i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/he

Review Article

Strategies for the sampling of hydrogen at

refuelling stations for purity assessment

Karine Arrhenius a,*, Thor Aarhaug b, Thomas Bacquart c, Abigail Morris c,

Sam Bartlett c, Lisa Wagner d, Claire Blondeel e, Bruno Gozlan e,
Yann Lescornez e, Nathalie Chramosta e, Christian Spitta f,
Etienne Basset g,h, Quentin Nouvelot g,h, Mathilde Rizand g,h
a
Research Institutes of Sweden AB (RISE), Frans Perssons Va € g, 412 76, G€oteborg, Sweden
b
SINTEF Industry, Richard Birkelandsvei 2b, 7034 Trondheim, Norway
c
National Physical Laboratory (NPL), Hampton Road, Teddington, Middlesex, TW11 0LW, United Kingdom
d
Linde GmbH, Gases Division, 85716, Unterschleissheim, Germany
e
Air Liquide, Innovation Campus Paris, Jouy en Josas, France
f
ZBT, Zentrum für BrennstoffzellenTechnik, Carl-Benz-Str. 201, 47057, Duisburg, Germany
g
ENGIE, Research Center ENGIE LAB CRIGEN, 4 Rue Josephine Baker, 93240, Stains, France
h
ENGIE, Research Center ENGIE LAB CRIGEN, 15 Quai Louis Aulagne, 69190, St-Fons, France

highlights

 Detailed description of different strategies for sampling hydrogen at the nozzle.

 Discussion on the representativeness of the sample collected.
 Results from different stability studies for gaseous species in different cylinders.
 Highlight the need of comparative studies to assess equivalence between strategies.

article info abstract

Article history: Hydrogen delivered at hydrogen refuelling station must be compliant with requirements
Received 26 April 2021 stated in different standards which require specialized sampling device and personnel to
Received in revised form operate it. Currently, different strategies are implemented in different parts of the world
4 August 2021 and these strategies have already been used to perform 100s of hydrogen fuel sampling in
Accepted 7 August 2021 USA, EU and Japan. However, these strategies have never been compared on a large sys-
Available online 4 September 2021 tematic study. The purpose of this paper is to describe and compare the different strategies
for sampling hydrogen at the nozzle and summarize the key aspects of all the existing
Keywords: hydrogen fuel sampling including discussion on material compatibility with the impurities
Hydrogen that must be assessed. This review highlights the fact it is currently difficult to evaluate the
Refuelling stations impact or the difference these strategies would have on the hydrogen fuel quality
assessment. Therefore, comparative sampling studies are required to evaluate the equiv-

* Corresponding author.
E-mail address: [email protected] (K. Arrhenius).
https://doi.org/10.1016/j.ijhydene.2021.08.043
0360-3199/© 2021 The Authors. Published by Elsevier Ltd on behalf of Hydrogen Energy Publications LLC. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
34840 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3

Sampling device alence between the different sampling strategies. This is the first step to support the
Fuel quality assessment standardization of hydrogen fuel sampling and to identify future research and develop-
ment area for hydrogen fuel sampling.
© 2021 The Authors. Published by Elsevier Ltd on behalf of Hydrogen Energy Publications
LLC. This is an open access article under the CC BY license (http://creativecommons.org/
licenses/by/4.0/).

Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34841
Definition of the different sampling strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34841
Strategies using the “gas serial” method” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
ASTM D7606-17 method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Description of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Sampling cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Information about the procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Air Liquide sampling device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Description of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Sampling cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Information about the procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34842
Japanese approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34843
Description of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34843
Sampling cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34843
Information about the procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34843
ENGIE sampling device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34843
Description of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34843
Sampling cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34844
Information about the procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34844
Strategies with the “gas parallel” method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34844
Methods using the Linde qualitizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34844
Description of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34844
Sampling cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34844
Information about the procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34845
Hy-SaM sampling device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34845
Description of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34845
Sampling cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34845
Information about the procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34845
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34846
Representativeness of the sample collected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34846
False positive or contamination of the sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34846
False negative of contaminant disappearing from the gas phase of the sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34846
Summary of the different sampling system for hydrogen fuel quality art HRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34846
Sampling device - material behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34846
Sampling cylinder e material behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34847
Validation of sampling Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34850
Sampling strategy and representativity of hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34850
Advantages/disadvantages offline methods against online methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34850
Safety of the sampling strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34851
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34851
Declaration of competing interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34852
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34852
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34852
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3 34841

