- Email: [email protected]
Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress B. Verdin a,b,c,*, F. Fouda-Onana a,b, S. Germe a,b, G. Serre a,b, P.A. Jacques a,b, P. Millet c Universite Grenoble Alpes, F-38000 Grenoble, France CEA-LITEN, F-38054 Grenoble, France c Universite Paris-Saclay, Institut de Chimie Moleculaire et des Materiaux d'Orsay, 91400 Orsay, France a
b
article info
abstract
Article history:
Proton Exchange Membrane Water Electrolysis (PEM WE) is considered as a key technology
Received 5 July 2017
for large scale storage of renewable electricity, using hydrogen as an energy vector. To
Received in revised form
achieve economically viable MW-scale storage units, there is a strong trend to increase
25 August 2017
both current density and electrode surface area. Such upscaling raises concerns regarding
Accepted 26 August 2017
the homogeneity of compression forces and current lines distribution throughout the
Available online xxx
electrolysis stack. Mechanical and thermal stresses resulting from any heterogeneity can result in faster local ageing and accelerated degradation. Thus, there is a major interest to
Keywords:
measure and understand the nature of such coupled heterogeneities. We report here on
PEM water electrolysis
results obtained with a non-disturbing tool for the current and the temperature mapping of
Current distribution mapping
PEM WE electrodes. A complete description of the setup is discussed in order to avoid some
Mechanical stress
mapping artefacts that could come from the measurement device. Measurements made on 250 cm2 surface area cells show that the clamping pressure homogeneity over the entire geometrical surface area of the cell is one of the first mandatory requirement for an appropriate operation of electrolysis cells. In particular, it was found that even homogeneous pressure distribution could lead to excessive current maldistribution in some cases, due to overriding inhomogeneous electric contacts. This situation has been corrected by using non-flat end plates and/or an accurate thickness of the gaskets. Current mapping has been performed. Results are discussed. © 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Hydrogen is considered as an appropriate energy vector to implement a carbon-neutral economy and to face to the demand for increasing amounts of clean primary energy sources [1]. Hydrogen can be used under various forms and for a wide panel of applications. However, today, hydrogen production is
not sustainable and compatible with this perspective of a carbon free energy society. Approximately 96% of the global hydrogen demand is satisfied by fossil fuel transformation, mainly Steam Methane Reforming (SMR) of oil and natural gas, and coal gasification. These processes are from far the main techniques used, but also the most polluting ones [2]. On the contrary, water electrolysis offers a sustainable and efficient way to produce hydrogen from abundant liquid water
* Corresponding author. 17 rue des Martyrs, 38054, Grenoble, Cedex 9, France. E-mail address: [email protected] (B. Verdin). http://dx.doi.org/10.1016/j.ijhydene.2017.08.189 0360-3199/© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
2
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
and clean energy sources. Accounting for the major part of hydrogen produced by water electrolysis, alkaline technology is the most mature, and commercially advanced [3], whereas Solid Oxide technology is mostly under development. But PEM (Proton Exchange Membrane) electrolysis is now on the way to become cost-competitive. The technical advances achieved up to now enable to reduce the system cost and to consider optimistic perspectives [4,5]. However, one of the critical point is related to system costs compared to life expectancy [6]. On one hand, the use of an acidic electrolyte and the high operating potential of the anode and temperature, and on the other hand, material optimization, especially membrane thickness, are now coming to a harsh point where a lot of efforts are required for small improvements. To optimize PEM water electrolysis technology and extend the lifetime of the cells, there is a need to better understand the ageing of individual cell components. First of all is the MEA (Membrane Electrode Assembly), in the closest conditions from the real stack operation. Then focusing on the R&D effort within the resulting conclusion to finally implement the more fundamental breakthroughs in terms of coatings, porous transport layers, membranes and catalysts accordingly. That is the way to make those advances significantly efficient. Having this in mind, we investigated the current distribution on a large surface area (250 cm2, circular shape) PEM water electrolysis single cell, a surface sufficiently large to analyze mechanical, thermal and fluidic management constraints that could be encountered on a large commercial stack [7e9], and subjects of intense developments from industrials. As explained above, particular focus will be made on ageing and degrading conditions. An important preliminary work has been done on the experimental set-up in order to implement the current distribution visualization device in the cell. The internal segmentation of the components has been first considered as the path that the electrons should follow. Quite inspiring work was done for fuel cells and we have found some correlations with our application. Various methods have been previously studied for PEM FC [10] (very few on PEM electrolyzers): MEA segmentation [11,12], flow field segmentation [13,14], shunt resistance [15], segmented PCBbased Bipolar Plates (BPP) [16e18], Sþþ® current measurement plates [19,20]. Especially a technique derived from the one of Ghosh et al. [15] has been considered to be adapted for our electrolysis purpose for its flexibility and easy evolution. But finally, given uncertainties of implementation, the most viable approach to stick as close as possible to real stack operation appears to be by using commercialized Sþþ® current measurement card. As far as we know this communication is the first to establish and use such a current and temperature mapping device for PEM WE. This communication will detail the experimental set-up used, and then describe an application of the tool for a local operando study of the cell dysfunction.
Experimental Cell specifications In order to perform reliable in-situ and operando measurements, we used a fragmented cell. Several specifications must
be fulfilled to perform not too intrusive measurements. From the mechanical viewpoint, the fragmented cell should be robust and rigid enough to support 1.8 MPa mechanical clamping pressure and also to transfer this compression constraint completely to the MEA, without absorbing any substantial part of the mechanical force. To perform testing under pressure, we fixed a minimum of 5 bars internal pressure at the cathode. From the fluidic viewpoint, the device must first be watertight with the ambient and internally: the introduced measurement device must not affect water- and gas-tightness at the risk of creating hazardous leakage and mixtures. Lastly, electrically speaking, it is needed to drain out the voltage for each segment independently of the running cell power.
Experimental set-up The single cell used to perform our measurements was initially designed for MEA characterization at low current densities (j < 0.5 A/cm2). Consequently, some mechanical optimizations were required to obtain a satisfying electrochemical behavior at higher current densities. The cell is 250 cm2 active area, and composed of various elements detailed in Fig. 1. End plates, current collectors and sealant gaskets have been resized and redesigned. Design is circular as in most cases for PEM WE mainly due to its robustness towards high gas pressures and potential leaks. Porous Transport Layers (PTLs) are made of two main pieces, eg meshed titanium wires layers and porous titanium sinters. A gold-plated electronic Printed Circuit Board from Sþþ® Simulation Services Company has been used for current density measurements. A specific circular design of the Sþþ® Current Scan Shunt measurement device was developed. This type of measurement relies on the measurement of the voltage drop over calibrated micro-resistances. The cell shown in Fig. 2 contains 64 electrically independent segments of equal surface area (3.89 cm2 each) positioned radially from the center. The interest of this design is that comparative measurement can be performed between inlet and outlet, center and edges, along different radii. In a stack configuration, current propagation is supposed to be isotropic along the stack and on the whole active area, thus current lines are supposed to be perpendicular to the plan of the assembly. This is not true at both stack ends, where we must consider a recombination of the current lines within the plan of the last Gas Diffusion Layer (GDL) and the last Bipolar Plate (BPP): because of current lines convergence in one particular point of the current collector. Comparatively, in a single cell, this recombination is likely to happen at both sides of the MEA, and may lead to inhomogeneous current lines distribution. To avoid this artefact while we effectively measure this distribution, 4 mm thick graphite plates, tailored by Fraunhofer ISE, are added on both sides of the Sþþ® device (Fig. 3). They are segmented with the same pattern than the Sþþ® card (cf Fig. 4), and are supposed to simulate upstream and downstream cells, without affecting the current lines distribution caused by the upstream part of the system, mostly the first MEA. Because of the elevated working potential of the anode during electrolysis, this set-up can be implemented only on the cathode side to avoid the graphite
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
3
Fig. 1 e Exploded view of full cell with internal components.
