Application of NDIR-TDLAS spectroscopy for rapid vapor and multi-gas analysis in high-temperature fuel cell/electrolyzer cells

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/382085182 Application of NDIR-TDLAS spectroscopy for rapid vapor and multi-gas analysis in high-temperature fuel cell/electrolyzer cells Conference Paper · July 2024 CITATIONS 0 READS 156 5 authors, including: Stefano Frangini ENEA 84 PUBLICATIONS1,485 CITATIONS SEE PROFILE Davide Pumiglia ENEA 28 PUBLICATIONS247 CITATIONS SEE PROFILE Simone Mataloni ENEA 5 PUBLICATIONS3 CITATIONS SEE PROFILE Francesca Santoni Parthenope University of Naples 22 PUBLICATIONS157 CITATIONS SEE PROFILE All content following this page was uploaded by Stefano Frangini on 09 July 2024. The user has requested enhancement of the downloaded file. EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 1/10 B0216 Application of NDIR-TDLAS spectroscopy for rapid vapor and multi-gas analysis in high-temperature fuel cell/electrolyzer cells Stefano Frangini (1), Massimiliano Della Pietra (1), Davide Pumiglia (1), Simone Mataloni (1,2), Francesca Santoni (1) (1) ENEA CR Casaccia, TERIN-DEC-H2V, Via Anguillarese 301, 00123 Rome, Italy (2) University of Rome " La Sapienza", Department of Chemical Engineering Materials and Environment, Via Eudossiana 18, 00184 Rome, Italy Tel.: +39-0630483138 stefano.frangin[email protected] Abstract Optimization of high-temperature fuel cell/electrolyzer systems is of greatest importance in order to improve their costs, safety, reliability and performance. Since these systems utilize complex humid gas mixtures, a rapid and reliable gas analysis can be undoubtedly one important tool to achieve these objectives. In this context, optical spectroscopic methods based on Non-Dispersive InfraRed (NDIR) technology may offer the possibility of a rapid, continuous and sensitive analysis of gases commonly present in fuel cell and electrolyzers such as CO2, CO and CH4 as compared to traditional non-optical methods with the further benefit of a more economical equipment and low maintenance costs. As further advantage, NDIR can be easily interfaced with Tuned Diode Laser Absorption Spectroscopy (TDLAS), one of the few techniques available for quantitative vapor analysis over a wide concentration range of up 30-35%. In spite of these premises, NDIR -TDLAS applications in this field are still very limited. In our laboratory a gas analysis system based on NDIR and TDLAS has been recently installed to test its suitability for rapid, quantitative multi-gas and vapor detection in Molten Carbonate (MC) and Solid Oxide (SO) devices at both single-cell and short-stack levels. Results obtained during the validation and verification experimental campaign will be reported in this work to document potentialities and limits of the NDIRTDLAS analytical method for the intended use. 1. Introduction Thanks to high efficiency and environmental sustainability, high-temperature electrochemical energy conversion and storage systems based on the fuel cell technology of Molten Carbonate (MC) and Solid Oxide (SO) electrolytes are poised to play a key role in facilitating the transition to future energy systems based on renewable or low carbon energy sources [1,2]. Although MC has the unique ability to capture CO2 during its electrical operations, SO cells have received the most attention in recent years because of their superior ability to operate in reversible mode (i.e., fuel cell or electrolysis) along with a better efficiency and durability [3]. Most current research is directed to optimizing cell performance and operations through appropriate testing and simulation tools in an attempt to further extending durability and reliability of the MC and SO systems and thus supporting their EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 2/10 commercial development [4]. Although improvements in analytical methods for rapid, continuous exhaust gas analysis is currently a neglected area of research, nevertheless it could provide a great optimization contribution in various respects, including, for example, to increase operational safety, protect system equipments, quickly control any compositional deviation from expected gas values or during dynamic or transient operations, where simulation tools and thermodynamics are much less effective in predicting exhaust gas compositions [5]. Many commercial gas analyzers based on Non-Dispersive Infra-Red (NDIR) spectroscopy have become available in these last years [6,7]. Due to their real-time analytical capabilities, low cost, maintenance and easy portability their application appears convenient particularly for analysis of