature controller and the Pt1000.
• a TDK-Lambda GEN 100-50-3P400 power supply to supply the stack while two different devices were used for heating depending on the test:
• A chiller FALC SB7 connected to the pump/stack with John Guest tubes and fittings.
• A WEST N6500 controller used for durability tests (the chiller water reservoir is not enough to provide heating during the whole week end without refilling).
The test bench used for the 19-cells stack consisted of a stack module equipped with a control module (see Figure 2), i. e. an EL250 system modified for INSIDE test. The main modifications were the following:
• The cooling fans control was modified in order to allow the stack temperature variation up to 55°C;
• The tank was directly connected to a chiller FALC SB 7 in order to allow the stack heating up to 55 °C and fast cooling;
• The “low stack pressure” alarm threshold was modified in order to decrease the pressure down to 10 bar.
Unfortunately major hydrogen leakages occurred during the tests performed on 19-cells stack operated at standard pressure, as well as in the 3-cells stacks, so the tests on the 19-cells stack were stopped waiting for the prototype 1.1 while tests on small stacks were performed at reduced pressure, 21 bar instead of standard 30 bar, given that the pressure is one of the parameters that shoes less influence in stack performances, in order to avoid leakages and possible damages to the PCB.
Tests in operating standard conditions (electrolyte temperature = 45 °C, electrolyte flow rate = 9, 5 L/min, hydrogen pressure = 21 bar and current = 32 A) were performed on a 3-cell stack without the PCB and on different 3-cells stack assembled with the PCB, in order to understand possible influence of the diagnostic tool on stack performances. Results in Figure 1 shows that the PCB assembled in the stack doesn’t introduce deviations from the standard stack behavior, and also that the assembly of the system integrated with PCB is reproducible, a result not necessarily predictable given that the additional sealing system is hand worked.
Results of tests on stacks assembled with the PCB (discussed in details in the next paragraphs) showed that the resolution of the obtainable data is not satisfactory, due to the lateral conductivity of the contact sheet between electrochemically active layer and bipolar plate, which lead to a prototype #1.1 with an integrated segment GDL. For technical reasons due to GDL cutting and GDL manufacturing tools, it was possible to integrate the segmented GDL only on the cathode side, so tests were performed with the PCB assembled on the cathode side in order to facilitate the comparison between the different prototypes.
5.3. Test of the Segmented Cell under normal and accelerated conditions (M27-M46)
Tests on the integration of a PCB in AEM stack according to the protocols described in D8.1 were performed, with minor changes from the protocols, on a 3-cells stack (AES565_180215) assembled with the PCB I (prototype 1) on the cathode side. The cathode side was chosen because the new generation PCB with integrated segmented GDL will be available only with GDL suitable for cathode side of the cell.
Respect to the test protocol foreseen in D8.1 the dynamical behavior of industrially hardened fats cycle was skipped because the stack load profile is not defined, while the dynamical behavior solar cycle was not performed because the power supply must be operated manually, so it is not possible for the moment to perform tests with a timeline longer than the personnel working hours. For the same reason the polarization curves were performed taking into account only the mandatory load values.
Again, for the same reason, during the test on operating parameters sensitivity the stack was put in off conditions (I = 0 A, p = 1,013 bar, no electrolyte circulation) overnight and switched on and preconditioned every morning.
As far as concerns the polarization curves, in order to obtain the stack pressurization it is needed to have hydrogen production and it is difficult to maintain the stack at the operative pressure when the production is stopped (i. e. at I = 0 A).
This is due to the use of a simple relief valve in the pressure gauge and also to the particular configuration of the stack: even if the 3 M tape can avoid electrolyte and external H2 leakages, still small internal H2 leakages are present, even if in the range of acceptance limits fixed internally for quality control purposes. During all the tests anyway the measured H2 flows were compared with the nominal flows calculated using Faraday’s Law in order to check the possible increase of internal leakages. For this reason the polarization curves were performed on reverse respect to the protocol, i. e. starting from 100% of maximum current density and going down to 0% before to raise again to 100%.