oxidation. This means that the current measurements are only performed downstream the MEA. The graphite plate in contact with the MEA is designed with linear channel-rib flow field to replace the mesh-sinter assembly which cannot be segmented (Fig. 4). Note from Fig. 4 that the 4 extreme top segments, and respectively bottom, are not fully in electric contact with MEA because of fluid distribution design and that current distribution measured in these segments is not comparable. In addition, for the contact resistance optimization as well as for maintaining a mechanical flexibility of the system and counter-balance the rigidity of graphite plates, in-house lasergraved segmented soft GDLs from Freudenberg (eventually chosen as ref. H23C2, cf 3.a.ii) were inserted between each
Fig. 2 e Schematic segmentation drawing of active surface area.
component to ensure flexibility and pressure absorption of the assembly to avoid damaging the graphite plate. Channels for fluid transport through the different layers are sealed with silicon gaskets. Sealing is proved up to 5 barg on the cathode side. Detailed stacking of layers is shown in Fig. 3 for summary. Both inlet and outlet for each side must be confined all the way through the multiple layers so as to avoid any leakage in between. Finally, the cell is assembled by applying a pressure of 4.5 tons (corresponding to 1.8 MPa clamping pressure on the active area) using a hydraulic press. Clamping is performed manually under pressure by tightening bolts to contact. Electrochemical cell characterization was performed using a Greenlight test station ETS P10-156 and a Biologic potentiostat HCP-803 80A. Standard test conditions are 80 C, 2 NLPM of deionized water flow and atmospheric pressure. Fig. 5 shows a photograph of the experimental setup. The cell voltage across the MEA is measured accurately using both the Sþþ® device and the test bench voltage sensors. Both measurements are in excellent agreement. Regarding current distribution over the segmented cell, data are provided by the software as 8 8 matrix and plotted accordingly (Fig. 6a), which is not exploitable as such. As a result, a post-treatment Matlab® program is realized to convert data to the desired shape and to obtain exploitable graphs (cf Fig. 6b) c). The MEAs used for the experiments are made in-house using Nafion® 117 as polymer electrolyte, IrO2 (1 mgIrO2/ cm2) as oxygen-evolving catalyst, commercial Tanaka Pt/C 46% cathode (0.5 mgPt/cm2) as hydrogen-evolving catalyst. The two catalytic layers were deposited by air-spray method [21], and assembled together in one step with membrane by decal method [22] after spray deposition on a PTFE flat substrate.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
4
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
Fig. 3 e Schematic of Sþþ® full assembly with graphite plates.
Experimental validation Full cell characterization was performed in 3 steps: first, the cell and its components are fully characterized without the current measurement system; then with the system (piece by piece) but no MEA, and finally with the full system and a MEA. This characterization includes: (i) mechanical characterization with a pressure sensitive film Fuji Prescale LLW
Fig. 4 e Graphite plate design with flow field and segmentation.
0.5e2.5 MPa from Fujifilm and numerical post-treatment to quantify the pressure distribution; (ii) electrical characterization with resistance measurement by electrochemical impedance spectroscopy (EIS) (from 200 kHz to 100 mHz) and linear scan voltammetry (LSV) from 0.1 to 0.1 V, 100 mV/s, and quantification of the ohmic contribution of every components; (iii) the electrochemical characterizations of the MEA by cyclic voltammetry to determine the electrochemical surface area (ECSA) of the anode, EIS and polarization curves at the initial state. Quick measurement of EIS and LSV allowed
Fig. 5 e Complete set-up implemented on the test bench.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
5
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
Fig. 6 e Current distribution over electro-active surface area: a) Row data from software; b) post-treated data; c) (x, y) coordinates equivalence (X-axis in a) corresponds to red columns in c), Y-axis to blue ones). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
for 2 measures for each situation, for reproducibility, and the average was calculated. EIS and LSV measures were very consistent, as shown in Table 1. Once validated, some results on current distribution related to pressure distribution are presented.
Results and discussion
well as current collectors in order to be restored. Table 2 shows the ohmic contribution of every components, measured by EIS, starting from current collector/current collector interface reference. It gives the individual ohmic resistance contribution for each components, but without discriminating bulk and interface resistance; which is useful to know the most resistive components in the assembly, and then correct the global resistance of the assembly to exactly determine the MEA resistance. It can be drawn that a 2 mm
Impact of set-up implementation Single cell Used as received, the cell resistance was above 1 U cm2. Aged and oxidized porous media are washed and acid-treated as
Table 1 e Comparative table of ohmic resistance values from LSV and EIS. RU (mU) GDL/Sþþ®/GDL Full assembly
Measured by LSV
Measured by EIS
3.59 ± 0,04 3.77 ± 0,04
3.69 ± 0.04 3.78 ± 0.04
Table 2 e Relative ohmic contribution of each cell component (CC refers to Current Collectors). Assembly BPP/BPP BPP/GDL/GDL/BPP BPP/sinter/sinter/BPP BPP/sinter/mesh/BPP BPP/sinter/GDL/mesh/BPP BPP/mesh/sinter/sinter/mesh/BPP BPP/mesh/sinter/sinter/mesh/BPP w. gaskets
RU/mU cm2 0 69 64,75 81,25 141,5 181,25 174
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
6
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
thick Ti sinter with interface resistance has the same resistance than a 200 mm GDL with interface resistance, meaning that interface resistance is predominant versus bulk resistance, at this compression level; thus that adding a GDL in between components must be weighted carefully since it seems to add some consequent resistance; and finally that the compression with or without the proper gasket sealants has no significant effect on the electric contact. In addition, the distribution of compression forces over the active surface area was measured using a pressure sensitive paper, then digitalized and calibrated. Pressure distribution was obviously non uniform and presented a huge gradient between edges and center. It has been showed in a paper from Al Shakhshir et al. [8] that mechanical constraints in a PEM WE cell are very influent in cell performances. After investigation it was assumed to be due to bended end plates, which are eventually rectified and leveled. Initial and final pressure distribution are shown in Fig. 7 a)ed) in the last configuration of Table 2 (full cell). Despite certain local heterogeneities, a minimum pressure is applied on the whole surface; and considering that the final system will involve several other components which are prone to add mechanical flexibility and consequently rub out these defaults, the pressure distribution obtained is assumed to be satisfying so far.
Table 3 e Relative ohmic contribution of each Sþþ® assembly component. RU/mU cm2
Assembly GDLs 2 X X X X
Sþþ X X X
Gr. Plates
X X
MEA
X
551,5 552,75 745 1025
Sþþ®-mounted cell Then the cell has been modified towards Fig. 3 and every additional components has been tested, electrically and mechanically. The resistance values are gathered in Table 3. The measuring Sþþ® card itself does not add any resistance, but graphite plates do, in the range of 100 mU cm2 each. MEA resistance alone, 470 mU cm2, is at first sight twice the literature values [23,24], but was obtained with partly dry MEA at room temperature, which is finally coherent. Current collectors were polished, PTLs were changed, GDLs were optimized among available references, and voltage measures were improved (location and wires connection). Afterwards, this allowed for much lower resistance, down to 489 mU cm2 with full system assembly and MEA (only the
Fig. 7 e Pressure distribution over active surface area (and gaskets) for respectively initial a); b) and final c); d) cell: a) c) measured by pressure sensitive film, b) d) digitalized and scaled (MPa). Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
7
Fig. 8 e Pressure distribution over active surface area (and gaskets) for full assembly: a) measured by pressure sensitive film, b) digitalized and scaled (scale in MPa).
