Project Topics
Get Complete Project Material File(s) Now! »
Internal cell instrumentation
The effort is also increasingly turned towards in-situ experiments and localized measurements, so that the results directly translate the real-system performance. Moreover local knowledge of the cell performance can be used for fine model validation and insightful optimization of industrial systems.
In the field of PEMFC, scientists have already investigated local phenomena. At the LEMTA laboratory, the team [98, 99] has developed a segmented fuel cell with local reference electrodes, in order to follow the current density and potential evolution inside the cell.
To do so, they use a shunt resistor network and local reversible hydrogen electrode references. These works have highlighted some heterogeneities during fuel cell operation and related local ageing effects.
Attempts to observe local behavior figure among the redox flow battery literature, but advanced set-ups such as the one cited above do not exist yet. Hsieh and coworkers [100] are the first to have measured local currents by comparing two designs of segmentation: one with segments in the current collector behind the graphite end plate and one with the graphite end plate segmented itself. They found that segmenting the end plate was necessary to avoid lateral current spread in it but the poor contact resistances between the divided plate and the connectors worsened the battery performance. Clement et al. [101] used a printed circuit board (PCB) placed behind a divided flow plate. Their set-up demands a perfect mirroring between the PCB and the divided flow plate to get correct in-plane current. The tiny segments machined in the graphite composite and only maintained by a thermoset resin can jeopardize a uniform compression and results in high contact resistances. However, the set-up proved its usefulness for the system study. They demonstrated that the distributed currents reflected mass transport within the cell; the locally-resolved current diagnostics is useful to assess electrode materials or optimize the operating conditions. To get relevant and easy-to-compare data, the characterization were performed at the mass-transport limiting point.
Segmentation of the end plate was also used to measure local potential and VOC measurements gave insight on the local SOC. The results were used to observe the electrode compression effect on the flow distribution of the electrolyte [102].
Becker et al. [103] used solid-phase potential probes made of carbon fibers to measure the local felt resistance at several points of the active area. They deduced the in-plane localized current densities during polarization curves. Another research group employed potential-probes between several layers of carbon paper to evaluate the through-plane current distribution inside the cell. They could deconvolute the resistances comprised in the PC for each haf-cell [104]. Gandomi et al. could validated their model accuracy with a similar set of potential probes through the thickness of the electrode [84].
Another crucial point to study cell operation is to identify the losses in the system. The contribution of each component can be determined by in-situ electrochemical techniques, such as electrochemical impedance spectroscopy (EIS), combined with a reference electrode to decouple the two half-cell processes [89, 105–107]. Thanks to a dynamic hydrogen reference electrode and a set-up where the state of charge of the cell is kept constant, Sun et al. [108] showed that in a vanadium redox flow battery the negative electrode is limiting in discharge. Later on [89], they built a symmetric cell: anolyte at a state of charge of 50% was flowed on both sides. The experiment revealed how the overall cell voltage losses can be split into distinct mechanisms: charge-transfer, mass-transfer and resistive overvoltages. Detailed literature data are given along the PhD thesis where necessary.
Methodology of the PhD study
The PhD research objective is to meet the need of KEMIWATT to build a thorough knowledge base of its system and refine the experimental methods by developing reliable analytical tools. The results seek to help with determining the optimal design and operating conditions. The starting point of the PhD work is the literature about redox flow batteries presented above, coupled with the expertise of the partner laboratory LEMTA in instrumentation of electrochemical systems. The PhD study is divided into 3 main axes of research:
c Development of specific experimental platforms to get an insight into the behavior of the system. Separate analysis (Chap. 3) and then half-cell characterization (by means of a symmetric cell layout, Chap. 4) permitted the deconvolution of the materials influence on the battery. A porous electrode model was developed and then combined so as to estimate key attributes of the system from the experimental data (Chap. 4).
c Development of an instrumented test bench for full cell experiments to investigate the local phenomena and get a spatial resolution of the current (Chap. 2). After validation, this tool was used to perform a fine parameter study by varying alternatively the current, the flow rate and the temperature. The different tests were correlated to build operational maps adaptable to real-stack conditions (Chap. 5).
c Investigation of industrial issues and particularly hydraulic-related aspects (Chap. 6).
A reduced clear flow cell enabled the flow visualization through a typical cell stack design. A hydrodynamic model completed the study.