permanent gases such as CO2, CO or CH4 as compared to traditional slow, off-line analytical techniques based on gas chromatography (GC). Another important advantage is that NDIR analyzers may be easily integrated with thermal and electrochemical sensors for the additional analysis of H2 and O2, respectively [8]. MC or SO applications of NDIR spectroscopy can be found in literature, but mostly limited to routine gas control tasks in auxiliary units (reformers, biomass gasifiers) of large-scale fuel cell plants [9,10]. Conversely, examples of NDIR gas analysis applied directly on MC or SO fuel cell modules are still very scarce [11]. Thus, a more systematic and fundamental research appears important at laboratory scale to develop optimized, standard procedures for a reliable NDIR analytical application to fuel cell gas products over long-term cell or stack operations. Although NDIR has its own merits, it has also some disadvantages. Apart a lower sensitivity as compared to GC, one serious limitation is that gas samples must be pre-treated with drying systems in order to avoid absorption interferences by water vapor with consequent loss of accuracy [12]. This means that, similarly to GC, NDIR cannot be used for determination of water vapor in gases. As MC and SO cells may contain large amounts of vapor, especially when operating in electrolysis mode, implementing a direct vapor analytical system would be another important step for testing and optimization of MC and SO cells, also in terms of operational safety. In fact, if one considers the high temperatures of these cells, a vapor analysis monitoring could be also a useful tool to control flammability of fuelcontaining gas streams. Tuneable Diode Laser Absorption Spectroscopy (TDLAS) is a recently-developed nearinfrared analytical technique allowing rapid, precise and accurate measurements of various gaseous species including water vapor in a wide concentration range from trace up to high levels [6,13]. Thanks to its robustness and nearly calibration-free characteristics, it is today a widely accepted method for demanding diagnostic applications in several industries such as in combustion [14], biomedical [15], sensors [16] and environment [17]. Although TDLAS is compatible with high temperatures and corrosive gas environments [13], no studies have been until now reported about TDLAS applied to vapor monitoring in high temperature fuel cells or electrolysers based on MC or SO electrolytes, at least to the best authors’ knowledge. Occasional research interest for TDLAS in the fuel cell area has been found yet limited to water transport studies in low temperature fuel cells [18,19]. Due to similar analysis times and gas flow conditions, TDLAS is well suited for direct coupling with NDIR analysis methods. With reference to above, a combined NDIR-TDLAS spectroscopy has been therefore setup in our laboratory to implement a rapid, in-line multi-gas as well as vapor monitoring system of MC and SO gas mixtures. In this work, the effectiveness of the NDIR-TDLAS method for the intended application will be illustrated in terms of some key performance indicators such as precision, accuracy and time of response at various gas conditions. Selected examples of NDIR-TDLAS spectroscopy applied to working MC/SO single cells as well as short-stacks will be also shown. EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 3/10 2. Scientific approach As mentioned above, NDIR and TDLAS have similar gas sampling flow rate requirements, being at 1-3 L/min for optimal results. These gas sampling conditions are compatible with the gas flow rates involved in short-stack cell experiments (1-1.5 L/min). However, problems may arise when applying NDIR-TDLAS to single cells, since gas flow rates in this case are significantly lower than 1 L/min (about 3-600 mL/min). Therefore, one primary objective of this work is to show the effect of lower than recommended gas flow rates on analytical performance of the NDIR-TDLAS technique. 