Fig. 9 e Pressure distribution over active surface area for cell assembly with GDL between sinters: a) measured by pressure sensitive film, b) digitalized and scaled (scale in MPa).
Fig. 10 e a) Polarization curve at 80 C and Patm, measured, and corrected from Sþþ® system ohmic drop. b) EIS Nyquist diagrams at 1.40 V, 1.45 V, 1.50 V, 1.60 V. Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
8
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
Fig. 11 e Current distribution for polarization a) at 350 mV for an empty cell (no MEA); b) at 1.6 V with an MEA.
upstream graphite plate out of the 2 is taken in the measured voltage, since the second one will add unnecessary ohmic drop), which is still high but acceptable. In particular when we consider that approximately 200 mU cm2 are due to the
graphite plate on the cathode side and one mesh-sinter assembly on the anode side. After little redesign of gaskets sealants and addition of mechanical wedges, we obtained the pressure distribution
Fig. 12 e a) Pressure distribution over active surface area for cell assembly, with MEA, digitalized and scaled; b) 2D and c) 3D resulting current distribution.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
shown in Fig. 8. Contrary to the porous media, we can see here the channel-rib structure of the graphite plate mentioned earlier. Pressure distribution is even on the whole contact area, but differences are locally important. Small areas, in a) white; b) blue, remain supposedly inactive. It is assumed that this shape is due to GDL deformation, as it has been previously observed in our study (see Fig. 9). We clearly see the same punctured shape, superposed with the flow field structure; this exists anyway with a GDL and is not particular to our assembly, and is attenuated in the presence of the membrane. The electrochemical performance of the instrumented cell is reported in Fig. 10 a) (polarization curves) and b) (EIS diagrams at different operating voltages).
Current distribution Once the set-up is implemented, the cell was fully characterized in order to better understand its behavior. Firstly without an MEA, secondly with an MEA and some perturbations we willingly introduce to see the influence on the current distribution response.
Repartition factor In a first experiment, we measured the distribution of current lines in an empty cell (ie. without MEA) in order to determine
9
the impact of electronic conductors. Due to the geometry of the cell (the two titanium collecting plates are connected to the external DC power source using two tips) the distribution of current over the cell section is non-homogeneous. A privileged path corresponding to the shorter electronic resistance exists in the neighborhood of the two tips (see Fig. 11-a). Since the electronic resistivity of the metallic components is low, the shorter electronic path is also the shorter distance. In a second experiment, we measured the distribution of current lines in a cell containing the MEA. The ionic resistance of the membrane and the charge transfer overvoltages induce a much higher cell resistance, and the titanium collecting plates tend to become equipotential. Consequently, current lines are quasi-homogeneously distributed over the entire surface (see Fig. 11-b).
Pressure distribution In a second set of experiments, we tried to correlate mechanical compression field and current distribution. For this, we used external plates of different shapes in order to apply different compression pattern to the cell. Fig. 12-a shows what happens when a radially symmetric pattern is used (eg. the center by misbalancing pressure distribution thanks to the addition of a piece of a PTFE film on the back side of the cell, between the insulator and the end plate,
Fig. 13 e Current distribution at 0.4 A cm¡2 a) without PTFE mechanical wedge; b) with mechanical wedge; Pressure distribution c) without mechanical wedge (color scale 2, 5e10 MPa); d) with mechanical wedge (color scale 0, 5e2, 5 MPa). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
10
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
at the center of the active area). In this case, the privileged path is intentionally created mechanically. The resulting current distribution is plotted in Fig. 12-b (2D plot) and 12-c (3D plot). We here also observe on Fig. 12 b) and c) that current distribution with an MEA (and without, not shown) follows a similar path, clearly visible. There is a clear mapping of compression forces and current lines. Local extra mechanical compression enhances and lowers electrical contact, providing faster electron passage and thus increasing local current density. This evident conclusion first of all validate the assembly and measurement method, and the fact that we are able to correlate and quantify pressure and current distributions. This is confirmed in a deeper way in the next study, illustrated Fig. 13. It shows that the outer part of the MEA is much less active than the inner part with many segments exceeding ±25% vs. average (top and bottom scale) despite a constant and equivalent pressure distribution (cf Fig. 13d)). Then we add a 50 mm thick PTFE overlayer exactly sized to this outer circle in order to counter balance current heterogeneities (see Fig. 13c)), as we did previously for Fig. 12. Pressure enhancement is visible on Fig. 13c). The current response was still misbalanced (see Fig. 13a)), but much less than without the overlayer (Fig. 13b)). Comparatively, the current is lower in the inner segments and higher in the outer ones, and every
current is maintained within the ±25% range. A second try with 2 layers showed much less attenuation. However, this confirms that the over pressure acts to a certain extent as a “current enhancer”, and is necessary in current lacking areas. But since an enhanced electrical contact does not completely correct the current density distribution, it means that design and/or electrochemistry issues might occur as well, for instance a lower peripheral water supply. To confirm this hypothesis, we aim at quantifying the electrical improvement in order to be able to state that other factors are contributing to current misdistribution. We run a polarization curve on the MEA, and plot it for each single segment (which are eventually controlled in potentiodynamic mode one by one in parallel). From each curve we extract ohmic resistance and overvoltage corresponding to each segment. Then both are plotted back in the segmented shape, so that we see exactly the ohmic resistance and overvoltage evolution between the 2 configurations for each segment (Fig. 14a) to d)). We see that overvoltage is quite uniform (proving uniformity of catalyst deposition) despite a high average value (0.4 V ± 20%), while resistance shows the exact opposite graph than current: it decreases drastically exactly on the peripheral area, from >0.25 mU/seg down to 0.06 mU/seg. Furthermore, the addition of the mechanical wedge made the performances completely homogeneous regarding local ohmic resistance and overvoltage, and quite
Fig. 14 e Ohmic resistance distribution a) without PTFE over layer b) with PTFE over layer. Top and bottom red/orange segments are due to less contact with graphite plate (see Fig. 10-c). Overvoltage distribution (in volts) c) without PTFE over layer d) with PTFE over layer. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
X X X X X X X X vs. None vs. None vs. None vs. None Central Peripheral Peripheral Peripheral Peripheral X X X X X X
Compression pressure distribution Compression pressure distribution UI þ EIS Repartition factor Repartition factor Impact of over compression Impact of wedge e current distrib. Impact of wedge e pressure distrib. Impact of wedge e RU distrib. Impact of wedge e overvoltage distrib.
