Hydrogen fuel quality from two main production processes: Steam methane reforming and proton exchange membrane water electrolysis

Journal of Power Sources 444 (2019) 227170

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Hydrogen fuel quality from two main production processes: Steam methane reforming and proton exchange membrane water electrolysis Thomas Bacquart a, *, Karine Arrhenius b, Stefan Persijn c, Andr�es Rojo d, Fabien Aupr^etre e, Bruno Gozlan f, Niamh Moore a, Abigail Morris a, Andreas Fischer b, Arul Murugan a, �ndez d, Sam Bartlett a, Guillaume Doucet e, François Laridant e, Eric Gernot e, Teresa E. Ferna d f f g �n Go �mez , Martine Carr�e , Guy De Reals , Frederique Haloua Concepcio a

Chemical, Medical, Environmental Science Department, National Physical Laboratory, Hampton Road, Teddington, Middlesex, TW11 0LW, United Kingdom RISE Research Institutes of Sweden, Box 857, 501 15, Borås, Sweden c VSL, Thijsseweg 9, Delft, the Netherlands d Centro Espa~ nol de Metrología (CEM), Alfar, 2, 28760, Tres Cantos, Madrid, Spain e AREVA H2Gen, 8 Avenue du Parana, 91940, Les Ulis, France f Air Liquide, Paris-Saclay Research Center, BP 126, 78353, Jouy en Josas, France g Laboratoire National de m�etrologie et d’Essais, 29 Avenue Roger Hennequin, F-78197, Trappes Cedex, France b

H I G H L I G H T S

� No contaminants above ISO 14687-2 threshold in H2 from SMR with PSA. � No contaminants above ISO 14687-2 threshold in H2 from PEMW with TSA. � Impact of TSA on contaminants from PEMW electrolyser. � Sampling contamination may lead to false positive. � Probability of contaminants presence in line with real hydrogen samples. A R T I C L E I N F O

A B S T R A C T

Keywords: Fuel cell electrical vehicles ISO14687 Gas analysis Hydrogen production Hydrogen quality

The absence of contaminants in the hydrogen delivered at the hydrogen refuelling station is critical to ensure the length life of FCEV. Hydrogen quality has to be ensured according to the two international standards ISO 14687–2:2012 and ISO/DIS 19880-8. Amount fraction of contaminants from the two hydrogen production processes steam methane reforming and PEM water electrolyser is not clearly documented. Twenty five different hydrogen samples were taken and analysed for all contaminants listed in ISO 14687-2. The first results of hydrogen quality from production processes: PEM water electrolysis with TSA and SMR with PSA are presented. The results on more than 16 different plants or occasions demonstrated that in all cases the 13 compounds listed in ISO 14687 were below the threshold of the international standards. Several contaminated hydrogen samples demonstrated the needs for validated and standardised sampling system and procedure. The results validated the probability of contaminants presence proposed in ISO/DIS 19880-8. It will support the implementation of ISO/ DIS 19880-8 and the development of hydrogen quality control monitoring plan. It is recommended to extend the study to other production method (i.e. alkaline electrolysis), the HRS supply chain (i.e. compressor) to support the technology growth.

1. Introduction The expansion of fuel cell electrical vehicles (FCEV) is a crucial step

to decarbonise the transport sector [1]. According to the European Commission, hydrogen could represent 32% of the European fuel mix in 2050 [2]. The European hydrogen market will increase significantly

* Corresponding author. E-mail address: [email protected] (T. Bacquart). https://doi.org/10.1016/j.jpowsour.2019.227170 Received 5 June 2019; Received in revised form 27 August 2019; Accepted 17 September 2019 Available online 22 October 2019 0378-7753/Crown Copyright © 2019 Published by Elsevier B.V. This is an open (http://creativecommons.org/licenses/by-nc-nd/4.0/).

access

article

under

the

CC

BY-NC-ND

license

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

over the few next years with the following objectives in Europe: France: 100 hydrogen refuelling stations (HRS) by 2023 and 400–1000 by 2028 [3]; Germany: 100 HRS in 2019 and 400 by 2023 [4]; the Netherlands: 50 HRS by 2025 [5] and United Kingdom: 20 HRS by 2020 [6]).FCEV are developing quickly across a wide range of vehicle types, from pas­ senger vehicles (i.e. Toyota Mirai [7,8]) to truck (i.e., Nikola [8], Hyundai [9]) or fork-lift (i.e. PlugandPower [8–10]). The absence of contaminants in the hydrogen delivered at the hydrogen refuelling sta­ tion (HRS) is critical to ensure the length life of FCEV [11] (dura­ bility > 5000 h [12]). The international standard ISO 14687:2012 defined the maximum amount fraction of each contaminant in hydrogen for FCEV applications [13]. Therefore, hydrogen suppliers and pro­ ducers need to ensure hydrogen quality defined in ISO 14687–2:2012 or SAE J2719:2015 [14] at hydrogen refuelling station (HRS). Currently the analytical cost (sampling and analysis) to measure all these con­ taminants on a routine scheme is a non-negligible cost. Decreasing hydrogen cost is a prerequisite for the development of mass expansion of hydrogen into transport sector [15]. European Directive 2014/94/EU on the deployment of an alternative fuels infrastructure [16] sets out that “The hydrogen purity dispensed by hydrogen refuelling points shall comply with the technical specifications included in the ISO 14687-2 standard (which will be replaced by the European standard EN 17124:2018 [17]). Based on the requirement of ISO/DIS 19880-8 [18] for fuel quality control for HRS and EN 17124:2018, a quality control plan has to be developed and implemented for each HRS. The quality control plan may be based on the probability risk of contaminants similar to risk assessment principle. The development of the probability risk of contaminant requires an understanding and evidence for each contaminant’s presence probability in each part of the supply chain. Scarce information on hydrogen fuel quality is currently available and most of the information were reported for hydrogen fuel quality sampled at the HRS nozzle. Aarhaug et al. [19,20] were not able to establish a link between feedstock and hydrogen contaminants results at the nozzle. The EMPIR 15NRM03 Hydrogen project [21] aimed to provide the first evidence for hydrogen fuel quality from two production processes: proton exchange membrane (PEM) water electrolysis with temperature swing adsorption (TSA) and steam methane reforming (SMR) with pressure swing adsorption (PSA). The results of analysis from several SMR with PSA and PEM water electrolysis with TSA support the recent study by Bacquart et al. [22] proposing a probability of presence for these production processes. The objective of this study is to demonstrate how hydrogen quality can be ensured according to the two international standards ISO 14687–2:2012 and ISO/DIS 19880-8 [23]. The report provides technical evidences for the development of quality control plan according to the methodology of risk assessment defined in ISO/DIS 19880-8. The study presents additional information on sampling issues, limit of detection from various analytical methods and impact of purification on the contaminant’s presence for PEM water electrolysis.