The polarization curves with the mandatory points listed in table 3 of D8.1 (in the following text called “long PC”) were performed after stack preconditioning, after every test on operating parameters sensitivity and before and after the durability and the ageing tests. For the test on operating parameters sensitivity a shorter polarization curve (in the following text called “short PC”) with points corresponding to 100%, 75%, 50%, 25% of maximum current density were recorded for each value of the parameter under study.
For every step of all polarization curves the step time was 6 minutes, as suggested in D8.1: 4 minutes of stabilization time and 2 of analysis time: only the data recorded in the last 2 minutes were considered in calculation of the parameters mean values.
The pressure was measured with a pressure gauge, so it was impossible to acquire the data number requested for standard deviation calculation. The same is valid for H2 flow, measured manually with a bubble flow meter, especially for low currents, because flows are so low that is impossible to have more than one measure in 2 minutes.
The stack was assembled using insulator plates on both endplates and with special bipolar plates on cells # 1 and 3 connected to the power supply: these bipolar plates have four loops distributed evenly on the circumference in order to distribute as homogeneously as possible the current supplied on all the bipolar surface. These bipolar plates in principle are useful to prevent damages to both the cell and PCB due to uneven current distribution.
The peculiar stack configuration with the PCB substituting a BP doesn’t allow the measure of single cells potentials, so only the total stack voltage was measured, via the PCB, during tests,
The stack was pressurized using a pressure gauge equipped with a Hamlet relief valve with pressure range from 3.4 to 24 bar and the flow was measured using a bubble flow meter.
The electrolyte circulation system and other hardware used for tests are described in the previous paragraph
In order to prevent stack damages during the test a safety potential limit was fixed on the power supply: the limit is the same used to protect the stack during the current start up ramp, and it is 2,3 V per cell, corresponding in this case at 6,9 V. The power supply stop to supply current when the voltage limit is reached.
The stack preconditioning is composed of a first heating step of one hour with the stack switched off and with electrolyte circulation on, in order to stabilize the stack temperature to the standard operative temperature of 45°C. When the temperature is stable the stack hydrogen production is started performing a current ramp up to 32 A with the potential limit described above. When the stack was operating at 32 A, it was again checked for external H2 leakages using a sensor (this checked was repeated daily for all the tests duration), and for internal leakages by measuring the flow with the bubble flowmeter, then the pressure was raised up to 21 bar. The stack potential was then allowed to stabilize for at least one hour before starting the test. After stack preconditioning a long PC was recorded. The first tests performed were the one related to operating parameters sensitivity, needed as stack characterization tests. The stack was preconditioned as described above before the test on each parameter.
After the recording of the first PC, the electrolyte temperature was lowered in order to have a stack temperature of 25°C, taking into account the stack self-heating due to the current. When the temperature and the other operating parameters were stabilized, a short PC was measured. The temperature was then raised to 35°C, 45°C, 55°C allowing the operation parameters to stabilize before the short PC. A long PC was performed again after the stack temperature was decreased to 45°C.
Given that the curves for ascending and descending power show no hysteresis, polarization curves comparison for every operation parameter considered are shown for ascending current only, in order to simplify the graphical rendering.
Polarization curves recorded at different temperatures show, as expected, a significant increase of performances increasing the temperature. The operation temperature of 45°C was chosen in order to find a better compromise between stack performances and durability, especially respect to membrane degradation.
The next test performed was the one on pressure variation sensitivity. Operation pressure for the current technology is 35 bar, but the peculiar configuration of the stack suggested to operate at lower pressure, 21 bar, in order to prevent damages to the PCB or hydrogen leakages. The pressure in the stack is checked and maintained using a pressure gauge connected to the hydrogen outlet and equipped with a relief valve and a two-way valve to pressurize and de-pressurise the stack.
Short PC was measured for each pressure value, 21, 14 and 7 bar, and a long PC was performed at the end of the pressure variation test after the stack was put again in standard operation conditions and allow to stabilize. Other operation parameters are standard.