- The procedure to prepare the sampling cylinders before

Introduction sampling (cleaning strategy).
- The procedure to purge the sampling device (for instance to
Hydrogen-powered vehicles are one of the most efficient op- remove air and water).
tions for decarbonizing long-distance and heavy-duty vehi- - The procedure to vent the device after sampling.
cles [1]. Fuel cell electric vehicles (FCEVs) are fully carbon
dioxide emission free, unlike other options such as biofuels, Currently, different strategies are implemented in different
natural gas fuels and hybrids. A complete comparison of parts of the world and so far (i.e. ASTM D7607-17 in North
emissions even including emissions from the manufacturing America or the direct method in Japan), these strategies have
process showed that FCEVs are also very competitive [2]. been used to perform already 100s of hydrogen fuel sampling
The development of FCEVs in our daily life requires the in USA [15], EU [16,17]and Japan. However, these strategies
deployment of a refuelling infrastructure with hydrogen refu- have never been compared on a large systematic study. Two
elling stations (HRSs) [3,4]. In 2020, the number of HRSs is above bilateral comparisons have been recently performed in Europe
140 in Europe [5] and 470 worldwide [6]. Regulatory re- (Hycora project [16,18]. It is of high importance to demonstrate
quirements are implemented together with the infrastructure that regardless of the strategy chosen, the outcomes of the
development. Hydrogen delivered at the HRSs must be hydrogen purity assessment are the same and that any bias
compliant with requirements which are stated in different due to the sampling strategy is avoided (i.e. the two scenarios
standards as for example the international standard where false results, negative or positive could occur must be
ISO14687:2019 [7], the European standard EN17124:18 [8] or SAE avoided). A false positive (over-estimation) would be the case
standard J2719 [9]. These requirements have been established where the hydrogen is sufficiently pure but the sampling
as a consensus based on extensive [10e12] yet ongoing research procedure itself contaminated the sample. A false negative
on the impact of contaminants on the fuel cell's ability to (under-estimation) would be the case where impurities in the
function properly. The most common method to monitor the hydrogen are lost either during the sampling or transport of
quality of hydrogen at a HRS according to these standards is the the sampling vessel.
so-called spot sampling which involves collecting a sample of The purpose of this paper is to describe and compare the
gas at the nozzle in a sampling cylinder that is subsequently different strategies for sampling hydrogen at the nozzle
sent to a laboratory for analysis. The method requires special- including summarizing all information available for the six
ized sampling equipment (often referred to as a sampling de- parts described above for each strategy. Some information
vice), and personnel to operate it [13]. Due to the complexity of about some strategies can be found in at least two standards
the measurement and the lack of online analysers to accurately and in a recent study from Bacquart et al. [19] which described
determine the hydrogen fuel quality according to ISO sampling procedures and purging for hydrogen samples taken
14687:2019, spot sampling is currently the only option to assess both at low pressure (10e80 bar) and at the nozzle. The stan-
the hydrogen fuel compliance. The advantage of spot sampling dard ASTM D7606-17 [20], describes in detail a sampling pro-
is that laboratory analysis can be performed on the sample cedure for high-pressure hydrogen at stations operating at 350
using a variety of analytical instruments. The disadvantages or 700 bar with a method called “gas serial method”. The
are that results are not obtained directly at the HRS and are standard ISO19880e1:2020, annex K [21] describes summarily
representative of a single point in time. Sampling cylinders three methods: “parallel method”, “gas serial method” and
have insufficiently been tested at ISO 14687 threshold levels to “direct method”. However, the strategies “gas serial method”
guarantee that analysis results are representative for the and “direct method” present technical similarities and in this
hydrogen delivered at the HRS for all components [14]. paper, they have been grouped under the methods “gas serial
Beyond the accuracy of the analytical method, taking a method”. Other strategies have recently been developed by
representative sample is of high importance for the hydrogen different organisations such as ENGIE, Air Liquide, Airborne
industry as important decisions (e.g. public access for refuel- Laboratories International and ZBT. Most of the strategies
ling) are based on the outcomes of the hydrogen quality presented in this paper aim at filling a cylinder for offline
assessment which includes the sampling procedure. This re- analysis. However, recently developed strategies (ENGIE and
quires using appropriated sampling strategies including Air Liquide) also include concepts to perform a partial onsite
appropriate materials for the sampling devices and cylinders. quality assessment (for example online analysis of water and
In this paper, strategies for sampling hydrogen at the oxygen).
nozzle of a HRS are defined as the list of components, re- The strategies using the “gas serial” method imply that
quirements and procedures needed to safely install, fill and hydrogen is filled in gas cylinder from the nozzle while the
disconnect a cylinder and includes: strategies using the “gas parallel” method include a tee-
connection as component to parallelly fill the sampling cyl-
- The design (including the components) of the sampling inder and a vehicle tank.
device which is dependent upon the sampling method (for
example, parallel or series).
- The sampling cylinder (one or two ended cylinder, size, Definition of the different sampling strategies
material, treatment, valves).
- The different requirements in term of filling pressure, This paper will consider two different sampling strategies of
safety, connection fitting … hydrogen fuel for offline analysis called here “gas parallel” and
34842 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3

“gas serial”. Both strategies concern sampling at the HRS is filled to a pressure of 69 bar and two to three sample cyl-
nozzle. inders shall be taken for a hydrogen sample at a HRS since the
“Gas serial”: in these strategies, the sample is taken directly analyses of two sampling cylinders for each sample may be
into a sampling cylinder. These sampling systems imply to necessary to prove the existence and estimate the amount of a
manage the hydrogen fuel conditions and may require oper- contaminant in a hydrogen fuel system.
ating the HRS in service mode. The sampling system may also
include a tank allowing to not override the protocol of the Information about the procedure
station (e.g. case of ENGIE method). A cleaning procedure aiming at removing traces of moisture in
“Gas parallel”: in these strategies, a tee-connection is used the HQSA, sampling line and sampling cylinder is performed
to parallelly fill the sampling cylinder and a FCEV or a recep- before sampling. Once the nozzle pressure has been regulated
tacle (larger than the sampling cylinder). These strategies do to 69 bar, the HQSA is cleaned by purging 1 kg of hydrogen fuel
not require to bypass the safety protocol of the station. through the HQSA (the hydrogen flow rate is approximately
33.3 g/s). After sampling, a vent procedure is performed by
opening the pressure release valve (14 on Fig. 1) before
Strategies using the “gas serial” method” removal of the sampling cylinder. The hydrogen is then
released through a check valve (16 on Fig. 1) and the ventila-
At least four different strategies are based on the principle of tion assembly.
the “gas serial method”. One of these strategies is described in
the standard ASTM D7606-7 [10] and two other strategies Air Liquide sampling device
based on this principle have been developed by Air Liquide
and ENGIE. The fourth strategy is the method currently used To sample at the HRS nozzle, Air Liquide has developed a “gas
in Japan which is summarised in ISO19880-1, Annex K [21]. serial” sampling system. The modular sampling device (Fig. 2)
has two functions: it allows to measure the humidity onsite
ASTM D7606-17 method and to sample hydrogen for offline quality control of all other
parameters.
This method is currently used in the USA by companies con-
ducting hydrogen fuel quality audits and is described in the Description of the system
standard ASTM D7606-17 [20]. The sampling is performed at The sampling is performed by connecting the HRS nozzle to
the nozzle and venting of hydrogen to atmosphere is per- the sampling device receptacle and by venting of hydrogen to
formed through an exhaust stack. The sampling device is atmosphere through a mobile vent. The device includes
referred to as the Hydrogen Quality Sampling Apparatus receptacle, quick connect fittings, pressure regulators, pres-
(HQSA). The method is adaptable for stations delivering sure relief valves, pressure gauge and mobile vents. The de-
hydrogen fuel both at 350 and 700 bar. Airborne laboratories vice is equipped with a parallel line to host a portable analyser
International have developed a commercial sampling device (usually a moisture analyser) which can be used during the
with Sulfinert® passivation (called NSP-7606) [22] compliant purge or the sampling phase (static or dynamic mode).
with ASTMD7606-17. It comes with 10-m flexible line con-
nected to the vent system of the HRS or a tripod for atmo- Sampling cylinder
spheric ventilation. It is also possible to install detector tubes The sampling cylinder is an aluminium double-ended valve 5-
(such as Draeger tubes or similar test tubes) for onsite L cylinder (no specific treatment) with double-ended stain-
screening of some impurities listed by SAE J2719. less-steel valves and is filled to 150 bar. Each cylinder un-
dergoes a cleaning procedure at the laboratory. The cylinder is
Description of the system first emptied and flushed with nitrogen. The cylinder is then
The sampling is performed by connecting the HRS nozzle to flushed with hydrogen (with a minimum of five pressurization
the sampling device receptacle and venting of hydrogen to and venting cycles) and then filled with residual hydrogen to
atmosphere through an exhaust stack. The components of the around 5 bar. The system is suitable for 5-L cylinders with type
sampling device with receptacle (J2799 compliant), exhaust E or DIN1 fittings. The use of other cylinders (volume or fit-
stack, quick connect fittings, pressure relief valves, valves and tings) is possible if they possess intermediate fittings. Treat-
regulators are shown on Fig. 1 which is taken from the stan- ments of the cylinders are not prohibited.
dard. The system has a high-pressure section (max 1000 bar)
and a lower pressure section after the 1000 bar regulator Information about the procedure
allowing a maximum pressure of 140 bar (the pressure relief Certain specifications must be followed to minimize safety
valve is set at 110 bar). The cylinder is located in the lower risks (pressure or electrical): 1. The device is directly con-
pressure section. The gas path in blue indicates the gas path nected to the station's vent or is equipped with a portable vent
during the filling procedure. pipeline, 2. The sampling device connected to the nozzle and
the flexible pipe connected to the sampling cylinder are
Sampling cylinder equipped with an anti-whip cable, 3 e the flow path is
The sampling cylinder is a 1-L stainless steel double-ended equipped with a check valve connected to the portable vent
valve cylinder. The sampling cylinder and the valves should system or the station's vent, 4 -each sampling cylinder is
be passivated (internally coated with silicon) to minimize equipped with a check valve and engraved with the letter “H”
adsorption of sulphur species. During sampling, the cylinder to respect the European agreement concerning the carriage of
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3 34843