hydrogen production plants in Europe (PEM water electrolyser, steam methane reformer) and one hydrogen refuelling station with hydrogen from chlor-alkali process. Thirteen samplings were performed on different PEM water electrolysers (eight with purification by tempera­ ture swing adsorption and five without purification) in Europe covering different manufacturers and technologies. Eleven samplings were per­ formed on different steam methane reformers with purification by pressure swing adsorption in Europe covering different manufacturers and technologies. 2.1. Hydrogen sampling from production process During the project, 25 independent samplings of hydrogen at various production process plants in Europe were performed using a sampling procedure [24]. The project partners and the participants were extremely careful to avoid contamination during samplings by following the exact protocol and ensuring preparation of the sampling cylinder by evacuation. The cylinders used for sampling were different depending on availability, safety and laboratories requirements. The type of cyl­ inders used are aluminium cylinder (5 and 10 L), aluminium cylinder (10 L) with SPECTRA-SEAL passivation (BOC, UK), stainless steel cyl­ inder (1 L) and stainless steel cylinders (1, 2 and 4 L) sulfinert® coated (Silcotek, US). The analyses were performed over a long period of time due to shipment delay between the sampling site and the partners. Even if the consortium minimised as much as possible the delay, it is impor­ tant to consider that there could be several weeks to months between the sampling date and the analysis completion date. It is currently difficult to assess the stability of the different contaminant over time (between cylinder reception and analysis) in hydrogen gas cylinder especially for reactive species (i.e. total sulphur, formaldehyde, formic acid and hydrogen chloride) due to the lack of stability studies. 2.2. Analytical methods A summary of the analytical methods used in the study is provided below with the reported limit of detection (Table 1). The limits of detection in the table correspond to the lowest limit of detection re­ ported in the sampling campaign by the national metrology institutes. Detail of the analytical methods, calibration strategy and gas stan­ dards are given for each method and each laboratory. 2.2.1. Optical Parametric Oscillator (OPO)-based cavity ring down (CRDS) spectrometer VSL, the Dutch Metrology institute (Delft, the Netherlands) measured 4 different analytes (formic acid, formaldehyde, hydrogen chloride and ammonia) with the same Optical Parametric Oscillator (OPO)-based Cavity Ring Down (CRDS) spectrometer [25,26]. Fig. 1 shows a schematic overview of experimental set-up based on an OPO as light source and CRDS used for the gas detection. The pump laser of the OPO consists of a narrow line width fibre laser (NKT Photonics, output power set at 4.4 mW) which is amplified in a fibre amplifier (IPG Pho­ tonics). This combination provides a wide mode-hop free tuning range of 100 GHz and an output power up to 10.5 W at 1064 nm. The output of the amplifier is coupled into the OPO cavity via a collimator (COL), Faraday isolator (FI) and AR-coated focusing lens (L1). The periodically poled crystal (PPLN) from HC-Photonics is contained in an oven and has AR coatings for signal, pump and idler wavelengths. The mirrors are highly transparent for both idler and pump wavelengths and highly reflective for the signal wavelength. Within the PPLN crystal the pump light is converted into the signal and idler with the signal resonating in the OPO cavity. The output of the OPO is collimated using a CaF2 lens (L2). Signal and residual pump are separated from the idler using a dichroic beam splitter (DBS). Part of the idler beam is directed to a wavelength meter (Bristol instruments) using a ZnSe window placed near the Brewster angle. Different PPLN crystals and mirror sets in the OPO cavity are used

2. Materials and methods In the EMPIR Hydrogen project, an analytical campaign has been organized to assess the level of contaminants listed in ISO 14687 in hydrogen from different production processes. In this report, analysis results of the steam methane reforming process (SMR) analytical campaign and the electrolysis process analytical campaign are presented for the contaminants analysed by four national metrology institutes: ~ ol de Metrología (CEM, Spain), National Physical Labora­ Centro Espan tory (NPL, United Kingdom), RISE Research Institutes of Sweden (RISE, Sweden) and Van Swinden Laboratory (VSL, The Netherlands). The scope of contaminants analysed corresponds to ISO standard 14687-2 except for particulates which is not included here [14]. The aim is to provide evidences to support the probability of presence of contami­ nants performed by process expert. In between 2016 and 2018, 25 hydrogen sampling was performed on 2

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

Table 1 Analytical techniques used by each national measurement institutes (NPL, VSL, RISE, CEM) for the analysis of the 13 gaseous contaminants in hydrogen fuel samples. The limits of detection are reported as indicative. Compounds

ISO 14687-2 threshold [μmol/ mol]

NPL

VSL

RISE

CEM

Water H2O

5

–

OFCEAS (LOD: 2 μmol/mol)

–

Methane CH4

2

–

GC-FID (LOD: 0.5 μmol/mol)

–

Non methane hydrocarbons Oxygen O2

2

–

–

–

5

Quartz crystal microbalance (LOD: 0.5 μmol/mol) CRDS (LOD: 0.5 μmol/ mol) GC-methaniser-FID (LOD: 0.01 μmol/ mol) GC-methaniser-FID (LOD: 0.01 μmol/ mol) GC-PDHID (LOD: 0.5 μmol/mol)

–

OFCEAS (LOD: 3 μmol/mol) GCTCD (Ar þ O2 LOD: 25 μmol/mol)

Helium He Nitrogen N2

300 100

– GC-PDHID (LOD: 1 μmol/mol)

– –

– GC-TCD (LOD: 25 μmol/mol)

Argon Ar

100

GC-PDHID (LOD: 0.5 μmol/mol)

–

GC-TCD (Ar þ O2 LOD: 25 μmol/ mol)

Carbon dioxide CO2

2

–

Carbon monoxide CO Total sulphur compounds Formaldehyde HCHO Formic acid HCOOH Ammonia NH3

0.2

–

OFCEAS (LOD: 0.1 μmol/mol) GC-TCD (LOD: 5 μmol/mol) OFCEAS (LOD: 0.02 μmol/mol)

0.004

GC-methaniser-FID (LOD: 0.01 μmol/ mol) GC-methaniser-FID (LOD: 0.01 μmol/ mol) GC-SCD (LOD: 0.002 μmol/mol)

GC-PDHID (LOD: 5 μmol/ mol) GC-TCD (LOD: 5 μmol/ mol) GC-TCD (LOD: 5 μmol/mol) GC-TCD(LOD: 25 μmol/mol) GC-PDHID (LOD: 25 μmol/ mol) GC-TCD(LOD: 50 μmol/mol) GC-PDHID (LOD: 25 μmol/ mol) –

–

OFCEAS (LOD: 0.004 μmol/mol)

–

0.01

–

–

–

0.2

–

–

–

0.1

–

–

–

0.05

–

–

–

2 2 2 2 2

– – – – –

CRDS (LOD: 0.005 μmol/mol) CRDS (LOD: 0.05 μmol/mol) CRDS (LOD: 0.1 μmol/mol) CRDS (LOD: 0.005 μmol/mol) – – – – –

GC-FID (LOD: 0.5 μmol/mol) GC-FID (LOD: 0.5 μmol/mol) GC-FID (LOD: 1 μmol/mol) GC-FID (LOD: 1 μmol/mol) TD-GC-FID/MS (LOD: 0.05 μmol/ mol)

– – – – –

Total halogenated (HCl) C2 hydrocarbons C3 hydrocarbons C4 hydrocarbons C5 hydrocarbons C6 – C18 hydrocarbons

–

around 2 km. Both the pressure regulator and the CRDS measurement cell are SilcoNert 2000 coated measurement cell to reduce adsorption and re­ action effects (important for HCl, ammonia and formic acid but less relevant for formaldehyde). For the tubings either polymer tubings or SilcoNert 2000 coated stainless steel tubing are used (tubings are kept as short as possible by placing sample cylinders on the optical table near the CRDS cell). Measurements are typically performed using a flow rate of 30 l/h set by a mass flow controller (coated with SilcoNert® 2000). The cell pressure is controlled using a combination of a pressure regu­ lator and a membrane pump. It is normally set at atmospheric pressure. The time response of the system is dependent on the compound: formaldehyde is very fast, followed by ammonia and finally hydrogen chloride and formic acid. As an example, the measurement reading for formaldehyde takes about 1 min to stabilize while ammonia takes about 12 min to stabilize (before this, the flow system and reducer were exposed to clean H2). Note that this time is amount fraction dependent for reactive compounds, the lowest amount fractions require the longest stabilization times. Depending on the compound and sample size the total measurement is typically 15–45 min per component (with form­ aldehyde and formic acid measured simultaneously). Static reference gas standards in nitrogen are available for ammonia (produced by VSL), hydrogen chloride (commercial supplier and certi­ fied by VSL and static reference gas standard from, CEM (Spain) and formaldehyde (commercial supplier and certified by VSL). These are diluted with high purity hydrogen for calibration. For formic acid a commercial mixture in a hydrogen matrix is available. This mixture has

Fig. 1. Schematics of the singly resonant cw OPO and CRDS spectrometer. The output of the seed fibre laser is amplified up to maximum 10.5 W and coupled into the OPO cavity. Part of the idler is directed to a wavelength meter and the rest is directed via an Acoustical Optical Modulator (AOM) to the cavity ring down measurement cell.

to cover the entire wavelength range of 2.3–5.1 μm. Formic acid, formaldehyde and hydrogen chloride are measured in a 10 cm 1 wavelength range where all 3 compounds strongly absorb (stronger absorption lines are available within the tuning range of the OPO but this would require significantly more time for the analysis). The back­ ground decay time is high (about 13.5 μs) which corresponds with a long effective absorption path length of about 4 km. Ammonia is measured at another crystal period and temperature. Here the background decay time is around 7 μs, corresponding to an absorption path length of 3

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

been diluted with nitrogen to enable a comparison with the formic acid spectrum contained in the PNNL spectral database [27].