The pressure, as shown in Figure 3, doesn’t affect stack performances. Higher pressure are preferred in order to reduce hydrogen humidity.
Tests on variation of electrolyte concentration was performed at standard concentration, i. e. 1 Wt % KOH, and at 0,5 Wt % concentration.
Similar results were obtained with a 2-cell stack (AES561_180108) assembled with PCB III (prototype 1.1). Unfortunately with prototype 1.1 it was not possible to complete the tests in accelerated conditions due to a failure in a PCB segment after about 172 hours of continuous operation in standard conditions, while with the stack AES565 the tests were continued as described in detail below before the same kind of failure happened.
Stack AES565 was tested for:
• about 240 hours durability test in standard operating conditions
• about 190 hours durability test at maximum operating current (I = 47 A)
• about 37 hours start/stop test as described in D8.1.
Figure 8 shows a comparison in PC recorded for stack AES565 after different test step: after stack preconditioning, i. e. with the stack operated for few hours (Ustack = 5,240 V), after the characterization tests (Ustack = 5,215 V) and after the durability test at high current (Ustack = 5,370 V). The test was supposed to be stopped when the potential limit of 6,9 V was reached, but degradation was to slow to allow the potential to reach the limit, so it was decided to switch to the start/stop test, including also a temperature shock, in order to try to accelerate the degradation. The manual operations on the power supply didn’t allow to perform cycling overnight and during the week end, the temperature controller used instead of the chiller didn’t allow the temperature to remain stable at 45°C overnight and during the week-end, thus affecting also the stack potential. Unfortunately the unstable potential value didn’t allow an evaluation of stack degradation, and the PCB failure during test didn’t allow the recording of a PC.
5.4. Identification of electrochemical behaviour under critical operating conditions (M29-M48)
The most critical condition for the stack is to operate at standard current and temperature without electrolyte flow, which is also acting as a coolant: without electrolyte flow stack temperature increases causing at the same time membrane drying, which can result in serious damage to the membrane itself. For this reason standard electrolysers are equipped with a safety control that switch off the current when a critical voltage (usually 2 V per cell) is reached. A starvation test was performed in AES561 stack (2-cells stack equipped with PCB III (prototype 1.1) on cathode side. For this test the potential limit was set higher, up to 2,3 V per cell, in order to preserve (if possible) the stack for further investigations. With the power supply used for the test the current is not abruptly switched off, but it is lowered by the power supply in order to allow the potential to decrease below the limit.
The stack was allowed to stabilize at standard operating conditions (in this shorter test the pressure was set at 30 bar) for about 3 hours before the starvation test, that was performed by switching off the electrolyte pump leaving the stack at standard pressure and current. The stack potential was not strongly affected by the lack of electrolyte until all the water remaining in the cells (basically trapped in GDLs) was evaporated. After about 45 minutes the potential started to raise up very quickly together with the temperature, until the potential limit of 4,6 V was reached and the power supply started to decrease the current. A graph showing stack voltage and temperature behavior during the starvation test is shown in Figure 10, while in Figure 11 is shown a current density map recorded before the start of the starvation test compared with one at the end, i. e. when the current started to decrease.
When the current started to decrease the pump was switched on again, and after few minutes, when the stack hydration was completed, the stack was operating again with the same parameters recorded after preconditioning (32 A, 30 bar, 44,9°C, V = 3,51 V after preconditioning vs 32 A, 30 bar, 44,4°C, V = 3,52 V after the pump restarting).