X 7
8 9 10 11a) 11b) 12 13a) vs 13c) 13b) vs 13d) 14a) vs 14c) 14b) vs 14d)
X Ohmic contribution of Sþþ components Compression pressure distribution 3
1 2
n
Link compression / current Mech. wedge improves current distribution at periphery Compression distribution is non homogeneous with wedge Mech. wedge improves RU distribution on whole surface Mech. wedge improves overvoltage distribution on whole surface
Sþþ components contribution are measured GDL contribution is observable Electochemical performances are good ^ le of MEA Homogenization ro X
GDL
Before/after mechanical corrections X
Sþþ components Mechanical wedge Comparison of measurement method Ohmic contribution of cell components
Subject
MEA
Cell configuration
Additional
EIS and LSV are equivalent Ohmic contribution of interfaces and components (mesh, sinter, GDL) Ohmic contribution of graphite plate (Sþþ plate one is negligeable) Cell is improved
sult Main re
Ref. Figure/Table
Table 4 e Summary of tests and main results with associated figures and configurations (n 1e3 refer to Tables number, others to Figures).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
11
homogeneous (but not completely satisfying) regarding current density, while pressure distribution is not. Thus, it confirms the hypothesis raised above regarding i) the influence of misbalanced electrical contact in the homogeneous compression case, ii) the evidence of other(s) non-electrical cause(s) to current density misdistribution in the heterogeneous compression case, that we can probably attribute to lower water supply to the catalyst layer in this region. Temperature difference must be investigated as well but is not likely to be large enough to explain such a current difference. These hypotheses are yet being studied to be confirmed and eventually mitigated by various design optimizations. As a summary of the study, Table 4 below provides a recap of all the figures and associated tests with cell configuration, remarks, and most important results.
Conclusion and perspectives We report on a methodology for measuring local current density distribution across a PEM water electrolysis cell. A similar approach has already been used to characterize fuel cells. We capitalized on technical developments made in the fuel cell developments to characterize in-situ and operando a PEM WE cell. Because operating constraints of fuel cells and water electrolysis cells are significantly different, the set-up had to be extensively reviewed. The technique was found well-adapted to the characterization of large surface-area cells (250 cm2 in this report) which are closer to current industrial stack design, whereas with small size MEAs it would enhance electrochemistry or process heterogeneities which are not the core issue from stack point of view. Experiments show that there is a clear correlation between the pattern of compression forces and the distribution of current lines across the cell. Pressure distribution is a key factor in current distribution and has to be accentuated on the peripheral areas to favor current passage, where it is deficient in our cell design. We assume that these heterogeneities would lead to faster localized ageing of the active area of the cell and this topic will be very soon addressed in future work. The validation of this protocol allows great perspective of measurements and hopefully a lot of advances in the comprehension of operando global behavior of large area PEM electrolyzer.
Acknowledgement The authors thank Arnaud Morin for his advices about Sþþ® plate implementation and all the fruitful discussions, and Laurent Jacqmin for his support during Sþþ® plate implementation.
references
rrez-Martı´n F, Ochoa-Mendoza A, Rodrı´guez-Anto n LM. [1] Gutie Pre-investigation of water electrolysis for flexible energy
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
12
[2]
[3]
[4] [5] [6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
storage at large scales: the case of the Spanish power system. Int J Hydrogen Energy Apr. 2015;40(15):5544e51. Cetinkaya E, Dincer I, Naterer GF. Life cycle assessment of various hydrogen production methods. Int J Hydrogen Energy Feb. 2012;37(3):2071e80. Zeng K, Zhang D. Recent progress in alkaline water electrolysis for hydrogen production and applications. Prog Energy Combust Sci Jun. 2010;36(3):307e26. Barbir F. PEM electrolysis for production of hydrogen from renewable energy sources. Sol Energy May 2005;78(5):661e9. Mittelsteadt CK. PEM electrolysis: ready for impact. ECS Trans Sep. 2015;69(17):205e11. Ursua A, Gandia LM, Sanchis P. Hydrogen production from water electrolysis: current status and future trends. Proc IEEE Feb. 2012;100(2):410e26. Alizadeh E, Barzegari MM, Momenifar M, Ghadimi M, Saadat SHM. Investigation of contact pressure distribution over the active area of PEM fuel cell stack. Int J Hydrogen Energy Jan. 2016;41(4):3062e71. Al Shakhshir S, Cui X, Frensch S, Kær SK. In-situ experimental characterization of the clamping pressure effects on low temperature polymer electrolyte membrane electrolysis. Int J Hydrogen Energy Aug. 2017;42:21597e606. Tijani AS, Rahim AHA, Hisam MKB. A study of the loss characteristic of a high pressure electrolyzer system for hydrogen production. J Teknol 2015;75(8):65e9. Wu J, Zi Yuan X, Wang H, Blanco M, Martin JJ, Zhang J. Diagnostic tools in PEM fuel cell research: part II. Int J Hydrogen Energy Mar. 2008;33(6):1747e57. Sun S, Xiao Y, Liang D, Shao Z, Yu H, Hou M, et al. Behaviors of a proton exchange membrane electrolyzer under water starvation. RSC Adv 2015;5(19):14506e13. Weng F-B, Hsu C-Y, Li C-W. Experimental investigation of PEM fuel cell aging under current cycling using segmented fuel cell. Int J Hydrogen Energy Apr. 2010;35(8):3664e75. Reshetenko TV, Bender G, Bethune K, Rocheleau R. Application of a segmented cell setup to detect pinhole and catalyst loading defects in proton exchange membrane fuel cells. Electrochim Acta Aug. 2012;76:16e25. Reshetenko TV, Bender G, Bethune K, Rocheleau R. A segmented cell approach for studying the effects of
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
serpentine flow field parameters on PEMFC current distribution. Electrochim Acta Jan. 2013;88:571e9. Ghosh PC, Wu¨ster T, Dohle H, Kimiaie N, Mergel J, Stolten D. In situ approach for current distribution measurement in fuel cells. J Power Sources Mar. 2006;154(1):184e91. Strickland DG, Litster S, Santiago JG. Current distribution in polymer electrolyte membrane fuel cell with active water management. J Power Sources Nov. 2007;174(1):272e81. Tang W, Lin R, Weng Y, Zhang J, Ma J. The effects of operating temperature on current density distribution and impedance spectroscopy by segmented fuel cell. Int J Hydrogen Energy Aug. 2013;38(25):10985e91. Phillips A, Ulsh M, Porter J, Bender G. Utilizing a segmented fuel cell to study the effects of electrode coating irregularities on PEM fuel cell initial performance. Fuel Cells Apr. 2017;17(3):288e98. Nandjou F, Poirot-Crouvezier J-P, Chandesris M, Blachot J-F, Bonnaud C, Bultel Y. Impact of heat and water management on proton exchange membrane fuel cells degradation in automotive application. J Power Sources Sep. 2016;326:182e92. Jabbour L, Robin C, Nandjou F, Vincent R, Micoud F, PoirotCrouvezier J-P, et al. Feasibility of in-plane GDL structuration: impact on current density distribution in large-area proton exchange membrane fuel cells. J Power Sources Dec. 2015;299:380e90. Martinez-Vazquez B, Sanchez DG, Castillo JL, Friedrich KA, Garcia-Ybarra PL. Scaling-up and characterization of ultralow-loading MEAs made-up by electrospray. Int J Hydrogen Energy 2015;40(15):5384e9. Wei Z, Su K, Sui S, He A, Du S. High performance polymer electrolyte membrane fuel cells (PEMFCs) with gradient Pt nanowire cathodes prepared by decal transfer method. Int J Hydrogen Energy Feb. 2015;40(7):3068e74. Sun S, Shao Z, Yu H, Li G, Yi B. Investigations on degradation of the long-term proton exchange membrane water electrolysis stack. J Power Sources Dec. 2014;267:515e20. Edmondson CA, Stallworth PE, Wintersgill MC, Fontanella JJ, Dai Y, Greenbaum SG. Electrical conductivity and NMR studies of methanol/water mixtures in Nafion membranes. Electrochim Acta Apr. 1998;43(10e11):1295e9.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress B. Verdin a,b,c,*, F. Fouda-Onana a,b, S. Germe a,b, G. Serre a,b, P.A. Jacques a,b, P. Millet c Universite Grenoble Alpes, F-38000 Grenoble, France CEA-LITEN, F-38054 Grenoble, France c Universite Paris-Saclay, Institut de Chimie Moleculaire et des Materiaux d'Orsay, 91400 Orsay, France a
b
article info
abstract
Article history:
Proton Exchange Membrane Water Electrolysis (PEM WE) is considered as a key technology
Received 5 July 2017
for large scale storage of renewable electricity, using hydrogen as an energy vector. To
Received in revised form
achieve economically viable MW-scale storage units, there is a strong trend to increase
25 August 2017
both current density and electrode surface area. Such upscaling raises concerns regarding
Accepted 26 August 2017