90 � C and oven temperature is 100 � C. Stainless steel tubing is used for sampling lines. Pressure regulators were used with samples and PRMs. Two bar outlet pressure was used for sampling. The lines are purged flushing between 5 and 8 times before each analysis. After gaining experience, three 5-ways-valves are added to the sampling rig and used for connecting the sampling cylinders and the standards with the ana­ lyser. With this configuration, purge is more efficient reducing the time of analysis and the chance of air leaking within sampling lines. For helium calibration, PRMs were prepared gravimetrically in accordance with ISO 6142–1:2015 [17] in helium matrix starting from pure nitrogen (N2) BIP® and pure helium (He) BIP ®. The calibration curve ranges from 5 to 100 μmol/mol of helium in nitrogen. For oxygen calibration, PRMs were prepared gravimetrically in accordance with ISO 6142–1:2015 [17] from some PRMs (CEM, Spain). Pure nitrogen (N2) BIP® was used as balance gases. The calibration curve goes from 5 to 120 μmol/mol of oxygen in nitrogen. For argon and nitrogen calibration, PRMs were prepared gravimetrically in accordance with ISO 6142–1:2015 [17] in helium matrix starting from pure nitrogen (N2) BIP®, pure argon (Ar) BIP® and pure helium (He) BIP® from 25 to 150 μmol/mol of argon and nitrogen in helium. During analysis of ox­ ygen, argon, and nitrogen, some data were rejected due to air leak in the system. The calibration curve, results of analysis and uncertainties associated were determined using the methodology of ISO 6143:2001 [18] and in some cases a more conservative approach was used.

2.2.2. Gas chromatography with pulsed discharge helium ionization detector (GC-PDHID) NPL measured nitrogen, oxygen, argon by gas chromatography (Agilent, United Kingdom) with pulsed discharge helium ionization detector (VICI, Switzerland) using pure helium as a carrier gas (Helium BIP®þ, United Kingdom). Gases are sampled directly from the gas cyl­ inder to the GC-PDHID, a pressure regulator (set at 20 psig outlet) and a needle valve were used to restrict the flow to 30 ml/min. The GC/PDHID sampling loop volume was 1 ml and the sampled was then transferred onto capillary column molsieve 5A plot (30 m � 0.53 mm x 50 μm) and a second capillary column molsieve 5A plot (50 m � 0.53 mm x 50 μm). The GC oven was set at 30 � C and the PDHID detector was set at 180 � C. NPL primary reference materials (PRMs) in hydrogen matrix containing nitrogen (N2) and oxygen (O2) amount fraction were used to calibrate the analyser. Static reference gas standards and dynamic dilution system using mass flow controller system (Bronkhorst, NL) were used to generate calibration curve ranging from 1 to 75 μmol/mol of oxygen and 2–150 μmol/mol of nitrogen. The method can separate argon from ox­ ygen. Argon amount fraction was calibrated using a NPL PRM in helium matrix with 5 μmol/mol of argon. The data were scrutinised however no result was discarded without a technical reason. The calibration curve, results of analysis and uncertainties associated were determined using NPL software XLGENline [28]. An expanded uncertainty using a coverage factor of 2 was used. In some cases, a more conservative un­ certainty was derived from scientist expertise. CEM measured oxygen, Argon and Nitrogen with an Agilent 6890 GC with a PDHID detector (Pulsed Discharge Helium Ionization Detector) from VICI-VALCO. The capillary column was a HP-MOLESIEVE 5A (30 m � 0.53 mm x 50 μm). The injection temperature was 50 � C and the GC oven was set at 26 � C. The PDHID detector was set at 100 � C. Helium BIP® (Air Products, Spain) was used as carrier gas with flow rate of 1.5 ml/min. Pressure regulators were used with samples and PRMs and 2 bar outlet pressure was used for injection. The lines and regulators are purged before each analysis (3 times). An automatic sampler (designed by VSL) is programmed using AGILENT-Chemstation software to inject all the samples and the PRMs in the same analytical sequence. Oxygen and argon are not completely separated but it is possible to asses if any of them is present in the sample at the required level. For oxygen calibration, the PRMs were prepared gravimetrically in accordance with ISO 6142–1:2015 [29] from some PRMs (CEM, Spain). Pure nitrogen (N2) BIP® (Air Products, Spain) was used as balance gases. The calibration curve goes from 5 to 15 μmol/mol of oxygen in nitrogen. For argon and nitrogen calibration, PRMs were prepared gravimetrically in accordance with ISO 6142–1:2015 [16] in helium matrix starting from pure nitrogen (N2) BIP®, pure argon (Ar) BIP® and pure helium (He) BIP® from 25 to 150 μmol/mol of argon and nitrogen in helium. During analysis of oxygen, argon, and nitrogen, some data were rejected due to air leak in the system. The calibration curve, results of analysis and uncertainties associated were determined using the meth­ odology of ISO 6143:2001 [30] and in some cases a more conservative approach was used.

2.2.4. Gas chromatography with methaniser and flame ionization detector (GC-methaniser-FID) NPL measured carbon monoxide, carbon dioxide, methane and nonmethane hydrocarbons using a GC-methaniser-FID (Peak Laboratories, USA). The measurement of carbon monoxide, carbon dioxide and methane was done by separating them on a packed column Hayesep D (60/80 mesh, length 186 inch). The non-methane hydrocarbons were back flushed after the elution of CO, CO2 and CH4. The non-methane hydrocarbons eluted as one peak. The carbon compounds were con­ verted into methane using a methaniser set at 270 � 1 � C. The detector is a flame ionization detector (FID). Gases are sampled directly from the gas cylinder to the analyser. A needle valve was used to restrict the flow to 30 ml/min. The gas chromatography oven is set at 65 � C and the in­ jection loop volume equals to 5 ml. NPL PRMs in hydrogen matrix containing nitrogen (N2), carbon monoxide (CO), carbon dioxide (CO2), methane (CH4), ethane (C2H6) and oxygen (O2) were used to calibrate the GC. The calibration curve was from 20 nmol/mol to 10 μmol/mol for CO, CO2, CH4 and non-methane hydrocarbons (reported on methane basis). The data were scrutinised however no result was discarded without a technical reason. The calibration curve, results of analysis and uncertainties associated were determined using NPL software XLGEN­ line [16]. An expanded uncertainty using a k value of 2 was used. In some cases, a more conservative uncertainty was derived from scientist expertise. 2.2.5. Gas chromatography with sulphur chemiluminescence detector (GCSCD) NPL measured total sulphur by gas chromatography with sulphur chemiluminescence detector. The analysis of the sample is performed on an Agilent 7890A gas chromatograph (Agilent, USA) equipped with two detectors, a flame ionization detector and sulphur chemiluminescence detector (SCD 355, Agilent Technologies, USA). The GC/SCD sampling loop volume was 1 ml and the sample was then transferred onto capil­ lary column used which is a HP-5, 30 m � 0.320 mm ID x 0.251 μm film thickness (Agilent, USA). The column program temperature is isothermal at 110 � C. Helium is used as a carrier gas at a flow rate of 20 ml/min. Gases are sampled directly from the gas cylinder to the analyser, a needle valve was used to restrict the flow rate to 20 ml/min. Total sulphur analysis was calibrated using dynamic dilution of a gravimetric gas standard of H2S in hydrogen. Sulphur compounds are unstable at low nmol/mol amount fraction. Dynamic standards were