Dangerous Goods for transport rules, 5 - the device is con- pressure of 120 bars is specified (the cylinder itself has a max
nected to electric ground to avoid any electrostatic discharge. capacity pressure of 147 bar according to the manufacturer), it
Before sampling, an onsite cleaning procedure consisting is only mentioned that the sampling is stopped when enough
of several pressurization e venting events (dynamic mode) or hydrogen has been sampled.
of a flush under a continuous flow (static mode) is performed
onsite. Once the sampling cylinder is filled, the valves are Information about the procedure
closed, and the sampling device is vented through a vent The sampling cylinder is cleaned with hydrogen (backfilling)
pipeline. and then kept under vacuum. The sampling cylinder is also
purged onsite together with the sampling adapter through the
Japanese approach vent system assembly. Typically, 1 kg of hydrogen is used for
the purge.
This method is currently used in Japan and is summarised in
the standard ISO19880-1 Annex K [21]. In this standard, this ENGIE sampling device
method is considered as an alternative to gas serial and gas
parallel methods but in practical terms, from the information To sample at the HRS nozzle, ENGIE has developed a device
available, the method is related to a “gas serial” method but adaptable for stations delivering hydrogen fuel both at 350 and
uses a single-ended valve cylinder. 700 bar and which also allows to perform online analysis of
oxygen and water during a refuelling event. On the contrary to
Description of the system the other “gas serial methods, this method doesn't require the
The sampling device consists of a receptacle (1), pressure overriding of the safety protocol of the station.
regulator (2), safety release valves (3) and the sampling cyl-
inder (4) as shown in Fig. 3 (from ISO19880-1). It also contains a Description of the system
pressure sensor and a temperature sensor positioned and The device (Fig. 4) consists of three lines; line A has a 55-L tank
operated in close proximity to the gas cylinder for safety to simulate the presence of a FCEV vehicle, line B is dedicated
reasons. to the online analysis of oxygen and water after pressure
reduction (from 700 to 1 bar) and linc C is dedicated to the
Sampling cylinder filling of the cylinder after pressure reduction (from 700 to
The cylinder is made of manganese steel and the volume is 90 bar). All the lines are connected to the vent system. The
typically 46.7 L [23]. It has polished inner surface (cylinder device has three inlets, one for sampling at 350 bar stations,
series. “SUMI-FINE”). The valves of the cylinders (type DSP21) one for sampling at 700 bar stations and an auxiliary inlet for
are made of stainless steel. In the standard, a maximum preparation and cleaning.

Fig. 1 e D7606-17 sampling device. Permission details.

34844 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3

vehicle and receptacle (6), a sampling cylinder with a pressure

regulator and a pressure relief valve (3, 1, 2 respectively). The
cylinder must have a DIN477 No1 valve to fit the pressure
regulator (referred as number 2 in Fig. 5) of the Qualitizer
device.

Sampling cylinder
The sampling cylinder is a one-ended valve 10-L cylinder in
aluminum with standard DIN 477 No1 stainless steel valve.
Linde uses aluminium cylinders with no specific treatment
and SINTEF uses SPECTRA-SEAL® cylinders. The treatment
consists of a proprietary process that renders the aluminum
surface chemically inert. Additional processes convert the
passivation layer into a surface with negligible adsorptive
properties [24]. From a technical point of view, any cylinder
suitable for hydrogen, equipped with DIN 477 No.1 valve and
approved for the respective working pressure can be used.
Fig. 2 e Air Liquide sampling system.
Different procedures are in use to prepare the sampling
cylinders before sampling.
Linde uses repeated pressure swing purges at 60  C with
Sampling cylinder nitrogen and then helium and subsequent evacuation.
Preferably, the sampling cylinder is a 1-L two ended valves The procedure followed by SINTEF in the HyCoRA project
coated stainless steel cylinder. Even if the ENGIE sampling [25] has been to first evacuate the cylinders to 1 mbar, and
device has been designed for the utilization of a two-ended then pressurize them to 10 bar using Ultra High Purity (UHP)
valve cylinder of 1 L, it is quite flexible and other volumes of hydrogen. This procedure is repeated three times before the
cylinders can be used (from 0.5 to 5 L). The device is fully cylinders are finally evacuated to 1 mbar prior to sampling
compatible with cylinders equipped with other coatings or use. The pressures (1 mbar and 10 bar) and number of cycles
materials. Nevertheless, it is not adapted to sample with one- were chosen arbitrary and has not been validated at the lab-
ended valve cylinders. The sampling cylinder is cleaned oratory but several sampling campaigns consisting of more
before sampling with nitrogen and a 500 mbar pressure is than 40 samplings showed no evidence of carry-over from one
maintained in the cylinder. sample to the next even if the cases where impurities were
found in the hydrogen fuel above the thresholds in the stan-
Information about the procedure dard ISO14687 [7].
During the onsite cleaning, the equivalence of 10 times to the NPL developed a seven steps method. The procedure is
volume of the cylinder is purged. After the vent, the device is explained in detail in the MetroHyVe report A4.1.7 [26] and
flushed with nitrogen to remove the hydrogen before requires the use of a roughing pump, a turbo pump and re-
transportation. sidual gas analysis combined into a ‘evacuation rig’. The
roughing pump is used to partially evacuate the cylinder
(around 1.1  10 1 mbar or less) and the evacuation is subse-
Strategies with the “gas parallel” method quently done using the turbo pump (1  10 7 mbar). The
outgassing of air, moisture and any remaining contaminants
At least two different sampling strategies are based on the is monitored with the residual gas analyser. If an expected
principle of the “gas parallel” method, one strategy using the impurity remains within the system this should be removed
Linde qualitizer and one strategy developed by ZBT. Sampling by heating or including a subsequent hydrogen purge step.
devices with the Linde qualitizer are used by several organi-
sations who then adopt different sampling strategies (for
example, slightly different types of cylinders with regards to
the internal treatment or different protocols to prepare the
cylinders and with or without methods to purge the sampling
device).