3. Experiments Implementation of the in-line analytical system was realized using a commercial NDIR analyzer (ETG model MGA 100 Syn P) and a customized TDLAS vapor analyzer (ETG model 6900 P). The NDIR analyzer is integrated with a TCD sensor for the H2 analysis and with an electrochemical (ECD) sensor for the O2 analysis. A Peltier-type gas treatment unit (model ETG PSS 100) was also used for gas sample drying before they are being sent to the NDIR analyzer. Gas sampling was obtained at free gas flow, i.e. without using aspiration methods in order to avoid any cell performance alteration during the gas analysis. Two different NDIR-TDLAS setup were adopted depending on the gas flow rates to be sampled as illustrated in Fig.1. The configuration (a) was used for high gas flow rate situations (1L/min and higher), i.e. in stacks. Briefly, in this configuration TDLAS is coupled with the NDIR analyzer. Thus, the gas outlet from the stack fuel cell module is sent firstly to the TDLAS analyzer through a heated insulated transfer line made of a 1 mt long stainless steel flexible tube. The gas sample is then sent into the NDIR multigas analyzer through the gas drying unit using simple flexible PVC tubes. However, when gas flow rates are much lower than 1L/min as in the case of single cells an excessive pressure drop along the gas sample path was observed to impede gas from entering into the NDIR analyser. Thus, TDLAS and NDIR analysis was necessarily performed separately according to the scheme of Fig. 1(b). Thus, a first gas sample is sent to the TDLAS analyzer through the heated transfer line. After completing the vapor analysis, a manual switching valve is used to send the sample gas to the NDIR analyzer via the gas drying station. This scheme was also used for the validation tests with dry and/or controlled humidified gases that were prepared with certified gas cylinders and gas/vapor Bronkhorst control systems. Fig. 1 – TDLAS-NDIR setup used for analysis at: (a) high gas sampling rates (>1L/ min), (b) low gas sampling rates (<1L/min). EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 4/10 4. Results The TDLAS analyzer was validated in a H2-N2 gas mixture at various vapor concentrations and at two different gas flow rates. A first series of TDLAS analysis was conducted at the minimum recommended gas flow rate, namely at 1L/min gas flow rate. A typical analyzer output obtained at various vapor concentrations up to 35% during wet-up and drying-down cycles are reported in Fig.2(a). In the figure, some outgassing spikes are also visible at higher vapor concentrations due to residual water vapor condensing inside the instrument optical path. Analytical results are summarized in Table 1. In general, accuracy is good (about ± 3%) being the measured values in very good agreement with the input vapor (see Fig. 2(b)). Results obtained during the wet-up and drying-down cycles are also comparable indicating good reproducibility and robustness even at high vapor concentrations. For every input, stable output signals (time of analysis) are rapidly obtained (a few tens of seconds), although times of analysis tend to slow down during the drying-down cycle. Fig. 3(a) reports a typical TDLAS output at lower than recommended gas flow rates, precisely at 400 mL/min. Output signals are not perfectly stable showing tendency to drift over time. Occasional outgassing spikes are also visible at high vapor concentrations. Additional presence of a transient signal after each change of vapor concentration is probably related to overshoot of the water mass flow PID controller. Analytical results are shown in Table 2. It is immediately seen that the TDLAS analyzer needs of longer times of analysis with respect to the high gas flow rate condition (ca. ten times more). Moreover, Fig. 3(b) shows that there are distinct accuracy issues, especially at higher vapor concentration conditions (>20%), where accuracy drops at ca. ±10%. Analyzing the wet-up and dryingdown cycle results, it can be said that precision remains good in the lower vapor concentration range (0-10%). Based on all these results, it can be concluded that gas flow rate has a strong impact on the TDLAS analysis and that 1L/min must be seen as a threshold level to get rapid and accurate results. NDIR validation was conducted in 50CO2-50N2 and in 50CO2-25N2-25H2 gas mixtures. The effect of gas flow rates in the range 0.1-1L/min was studied alternating a measurement with an air purging step, during which the N2 concentration raises to 80% as seen in Fig.4, which shows a typical NDIR output obtained in the binary 50CO2-50N2 gas mixture. It is immediate to see that CO2 accuracy and time of analysis are all parameters strongly affected by the gas flow rate and that a minimum gas flow rate of 0.4L/min is required for accurate CO2 results (better than ±2%) and rapid times of analysis (16s or less). Fig. 2 – TDLAS vapor validation analysis at 1 L/min gas sampling rate: (a) typical output obtained during a wet-up and drying-down cycle, (b) Accuracy plot built by comparing the analytical results with the theoretical amount of vapor (reference line). 