2.2.3. Micro gas chromatography with thermal conductivity analyser (μGCTCD) CEM measured helium, oxygen, argon and nitrogen using a micro gas chromatograph Agilent 3000A with thermal conductivity detector (Agilent, Spain). The micro-GC was equipped with a PLOT-U (3 m � 0.30 mm) pre-column and a Molsieve 5A column (10 m � 0.30 mm). A backflush injector type of 1 μL is installed. For Helium analysis, Argon BIP® is used as carrier gas, injection tempera­ ture is 55 � C and oven temperature is 50 � C. For oxygen, argon and ni­ trogen, He BIP® is used as carrier gas (in some occasions Argon BIP® was used as carrier gas for nitrogen analysis), injection temperature is 4

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

generated using a dilution system based on calibrated orifices and mass flow controller system (Bronkhorst, NL). The critical orifices dynamic system was validated using Molbloc system (Molbox, Fluke, NL). The traceable dynamic dilution was done using pure hydrogen (BIP®þ quality, Air Products, UK) and NPL gas standards (40, 400 and 1000 nmol/mol). The calibration curve was from 2 nmol/mol to 20 nmol/mol for total sulphur.

2.2.9. Gas chromatography with flame ionization detector (GC-FID) RISE measured Total hydrocarbons compounds as methane basis C by GC-FID. The analytical method analysed methane and other light hydrocarbons (with 2–5 carbon atoms). The GC/FID uses helium as a carrier gas and a Porous Layer Open Tubular (PLOT) columns which are well suited for the analysis of light hydrocarbons. The gas is sampled in bags as contamination with air has been shown to be less than 100 μmol/mol with a proper handling of the bag. This level of contamination does not affect the concentration of methane and other hydrocarbons. A standard gas containing 104.6 μmol/mol of methane (CH4), 99.8 μmol/mol ethane (C2H4), 76.4 μmol/mol propane (C3H8), 94.9 μmol/mol of isobutane and 101.6 μmol/mol of butane (C4H10), 101.4 μmol/mol of pentane, 93.8 μmol/mol of isopentane (C5H12) was used to calibrate the instrument. For calibration purposes, 50 and 20 times dilution of the standard were also used (corresponding to calibration from 1 to 2–100 μmol/mol for each compound).

2.2.6. Quartz crystal microbalance (QMA) NPL measured water amount fraction in hydrogen using quartz crystal microbalance, QMA401 and QMA (Michell, USA) and using cavity ring down spectroscopy (Tiger Optics, USA), for water amount fraction below 2 μmol/mol. Gases are sampled directly from the gas cylinder to the analyser, a valve was used to restrict the flow to 0.333 L/ min for the QMA and to 1 L/min for the CRDS. The instruments were calibrated against NPL PRMs. Static gas standards of water amount fraction in hydrogen were used as quality control check. The gas line was extensively purged with high purity nitrogen (BIP® quality, Air Prod­ ucts, BE) prior to analysis in order to remove any moisture from the tubing.

2.2.10. Thermal desorption - gas chromatography/mass spectrometry flame ionization detection (TD-GC-FID/MS) RISE measured hydrocarbons (�C6) and possibly other impurities by TD-GC-FID/MS. Sampling from each cylinder was also performed at controlled flow rates (50 ml/min) onto stainless steel sorbent tubes packed with Tenax TA (200 mg, 60–80 Mesh) during 1 min (50 ml totally) respective 2 min (100 ml totally). The Tenax TA tubes are sub­ jected to a two-stage thermal desorption process using a Thermal Desorption Unit (Markes TD100 desorber), where the adsorbed sub­ stances are released by heating the sorbent tubes at 275 � C for 7 min and then transferred to a cold trap packed with graphitised carbon for focusing at 10 � C. The trap is then rapidly heated up to 300 � C and components are released and reached the gas chromatography (GC) column for separation. The column effluent is split into two streams for the detection of individual components, one stream passing through a flame ionization detector and the other stream through a mass spec­ trometer. The analysis of the sample is performed on an Agilent 6890 N gas chromatograph equipped with two detectors, a flame ionization detector and a mass spectrometer 5975C inert mass selective detector operated in the electron impact mode under standard conditions (ionizing electron energy 70 eV, masses scanned from 29 to 300 amu). The column used is a BPX5, 50 m � 32 mm ID x 1 μm film thickness (5% phenyl (equivalent) polysilphenylene-siloxane). The column program temperature is monitored from 35 � C (hold 4 min) to 100 � C at 3 � C/min, to 220 � C at 8 � C/min and then to 300 � C at 15 � C/min (hold 10 min). Helium is used as a carrier gas at a flow rate of 2.6 ml/min. The FID detector temperature is set at 300 � C.

2.2.7. Gas chromatography with thermal conductivity detector (GC-TCD) RISE measured nitrogen, oxygen, argon and carbon dioxide by GC/ TCD. From the cylinders, gas sampling bags (3 L Multi-layer foil, Restek) were filled. The bags were first washed out using RISE bag washout system using helium in order to reduce the quantity of nitrogen and oxygen in the bag to as low as possible. The system automatically per­ formed the following sequence of events: emptying the bag during 20 s, filling the bags with Helium (with a piston of 200 ml which is filled 5 times with helium at 1.8 bar), emptying the bags in 20 s, the last two operations are repeated 3 times. The analysis by GC/TCD was done using helium as a carrier gas. Gases are sampled in special gas bags and a known volume of gas (typically 30 ml) is withdrawn from the bag by a pump to fill a loop of 100 μl with the gas to analyse which is subsequently introduced in the columns of the GC/TCD. The gas chromatography columns were Mo­ lecular Sieve 5A and Hayesep N. The GC oven was maintained at 80 � C. The detector temperature was above 110 � C with a filament at 320 � C. A standard gas in hydrogen matrix containing 100 μmol/mol of nitrogen, 100 μmol/mol of argon, 10 μmol/mol of carbon dioxide and 10 μmol/ mol oxygen was used to calibrate the GC. With this method, argon cannot be separated from oxygen. So, the reported results correspond to the sum of argon and oxygen amount fraction. 2.2.8. Optical feedback cavity enhanced absorption spectroscopy (OFCEAS) RISE measured carbon monoxide, carbon dioxide, hydrogen sulfide (oxygen and water) by optical feedback cavity enhanced absorption spectroscopy (Proceas®, AP2E, France). OFCEAS is a direct intensity measurement scanning spectroscopy technique. The use of a resonating cavity – a multipath gas cell using hyper-reflective mirrors – enables a path length up to 10 km. The instrument used measures low μmol/mol levels of carbon monoxide (0.002–20 μmol/mol), carbon dioxide (0.2–2000 μmol/mol), hydrogen sulphide (0.001–0.2 μmol/mol), water (0.05–500 μmol/mol) and oxygen (1–2000 μmol/mol). The instrument is pre-calibrated and does not require daily calibration with certified gas mixtures. The cylinder containing the sample is connected to a transfer line consisting of a stainless steel particle filter with 7 μm pore size and a stainless steel restrictor. The sample is withdrawn at 3–5 l/h using an internal pump. When using cylinders, a split is needed before the filter. When an analysis is started, the concentration of the targeted compound rises until it stabilizes. The stable part of the data is used for the final data treatment.