the homogeneity of compression forces and current lines distribution throughout the
Available online xxx
electrolysis stack. Mechanical and thermal stresses resulting from any heterogeneity can result in faster local ageing and accelerated degradation. Thus, there is a major interest to
Keywords:
measure and understand the nature of such coupled heterogeneities. We report here on
PEM water electrolysis
results obtained with a non-disturbing tool for the current and the temperature mapping of
Current distribution mapping
PEM WE electrodes. A complete description of the setup is discussed in order to avoid some
Mechanical stress
mapping artefacts that could come from the measurement device. Measurements made on 250 cm2 surface area cells show that the clamping pressure homogeneity over the entire geometrical surface area of the cell is one of the first mandatory requirement for an appropriate operation of electrolysis cells. In particular, it was found that even homogeneous pressure distribution could lead to excessive current maldistribution in some cases, due to overriding inhomogeneous electric contacts. This situation has been corrected by using non-flat end plates and/or an accurate thickness of the gaskets. Current mapping has been performed. Results are discussed. © 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Hydrogen is considered as an appropriate energy vector to implement a carbon-neutral economy and to face to the demand for increasing amounts of clean primary energy sources [1]. Hydrogen can be used under various forms and for a wide panel of applications. However, today, hydrogen production is
not sustainable and compatible with this perspective of a carbon free energy society. Approximately 96% of the global hydrogen demand is satisfied by fossil fuel transformation, mainly Steam Methane Reforming (SMR) of oil and natural gas, and coal gasification. These processes are from far the main techniques used, but also the most polluting ones [2]. On the contrary, water electrolysis offers a sustainable and efficient way to produce hydrogen from abundant liquid water
* Corresponding author. 17 rue des Martyrs, 38054, Grenoble, Cedex 9, France. E-mail address: [email protected] (B. Verdin). http://dx.doi.org/10.1016/j.ijhydene.2017.08.189 0360-3199/© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
2
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
and clean energy sources. Accounting for the major part of hydrogen produced by water electrolysis, alkaline technology is the most mature, and commercially advanced [3], whereas Solid Oxide technology is mostly under development. But PEM (Proton Exchange Membrane) electrolysis is now on the way to become cost-competitive. The technical advances achieved up to now enable to reduce the system cost and to consider optimistic perspectives [4,5]. However, one of the critical point is related to system costs compared to life expectancy [6]. On one hand, the use of an acidic electrolyte and the high operating potential of the anode and temperature, and on the other hand, material optimization, especially membrane thickness, are now coming to a harsh point where a lot of efforts are required for small improvements. To optimize PEM water electrolysis technology and extend the lifetime of the cells, there is a need to better understand the ageing of individual cell components. First of all is the MEA (Membrane Electrode Assembly), in the closest conditions from the real stack operation. Then focusing on the R&D effort within the resulting conclusion to finally implement the more fundamental breakthroughs in terms of coatings, porous transport layers, membranes and catalysts accordingly. That is the way to make those advances significantly efficient. Having this in mind, we investigated the current distribution on a large surface area (250 cm2, circular shape) PEM water electrolysis single cell, a surface sufficiently large to analyze mechanical, thermal and fluidic management constraints that could be encountered on a large commercial stack [7e9], and subjects of intense developments from industrials. As explained above, particular focus will be made on ageing and degrading conditions. An important preliminary work has been done on the experimental set-up in order to implement the current distribution visualization device in the cell. The internal segmentation of the components has been first considered as the path that the electrons should follow. Quite inspiring work was done for fuel cells and we have found some correlations with our application. Various methods have been previously studied for PEM FC [10] (very few on PEM electrolyzers): MEA segmentation [11,12], flow field segmentation [13,14], shunt resistance [15], segmented PCBbased Bipolar Plates (BPP) [16e18], Sþþ® current measurement plates [19,20]. Especially a technique derived from the one of Ghosh et al. [15] has been considered to be adapted for our electrolysis purpose for its flexibility and easy evolution. But finally, given uncertainties of implementation, the most viable approach to stick as close as possible to real stack operation appears to be by using commercialized Sþþ® current measurement card. As far as we know this communication is the first to establish and use such a current and temperature mapping device for PEM WE. This communication will detail the experimental set-up used, and then describe an application of the tool for a local operando study of the cell dysfunction.
Experimental Cell specifications In order to perform reliable in-situ and operando measurements, we used a fragmented cell. Several specifications must
be fulfilled to perform not too intrusive measurements. From the mechanical viewpoint, the fragmented cell should be robust and rigid enough to support 1.8 MPa mechanical clamping pressure and also to transfer this compression constraint completely to the MEA, without absorbing any substantial part of the mechanical force. To perform testing under pressure, we fixed a minimum of 5 bars internal pressure at the cathode. From the fluidic viewpoint, the device must first be watertight with the ambient and internally: the introduced measurement device must not affect water- and gas-tightness at the risk of creating hazardous leakage and mixtures. Lastly, electrically speaking, it is needed to drain out the voltage for each segment independently of the running cell power.
Experimental set-up The single cell used to perform our measurements was initially designed for MEA characterization at low current densities (j < 0.5 A/cm2). Consequently, some mechanical optimizations were required to obtain a satisfying electrochemical behavior at higher current densities. The cell is 250 cm2 active area, and composed of various elements detailed in Fig. 1. End plates, current collectors and sealant gaskets have been resized and redesigned. Design is circular as in most cases for PEM WE mainly due to its robustness towards high gas pressures and potential leaks. Porous Transport Layers (PTLs) are made of two main pieces, eg meshed titanium wires layers and porous titanium sinters. A gold-plated electronic Printed Circuit Board from Sþþ® Simulation Services Company has been used for current density measurements. A specific circular design of the Sþþ® Current Scan Shunt measurement device was developed. This type of measurement relies on the measurement of the voltage drop over calibrated micro-resistances. The cell shown in Fig. 2 contains 64 electrically independent segments of equal surface area (3.89 cm2 each) positioned radially from the center. The interest of this design is that comparative measurement can be performed between inlet and outlet, center and edges, along different radii. In a stack configuration, current propagation is supposed to be isotropic along the stack and on the whole active area, thus current lines are supposed to be perpendicular to the plan of the assembly. This is not true at both stack ends, where we must consider a recombination of the current lines within the plan of the last Gas Diffusion Layer (GDL) and the last Bipolar Plate (BPP): because of current lines convergence in one particular point of the current collector. Comparatively, in a single cell, this recombination is likely to happen at both sides of the MEA, and may lead to inhomogeneous current lines distribution. To avoid this artefact while we effectively measure this distribution, 4 mm thick graphite plates, tailored by Fraunhofer ISE, are added on both sides of the Sþþ® device (Fig. 3). They are segmented with the same pattern than the Sþþ® card (cf Fig. 4), and are supposed to simulate upstream and downstream cells, without affecting the current lines distribution caused by the upstream part of the system, mostly the first MEA. Because of the elevated working potential of the anode during electrolysis, this set-up can be implemented only on the cathode side to avoid the graphite
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
3
Fig. 1 e Exploded view of full cell with internal components.