Methods using the Linde qualitizer

The sampling is performed using a device called the Qualitizer

manufactured by Linde. The filling of the cylinder is per-
formed while a FCEV car is being refuelled.

Description of the system

The sampling device presented in Fig. 5 (from the standard
ISO19880-1 [21] consists of a nozzle (4), a tee fitting (5), a Fig. 3 e Sampling device used in Japan.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3 34845

Information about the procedure overriding the refuelling protocol. It also offers the possibility
The filling of the cylinder is performed while a FCEV car is of venting to atmosphere rather than using a FCEV.
being refuelled. The tank of this car needs to be almost empty
[19] so the refuelling process takes enough time so the pres- Description of the system
sure in the sampling cylinder reaches at least 50 bar. The The system is divided into three modules. Module 1 contains
pressure is limited by the pressure reducer to a maximum of all parts for simultaneously sampling up to three cylinders
150 bar. The coupling of the sampling and refuelling of a car (2.25 or 10 l) parallel to fuelling a FCEV. The complete sampling
takes 3e5 min. line of module 1 including quick connectors for the cylinders
In ISO19880-1 Annex K, two procedures are proposed to is coated (Sulfinert® also called Silconert® 2000). Module 2 is
purge the Qualitizer sampling device: the mobile vent. Module 3 contains a buffer tank including the
necessary safety components. By using module 2, sampling
1) By Initiating a sampling but aborting within 15 s in order to can be performed without refuelling a FCEV. Optionally, the
isolate the test pulse and then depressurizing the sampling vent lines (including safety relief valves) from modules 1 and 3
device with the bleed valve. The procedure is described in can be connected to module 2 or the HRS vent line. Also
detail in Ref. [19]. optional is the simultaneous sampling for particles.
2) By performing the operational procedure without con-
necting the sampling device to FCEV. The HRS safety will Sampling cylinder
shut off hydrogen dispensing and depressurizes he bleed The sampling system can accommodate Spectra-seal treated
valve. 10 l aluminium cylinders with DIN 477 N.1 valve outlet or 2.25 l
double-ended stainless-steel cylinders with internal coating
NPL procedure to purge the sampling device is explained in (Sulfinert®). For connection to the sampling device, both cyl-
detail in the report A4.1.7 [26]. In this approach the sampling inder types have quick connectors. All valves and connectors
cylinders have been pre-filled with ultra-high purity (UHP) ae coated (Sulfinert®). The cylinders are usually filled to about
hydrogen to a pressure of 2 bar. This slight over-pressure of 90 bar. Cylinders are conditioned by evacuation down to
UHP hydrogen is then used to purge the sampling apparatus, 10 7 mbar. The cylinders can be used evacuated or are pres-
specifically the Linde Qualitizer to remove residual air present surized with e.g. 300 mbar hydrogen (quality 9.0 which is ob-
within the dead volume between the sampling apparatus and tained by when H2 (5.0) passes through a palladium
the sampling cylinder. The procedure requires to use a membrane purifier).
correction factor to account for the dilution due to the
hydrogen in the cylinder. Information about the procedure
The nozzle of the HRS is connected to the receptacle of
Hy-SaM sampling device module 1. In case of sampling in parallel to a refuelling
(without module 3), the module 1 nozzle is connected to the
Within the German Hy-Lab project, ZBT and ZSW developed a FCEX. Flushing of the hydrogen line is done by using the
sampling device called Hy-SaM (Fig. 6) which allows sampling overpressure of an aborted refuelling and with the slight
according to the ISO19880-1 [21]and SAE J2601 [27] without overpressure of the cylinder.

Fig. 4 e ENGIE sampling device.

34846 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3

False negative of contaminant disappearing from the gas phase

of the sample
Loss of chemical compounds from the gas phase during the
sampling can occur if species present in the hydrogen fuel are
adsorbed, absorbed or react for example by adsorption onto
the wall of the cylinder or onto the sampling lines leading to
the cylinder.
Materials are more or less prone to absorbing reactive
compounds such as sulphur compounds, ammonia, formal-
dehyde, formic acid onto their surfaces. This problem is highly
relevant for hydrogen fuel sampling as sulphur and other
active compounds need to be quantified at trace levels (as low
as 4 nmol/mol for sulphur). Therefore a defect of materials
may hinder few nmol/mol of reactive compounds. It can be
expected that many parameters such as the number of com-
ponents of the sampling device, the function of the compo-
nents used (for instance pressure reduction), the pressure, the
Fig. 5 e “Gas parallel” sampling device from ISO19880-1. temperature, the materials chosen may have an impact of the
amount fraction of reactive compounds in the hydrogen fuel
and in consequence to the representativeness of the sample
collected.
Discussion It is worth noting that the different strategies presented
here used different types of cylinders (stainless steel or
The discussion will present the different issue related to the
aluminum, treated or untreated) and sampling devices with
representativity of the hydrogen fuel sampling with regards to
different components such as regulators, relief valves, pres-
chemical composition (false positive and negative), then a
ence (“parallel method”) or absence (“gas serial method”) of
summary of the different sampling strategies will be pre-
tee-connection. Some of the sampling devices have many
sented and the different parameters affecting the sampling
components such as the ASTM D7606-17 method. The mate-
discussed (sampling material behaviour, sampling cylinder,
rials used are mainly stainless steel for the sampling device
procedure). Finally, discussions about the safety of the sam-
and aluminium or stainless steel for the cylinder. It is critical
pling strategy, validation of the system and future online
to assess each sampling strategy to understand the likelihood
monitoring will bring a different perspective to the sampling
of false positive and false negative occurrence for each of
strategy comparison.
them.

Representativeness of the sample collected

Summary of the different sampling system for hydrogen fuel
The main purpose of all sampling strategies is to collect a quality art HRS
sample of hydrogen that is representative of the hydrogen fuel
composition dispensed at the station. This requires that no Table 1 presents the key aspects of all the existing hydrogen
chemical compounds are added (false positive) nor removed fuel sampling system. From the table, it is clear that there are
(false negative) from the hydrogen fuel during the sampling. difference and similarities between the approaches.