6. Investigation of Failure Mechanisms (Work Package 7)
Ru-, Ir, and mixed Ir-Ru anode test materials were successfully prepared and integrated in the MEAs of a PEM electrolyzer
Anode materials for a PEM electrolyser were characterized ex situ before and after water electrolysis with XRD, TEM, STEM, XPS
The experimental tools for in situ studies of electrochemical systems with Near-Ambient Pressure X-Ray Photoelectron Spectroscopy (NAP-XPS) have been successfully developed and approved using model systems
Application of NAP-XPS to the anode of a PEM electrolyser allowed us to unveil the degradation mechanism during electrolysis and suggest the stabilization mechanism
Preliminary corrosion tests of raw FR4 material for PCB production were carried out under AE conditions
PEM-based MEAs comprising various unsupported and supported Ru-, Ir, and mixed Ir-Ru anode test materials were prepared by the DLR partner and investigated at the beginning of life and after ageing by CNRS-UdS and UAES partners using various ex situ (XRD, SEM, EDX,TEM, STEM, XPS, AFM, Kelvin probe force microscopy –KPFM, etc.) and quasi in situ approaches (conducting AFM, NAP-XPS, soft X-ray Absorption Spectroscopy);
Application of NAP-XPS and soft X-ray absorption to various types of Ir-based anodes of a PEM electrolyser allowed us to unveil the anion-type (OII-/OI-) red-ox oxygen evolution reaction mechanism and propose a tentative origin of higher stability of Ir- relative to Ru-based anodes, the latter operating through cation red-ox (as proven during the 1st reporting period);
Degradation/restructuring of ionomer in the PEM-based MEAs after (during) operation has been evidenced with material-sensitive, conductive AFM measurements performed by UAES partner and NAP-XPS measurements performed by CNRS-UdS partner;
Ionomer loss in the electrodes after operation has been detected by UAES partner using material-sensitive, conductive AFM.
UAES partner set up a fluid cell measurements technique to monitor the material/component degradation of membranes, electrodes or half-MEAs, and tested this approach on model proton-exchange membranes/MEAs;
UAES partner investigated thin films of anion-exchange ionomer with conductive AFM tapping mode (TUNA) measurements and discovered swelling and increasing conductivity of the film caused by the current flow.
The AWE conditions require the protection of any copper surface and the protection of cut, drilled or machined PCB edges against the invasion KOH lye. The PEMWE and AEMWE conditions did not show problematic corrosion.
Nickel contaminations in the membranes of operated PEMWE could be locally correlated to the reduced activity of this cell.
Ageing of industrial PEMWE samples could be attributed to ionomer distribution and mechanically induced cracks.
The stability of segmented PCB was investigated. The corrosion in heavy alkaline ambience was a triking problem, but only when cut edges oft he PCB material is in contact. In dilute alkaline ambience, this was not a problem, but Rhodium traces, presumably from the diffusion barrier below the gold coating, was found on teh segmented bippolar plate for AEMWE.
7. Test Protocols and Control Strategy (Work Package 8)
A literature study was performed what testing protocols are used in research groups over the world for alkaline, PEM and AEM electrolysers. Existing international standards for electrolysers were evaluated if they could provide descriptions of tests. In addition previous FCH-JU projects stating that they had developed their own testing procedures were contacted and asked if these procedures could be made available. However these procedures were usually not publically available. A contact with European international activities on harmonisation of electrolyser testing protocols was established, especially IEA Annex 30 and JRC electrolyser testing harmonisation activity.
The only already established international standards for electrolysers are two international standards for water electrolysers released by ISO: ISO 22734-1 Hydrogen generators using water electrolysis process — Part 1: Industrial and commercial applications and ISO 22734-2 ... Part 2: Residential applications. The tests described there are for verifying the reliability of the electrolyser system. All tests are described in a very short form and cannot be well used for the present case of collecting information from processes in the electrolyser.
Searching for some typical electrolyser application power profile two cases were identified: industrial electrolysers are operated according to the hydrogen demand following a daily/weekly profile. This applies e.g. for electrolysers used in the industrially hardened fat industry. Electrolysers fed by renewable energy have to follow the power supply profile which is quite dynamic and shows also long periods without operation. As an example the power profile of a photovoltaics field on an April day in Germany is given assuming direct use of the solar power respectively testing the maximum dynamics required for applications in this field.
The primary degradation mechanisms reported by researchers are electrodes degradation for all types of electrolysers, membrane degradation (thinning, ion exchange, loss of conductivity) for PEM and AEM, anode corrosion including diffusion layer, connection and bipolar plate corrosion and temperature variations causing leaking, corrosion and failure in BOP. High temperatures, periods of OCV, high current density and temperature variations are named as causes for degradation.