oxidation. This means that the current measurements are only performed downstream the MEA. The graphite plate in contact with the MEA is designed with linear channel-rib flow field to replace the mesh-sinter assembly which cannot be segmented (Fig. 4). Note from Fig. 4 that the 4 extreme top segments, and respectively bottom, are not fully in electric contact with MEA because of fluid distribution design and that current distribution measured in these segments is not comparable. In addition, for the contact resistance optimization as well as for maintaining a mechanical flexibility of the system and counter-balance the rigidity of graphite plates, in-house lasergraved segmented soft GDLs from Freudenberg (eventually chosen as ref. H23C2, cf 3.a.ii) were inserted between each
Fig. 2 e Schematic segmentation drawing of active surface area.
component to ensure flexibility and pressure absorption of the assembly to avoid damaging the graphite plate. Channels for fluid transport through the different layers are sealed with silicon gaskets. Sealing is proved up to 5 barg on the cathode side. Detailed stacking of layers is shown in Fig. 3 for summary. Both inlet and outlet for each side must be confined all the way through the multiple layers so as to avoid any leakage in between. Finally, the cell is assembled by applying a pressure of 4.5 tons (corresponding to 1.8 MPa clamping pressure on the active area) using a hydraulic press. Clamping is performed manually under pressure by tightening bolts to contact. Electrochemical cell characterization was performed using a Greenlight test station ETS P10-156 and a Biologic potentiostat HCP-803 80A. Standard test conditions are 80 C, 2 NLPM of deionized water flow and atmospheric pressure. Fig. 5 shows a photograph of the experimental setup. The cell voltage across the MEA is measured accurately using both the Sþþ® device and the test bench voltage sensors. Both measurements are in excellent agreement. Regarding current distribution over the segmented cell, data are provided by the software as 8 8 matrix and plotted accordingly (Fig. 6a), which is not exploitable as such. As a result, a post-treatment Matlab® program is realized to convert data to the desired shape and to obtain exploitable graphs (cf Fig. 6b) c). The MEAs used for the experiments are made in-house using Nafion® 117 as polymer electrolyte, IrO2 (1 mgIrO2/ cm2) as oxygen-evolving catalyst, commercial Tanaka Pt/C 46% cathode (0.5 mgPt/cm2) as hydrogen-evolving catalyst. The two catalytic layers were deposited by air-spray method [21], and assembled together in one step with membrane by decal method [22] after spray deposition on a PTFE flat substrate.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
4
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
Fig. 3 e Schematic of Sþþ® full assembly with graphite plates.
Experimental validation Full cell characterization was performed in 3 steps: first, the cell and its components are fully characterized without the current measurement system; then with the system (piece by piece) but no MEA, and finally with the full system and a MEA. This characterization includes: (i) mechanical characterization with a pressure sensitive film Fuji Prescale LLW
Fig. 4 e Graphite plate design with flow field and segmentation.
0.5e2.5 MPa from Fujifilm and numerical post-treatment to quantify the pressure distribution; (ii) electrical characterization with resistance measurement by electrochemical impedance spectroscopy (EIS) (from 200 kHz to 100 mHz) and linear scan voltammetry (LSV) from 0.1 to 0.1 V, 100 mV/s, and quantification of the ohmic contribution of every components; (iii) the electrochemical characterizations of the MEA by cyclic voltammetry to determine the electrochemical surface area (ECSA) of the anode, EIS and polarization curves at the initial state. Quick measurement of EIS and LSV allowed
Fig. 5 e Complete set-up implemented on the test bench.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
5
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
Fig. 6 e Current distribution over electro-active surface area: a) Row data from software; b) post-treated data; c) (x, y) coordinates equivalence (X-axis in a) corresponds to red columns in c), Y-axis to blue ones). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
for 2 measures for each situation, for reproducibility, and the average was calculated. EIS and LSV measures were very consistent, as shown in Table 1. Once validated, some results on current distribution related to pressure distribution are presented.
Results and discussion
well as current collectors in order to be restored. Table 2 shows the ohmic contribution of every components, measured by EIS, starting from current collector/current collector interface reference. It gives the individual ohmic resistance contribution for each components, but without discriminating bulk and interface resistance; which is useful to know the most resistive components in the assembly, and then correct the global resistance of the assembly to exactly determine the MEA resistance. It can be drawn that a 2 mm
Impact of set-up implementation Single cell Used as received, the cell resistance was above 1 U cm2. Aged and oxidized porous media are washed and acid-treated as
Table 1 e Comparative table of ohmic resistance values from LSV and EIS. RU (mU) GDL/Sþþ®/GDL Full assembly
Measured by LSV
Measured by EIS
3.59 ± 0,04 3.77 ± 0,04
3.69 ± 0.04 3.78 ± 0.04
Table 2 e Relative ohmic contribution of each cell component (CC refers to Current Collectors). Assembly BPP/BPP BPP/GDL/GDL/BPP BPP/sinter/sinter/BPP BPP/sinter/mesh/BPP BPP/sinter/GDL/mesh/BPP BPP/mesh/sinter/sinter/mesh/BPP BPP/mesh/sinter/sinter/mesh/BPP w. gaskets
RU/mU cm2 0 69 64,75 81,25 141,5 181,25 174
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
6
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
thick Ti sinter with interface resistance has the same resistance than a 200 mm GDL with interface resistance, meaning that interface resistance is predominant versus bulk resistance, at this compression level; thus that adding a GDL in between components must be weighted carefully since it seems to add some consequent resistance; and finally that the compression with or without the proper gasket sealants has no significant effect on the electric contact. In addition, the distribution of compression forces over the active surface area was measured using a pressure sensitive paper, then digitalized and calibrated. Pressure distribution was obviously non uniform and presented a huge gradient between edges and center. It has been showed in a paper from Al Shakhshir et al. [8] that mechanical constraints in a PEM WE cell are very influent in cell performances. After investigation it was assumed to be due to bended end plates, which are eventually rectified and leveled. Initial and final pressure distribution are shown in Fig. 7 a)ed) in the last configuration of Table 2 (full cell). Despite certain local heterogeneities, a minimum pressure is applied on the whole surface; and considering that the final system will involve several other components which are prone to add mechanical flexibility and consequently rub out these defaults, the pressure distribution obtained is assumed to be satisfying so far.
Table 3 e Relative ohmic contribution of each Sþþ® assembly component. RU/mU cm2
Assembly GDLs 2 X X X X
Sþþ X X X
Gr. Plates
X X
MEA
X
551,5 552,75 745 1025
Sþþ®-mounted cell Then the cell has been modified towards Fig. 3 and every additional components has been tested, electrically and mechanically. The resistance values are gathered in Table 3. The measuring Sþþ® card itself does not add any resistance, but graphite plates do, in the range of 100 mU cm2 each. MEA resistance alone, 470 mU cm2, is at first sight twice the literature values [23,24], but was obtained with partly dry MEA at room temperature, which is finally coherent. Current collectors were polished, PTLs were changed, GDLs were optimized among available references, and voltage measures were improved (location and wires connection). Afterwards, this allowed for much lower resistance, down to 489 mU cm2 with full system assembly and MEA (only the
Fig. 7 e Pressure distribution over active surface area (and gaskets) for respectively initial a); b) and final c); d) cell: a) c) measured by pressure sensitive film, b) d) digitalized and scaled (MPa). Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
7
Fig. 8 e Pressure distribution over active surface area (and gaskets) for full assembly: a) measured by pressure sensitive film, b) digitalized and scaled (scale in MPa).