False positive or contamination of the sample - All the approaches use sampling systems in stainless steel
Addition of compounds during the sampling can occur if the however it is not always clear if specific treatments are
cylinder already contains impurities. It can be air and water if applied to the sampling systems. Additional components
the cylinder or the sampling system has not been properly as pressure regulators, pressure gauges have not been
purged but also contaminants originating from previous uti- sufficiently described to determine their potential influ-
lizations of the cylinder if it has not been properly cleaned. ence on the amount fraction of the reactive species.
The effectiveness of the procedure chosen to prepare the - All strategies imply a cleaning procedure of the cylinders
cylinders as well as the procedure to purge the sampling de- and a sampling procedure involving purging to avoid false
vice before the sampling is of high importance and needs to be positive contamination.
demonstrated. - The sampling systems are using a large variety of gas cyl-
All the sampling strategies mentioned in this study include inders. It is often not specified what is the rationale behind
a step for preparation of the cylinders before sampling. the cylinder's selection.
Therefore, it is expected that contamination from previous
use of the sampling cylinders should not occur. However,
there is no published study available on this carry-over effect, Sampling device - material behaviour
it is then important to log information about each sampling The sampling devices are mainly made of stainless-steel
cylinder history. The purging procedure of the sampling de- components. However, there are various types of stainless-
vice is also important to avoid false positives. steel (i.e. SS316, SS304) and different treatments (i.e.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3 34847

Fig. 6 e Hy-SaM sampling device.

electropolished, passivated Sulfinert®, silconert 1000) that structure. Most of the reactions that take place on the internal
make difficult to predict the actual behaviour of contaminants surface of an aluminium cylinder are not direct reactions with
only based on the material. the aluminium (oxide) surface, they are cavity enhanced re-
Moreover, the different sampling system presented in this actions catalysed by the aluminium oxide structure. Reactive
study has additional components (i.e. pressure regulator, compounds are absorbed onto stainless steel through chem-
pressure relief valves, pressure gauge), these compounds may isorption. The addition of manganese into the steel improves
have additional parts (sealent, gaskets, movable part) made of strength and corrosion resistance of the alloy however no
other materials which may have an impact on the reactive study has yet investigated its effect on contaminant stability.
compounds. It is worth noticing that two systems for “gas Additionally to the cylinder type, there are several types of
serial” and “gas parallel” are currently Sulfinert® passivated: passivations or surface treatments existing: polishing or
NSP 7606 and Hy-SaM. The comparison with untreated sys- electropolishing of the internal surface (chemical or me-
tems would be interesting to evaluate the criticality of the chanical treatment), coating or passivation using inert silica
treatement. There is limited literature on the interaction of layer (e.g. Sulfinert®), carboxylane coating (e.g. SilcoTek Dur-
reactive compounds at nmol/mol level with materials. There san) [29] However chemical passivation or coating treatment
are currently no easy ways to assess the performance of a tend not to be feasible for each cylinder type: Sulfinert®
sampling device regarding the representativeness of the passivation can only be applied to stainless steel cylinder and
sample collected. can't be applied to aluminium cylinder due to material spec-
The best way would be to build an experimental facility ification and safety. Other treatment as SPECTRA-SEAL (BOC)
simulating a HRS (and including a nozzle) where hydrogen or Performax (Effectech) have been applied only to aluminum
fuel could be contaminated by known fraction amounts of cylinder [29].
species including reactive species. Another possibility would Some information about the impact on the material se-
be to use several sampling devices with defined sampling lection for the sampling cylinder on the representativeness of
strategies at the same HRS and then compare the results of the the sample collected can be found in literature, rarely in
analyses of the hydrogen fuel collected with different sam- hydrogen matrix, but mainly in air or nitrogen. It can anyway
pling strategies. However, if results differ from each other, it provide information if stability when using different types of
may be proven difficult to find the causes leading to these cylinder or internal cylinder passivation were observed. Satar
discrepancies as the composition of the hydrogen fuel is un- et al. [30] did not observed significant difference in water
known. Finally, if reliable online instruments are used at the amount fraction in nitrogen in various materials and treat-
stations, the analysis report obtained while sampling with a ment (including Sulfinert®®, aluminium or steel). Several
defined strategy can be compared to the online measure- compounds as helium, nitrogen, argon are almost considered
ments. However, as the online instruments are most likely not as inert and no unstability is expected in any gas cylinder type.
installed at the nozzle, once again, if discrepancies are However, there is a lack of literature demonstrating it in
observed, they cannot only be attributed to the sampling hydrogen matrix. One recent study demonstrated stability of
strategies. argon, nitrogen, helium, methane, ethane and carbon dioxide
but only in SPECTRA-SEAL (BOC) and SGS (Luxfer) aluminium
Sampling cylinder e material behaviour cylinders [31]. It is currently possible to buy these compounds
There are three main types of cylinders in use for hydrogen in hydrogen gas cylinder through gas supplier (i.e. Air Liquide
fuel sampling; aluminium, stainless steel and manganese or Linde) with stability of 1 year minimum.
steel. Total hydrocarbons cover a wide range of molecules from
By contact with air, aluminium surfaces [28] are passivated light linear hydrocarbon chains (e.g. ethane, butane) to long
due to the development of an oxide layer and inertness can be and heavy compounds (e.g. toluene, naphthalene) or highly
improved by increasing the thickness of this oxide layer. polar compounds (e.g. ethanol, methanol) exhibiting different
However, aluminium oxide has an inherent, honeycomb like behaviour in gas phase. Therefore, it is important to
 34848
Table 1 e Sampling strategies characteristics.
Gas serial Gas parallel
Air Liquide ASTM D7606:17 Gas direct Engie Device Linde qualitizer Hy-SaM
Equipment (sampling Quick connect fittings, Receptacle, main ball valve, Receptacle, Three lines for spot Tee-fitting, vehicle and Tee fitting, sampling
devicerowhead modules with pressure manometers, reduction valve, 4 decompression sampling, online receptacle, sampling cylinder, cylinder(s), pressure
regulators and way cross, pressure release measure), safety analysis of O2 and pressure regulator, pressure regulator, pressure relief
manometers, mobile vents valves, sampling cylinder, measures, sampling H2O and a line with a relief valve valve
and allows for sampling cylinder inlet and outlet valves, cylinder (T, P 55- liter tank to