These three approaches are complementary and interrelated in the entire work presented in the following pages. All the diagnostics tools were developed within the framework of the PhD, and when possible, were subsequently integrated to the standard R&D procedures at the company KEMIWATT.
Polarization curves
The Polarization curve, noted PC, gives an interesting depiction of the system behavior. This characterization particularly insightful during material benchmarking or cell architecture development [90, 125]. The general description of the PC is made in subsection 1.3.1. One consideration during the acquisition of the PC is the constant SOC of the battery; this assumption is actually not strictly met in a RFB, where the electrolyte is recirculated during characterization. To circumvent this issue, the experimental layout can be adapted to e.g. a cell-in-series or a 4-tank configuration [27], but these are constraining options. With an accurate method of acquisition, it is possible to hold the SOC almost constant during the course of the PC experiment. A number of techniques are described in the literature: Û Stepwise method: The first one consists of applying current steps of increasing intensity until a limiting current level, and may include a rest period between each step to return to equilibrium. This must be done separately in charge and discharge to obtain the two parts of the PC. Aaron et al. chose to hold current during 30 s and then allowed a 2 min rest [79]. They averaged the cell potential measurements over 30 s. This method induces a continuous variation of SOC over the increase of current. The current range in which one of the overpotential phenomena dominates depends on the SOC, thus the critical domains of activation, ohmic or mass transport losses (depicted in Fig. 1.8) cannot be clearly delineated with this technique. Another downside is the determination of the voltage for each current: the average value is strongly linked to the specified current step length. This relatively arbitrary criterion of step length can strongly affect the results of the polarization curve.
Û Alternating stepwise method: Becker et al. modified the method by using a stepwise potentiostatic mode that alternated the polarization sign at each step [103]. The cell was returned to its initial SOC after each value of overpotential tested. It constitutes an improvement in preserving a constant state during the characterization but does not overcome the problem of setting the best step length criterion. Additionally, switching the polarization from positive to negative direction might be traumatic for the system.
Û Ramp method: Another approach is to perform a ramp of potential or current at a given rate. The polarization sign is unchanged throughout the course of the experiment and the instantaneous cell response is taken to plot the curve. One could criticize that the transient behaviour affects the results because it includes the unwanted capacitive currents. Chen et al. [104] nevertheless asserted that the linear galvanic ramping yielded the same battery response as holding current steps in their set-up.
A comparison of the alternating stepwise and ramp modes was performed on the battery studied in this PhD, to have a clear view of how each impacted polarization curve recording. The PC study was carried out in the segmented cell (described below) with electrolytes precharged at SOC 50, and a flow rate of 100mLmin−1.
The current step method was performed galvanostatically with a current sign being alternated at each step. This protocol is illustrated in Fig. 2.4a by the temporal variation of U and I signals with steps of 60 s.
Impedance variation with SOC
An innovative test developed during this study was the monitoring of the impedance variation of the symmetric cell during the charge. This particular configuration allowed fitting of the resulting curve to an analytic model in order to determine key parameters (Sections 4.3.3 and 4.4.3). To this end, the cell was sequentially charged at 40mAcm−2; the step lengths were adapted to the state of progress of the charge to record enough points and have a fine resolution, without having a test that last too long. The charging steps (increasing the SOC by 5-10 % or less at the end of charge) were followed by a rest of 2 minutes and a GEIS measurement at high-medium frequencies: f=[1 kHz; 20 Hz]. A voltage limit of 0.9V secured the charge before side reactions are triggered within the symmetric cell. The cell was then discharged without interruption to recover the symmetric state. This method of cycling based on charging and discharging the cell for an identical period of time is the only way to return to a similar state after each cycle. The two electrolytes showed an acceptable capacity retention during a few cycles. This allowed the assertion that each charge started from the same conditions.