Fig. 9 e Pressure distribution over active surface area for cell assembly with GDL between sinters: a) measured by pressure sensitive film, b) digitalized and scaled (scale in MPa).
Fig. 10 e a) Polarization curve at 80 C and Patm, measured, and corrected from Sþþ® system ohmic drop. b) EIS Nyquist diagrams at 1.40 V, 1.45 V, 1.50 V, 1.60 V. Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
8
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
Fig. 11 e Current distribution for polarization a) at 350 mV for an empty cell (no MEA); b) at 1.6 V with an MEA.
upstream graphite plate out of the 2 is taken in the measured voltage, since the second one will add unnecessary ohmic drop), which is still high but acceptable. In particular when we consider that approximately 200 mU cm2 are due to the
graphite plate on the cathode side and one mesh-sinter assembly on the anode side. After little redesign of gaskets sealants and addition of mechanical wedges, we obtained the pressure distribution
Fig. 12 e a) Pressure distribution over active surface area for cell assembly, with MEA, digitalized and scaled; b) 2D and c) 3D resulting current distribution.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
shown in Fig. 8. Contrary to the porous media, we can see here the channel-rib structure of the graphite plate mentioned earlier. Pressure distribution is even on the whole contact area, but differences are locally important. Small areas, in a) white; b) blue, remain supposedly inactive. It is assumed that this shape is due to GDL deformation, as it has been previously observed in our study (see Fig. 9). We clearly see the same punctured shape, superposed with the flow field structure; this exists anyway with a GDL and is not particular to our assembly, and is attenuated in the presence of the membrane. The electrochemical performance of the instrumented cell is reported in Fig. 10 a) (polarization curves) and b) (EIS diagrams at different operating voltages).
Current distribution Once the set-up is implemented, the cell was fully characterized in order to better understand its behavior. Firstly without an MEA, secondly with an MEA and some perturbations we willingly introduce to see the influence on the current distribution response.
Repartition factor In a first experiment, we measured the distribution of current lines in an empty cell (ie. without MEA) in order to determine
9
the impact of electronic conductors. Due to the geometry of the cell (the two titanium collecting plates are connected to the external DC power source using two tips) the distribution of current over the cell section is non-homogeneous. A privileged path corresponding to the shorter electronic resistance exists in the neighborhood of the two tips (see Fig. 11-a). Since the electronic resistivity of the metallic components is low, the shorter electronic path is also the shorter distance. In a second experiment, we measured the distribution of current lines in a cell containing the MEA. The ionic resistance of the membrane and the charge transfer overvoltages induce a much higher cell resistance, and the titanium collecting plates tend to become equipotential. Consequently, current lines are quasi-homogeneously distributed over the entire surface (see Fig. 11-b).
Pressure distribution In a second set of experiments, we tried to correlate mechanical compression field and current distribution. For this, we used external plates of different shapes in order to apply different compression pattern to the cell. Fig. 12-a shows what happens when a radially symmetric pattern is used (eg. the center by misbalancing pressure distribution thanks to the addition of a piece of a PTFE film on the back side of the cell, between the insulator and the end plate,
Fig. 13 e Current distribution at 0.4 A cm¡2 a) without PTFE mechanical wedge; b) with mechanical wedge; Pressure distribution c) without mechanical wedge (color scale 2, 5e10 MPa); d) with mechanical wedge (color scale 0, 5e2, 5 MPa). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
10
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
at the center of the active area). In this case, the privileged path is intentionally created mechanically. The resulting current distribution is plotted in Fig. 12-b (2D plot) and 12-c (3D plot). We here also observe on Fig. 12 b) and c) that current distribution with an MEA (and without, not shown) follows a similar path, clearly visible. There is a clear mapping of compression forces and current lines. Local extra mechanical compression enhances and lowers electrical contact, providing faster electron passage and thus increasing local current density. This evident conclusion first of all validate the assembly and measurement method, and the fact that we are able to correlate and quantify pressure and current distributions. This is confirmed in a deeper way in the next study, illustrated Fig. 13. It shows that the outer part of the MEA is much less active than the inner part with many segments exceeding ±25% vs. average (top and bottom scale) despite a constant and equivalent pressure distribution (cf Fig. 13d)). Then we add a 50 mm thick PTFE overlayer exactly sized to this outer circle in order to counter balance current heterogeneities (see Fig. 13c)), as we did previously for Fig. 12. Pressure enhancement is visible on Fig. 13c). The current response was still misbalanced (see Fig. 13a)), but much less than without the overlayer (Fig. 13b)). Comparatively, the current is lower in the inner segments and higher in the outer ones, and every
current is maintained within the ±25% range. A second try with 2 layers showed much less attenuation. However, this confirms that the over pressure acts to a certain extent as a “current enhancer”, and is necessary in current lacking areas. But since an enhanced electrical contact does not completely correct the current density distribution, it means that design and/or electrochemistry issues might occur as well, for instance a lower peripheral water supply. To confirm this hypothesis, we aim at quantifying the electrical improvement in order to be able to state that other factors are contributing to current misdistribution. We run a polarization curve on the MEA, and plot it for each single segment (which are eventually controlled in potentiodynamic mode one by one in parallel). From each curve we extract ohmic resistance and overvoltage corresponding to each segment. Then both are plotted back in the segmented shape, so that we see exactly the ohmic resistance and overvoltage evolution between the 2 configurations for each segment (Fig. 14a) to d)). We see that overvoltage is quite uniform (proving uniformity of catalyst deposition) despite a high average value (0.4 V ± 20%), while resistance shows the exact opposite graph than current: it decreases drastically exactly on the peripheral area, from >0.25 mU/seg down to 0.06 mU/seg. Furthermore, the addition of the mechanical wedge made the performances completely homogeneous regarding local ohmic resistance and overvoltage, and quite
Fig. 14 e Ohmic resistance distribution a) without PTFE over layer b) with PTFE over layer. Top and bottom red/orange segments are due to less contact with graphite plate (see Fig. 10-c). Overvoltage distribution (in volts) c) without PTFE over layer d) with PTFE over layer. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
X X X X X X X X vs. None vs. None vs. None vs. None Central Peripheral Peripheral Peripheral Peripheral X X X X X X
Compression pressure distribution Compression pressure distribution UI þ EIS Repartition factor Repartition factor Impact of over compression Impact of wedge e current distrib. Impact of wedge e pressure distrib. Impact of wedge e RU distrib. Impact of wedge e overvoltage distrib.