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3
without vehicle and safety inlet and outlet quick valve, monitoring) simulate a FCEV car
measures. quick connection, check valve
Sampling vessels
Air Liquide ASTM D7606:17 Gas direct Engie Device Linde qualitizer Hy-SaM
Size and configuration Two-ended valve cylinder Two-ended valve cylinder of 0.5 One-ended valve Two-ended valve One-ended valve cylinder of One-ended valve cylinder of
of 5-L e2 L cylinder of 47 L cylinder of 1 L 10 L 2.25-L or 10-L
Materials Aluminium (cylinder)/ Stainless steel Manganese steel Stainless steel Aluminium Aluminium/Stainless steel
stainless steel (Valve)
Treatment Internally coated with silicon Polished Coated (Sulfinert®) Ex: SPECTRA-SEAL SPECTRA-SEAL/Sulfinert®
coated
Requirements
Air Liquide ASTM D7606:17 Gas direct Engie Device Linde qualitizer Hy-SaM
HRS override manual operation yes yes no no no/yes
Filling pressure ~150 bar 69 bar max 120 bar 90 bar 90e130 bar ~90 bar
Sampling duration A) < 1 min/cylinder <1 min/cylinder <3 min <3 min 2e5 min
B) ~ 30min/installation
Maximum rated 200 bar 147 bar 100 bar 160 bar 150 bar
pressure
Connection fitting Quick connect fitting Quick connect fitting e Quick connect fitting Quick connect fitting Quick connect fitting
Venting Yes (mobile or HRS) yes yes no (FCEV) yes/no (FCEV)

Preparation procedures

Cylinder Cleaning Standard cleaning The sampling cylinder is The sampling Several procedures exist: ex: The cylinders are previously
procedure (lab) procedure (on site) Several prepared by pulling a cylinder is cleaned repeated pressure swing purges evacuated down to
compression hard vacuum after a with nitrogen. A at elevated temperature with 10 7 mbar with a pump. The
decompression pure hydrogen backfill. residual pressure of nitrogen and helium with cylinders can be used
Initial cleaning procedure The sampling cylinder is 500 mbar is subsequent evacuation (LINDE) evacuated or are
for each new sampling purged with the maintained in the pressurized with 300 mbar
cylinder or for those from a hydrogen to be sampled cylinder hydrogen.
Cylinder cleaning previous sampling with an Together with the sampling Together with the Together with the no no
procedure onsite excessive amount of device sampling device sampling device
Procedure to purge the impurities (lab) 1) Emptying With 1 kg hydrogen through the With 1 kg hydrogen So as 10 times the Several procedures exist: ex: By By initiating a sampling,
sampling devices 2) Steaming with N2 3) device through the device volume of the Initiating a sampling/aborting/ aborting, depressurizing the
Rinsing with H2 (minimum cylinder is purged depressurizing or by system and depressurizing
5 cycles) 4) Filling with H2 performing the operational the cylinders
residual (pressure ~ 5 bar) procedure without connecting
the sampler
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3 34849

investigate more than one type of hydrocarbons to ensure that itself. Stability issues are expected in hydrogen independently
the cylinder type is fit for purpose. on the passivation treatment used due to the reaction with
The stability of carbon monoxide at 100 nmol/mol amount hydrogen as the surface of the cylinders was suspected to act
fraction was studied in aluminium cylinder (untreated, as catalyst [37]. Other cylinder pre-treatments (Stainless steel
SPECTRA-SEAL and SGS) [31,32], even if SPECTRA-SEAL and with Sulfinert® passivation and aluminium with performax)
SGS aluminium cylinder showed a good stability, some issues are reported to show a better stability than aluminium cylin-
were observed in untreated aluminium cylinder over long der with SPECTRA-SEAL. These cylinder types are reported to
period of time (2 years). Considering the short duration be- achieve a stability of 80% over 1 month at 1 mmol/mol form-
tween sampling and analysis, this type of untreated aldehyde in hydrogen [37]. A study [38] in nitrogen matrix
aluminum cylinder can be considered acceptable however it is demonstrated that Aculife VIII (Scott Specialty Gases) shows
not the best option to consider. good stability down to 500 nmol/mol of formaldehyde.
For very low amount fractions of reactive contaminants, Recent study showed instability of low amount fraction of
like those specified in ISO 14687, there are very few studies on formaldehyde in aluminium cylinder (SPECTRA-SEAL and
stability in different cylinders due to the complexity of pre- SGS) in less than 24 h [31]. In the same study, formic acid was
paring such gas mixtures. For hydrogen sulphide, stability in proven stable in SPECTRA-SEAL aluminium cylinder at
different cylinders have been the subject of several studies approximately 400 nmol/mol while significant instability was
(however rarely in hydrogen matrices). Most of the studies on observed in aluminium SGS for similar concentration over a
hydrogen sulphide tend to show that some kind of treatment short period. Other reports mentioned that formic acid is
of stainless steel or aluminium cylinders is required. For stable in aluminium cylinder at amount fraction from 10 to
example, a study showed that hydrogen sulphide at a con- 100 mmol/mol in hydrogen for 1e5 years’ time however it has
centration of 17 nmol/mol [33] was totally lost after a day been reported to be more challenging at nmol/mol amount
when stored in non-treated stainless-steel cylinders while fraction [37].
hydrogen sulphide concentration remained stable for a period Ammonia has been studied in nitrogen matrix at amount
of at least 7 days when stored in Sulfinert® coated cylinders. fraction 10e100 mmol/mol and good stability was observed for
Another study has shown that 1.5 nmol/mol of hydrogen stainless steel Sulfinert® coated cylinders for which no losses
sulphide in air remained stable in Sulfinert® treated canisters were observed. Relatively good performance was also ob-
[34]. Few stability results could be found for hydrogen sul- tained for SPECTRA-SEAL cylinders (BOC) and cylinders from
phide at low nmol/mol in hydrogen matrix in aluminium Takachiho [39]. Recent study showed instability of low
cylinders. These studies pointed out a decay at concentration amount fraction of ammonia (~200 nmol/mol) in aluminium
of 7e40 nmol/mol in aluminium cylinder with SPECTRA-SEAL cylinder (SPECTRA-SEAL and SGS) in less than 24 h [31].
treatment [31,32]. However, concentration of 5e15 nmol/mol For HCl at approximately 400 nmol/mol amount fraction,
was stable over few months in aluminium Superior gas sta- only one cylinder type (aluminium Acculife IV) has been
bility - SGS cylinder [31]. The stability of hydrogen sulphide at tested and the results showed that the measurement stabi-
500 nmol/mol in synthetic air has been studied in different lization is slow with measured value significantly below the
types of alumimiun cylinders [16] (“A”: aluminium SGS cylin- expected value [29]. Regarding halogenated species, it may be
ders, “B”: basic alumimium alloy cylinders and AW: acid impossible to ensure that all chemical compounds are tested
washed cylinders), for each type of cylinders, several cylinders for their stability in gas cylinder. It may require extrapolating
were tested. Fast decay was observed in AW cylinders (total stability from few halogenated compounds (i.e. dichloro-
loss after 2 days) and in “B” cylinders (total loss after 10 days in methane, chloroform, tetrachloroethylene, 1,2,3,4-C4Cl4F6,
most cylinders tested) while hydrogen sulphide concentration dichlorobenzene, dichlorohexafluoro-2-butene). A recent
remained more stable in “A” cylinders. However even in these study on dichloromethane showed this compound was sta-
cylinders, a certain loss (up to 20% for one cylinder) is ble in both SPECTRA-SEAL and SGS aluminium cylinders at
observed mostly at the beginning of the tests. Similar to the approximately 50 nmol/mol over the period of a few months
total hydrocarbons, total sulphur requires to investigate more [31].
than one sulphur molecule as dimethyl-sulphide (DMS). A recent study [31] demonstrated water was stable at
Recent study from the Korea Research Institute of Standards around 5 mmol/mol in SPECTRA-SEAL and SGS cylinders for a
and Science (KRISS) achieved to 0.5e7.0 nmol/mol of DMS in few months, however, the same study mentioned that decay
nitrogen stable for at least a year in aluminum cylinder treated of any oxygen that could be present in the cylinder by reaction
with Experis treatment (Air Products) [35]. Similarly, to total with the matrix hydrogen could cause instability of water. The
hydrocarbons, it is important to investigate more than study also showed that oxygen could be kept stable in both
sulphur compounds to ensure that the cylinder type is fit for SPECTRA-SEAL and SGS cylinder for a few months but noted
purpose. that differences had been seen in stability of oxygen between
Other reactive species such as formic acid or formaldehyde cylinders of the same internal passivation treatment.
have not been studied to the same extent as hydrogen sul- In Table 2, an attempt to summarize the results from the
phide and the implication of the sampling of a gas potentially different stability studies is done. It is important to notice that
containing several reactive species at trace levels have rarely the evaluation is based on results not always performed in a
been studied. Formaldehyde in hydrogen has been studied at hydrogen matrix. The time-period of testing is not standard-
10 mmol/mol in aluminum SPECTRA-SEAL cylinder and has ized so some studies were performed over months while
showed decay over time [36]. The authors suggested that others were performed over weeks. Finally, the definition of
decay was strongly dependent on the SPECTRA-SEAL cylinder the term “suitable” would need to be defined quantitatively
34850 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3