Based on results from literature and the project participicants’ own experience basic testing protocols were suggested and discussed among the partners. In line with previous FCH-JU project testing protocol activities established for PEM fuel cells and using the experience and verified wording from these projects, test modules and test procedures were defined. A test module is a basic piece of test described well in detail. A test procedure is a sequence of test modules making up a test in a practical experiment.
The basic test modules defined are
• Polarisation Curve
• Operating parameter X sensitivity (with X standing alternatively for temperature, pressure, differential pressure, liquid flow anode, liquid flow cathode, concentration)
• Steady State Endurance
• Dynamic Behaviour Industrially Hardened Fats Cycle
• Dynamical behaviour Solar Cycle
The test modules are written in such a way that they can be applied to all types of electrolysers differentiating for the different way of operation wherever necessary within the test module description.
These test modules are then combined into suggested procedures for basic characterisation for the different types of electrolysers. The purpose of this basic characterisation is to evaluate the quality of performance of the electrolyser and to collect data for internal processes (using the new current density distribution measurement) during normal operation of an electrolyser.
A literature study was performed what degradation mechanisms are especially relevant for which type of electrolyser, how these are investigated and if there are accelerated tests available.
From the literature survey it is not exactly clear what states of the electrolyser will contribute to what degradation effect. Also no public accelerated stress tests (AST) are reported. However it seems to be clear that several conditions can accelerate degradation in general: these are prolonged operation at high current/power density, voltage-controlled dynamic operation (reported for PEM electrolysers), increased impurity levels in the water (for membrane electrolysers), long periods of OCV for the stack (especially at higher temperature) and frequent temperature variations as well as higher temperature operation. Furthermore during highly dynamic operation fast fluctuations in the system might be induced by inadequate control (e.g. in relative pressure, KOH supply) that might as well induce degradation. Tests along this line are suggested for this project. For different types of electrolysers different degradation mechanisms seem to be more relevant giving somewhat different accelerated testing protocols suggested for the electrolysers.
Based on results from literature and the project participicants’ own experience accelerated testing protocols were suggested and discussed among the partners. It was decided that it is necessary to show a clear warning that these tests should not be applied to expensive equipment that should still be used after the test because the tests might be very destructive to the performance. Preferentially such tests should only be applied to single cells and small stacks.
For alkaline electrolysers the following accelerated testing protocols are suggested and written:
• Long term operation at constant maximum current (several periods of steady state endurance test at standard operating conditions for 200 h at high current density alternating with polarisation curve characterisation)
• Accelerated start-stop test (repeating 30min on, 30 min off with accelerated heating/cooling)
• Long term operation at increased temperature (several periods of steady state endurance test at maximum temperature for 200 h alternating with polarisation curve characterisation)
• OCV at increased temperature (several periods of steady state endurance test at 0 A for 200 h alternating with polarisation curve characterisation)
For PEM electrolysers the following accelerated testing protocols are suggested and written:
• Long term operation at constant maximum current
• Voltage-controlled dynamical stress test (30 seconds at voltage corresponding to a cell current of 1 A/cm2 during the initial polarisation curve; 30 seconds 0.1 V; sets of 200 cycles)
• Long term operation with increased water impurity levels (using redistilled water with 1 mg/l Ca2+, 0.1 mg/l Cu2+ and 0.1 mg/l Fe2+ added)
• OCV at increased temperature
• Accelerated start-stop test
• Long term operation at increased temperature
For AEM electrolysers the following accelerated testing protocols are suggested and written:
• Long term operation at constant maximum current
• Accelerated start-stop test
The test procedures have not yet been verified. Verification experiments will show if the tests can be operated as defined here in practical laboratory operation, if they will give reproducible results and what type of degradation mechanism is accelerated by these tests. For final evaluation of such a test a very long term operation of an electrolyser at normal conditions and extensive characterisation during operation as well as after use would be necessary to compare the accelerated degradation achieved to the regular degradation phenomena. However such a comparison cannot sufficiently be done during this project due to limited time and different focus of the project.