Segmentation and local currents
Access to the locally-resolved current density requires the measurement of the real throughplane component generated within the cell, while not disturbing the cell operation. As it was numerically demonstrated by Eckl et al. in the case of fuel cells, lateral current spread can be substantial if the solid plate behind the electrode is not segmented [128]. This was confirmed experimentally in the case of a RFB by comparing the probed internal current with or without full segmentation of the end plates [100, 129]. The accurate estimation of internal current must go along with segmentation of the solid plates constituting the back of the cell. It was asserted in former PEMFC studies that the segmentation does not need to be extended to the electrodes [130]. Nevertheless, the RFB differs from fuel cell by the presence of conductive liquid electrolytes inside the porous electrode. As a consequence, some current might spread laterally through the liquid phase [129]. Bhattarai et al. were the first ones to implement the porous electrode segmentation in a RFB [131]. They obtained slightly improved resolution with electrode segmentation. Apart from this gain, the electrode segmentation is difficult to execute and affects the internal porous structure. For these reasons this option was not considered in the PhD study.
Since the current applied on one side crosses the entire cell, it is sufficient to segment only the current collector at one end. For the same reason, the half-cell used for the localized measurements does not affect the current [101].
The in-plane distribution of current can be collected by several ways. The four most common techniques described in the literature are the resistor network, the printed circuit board, the Hall effect sensors and the potential probes.
Table of contents :
1 Redox flow battery technology
1.1 The energy storage challenge
1.2 The Redox Flow Battery solution
1.2.1 Concept and advantages
1.2.2 Existing chemistries
1.2.3 Kemiwatt’s industrial challenge and opportunity
1.3 General understanding of the Redox Flow Battery
1.3.1 Fundamentals
1.3.2 Modeling
1.3.3 Analytical platforms
1.3.4 Internal cell instrumentation
1.4 Methodology of the PhD study
2 Diagnostic tools
2.1 Materials and pretreatments
2.1.1 Electrolytes
2.1.2 Standard test cell
2.2 Battery tests
2.2.1 Assembly
2.2.2 Cycling
2.2.3 Polarization curves
2.2.4 Electrochemical impedance spectroscopy
2.3 Symmetric cell
2.3.1 Principle
2.3.2 Impedance variation with SOC
2.4 Segmented cell
2.4.1 Internal design
2.4.2 Segmentation and local currents
2.4.3 Challenges of local potential probing
2.4.4 Assessment of the local RHE
2.4.5 Test bed
2.5 Conclusion
3 Separate characterization of the components
3.1 Electrochemical analysis
3.1.1 Electrolytes and assumptions
3.1.2 Electrochemical cell
3.1.3 Potential window and reversibility of a redox couple
3.1.4 RDE voltammetry
3.1.5 Cyclic voltammetry
3.1.6 Discussion of the results
3.2 Physico-chemical electrolyte properties
3.2.1 Conductivity
3.2.2 Viscosity
3.2.3 Material compatibility and photodegradation of catholyte
3.3 Membrane characterization
3.3.1 Membrane pretreatments: FTIR-ATR study
3.3.2 Membrane affinity with solutions
3.4 Porous electrode characterization
3.4.1 Structural observation
3.4.2 NMR analysis
3.4.3 Comparison of two materials by blocking electrode model
3.5 Conclusion
4 Half-cell characterization
4.1 Stationary porous electrode model
4.1.1 Fundamental relations
4.1.2 Analytical expression with infinite electronic conductivity
4.1.3 Electrode impedance
4.1.4 Electrode impedance in blocking electrode conditions
4.2 Global impedance of the symmetric cell: methodology
4.3 Catholyte symmetric cell
4.3.1 Effect of the light exposure of the catholyte on cell impedance
4.3.2 Effect of the electrode material on cell impedance
4.3.3 Rcell vs SOC curves: Identification of the degradation mechanisms
4.4 Anolyte symmetric cell
4.4.1 Impedance evolution in circulation
4.4.2 Analysis of the cell overcharge
4.4.3 Rcell vs SOC curves
4.5 Conclusion
5 Full cell study
5.1 Standard cycling
5.1.1 Comparison standard / segmented cell
5.1.2 Parameters evolution during cycling
5.2 Influence of operating conditions
5.2.1 Strategy to investigate internal heterogeneity of battery operation
5.2.2 Current density
5.2.3 Flow rate
5.2.4 Temperature
5.3 Summary of the parameter study
5.3.1 Comparison of the impact of the three parameters
5.3.2 Development of an operational map
6 Industrial cell study
6.1 Hydraulic study
6.1.1 Model development
6.1.2 Computational results
6.1.3 Clear flow cell experiment
6.2 Flow rate optimization
6.2.1 Strategy
6.2.2 Results
6.3 Conclusion
Conclusions
Appendices
Bibliography