which is not the case in these studies. Finally, results from closely represents the actual hydrogen fuel received by the
different studies using the same type of cylinders reach FCEV. However, it represents the situation of all the storage
different conclusions. banks at the HRS so no information is obtained for one specific
storage bank.
Validation of sampling Strategy The serial sampling requires the station to be set in
maintenance mode and the volume of gas sampled is lower
The comparison and validation of sampling strategies based due to the cylinder volume and its pressure rating. Therefore,
on bilateral sampling and analysis is an important part of a hydrogen from only one storage bank may be sampled. It is
method validation. Several activities to compare sampling important for the sampling operator to understand this spe-
strategies are ongoing in European projects. HYDRAITE proj- cific feature as it may require multiple samplings if there are
ect organises a bilateral comparison between HySaM sam- multiple storage banks (involving different processes as
pling device and Qualitizer procedure in 2021 at a hydrogen compression). The serial sampling will however allow to
refuelling station to evaluate any agreement or disagreement clearly evaluate the hydrogen fuel quality from this particular
between the two strategies. storage bank.
The strategies with the Qualitizer and Air Liquide sampler
were compared at a HRS [19]. This study with the first bilateral Advantages/disadvantages offline methods against online
comparison of two sampling systems emphasised that all methods
contaminants were found to be below limit of detection. The
two sampling methodologies agreed except on nitrogen Other aspects to consider are the time and the costs aspects.
amount fraction however it was suspected to be due to the Once the sample of hydrogen has been collected, it must first
hydrogen fuel at the HRS. As the HRS was previously in be transported to the analysis laboratory following rules that
maintenance mode, it may be possible that the nitrogen are stringent due to hydrogen itself and then analysed before
amount fraction was linked to HRS infrastructure purging. any conclusion can be drawn regarding the quality of the fuel
Therefore, it is difficult to determine if there was any false dispensed. These two steps are time consuming and costly.
negative effect due to the sampling system material or Depending on the location of the HRS station and the location
procedure. of the analysis laboratory, the hydrogen purity assessment
Comparing sampling strategies at a HRS is therefore report can be delivered in the best-case scenario after a few
important to evaluate potential false positive however, workdays and up to after a month or so in the worst-case
studies using contaminated hydrogen would also be impor- scenario.
tant if technically possible. Currently, the cost to analyse one sample of hydrogen is
A recent campaign on 28 European HRS showed that most evaluated to be between V6000 to V11,000 depending on the
of the HRS are free of contaminants [18]. However, the study number of samples analysed at the same occasion. To the cost
highlighted some limitations of the current quality control of analysis, the cost for sampling evaluated to be around
tools available for hydrogen fuel quality (lack of reference V4000 needs to be added [41].
materials, standardized methods or inter-laboratory com- Once the results are known, the data interpretation should
parisons). Moreover, the limitations due to material compat- be a collaborative process between the HRS, the analysis lab-
ibility can lead to false-negative results. If no contaminants oratory and the personnel having performed the sampling,
are present in the hydrogen, the validation exercise will mostly if critical results are reported. For example, it will
mainly assess the presence of false positive. The validation of require investigating if the species found above thresholds
the sampling strategy for all the compounds including reac- originate from the hydrogen fuel itself or from one part of the
tive compounds at low amount fraction may be difficult to sampling strategy, especially in the cases where oxygen, ni-
scientifically validate due to their absence in the fuel. There- trogen and/or water are found. If online measurements of
fore, the development of alternative validation strategy oxygen and water were to be performed at the station, this
involving synthetically contaminated fuel will be beneficial to issue could be prevented. To this purpose, some of the sam-
assess the false negative or the impact of material on the fuel pling strategies presented here have included online analy-
composition. sers (Air Liquide method and ENGIE method). This feature also
presents the advantage to directly give information on
Sampling strategy and representativity of hydrogen hydrogen fuel quality with regards to these species leading to
immediate decisions to be taken (for instance shut down).
HRS have different designs that may involve several storage However, it is not yet probable that all species can be moni-
banks and different compression methods (for instance me- tored online due both to the lack of instruments; no instru-
chanical compressors or metal hydride compressors [40]). The ment can so far monitor all the gaseous species at the required
two sampling approaches may differ regarding the represen- levels [42,43] and the high costs of instruments if all species
tativeness of the sample collected It is therefore important were to be monitored online. Therefore, reliable sampling
that the sampling operator has an understanding of the HRS strategies for offline monitoring of the hydrogen fuel quality
design to determine what would be the most relevant remain essential for the hydrogen industry.
approach. Online monitoring at the HRS for a selection of species can
The parallel sampling follows the refuelling protocol and be strategically implemented. For example, it can have the
will therefore collect a fraction of the hydrogen fuel repre- goal to target species with probability of presence “frequent”
sentative of the whole process at the HRS. This sampling then and “possible” [44,45] for the relevant hydrogen production
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3 34851

Table 2 e Cylinder suitability for a time period of 4 months.