2000 3000 4000 5000 6000 0 5 10 15 20 25 30 35 40 45 50 * * * * * * * 5 % 10 % 15 % 20 % 30 % 35 % 25 % 20 % 15 % 10 % vapor conc. / % time / s vapor additions 5 % ** outgassing spikes 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 vapor output / % vapor input / % reference measured values, wet-up cycle measured values, drying-down cycle accuracy =± 3% EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 5/10 Table 1 – Summary of the TDLAS results obtained during the wet-up and drying-down cycle at 1 L /min gas sampling rate. Fig. 3 – Vapor TDLAS validation analysis at 0.4 L/min gas sampling rate: (a) typical output obtained during a wet-up and drying-down cycle, (b) corresponding accuracy plot. Table 2 – Summary of the TDLAS results obtained during a wet-up and drying-down cycle at 0.4 mL/min. Fig.5 shows the NDIR output in the ternary 50CO2-25N2-25H2 gas (Fig. 5 (a)) and the corresponding accuracy results for CO2 and H2 (Fig. 5(b)). Conversely to the previous case, a ternary matrix gas requires minimum a 0.8L/min gas flow rate to get rapid stable results with ca. a ±2 % accuracy suggesting that the NDIR analysis times depend also on the number of analyzed gas components. Repeated measurements (results not shown) with both the binary and ternary gases indicate that precision is less affected than accuracy by the gas flow rate remaining at a good 1-2 % in the range 0.4-1L/min. Due to the complex MC/SO fuel cell gas environment (usually, ternary or quaternary mixtures), gas flow rates higher than 0.8L/min are always required if one needs highly accurate NDIR results. Wet-up cycle Drying-down cycle Vapor % Time of analysis (s) Vapor % Time of analysis (s) input output input output 5 5 200 30 28.5 60 10 10 60 20 22 120 15 15 60 15 16 120 20 21 70 10 11 80 25 26 40 5 5 140 35 34 120 0 0.2 60 Wet-up cycle Drying-down cycle Vapor % Time of analysis (s) Vapor % Time of analysis (s) input output input output 5 4 500 10 11 700 10 10.5 800 0 0.5 300 15 15.5 500 20 19 700 25 23 900 30 27.5 1000 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 vapor output / % vapor input / % reference measured values, wet-up cycle measured values, drying-down cycle accuracy = ± 10% 4000 6000 8000 10000 0 5 10 15 20 25 30 35 40 45 50 # # # # # # # * * * 5 % 10% 30% 25% 20% 15% vapor conc. / % time / s 10% vapor additions * outgassing spikes # PID control overshoot # EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 6/10 Fig. 4 – Typical NDIR output obtained during validation analysis in binary 50CO2-50N2 gas mixture at increasing gas flow rates from 0.1 up to 1 L/min. Fig. 5 – NDIR validation analysis in ternary 50CO2-25N2-25H2 gas mixture: (a) typical NDIR output at increasing gas flow rates from 0.2 up to 1 L/min and (b) accuracy results. Verification studies were made with real gas samples of MC/SO fuel cells at both single-cell and stack level. For sake of brevity, only analysis of anode gas exhausts will be reported in this work. In a first series of measurements, NDIR-TDLAS analysis was applied to a MC single cell fueled at the anode with 20H2-20H2O-20CO2-40N2 gas at a total 0.4L/min flow rate. Consequently, the measurements were made according to the scheme of Fig. 1(b). Fig.6 reports a typical TDLAS output obtained during the vapor analysis of the MC single cell operating at 650°C without current load (Open Circuit Voltage, OCV). Nearly stable results are obtained after ca. 500 seconds, which is in accordance with the time of analysis measured during the validation tests as reported in Table 2. Fig. 6 – Typical TDLAS output obtained during vapor analysis at the anode exhaust of a MC cell heated at 650°C, at OCV conditions. Gas inlet was: 20H2O-20H2-20CO2-40N2 with a total gas flow rate of 0.4 L/min. (b) Gas flow rate (L/min) CO2 accuracy (± %) H2 accuracy (± %) 0.2 14 40 0.4 6.7 14 0.6 4 8 0.8 2 2.5 1.0 0.4 1.6 480 540 600 660 720 780 840 900 960 0 10 20 30 40 50 60 70 80 90 100 measured CO2 % 0.5 0.5 1 2 6 12 accuracy (± %) 50.3% 50.6% 51.2% 53.1% 5 7 10 16 26 1 0.8 0.6 0.4 0.2 concentration / % time / s CO2 N2 0.1 50 flow rate (L/min) time of analysis (s) 56% input : 50CO2-50N2 50.3% 1000 1250 1500 1750 2000 0 10 20 30 40 end of gas sampling vapor conc. / % time / s start of gas sampling stable value =21-22% 180 240 300 360 420 480 540 0 10 20 30 40 50 60 70 80 90 100 49.8 48.9 24.725.6 48.0 27.029.7 46.7 35.9 42.9 measured H2 % measured CO2 % 10 14 18 20 1 0.8 0.6 0.4 concentration / % time / s CO2 H2 N2 0.2 input: 50CO2-25N2-25H2 flow rate (L/min) 32 time of analysis (s) (a) EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 7/10 The corresponding NDIR output is shown in Fig. 7. Stable results are obtained after 40-50 seconds of continuous gas sampling. Due to CO production from the CO2 reverse gas shift reaction, the anode exhaust is composed of a quaternary gas mixture. Consequently, times of analysis are a