3. Results and discussion 3.1. Steam methane reforming with PSA sampling campaign results The results were reported for each national metrology institute involved in the campaign in Table 2. Most of the compounds were re­ ported below the detection limit of the analytical methods used. No significant difference was observed between the laboratories performing similar analysis. The results of analysis from SMR with PSA are presented as a range between the highest and the lowest values obtained from eight different and independent samplings at various SMR plants in Europe (Table 3). The results covered a range of technology and system with the objective to provide a general overview of hydrogen quality from SMR with PSA. The results showed that no contaminants were above the threshold of ISO 14687-2. Even if the number of sampling is low, the complete absence of contaminant can be used to confirm the results of the prob­ ability of contaminant presence.

5

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

Table 2 Results of analysis for eight hydrogen fuel samples sampled at European steam methane reforming plant after pressure swing adsorption purification. The results are presented for all the gaseous contaminants listed in EN 17124:2018 and ISO/FDIS 14687. The results are reported for each national metrology institutes (1): NPL; (2): RISE; (3): VSL and (4): CEM. CO (1) CO (2) CO2 (2) CO2 (1) CH4 (1) CH4 (3) CH4 (2) Non CH4 hydrocarbons (1) H2O (1) H2O (2) Total sulphur compounds (1) H2S (2) O2 (4) O2 þ Ar (2) O2 (1) N2 (4) N2 (2) N2 (1) Ar (2) Ar (4) Ar (1) Total halogenated (HCl) (3) CH2O (3) CH2O2 (3) NH3 (3) He (4) C2 hydrocarbons (2) C3-hydrocarbons (2) C4-hydrocarbons (2) C5-hydrocarbons (2) C6 – C18 hydrocarbons (2)

Sample 1

Sample 2

Sample 3

Sample 4

Sample 5

Sample 6

Sample 7

Sample 8

<0.053 <0.2 <0.1 0.042 � 0.016 0.044 � 0.007 n.a. <0.5 <0.05 <0.6 n.a <0.002 <0.04 n.a. n.a. 0.39 � 0.13 n.a. n.a. 1.5 � 0.6 n.a. n.a. 2.8 � 0.1 <0.005 <0.005 <0.1 <0.1 20 � 4 <0.5 <1 <1 <1 <0.050

<0.053 <0.2 <0.1 <0.02 <0.02 n.a. <0.5 <0.05 <0.6 n.a. <0.002 <0.04 <5 n.a. 0.39 � 0.13 n.a. <25 <1.0 n.a. <25 <0.5 <0.005 <0.005 <0.1 <0.1 12 � 5 <0.5 <1 <1 <1 <0.050

<0.01 n.a. <5 <0.01 <0.01 ~0.01 <0.5 <0.01 <0.5 n.a. <0.0036 n.a. n.a. n.a. <0.5 <100 n.a. <1.2 <30 n.a. <0.5 <0.005 <0.005 <0.1 n.a. n.a. <0.5 <1 <1 <1 <0.050

<0.01 n.a. <5 <0.01 <0.01 ~0.01 <0.5 <0.01 <0.5 n.a. <0.0036 n.a. <5 n.a. <0.5 <50 n.a. <1.2 <30 n.a. <0.5 <0.005 <0.005 <0.1 <0.1 n.a. <0.5 <1 <1 <1 <0.050

<0.01 n.a. <5 <0.01 <0.01 ~0.01 <0.5 <0.01 <0.5 n.a. <0.0036 n.a. <5 n.a. <0.5 <60 <80 <1.2 <30 <80 <0.5 <0.005 <0.005 <0.1 <0.1 <50 <0.5 <1 <1 <1 <0.050

<0.02 n.a. <0.5 <0.02 <0.02 n.a. <0.5 <0.02 <1.8 <2 <0.002 n.a. <25 <50 <0.5 <50 <25 5.2 � 0.6 n.a. <25 1.00 � 0.10 <0.005 <0.005 <0.1 <0.1 44 � 10 <0.5 <1 <1 <1 <0.050

<0.02 n.a. n.a. <0.02 <0.02 n.a. <0.5 <0.02 <1.5 n.a. <0.002 n.a. <25 <50 <1.0 <50 <25 10.4 � 1.1 n.a. <25 1.30 � 0.10 <0.005 <0.005 <0.1 <0.1 43 � 10 <0.5 <1 <1 <1 <0.050

<0.02 n.a. n.a. <0.02 <0.02 n.a. <0.5 <0.02 <1.2 n.a. <0.002 n.a. <25 <50 <0.5 <50 <25 5.5 � 0.6 n.a. <25 1.11 � 0.10 <0.005 <0.005 <0.1 <0.1 43 � 8 <0.5 <1 <1 <1 <0.050

3.2. PEM water electrolyser with TSA sampling campaign results

Table 3 Range amount fraction of ISO 14687-2 contaminants in hydrogen from steam methane reforming with pressure swing adsorption. The results correspond to the range from the lowest to the highest values obtained in eight different samples from different SMR with PSA in Europe.

The results were reported for each national metrology institute involved in the campaign in Table 4. Most of the compounds were re­ ported below the detection limit of the analytical methods used. The results of analysis from PEM Water electrolyser with TSA are presented as a range between the highest and the lowest values obtained from eight different and independent samplings at various PEM Water electrolyser plants in Europe. The results covered a range of technology and system with the objective of providing a general overview of hydrogen quality from PEM water electrolyser with TSA. The results showed that no contaminants were above the threshold of ISO 14687-2. Even if the number of sampling is low, the complete absence of con­ taminants can be used to confirm the results of the probability of contaminant presence from Bacquart et al. [16].

Compounds

ISO 14687-2 threshold [μmol/ mol]

SMR with PSA (8 samples) Results [μmol/mol]

Probability of occurrence [16]

Water H2O Methane CH4 Non methane hydrocarbons Oxygen O2 Helium He Nitrogen N2 Argon Ar Carbon dioxide CO2 Carbon monoxide CO Total sulphur compounds Formaldehyde HCHO Formic acid HCOOH Ammonia NH3 Total halogenated C2 hydrocarbons C3 hydrocarbons C4 hydrocarbons C5 hydrocarbons C6 – C18 hydrocarbons

5 2 2

<2 <0.02 to 0.05 <0.05

Unlikely (0) Rare (2) Unlikely (0)

5 300 100 100 2

<1.0 <54 <1.2 to 11 <0.5 to 1.3 <0.02 to 0.45

Unlikely (0) Unlikely (0) Possible (3) Rare (2) Unlikely (0)

0.2

<0.02

Frequent (4)

3.3. Additional findings during the sampling campaign

0.004

<0.0036

Unlikely (0)

0.01

<0.005

Very rare (1)

0.2

<0.1

Unlikely (0)

0.1 0.05 2 2 2 2 2

<0.1 <0.005 <0.5 <1 <1 <1 <0.05

Unlikely (0) Unlikely (0) Unlikely (0) Unlikely (0) Unlikely (0) Unlikely (0) Unlikely (0)

3.3.1. Analytical method limit of detection The national measurement institutes reported the limit of detection for each analytical methods used. Most of the methods proposed are compliant with the threshold required by ISO 14687-2 except GC-TCD for oxygen. In some cases, it is clear that the analytical technique re­ ported was not sufficient to determine if the method was compliant. For example, GC-PDHID method from CEM had a LOD of 5 μmol/mol and the GC-PDHID method from NPL had a LOD of 0.5 μmol/mol. The analytical instrument (i.e. GC-PDHID) is not sufficient to determine if an analytical method will be compliant with ISO 14687. It demonstrates the requirement to validate the analytical method as NPL GC-PDHID method is compliant ISO 14687 for oxygen while CEM GC-PDHID method is not compliant. Some methods mentioned (formaldehyde, ammonia by CRDS and OFCEAS for oxygen or water) have a LOD really close to the ISO 14687-2 6