The combined evaluation of these protocols was not possible to this point.
As a contribution to the JRC´s activities in harmonised electrolyser testing, the consortium shared their generic testing protocols with the JRC as concrete starting point for more detailed definitions.
Potential Impact:
The project is directly aimed towards improved performance and lifetime of the state-of-the-art water electrolysis technologies available. The creation of a tool which monitors the electrochemistry in-situ and on-line will enable the manufacturers to find improved operations conditions with higher efficiencies. Moreover, hazardous operation mode can be identified and methods can be developed to anticipate and avoid critical conditions, extending the lifetime and sustainability of water electrolysers.
The changes currently going on in the European electricity market will demand increasing decentralised capacities for energy storage. The use of hydrogen as an energy carrier will not only reduce the dependence of the European countries on oil and gas exporters. It will also provide a location independent method of storing excess electricity, and in combination with fuel cell and turbine power plant it will provide even a method of redistributing the stored energy. The increasing incorporation of solar, hydroelectric, and windmill plants with their unsteady output profiles into the European electricity grid will require the ability to operate with variable load in a reliable manner. In particular, frequent load changes are a challenge to the lifetime and performance stability of electrolysers, and thus the monitoring of these systems is an existential issue for the reliability and tunability. The capability of the segmented cell tool for on-line monitoring the internal electrochemistry is the key property for the diagnostics of load changes.
The creation of the in-situ tool for on-line monitoring the electrolyser performance will improve the overall efficiency of electrolysers and will lay path for a decentralised use of electrolyser technology for the stabilisation of the European electricity grid and the provision of hydrogen as an environmentally friendly fuel. This development and its impact on the energy policy will emphasise the leading role the European Union in the world in sustainable energy technologies.
Wider societal impacts are indirect and related to the general willingness of European society and policy makers to push forward the transition from fossile to renewable energies and sustainable technologies in general. This project is adding another bit to this big puzzle.
The dissemination activities for this project are mostly research and industry related. On top of publications, conferences and workshops, the winning of a an award and the corresponding ceremony during the f-cell conference soirée was probably the most public dissemination event, and was attended attended by representatives from industry and research, and also from politics and press. Another unexpected dissemination activity, a page in the official calendar of the German Aerospace Center was unfortunately stopped by a decision of the headquarters.
The exploitation of results by the individual beneficiaries is mostly depending on the nature of their institutions/ companies:
DLR
The German Aerospace Centre as holder of patents for segmented boards for PEM fuel cell diagnostics and grantor of licenses to industrial companies, e.g. S++ Simulation Services, has a very high interest to continue this by creating patents and grating licences on similar developments in the field of electrolysis technology. It is the task of the German Aerospace Centre to advance technologies towards usability, and not to bring them onto the market itself, but enable and motivate industry to do so. In addendum to the financial flowback by licence costs, which may be ideally used for initial research, the patents are used as performance figures of department and institute. Patents also secure the high-tech position of Europe and Germany, which is another mission of DLR. Concretely, two technological advances in the field of monitoring of electrochemical converters are envisioned.
The technological advance created in this project is always used as basis for future R&D activities. As DLR is active in adapting the segmented board technology also to other electrochemical converters, each bit of experience will carry to further activities. The project INSIDE is a part of DLR´s strategy to extend and strengthen the position in the field of in-situ and in-operando analytics.
As a research institution within the Helmholtz Association, DLR has the interest of publishing and presenting created foreground. Successful high quality dissemination is internally a performance figure and externally a medium for developing collaborations and new projects, and based on that, funding possibilities. Thus, it is essential for the existence as a research institution to address the scientific public and the non-scientific public as well. Concrete research ideas are the collaboration with the University of Aarhus with the goal to correlate modelling results on PEMWE hardware with experimental in-operando data, and to evaluate operation strategies together with the University of Lappeenranta.