bit longer than those measured during the validation tests made in the ternary gas (see Fig.5). Fig. 7 – Typical NDIR output obtained during the multi-gas analysis at the anode exhaust of a MC single cell heated at 650°C, at OCV conditions. Gas inlet was: 20H2O-20H2-20CO240N2 with a total gas flow rate of 0.4 L/min. Combining the TDLAS and NDIR results, the Table 3 reports the gas composition of the anode exhaust given on both dry and wet basis (%). The equilibrium gas composition computed with a thermochemical software (HSC Chemistry 5) is also reported in the table. Comparison between the analyzed and the computed data allows to confirm that gas flow rate has a significant impact on the accuracy of NDIR-TDLAS measurements. Under these gas flow conditions (0.4 L/min), accuracy was similar for both the analyzers (± 5-10 %). To note that, in contrast to the equilibrium composition, the measured CO2 concentration is significantly higher than H2. In accordance with [20,21], it is believed that the CO2 utilization at the anode by the reverse gas shift reaction may cause some CO2 transfer from cathode to anode even at OCV, as also the cathode gas of a MC cell contains CO2. Overall, these results indicate that acceptable results are obtainable even at such low gas flow rates at the expense of only some loss in rapidity and measurement accuracy. Table 3 – Experimental anode exhaust composition of the MC single cell given on both dry and wet basis. Theoretical equilibrium composition is also given for comparison. The results of gas/vapor analysis from the anode exhaust of a SO multi-stack module at OCV conditions and 800°C is illustrated in the next two figures. Due to high gas flow rates (> 1L/ min), the scheme of Fig.1 (a) was employed. Inlet anode gas composition was as follows: 55H2-5H2O-45CO2. Fig. 8 (a) and (b) shows the TDLAS and the NDIR output, Gas component Input composition (%) Analyzed composition (%) Calculated composition (%) Dry basis Wet basis Wet basis H2O 20 ___ 21.5 24.7 H2 20 20.2 15.8 15.3 CO2 20 25.5 19.9 15.3 CO __ 4.2 3.3 4.7 N2 40 49.5 38.6 40 2900 2925 2950 2975 3000 3025 3050 0 10 20 30 40 50 60 70 80 90 100 CO=4.2% H2=20.2% CO2=25.5% concentration / % time / s start of gas sampling N2=49.5% EFCF 2024: 16th European SOFC & SOE Forum 2 – 5 July 2024, Lucerne, Switzerland ThemeChapter xx - Session Xyy - 8/10 respectively. As shown in Fig. 8 (a), the vapor analysis was repeated twice for reproducibility. Despite some outgassing peaks, stable result (vapor=25.5%) was achieved after ca. 2-300 seconds of gas sampling. In comparison, the NDIR multi-gas analysis was more rapid (time of analysis < 10 seconds), as visible in Fig. 8 (b). Overall analytical results are presented in Table 4 along with the theoretical equilibrium gas composition. Excellent correlation exists between analytical and theoretical values resulting in accuracy generally better than ± 2 %. Further, conversely to MC, optimal correspondence of measured and theoretical CO2 can be also observed. This is because the cathode gas of a SO cell usually does not contain CO2, so CO2 accumulation at anode due to transfer effects is not possible in such systems. Fig. 8 – NDIR-TDLAS analysis for the anode gas exhaust of a SO stack heated at 800°C, at OCV conditions. Gas flow rates > 1 L/min. Anode gas inlet was: 55H2-5H2O-45CO2. Typical output chart of the: (a) TDLAS and (b) NDIR analyzer. Gas component Input composition (%) Analyzed composition (%) Calculated composition (%) Dry basis Wet basis Wet basis H2O 5 ___ 25.5 26 H2 55 45 33.4 33 CO2 40 24 18.3 18 CO __ 30 22.8 21 Table 4 – Measured anode exhaust composition of the SO stack module given on both dry and wet basis. Theoretical equilibrium composition is also given for comparison. 5. Summary This work described the implementation of an inline NDIR-TDLAS spectroscopy method for a rapid and reliable gas/vapor analysis in MC/SO single cell and stack laboratory stations. Our results showed that gas flow rate has a strong impact on the rapidity and accuracy of the NDIR-TDLAS measurements and that best results were obtained in analysis of stack gases, where sufficiently high gas flow rates are available (≥ 0.8-1 L/min). Vapor/gas analysis in single cells is also possible. However, a minimum gas flow rate of 0.4 L/min is required to limit NDIR-TDLAS performance losses at acceptable values. 8000 9000 10000 11000 12000 13000 0 5 10 15 20 25 30 35 40 45 start of measurement vapor concentration / % time / s start of measurement time of analysis =2/300 s (a) 1180 1200 1220 1240 1260 1280 1300 0 10 20 30 40 50 60 70 concentration / % time / s N2 H2= 45 % CO= 30 % CO2= 24% time of analysis < 10s (b)