Journal of Power Sources 444 (2019) 227170

T. Bacquart et al.

Table 4 Results of analysis for eight hydrogen fuel samples sampled at European PEM water electrolyser plant after temperature swing adsorption purification. The results are presented for all the gaseous contaminants listed in EN 17124:2018 and ISO/FDIS 14687. The results are reported for each national metrology institutes (1): NPL; (2): RISE; (3): VSL and (4): CEM. Amount fraction [μmol/mol] Compounds

Sample 1

Sample 2

Sample 3

Sample 4

Sample 5

Sample 6

Sample 7

Sample 8

CO (1) CO (2) CO2 (2) CO2 (1) CH4 (1) Non CH4 hydrocarbons (1) CH4 (2) H2O (1) H2O (2) Total sulphur compounds (1) H2S (2) O2 (4) O2 þ Ar (2) O2 (1) N2 (2) N2 (4) N2 (1) Ar (4) Ar (1) Total halogenated (HCl) (3) CH2O (3) CH2O2 (3) NH3 (3) He (4) C2 hydrocarbons (2) C3-hydrocarbons (2) C4-hydrocarbons (2) C5-hydrocarbons (2) C6 – C18 hydrocarbons (2)

<0.053 <0.2 <0.1 0.443 � 0.010 0.031 � 0.006 <0.05 <0.5 <0.6 n.a. <0.002 <0.04 <5 n.a. 0.45 � 0.13 n.a. <25 2.0 � 0.5 <25 <0.5 <0.005 <0.005 <0.1 <0.1 34 � 5 <0.5 <1 <1 <1 <0.050

<0.02 n.a. <5 0.245 � 0.010 <0.01 <0.02 <0.5 <0.8 n.a. <0.002 <5 <5 <0.5 <40 <50 4.6 � 0.3 <50 <0.5 <0.005 <0.005 <0.1 <0.1 <5 <0.5 <1 <1 <1 <0.05

<0.02 <0.02 <5 0.229 � 0.08 <0.01 <0.02 <0.5 <1.4 n.a. <0.002 <5 <11 <0.5 <70 <50 4.2 � 0.4 <50 <0.5 <0.005 <0.005 <0.1 <0.1 <5 <0.5 <1 <1 <1 <0.05

<0.02 <0.02 <0.4 <0.02 <0.02 <0.02 <0.5 <3 <3 <0.002 <0.004 <5 <5 <0.6 <50 <1.5 <50 <0.5 <0.005 <0.005 <0.1 <0.1 15–45 <0.5 <1 <1 <1 <0.05

<0.02 <0.02 <0.4 <0.02 <0.02 <0.02 <0.5 <3 <5 <0.002 <0.004 <5 <3 <0.6 <50 <1.5 <50 <0.5 <0.005 <0.005 <0.1 <0.1 <5 <0.5 <1 <1 <1 <0.05

<0.02 n.a. <5 <0.01 <0.01 0.156 � 0.030 n.a. <0.8 n.a. <0.0030 n.a. <5 <25 1.39 � 0.36 <100 <80 1.51 � 0.2 <80 <0.5 n.a. <0.005 <0.1 n.a. <9 n.a. n.a. n.a. n.a. n.a.

<0.02 n.a. n.a. <0.01 <0.01 0.126 � 0.026 n.a. <1.2 n.a. <0.0030 n.a. n.m. n.a. <0.5 n.a. n.a. <1.0 n.a. <0.5 <0.005 <0.005 <0.1 n.a. <9 n.a. n.a. n.a. n.a. n.a.

<0.02 n.a. <5 <0.01 <0.01 0.111 � 0.024 n.a. <3 n.a. <0.0030 n.a. <5 <25 1.59 � 0.45 <100 n.a. 1.86 � 0.2 n.a. <0.5 <0.005 <0.005 <0.1 n.a. <9 n.a. n.a. n.a. n.a. n.a.

threshold (less than 50% relative). It raised the requirement for the development of analytical methods with improved limit of detection or limitation of some techniques.

Table 5 Range amount fraction of ISO 14687-2 contaminants in hydrogen from PEM water electrolysis with and without temperature swing adsorption. The results correspond to the range from the lowest to the highest values obtained in eight different samples from different PEM water electrolyser in Europe. Compounds

ISO 14687-2 threshold [μmol/mol]

PEM water electrolysis with TSA Results on 8 samples [μmol/mol]

Probability of occurrence [16]

Water H2O Methane CH4 Non CH4 hydrocarbons Oxygen O2 Helium He Nitrogen N2 Argon Ar Carbon dioxide CO2 Carbon monoxide CO Total sulphur compounds Formaldehyde HCHO Formic acid HCOOH Ammonia NH3 Total halogenated C2 hydrocarbons C3 hydrocarbons C4 hydrocarbons C5 hydrocarbons C6 – C18 hydrocarbons

5 2 2

<3 <0.02 0.08 to 0.2

Rare (2) Unlikely (0) Unlikely (0)

5 300 100 100 2

<0.5–2 <9 to 45 <1.0 to 4.6 <0.5 <0.02 to 0.25

Rare (2) Unlikely (0) Rare (2) Unlikely (0) Very rare (1)

0.2

<0.02

Unlikely (0)

0.004

<0.0036

Unlikely (0)

0.01

<0.005

Unlikely (0)

0.2

<0.1

Unlikely (0)

0.1 0.05

<0.1 <0.005

Unlikely (0) Unlikely (0)

2 2 2 2 2

<0.5 <1 <1 <1 <0.05

Unlikely (0) Unlikely (0) Unlikely (0) Unlikely (0) Unlikely (0)

3.3.2. Sampling issues The sampling of the different hydrogen production processes is not standardised. The sampling campaign presented in this study followed a specific procedure. However in few cases, the procedure was amended due to specific requirements (operator’s specific sampling procedure, low pressure). Sampling of chlor-alkali feedstock is a challenge as the pressure of the system is really low (P < 0.2 MPa). Therefore, the sam­ pling procedure applied in this study was not suitable. Chlor-alkali process was sampled at a hydrogen refuelling station. Hydrogen refu­ elling station may be the potential origin of the contaminants. The production process sampling aims to understand the contaminants present or absent at the outlet of the production process. It should avoid having additional contamination sources from hydrogen supply chain (i. e. pipeline, transport) or from the HRS infrastructure (i.e. compressor, buffer tank). The results showed some issues that could be linked with sampling procedure (insufficient purging and additional contamination from supply chain). The results presented in Table 6 were considered contaminated and therefore not associated to the results of analysis per feedstock. The contaminated sample 1 was obtained from HRS with chlor-alkali feedstock. The results showed presence of high amount fraction of nonmethane hydrocarbons, nitrogen and water. It is unexpected that chloralkali feedstock produced high amount of hydrocarbons however the HRS itself or the supply chain could led to high amount of hydrocarbons (i.e. oil, grease). High level of nitrogen may be linked to HRS purging or maintenance. Therefore the sample was considered not representative of the chlor-alkali membrane electrolysis process. It should be noted that no halogenated compounds were found. All compounds except nitrogen, non-methane hydrocarbons and water were found below the ISO 14687 threshold. 7

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

Table 6 Results of analysis from three contaminated samples. It presents example of the typical contamination observed within the study. The results are presented for all the gaseous contaminants listed in EN 17124:2018 and ISO/FDIS 14687. Compounds

Water H2O Methane CH4 Non CH4 hydrocarbons Oxygen O2 Helium He Nitrogen N2 Argon Ar Carbon dioxide CO2 Carbon monoxide CO Total sulphur compounds Formaldehyde HCHO Formic acid HCOOH Ammonia NH3 Total halogenated

ISO 14687-2 threshold [μmol/mol]

5 2 2 5 300 100 100 2 0.2 0.004 0.01 0.2 0.1 0.05

Contaminated sample 1

Contaminated sample 2

Contaminated sample 3

Chlor-alkali membrane electrolysis feedstock

SMR feedstock

SMR feedstock

Results [μmol/mol]

Results [μmol/mol]

Results [μmol/mol]