X 7
8 9 10 11a) 11b) 12 13a) vs 13c) 13b) vs 13d) 14a) vs 14c) 14b) vs 14d)
X Ohmic contribution of Sþþ components Compression pressure distribution 3
1 2
n
Link compression / current Mech. wedge improves current distribution at periphery Compression distribution is non homogeneous with wedge Mech. wedge improves RU distribution on whole surface Mech. wedge improves overvoltage distribution on whole surface
Sþþ components contribution are measured GDL contribution is observable Electochemical performances are good ^ le of MEA Homogenization ro X
GDL
Before/after mechanical corrections X
Sþþ components Mechanical wedge Comparison of measurement method Ohmic contribution of cell components
Subject
MEA
Cell configuration
Additional
EIS and LSV are equivalent Ohmic contribution of interfaces and components (mesh, sinter, GDL) Ohmic contribution of graphite plate (Sþþ plate one is negligeable) Cell is improved
sult Main re
Ref. Figure/Table
Table 4 e Summary of tests and main results with associated figures and configurations (n 1e3 refer to Tables number, others to Figures).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
11
homogeneous (but not completely satisfying) regarding current density, while pressure distribution is not. Thus, it confirms the hypothesis raised above regarding i) the influence of misbalanced electrical contact in the homogeneous compression case, ii) the evidence of other(s) non-electrical cause(s) to current density misdistribution in the heterogeneous compression case, that we can probably attribute to lower water supply to the catalyst layer in this region. Temperature difference must be investigated as well but is not likely to be large enough to explain such a current difference. These hypotheses are yet being studied to be confirmed and eventually mitigated by various design optimizations. As a summary of the study, Table 4 below provides a recap of all the figures and associated tests with cell configuration, remarks, and most important results.
Conclusion and perspectives We report on a methodology for measuring local current density distribution across a PEM water electrolysis cell. A similar approach has already been used to characterize fuel cells. We capitalized on technical developments made in the fuel cell developments to characterize in-situ and operando a PEM WE cell. Because operating constraints of fuel cells and water electrolysis cells are significantly different, the set-up had to be extensively reviewed. The technique was found well-adapted to the characterization of large surface-area cells (250 cm2 in this report) which are closer to current industrial stack design, whereas with small size MEAs it would enhance electrochemistry or process heterogeneities which are not the core issue from stack point of view. Experiments show that there is a clear correlation between the pattern of compression forces and the distribution of current lines across the cell. Pressure distribution is a key factor in current distribution and has to be accentuated on the peripheral areas to favor current passage, where it is deficient in our cell design. We assume that these heterogeneities would lead to faster localized ageing of the active area of the cell and this topic will be very soon addressed in future work. The validation of this protocol allows great perspective of measurements and hopefully a lot of advances in the comprehension of operando global behavior of large area PEM electrolyzer.
Acknowledgement The authors thank Arnaud Morin for his advices about Sþþ® plate implementation and all the fruitful discussions, and Laurent Jacqmin for his support during Sþþ® plate implementation.
references
rrez-Martı´n F, Ochoa-Mendoza A, Rodrı´guez-Anto n LM. [1] Gutie Pre-investigation of water electrolysis for flexible energy
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189
12
[2]
[3]
[4] [5] [6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 7 ) 1 e1 2
storage at large scales: the case of the Spanish power system. Int J Hydrogen Energy Apr. 2015;40(15):5544e51. Cetinkaya E, Dincer I, Naterer GF. Life cycle assessment of various hydrogen production methods. Int J Hydrogen Energy Feb. 2012;37(3):2071e80. Zeng K, Zhang D. Recent progress in alkaline water electrolysis for hydrogen production and applications. Prog Energy Combust Sci Jun. 2010;36(3):307e26. Barbir F. PEM electrolysis for production of hydrogen from renewable energy sources. Sol Energy May 2005;78(5):661e9. Mittelsteadt CK. PEM electrolysis: ready for impact. ECS Trans Sep. 2015;69(17):205e11. Ursua A, Gandia LM, Sanchis P. Hydrogen production from water electrolysis: current status and future trends. Proc IEEE Feb. 2012;100(2):410e26. Alizadeh E, Barzegari MM, Momenifar M, Ghadimi M, Saadat SHM. Investigation of contact pressure distribution over the active area of PEM fuel cell stack. Int J Hydrogen Energy Jan. 2016;41(4):3062e71. Al Shakhshir S, Cui X, Frensch S, Kær SK. In-situ experimental characterization of the clamping pressure effects on low temperature polymer electrolyte membrane electrolysis. Int J Hydrogen Energy Aug. 2017;42:21597e606. Tijani AS, Rahim AHA, Hisam MKB. A study of the loss characteristic of a high pressure electrolyzer system for hydrogen production. J Teknol 2015;75(8):65e9. Wu J, Zi Yuan X, Wang H, Blanco M, Martin JJ, Zhang J. Diagnostic tools in PEM fuel cell research: part II. Int J Hydrogen Energy Mar. 2008;33(6):1747e57. Sun S, Xiao Y, Liang D, Shao Z, Yu H, Hou M, et al. Behaviors of a proton exchange membrane electrolyzer under water starvation. RSC Adv 2015;5(19):14506e13. Weng F-B, Hsu C-Y, Li C-W. Experimental investigation of PEM fuel cell aging under current cycling using segmented fuel cell. Int J Hydrogen Energy Apr. 2010;35(8):3664e75. Reshetenko TV, Bender G, Bethune K, Rocheleau R. Application of a segmented cell setup to detect pinhole and catalyst loading defects in proton exchange membrane fuel cells. Electrochim Acta Aug. 2012;76:16e25. Reshetenko TV, Bender G, Bethune K, Rocheleau R. A segmented cell approach for studying the effects of
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
serpentine flow field parameters on PEMFC current distribution. Electrochim Acta Jan. 2013;88:571e9. Ghosh PC, Wu¨ster T, Dohle H, Kimiaie N, Mergel J, Stolten D. In situ approach for current distribution measurement in fuel cells. J Power Sources Mar. 2006;154(1):184e91. Strickland DG, Litster S, Santiago JG. Current distribution in polymer electrolyte membrane fuel cell with active water management. J Power Sources Nov. 2007;174(1):272e81. Tang W, Lin R, Weng Y, Zhang J, Ma J. The effects of operating temperature on current density distribution and impedance spectroscopy by segmented fuel cell. Int J Hydrogen Energy Aug. 2013;38(25):10985e91. Phillips A, Ulsh M, Porter J, Bender G. Utilizing a segmented fuel cell to study the effects of electrode coating irregularities on PEM fuel cell initial performance. Fuel Cells Apr. 2017;17(3):288e98. Nandjou F, Poirot-Crouvezier J-P, Chandesris M, Blachot J-F, Bonnaud C, Bultel Y. Impact of heat and water management on proton exchange membrane fuel cells degradation in automotive application. J Power Sources Sep. 2016;326:182e92. Jabbour L, Robin C, Nandjou F, Vincent R, Micoud F, PoirotCrouvezier J-P, et al. Feasibility of in-plane GDL structuration: impact on current density distribution in large-area proton exchange membrane fuel cells. J Power Sources Dec. 2015;299:380e90. Martinez-Vazquez B, Sanchez DG, Castillo JL, Friedrich KA, Garcia-Ybarra PL. Scaling-up and characterization of ultralow-loading MEAs made-up by electrospray. Int J Hydrogen Energy 2015;40(15):5384e9. Wei Z, Su K, Sui S, He A, Du S. High performance polymer electrolyte membrane fuel cells (PEMFCs) with gradient Pt nanowire cathodes prepared by decal transfer method. Int J Hydrogen Energy Feb. 2015;40(7):3068e74. Sun S, Shao Z, Yu H, Li G, Yi B. Investigations on degradation of the long-term proton exchange membrane water electrolysis stack. J Power Sources Dec. 2014;267:515e20. Edmondson CA, Stallworth PE, Wintersgill MC, Fontanella JJ, Dai Y, Greenbaum SG. Electrical conductivity and NMR studies of methanol/water mixtures in Nafion membranes. Electrochim Acta Apr. 1998;43(10e11):1295e9.
Please cite this article in press as: Verdin B, et al., Operando current mapping on PEM water electrolysis cells. Influence of mechanical stress, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.08.189