Both the technological basis and the scientific reputation from the created foreground will be the background for acquisition of third-party funding and thus one of the supporting pillars of the German Aerospace Centre and its future activities. Concretely, a R&D project is planned, together with Enapter S.r.l. (the beneficiary initially named Heliocentris Italy in this project), which will make a direct link from INSIDE to the public and is expected to generate a significant societal impact.
NEL
NEL will use knowledge developed during the project in its R&D activity focused on development of new extremely flexible alkaline water electrolysers with improved energy efficiency. The tool will serve the development of new generation of electrocatalytic compositions based on recent achievements of nanotechnology. The tool will be an instrument for the detection of possible failures in advanced cells in-situ instead of making several post mortem analyses and therefore will reduce time and cost of development of new products.
Utilization of the diagnostic tools will allowed to reduce development of new products in 2-3 times, by reduction of time for ex-situ testing of materials for both electrodes and diaphragms, reduction of expenses for verification of mathematical modelling of heat and flow distribution along and across of the cells. So, direct reduction of the cost is expected at the level of 0.4 million euro/year. Share of new products in the profit will be increase and generate profit increase of app. 20%.
Enapter S.r.l. (formerly Heliocentris ITaly, S.rl.l. HELIO)
Heliocentris Italy S.r.l. offers a unique technology and very innovative system devoted to UPS saving system and a variety of off-grid applications. The devices can be coupled with a variety of renewable energy sources and must be able to resist also to adverse environmental conditions such as high temperatures and/or humidity. Given that the technology is new, a further insights on lifetime and failures modes are needed in order to improve devices durability and reliability in stress conditions described above (environmental conditions but also operation with discontinuous and non-stable energy supply form renewable sources), and the diagnostic tool developed during the project is a very powerful tool to obtain the information needed to reach the goal, in addition reducing times devoted to development. Information obtained with the diagnostic tool will be also relevant to improve stack performances reducing at the same time to production and material costs, especially in view of the development of a new stack generation starting from the beginning of 2018. The new generation and the cooperation with DLR is currently going into anoher project with an expected societal impact.
CNRS / UdS
CNRS-UdS is a research institution and will participate in dissemination measures, including any scientific publications relating to foreground and its content will be made available in the public domain thus demonstrating the added-value and positive impact of the project on the European Community. CNRS-UdS has the interest of publishing and presenting results of the research and development activities in this project in international scientific journals, such as Electrochimica Acta, Electrochemical Communications, Journal of Power Sources, Journal of Solid State Electrochemistry, etc., and conferences and exhibitions, such as meetings of the Electrochemical Society (ECS), International Society of Electrochemistry (ISE), Gordon Research Conferences (where E. Savinova will act as a chair of GRC “Fuel Cells” in 2014). Technology improvements resulting from this project will be protected by patents jointly with the industrial partners.
UAES
UAES is a University of Applied Sciences and dedicated to teaching and applied research. UAES has strong interest in publishing the scientific results (considering the confidentiality issue) in peer reviewed international journals for driving the expertise in the research area of ion exchange membranes and catalysts. A successful cooperation with the DLR where nanoscopic analysis of MEAs was performed by UEAS before and after Electrolyser operation has led to several common peer reviewed papers. Further publications on degradation detected in a cell with segmented PCB are in preparation. Talks at international conferences, i.e at the ISE Spring Meeting in Buenos Aires and the Hyceltech in Porto, strengthen the scientific landscape and the institutes profile and were used to install further international cooperation. .
As a second goal, UAES is dedicated to education, also in the field of renewable energy. Since 3 semesters, a lecture including experiments has been established within a master course for students interested in this field. The project activities provide opportunities for young researchers and students. This includes an outstanding education leading to PhD degrees, scientific international exchange, and excellent job opportunities. A facebook contribution and a video of INSIDE activities dedicated to reach the younger generation via youtube have been produced and will be updated and complemented.
List of Websites:
fb.me/insideelectrolysis.eu
The original website is temporarily unavailable.
Further contact details:
Indro Biswas
+49 711 6862 603
indro.biswas@dlr.de