13.2 � 1.7 14.28 � 0.07 >200 <0.5 <20 579 � 23 <1.0 0.316 � 0.007 <0.02 <0.0036 <0.005 <0.1 <0.1 <0.005

2.48 � 0.25 <0.02 <0.05 35 � 2 12 � 5 134 � 2 1.43 � 0.10 0.101 � 0.004 <0.053 <0.002 <0.005 <0.1 <0.1 <0.005

17.1 � 3.5 0.038 � 0.004 <0.040 1.35 � 0.07 n.a. 14.6 � 0.8 4.2 � 0.3 <0.04 <0.02 <0.002 <0.005 <0.1 n.a. <0.005

Table 7 Results of analysis from PEM water electrolyser with and without purification. It presents the dif­ ference of contaminants amount fraction due to the purification process (highlighted in red). The results are presented for all the gaseous contaminants listed in EN 17124:2018 and ISO/FDIS 14687.1

8

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

The contaminated sample 2 was obtained from a steam methane reformer. The results for Ar/O2/N2 amount fraction showed a ratio Ar/ O2/N2 equal to 0.83/20.4/78.1 which is extremely close to air ratio. It is highly possible that small air ingress happened during the sampling procedure and biased the results of the hydrogen quality sample. Therefore, the result of this sample was considered not representative of the SMR production process. In this specific case, the sampling team was unable to notice the contamination during the sampling. It is important to notice that the results reported were not compliant with ISO 14687 for oxygen and nitrogen. The other compounds were below the ISO 14687 threshold. It raised the importance of proper sampling equipment and procedure to avoid false positive. The contaminated sample 3 was obtained from a steam methane reformer. The result for water amount fraction was overlapping the threshold of 5 μmol/mol. The sampling was not performed by the project team and it raised a concern towards proper sampling procedure and sufficient sampling equipment purging (especially to remove all traces of water vapour). Therefore, the result of this sample was considered not representative of the SMR production process. In this specific case, it raised the importance of proper sampling equipment, trained staff and procedure to avoid false positive.

No CO amount fraction was measured at level 10 times lower than the ISO 14687-2 threshold. Nitrogen probability was proposed as possible yet nitrogen amount fractions were measured 10 times lower than the ISO 14687-2 threshold likewise argon and methane. The probability of contaminants presence was considered conservative. The low number of samplings in this study did not allow observing all possible situations. It is recommended to keep this probability of presence until larger sam­ pling campaign supports a revision. For formaldehyde, revision of the ISO 14687-2 threshold may support a revision of the probability of presence supported by the measurement in this study (all samples had formaldehyde amount fraction below 0.005 μmol/mol). 3.4.3. Chlor-alkali membrane electrolysis with TSA The result of the contaminated sample 1 was obtained from Chloralkali membrane electrolysis. Even if the sample was contaminated, the analysis of HCl showed that the amount fraction was below 5 nmol/ mol. This result provided the first evidence of the absence of HCl in hydrogen produced from chlor-alkali membrane electrolysis with TSA. This result was obtained from only one sample and will need additional sampling and analysis to provide more confidence for this process. 4. Summary

3.3.3. Impact of purification step on PEM water electrolyser hydrogen quality During the study, the impact of purification process of PEM water electrolyser was investigated by sampling hydrogen before and after the temperature swing adsorption processes. Five samples were obtained from various European electrolysers. The results presented in Table 7 summarised the difference in the contaminants found with or without temperature swing adsorption. Oxygen, water and carbon dioxide were the three compounds found above ISO 14687 threshold in hydrogen produced from PEM water electrolysis without TSA. As expected, the TSA was reducing the oxygen, water and carbon dioxide amount fraction below ISO 14687 threshold. It is important to notice that all the other compounds were below ISO 14687 thresholds before the purification.

This study presents the first results of hydrogen quality from two production processes in the supply chain of hydrogen fuel for FCEV: PEM water electrolysis with TSA and SMR with PSA. The results on more than 16 different plants or occasions demonstrated that in all cases the 13 compounds listed in ISO 14687 were below the threshold of the in­ ternational standards. Secondly, the results validated the probability of contaminants presence of Bacquart et al. [16] for PEM water electrol­ ysis. A clear link can be established between contaminants probability and TSA purification presence. For SMR with PSA, the probability of contaminants presence of Bacquart et al. [16] was conservative as several contaminants are probable while this study did not measure any of them at significant level. Without large amount of data on hydrogen fuel quality, the present study provided confirmation that the proba­ bility of presence [16] is conservative and coherent (no contaminants in improbable section were detected). The study raised the attention on the sampling protocol which was the source of two false positives. Water and air were the main contam­ inants observed above the ISO 14687 threshold and clearly unrelated to the production methods. The study highlighted the fact that it is critical to divide the supply chain into separate parts to understand the contaminant source path and develop a quality control plan with mea­ surement or control at the right position of the supply chain. The MPIR 15NRM03 Hydrogen project provided the first analytical results for two hydrogen production methods. The number of samples may be considered low, but it provides the first open access data on hydrogen quality at production processes. It is recommended to extend the study to other production methods (i.e. alkaline electrolysis) or in­ crease the number of production plant tested. The other parts of the HRS supply chain (e.g., compressor and pipeline) should follow similar evaluation involving the probability of presence evaluation by experts and an analysis campaign to support the findings.

3.4. Recommendations to the hydrogen industry 3.4.1. PEM water electrolyser The hydrogen quality from PEM water electrolyser with temperature swing adsorption was in agreement with ISO 14687 in all European sites sampled in this study (n ¼ 8). The importance of purification system is an important point regarding potential contaminants. Without the TSA purification process, only three compounds were above the threshold: oxygen, water and carbon dioxide. These results agree with the probability of contaminant presence as presented in Table 5. It highlights the list of contaminants that requires close monitoring for this production process and will support the development of dedicated quality control monitoring plan. One differ­ ence between PEM water electrolyser and SMR processes is about the production location. One advantage of PEM water electrolyser is on-site production; however the local production processes may not have staff on site that can perform maintenance in a short time frame. Therefore the impact of purification performance will support the choice of the contaminants to monitor in the hydrogen produced on site. The study would support regular or online monitoring of oxygen, water and carbon dioxide for PEM water electrolyser. The other contaminants listed in ISO 14687 are unlikely to be present at the process outlet.

Acknowledgements The study was funded by The Joint Research Project « Metrology for sustainable hydrogen energy applications » HYDROGEN which is sup­ ported by the European Metrology Programme for Innovation and Research (EMPIR). The EMPIR initiative is co-funded by the European Union’s Horizon 2020 research and innovation programme and the EMPIR Participating States.

3.4.2. Steam methane reforming The hydrogen quality from SMR with pressure swing adsorption was in agreement with ISO 14687 in all European sites sampled in this study (n ¼ 8). These results agree with the probability of contaminant pres­ ence proposed by Bacquart et al. [16] even if the proposal highlighted probability to have frequently CO above the threshold of ISO 14687-2. 9

T. Bacquart et al.

Journal of Power Sources 444 (2019) 227170

References

[16] DIRECTIVE 2014/94/EU OF, The European Parliament and of the Council of 22 October 2014 on the Deployment of Alternative Fuels Infrastructure, Official Journal of the European Union, 2014. http://eur-lex.europa.eu/legalcontent/EN/ TXT/PDF/?uri¼CELEX:32014L0094&from¼EN. accessed on the 10/02/2019. [17] EN 17124:2018, Hydrogen Fuel. Product Specification and Quality Assurance. Proton Exchange Membrane (PEM) Fuel Cell Applications for Road Vehicles, European Committee on Standardisation, Bruxelles, BE, 2018. [18] ISO/DIS 19880-8(E), Gaseous Hydrogen - Fueling Stations - Part 8: Fuel Quality Control. International Organization for Standardization, Geneva, Switzerland. [19] T.A. Aarhaug, Deliverable 3.1 Deliverable 3.2 Measurement of hydrogen quality variation at various HRS with different fuel feedstock Grant agreement no: 621223. http://hycora.eu/deliverables/D%203.2%20Measurement%20of%20hydrogen% 20quality%20variation%20at%20various%20HRS%20with%20different%20fuel% 20feedstock.pdf accessed on the 04/05/2019]. [20] T.A. Aarhaug, Summary of HRS sampling and analysis in HyCoRA. HyCoRA OEM workshop trondheim (Norway) 2017-06-07. http://hycora.eu/workshops/060620 17/HyCoRA_WS_201706_Aarhaug_SA_v2.pdf, 2017 accessed on the 04/05/2019. [21] F. Haloua, T. Bacquart, K. Arrhenius, B. Delobelle, H. Ent, Metrology for hydrogen energy applications: a project to address normative requirements, Meas. Sci. Technol. 29 (2018), 034001. https://doi.org/10.1088/1361-6501/aa99ac. [22] T. Bacquart, A. Murugan, M. Carr�e, B. Gozlan, F. Aupr^etre, F. Haloua, T. A. Aarhaug, Probability of occurrence of ISO 14687-2 contaminants in hydrogen: principles and examples from steam methane reforming and electrolysis (water and chlor-alkali) production processes model, Int. J. Hydrogen Energy 43 (26) (2018) 11872–11883. [23] ISO/DIS 19880-8:2017(E), Gaseous Hydrogen - Fueling Stations - Part 8: Fuel Quality Control. International Organization for Standardization, Geneva, Switzerland. [24] T. Bacquart, A. Murugan, S. Bartlett, N. Moore, A. Morris, B. Gozlan, M. Carr� e, G. De Reals, G. Doucet, F. Laridant, E. Gernot, F. Aupr^ etre, I. Profatilova, P.A. Jacques, A. Rojo, K. Arrhenius, A. Fischer, S. Persijn, Report on Risk Assessment of Impurities in Hydrogen for Fuel Cells and Recommendations on Maximum Concentration of Individual Compounds Based on the New Fuel Cell Degradation Studies and on the Probability of Presence, 2019. EMPIR Grant Agreement Number: 15NRM03 Hydrogen, Deliverable D1. [25] S. Persijn, Purity analysis of gases used in the preparation of reference gas standards using a versatile OPO-based CRDS spectrometer, J. Spectrosc. 7 (2018). https://doi.org/10.1155/2018/9845608. [26] S. Persijn, F. Harren, A. van der Veen, Quantitative gas measurements using a versatile OPO-based cavity ring down spectrometer and the comparison with spectroscopic databases, Appl. Phys. B 100 (2) (2010) 383–390. [27] S.W. Sharpe, R.L. Sams, T.J. Johnson, The PNNL Quantitative IR Database for Infrared Remote Sensing and Hyperspectral Imaging, Applied Imagery Pattern Recognition Workshop, 2002. Proceedings, Washington, DC, USA, 2002, pp. 45–48, https://doi.org/10.1109/AIPR.2002.1182253, 2002. [28] I.M. Smith, F.O. Onakunle, SSfM-3 1.6.1—XLGENLINE, Software for Generalised Least-Squares Fitting, Developed by the (NPL), Teddington, UK, NPL Document Reference, National Physical Laboratory, Teddington, 2007. CMSC/M/06/657. [29] ISO6142, Gas Analysis – Preparation of Calibration Gas Mixtures – Part 1. Gravimetric Method for Class I Mixtures, International Organization for Standardization, Geneva, Switzerland, 2015. [30] ISO6143, Gas Analysis. Comparison Methods for Determining and Checking the Composition of Calibration Gas Mixtures, International Organization for Standardization, Geneva, Switzerland, 2001.

[1] G. Marb� an, T. Vald�es-Solís, Towards the hydrogen economy? Int. J. Hydrogen Energy 32 (12) (2007) 1625–1637. https://doi.org/10.1016/j.ijhydene.2006.1 2.017. [2] European commission, Directorate-General for Research and Innovation, Final Report of the High-Level Panel of the European Decarbonisation Pathways Initiative European Commission, Publications Office of the European Union, Luxembourg, 2018, https://doi.org/10.2777/636. https://ec.europa.eu/info/site s/info/files/ec-18-002-decarbonisation_booklet_27112018.pdf. accessed 02/01/ 2019. [3] Ministere de la transition ecologique et solidaire, Plan de deploiement de l’hydrogene pour la transition energetique. https://www.ecologique-solidaire.gou v.fr/sites/default/files/Plan_deploiement_hydrogene.pdf, 2018 accessed 03/01/ 2019. [4] NOW GmbH National Organisation, Electric Mobility with Hydrogen and Fuel Cells State of development and market introduction in the area of passenger vehicles in Germany. https://www.now-gmbh.de/content/service/3-publikationen/1-ni p-wasserstoff-und-brennstoffzellentechnologie/electric-mobility-with-hydrogen -2017_en_310817.pdf, 2017 accessed 03/01/2019. [5] H2 Platform, Op weg naar meer waterstofstations. https://opwegmetwaterstof. nl/2018/12/13/op-weg-naar-meer-waterstoftankstations-nederland-video/, 2018 accessed 04/02/2019. [6] ICCT, Developing hydrogen fueling infrastructure for fuel cell vehicles: a status update. https://www.theicct.org/sites/default/files/publications/Hydrogen-inf rastructure-status-update_ICCT-briefing_04102017_vF.pdf, 2017 accessed on the 10/02/2019. [7] Toyota, The Toyota Mirai. https://www.toyota.co.uk/new-cars/new-mirai/ accessed on the 02/01/2019. [8] I. Staffell, D. Scamman, A. Velazquez Abad, P. Balcombe, P.E. Dodds, P. Etkins, N. Shahdand, K.R. Ward, The role of hydrogen and fuel cells in the global energy system, Energy Environ. Sci 12 (2018) 463–491, https://doi.org/10.1039/ c8ee01157e, 2019. [9] ch H2energy, Hyundai motor and H2 energy sign Joint venture contract to spearhead hydrogen mobility in Europe. https://h2energy.ch/en/hyundai-motorand-h2-energy-sign-joint-venture-contract-to-spearhead-hydrogen-mobility-in-e urope/ accessed on the 05/05/2019]. [10] Plugpower, GenDrive. https://www.plugpower.com/wp-content/uploads/2018/0 6/2018GenDriveBrochure_F1Digi.pdf, 2018 accessed on the 05/05/2019. [11] K. Narusawa, M. Hayashida, Y. Kamiya, H. Roppongi, D. Kurashima, K. Wakabayashi, Deterioration in fuel cell performance resulting from hydrogen fuel containing impurities: poisoning effects by CO, CH4, HCHO and HCOOH, JSAE Rev. 24 (1) (2003) 41–46. [12] P. Caloprisco, Research Activities for Transport Applications: MEAs, Components, Stacks and Subsystems, HRS, 2018, 11 15. Retrieved 12/31/2018, from, https ://www.fch.europa.eu:https://www.fch.europa.eu/sites/default/files/documents/ ga2011/1_Session%204_Caloprisco%20%28ID%204811766%29.pdf. [13] ISO 14687-2:2012, Hydrogen Fuel—Product Specification— Part 2: Proton Exchange Membrane (PEM) Fuel Cell Applications for Road Vehicles, International Organization for Standardization, Geneva, Switzerland, 2012. [14] SAE J2719:2015, Hydrogen Fuel Quality for Fuel Cell Vehicles, SAE International, Warrendale, USA, 2015. https://doi.org/10.4271/J2719_201511. [15] M. Carr�e, How to ensure H2 quality without increasing H2 analysis cost? International workshop Metrology for sustainable hydrogen energy applications, Versailles (France) 2018-11-08. http://projects.lne.eu/jrp-hydrogen/download /2083, 2018 accessed on the 05/05/2019.

10