Stainless steel aluminium
Untreated Sulfinert® Untreated Aculife VII Performax SPECTRA-SEAL Untreated SGS
a b a b a b a b a b a b a b
C2H6 X X X X X X X X X X S S S S
He X X X X X X X X X X S S S S
N2 X X X X X X X X X X S S S S
Ar X X X X X X X X X X S S S S
CO2 X X X X X X X X X X S S S S
CO i.d. S i.d. S S S i.d. i.d. i.d. i.d. S S S S
H 2S i.d. I/S X. S i.d. I i.d. I i.d. i.d. I I S i.d.
HCl i.d. i.d. i.d. I i.d. i.d. i.d. I i.d. i.d. i.d. i.d. i.d. i.d.
CH2O i.d. i.d. i.d. S* i.d. i.d. i.d. i.d. i.d. S* I I I i.d.
CH2OH i.d. i.d. i.d. i.d. i.d. X. i.d. i.d. i.d. i.d. S S I i.d.
NH3 i.d. i.d. i.d. X i.d. i.d. i.d. i.d. i.d. i.d. I X I i.d.
O2 i.d. i.d. i.d. i.d. i.d. i.d. i.d. i.d. i.d. i.d. Sa Sa Sa Sa
H 2O i.d. i.d. Xb Xb i.d. i.d. i.d. i.d. i.d. i.d. Sb Sb Sb Sb

a: at ISO14687:2019 threshold.
b: at Higher concentrations (i.e. 50 times ISO14687).
X: should be suitable.
S: suitability demonstrated (* more than 80% stability over at least a month).
I: Issues were found (ex. of issues: need careful selection of the cylinder, initial loss …).
i.d.: Insufficient data.
a
Oxygen stability seems to vary between cylinders of same internal treatment.
b
Oxygen reactivity may affect the amount fraction of water through the reaction in hydrogen matrix.

method(s); CO and N2 for steam methane reforming with PSA, building openings and overhangs [25]. Moreover, hydrogen
and O2 for chlor-alkali process) or to identify leaks and/or to refuelling stations are regulated environment with ATEX
control that the station has been properly purged after zone. The venting location, volume or flow needs to be agreed
maintenance (by monitoring H2O and O2). beforehand with the operators if a mobile vent facility is
required. All systems have pressure relief valves, these safety
Safety of the sampling strategy valves may require to be connected to a safe vent. Otherwise,
the hydrogen released in the event of an incident will be close
It is critical that the components of the sampling device for to the nozzle with risk associated to the ATEX zone. Another
hydrogen fuel sampling at the HRS nozzle comply with local possibility would be to connect sampling systems to the HRS
and international regulations. They should be operated by safety vent. In this case, it would be important to standardize
trained staff and follow regular maintenances and audits to the connection to the HRS safety vent to allow all sampling
ensure the safety of the system while in operation. equipment to be compatible to the HRS safety vent.
A critical aspect of hydrogen fuel sampling is the venting of
hydrogen prior, during and after the sampling event. As
explained when describing the different strategies, it is Conclusion
required to purge several times the sampling system with
large volumes of hydrogen (i.e. 1 kg), or at various locations of This study describes different sampling strategies to safely and
the sampling system (i.e. purge valve of the Qualitizer). representatively sample hydrogen at the nozzle of a HRS. Stra-
Therefore, an important aspect to consider is how to safely tegies consist of the choice of components for the sampling de-
perform hydrogen venting at HRS during a sampling event. vice inclusive the sampling cylinder, the design of the sampling
Depending on the device used, the quantity of hydrogen to be device, the requirements in terms of filling pressure, safety,
vented differs significantly. Moreover, some of the strategies connection and fitting, the procedure to prepare the sampling
already imply venting at the beginning of the sampling cylinders before sampling, the procedure to purge the sampling
strategy to purge the sampling device; for instance the HQSA device and the procedure to vent the device after sampling.
in the ASTM D7607 method is cleaned by purging 1 kg of The strategies using the “gas serial” method imply that
hydrogen fuel through the HQSA. As hydrogen's flammability hydrogen is filled in gas serial from the nozzle in a sampling
range is very wide, with a lower explosive limit (LEL) of about cylinder/and may require a tank) while the strategies using
4% and an upper explosive limit (UEL) of about 75%, hydrogen the “gas parallel” method include as component a tee-
from vents and safety relief equipment shall be piped out- connection to parallelly fill the sampling cylinder and a car
doors to a safe location where they do not generate a hazard or a tank. The main purpose of all sampling strategies is to
for persons or neighbouring structures, away from personnel collect a sample of hydrogen that reflects the hydrogen
areas, electrical lines and other ignition sources, air intakes, dispensed at the station.
34852 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3

As materials are prone to absorbing reactive compounds [5] Filling up with H2 [Online]. Available, https://h2.live/en.
onto their surfaces, it can be expected that many parameters [Accessed 6 April 2021].
such as the number of components of the sampling device, the [6] Hydrogen, more efforts needed [Online]. Available, https://
www.iea.org/reports/hydrogen. [Accessed 6 April 2021].
function of the components used, the pressure, the tempera-
[7] ISO 14687. 2019 - Hydrogen fuel quality d product
ture and specially the materials chosen may have an impact of specification. Geneva, Switzerland: International
the representativeness of the sample collected. It is of high Organization for Standardization; 2019.
importance to demonstrate that regardless of the strategy [8] EN 17124. 2018 - Hydrogen fuel - product specification and
chosen, the outcomes of the hydrogen purity assessment are quality assurance - proton exchange membrane (PEM) fuel
the same. cell applications for road vehicles. Brussels, Belgium:
As it can be seen in this study, strategies currently imple- European Committee for Standardization; 2018.
[9] Hydrogen fuel quality for fuel cell vehicles J2719_201109.
mented used different types of cylinders (stainless steel,
Warrendale, PA, USA: SAE International; 2011.
manganese steel or aluminum, treated or untreated) and [10] Wang X, Baker P, Zhang X, Garces HF, Bonville LJ,
sampling devices with different components. A recent devel- Pasaogullari U, Molter TM. An experimental overview of the
opment is to implement online monitoring for a selection of effects of hydrogen impurities on polymer electrolyte
species as part of the sampling strategy. Due to the lack of membrane fuel cell performance. Int J Hydrogen Energy
evidence on compounds stability in sampling gas cylinder, it is 2014;34:19701e13.
[11] Tuominen R, Helppolainen N, Ihonen J, Viitakangas J.
currently very challenging to demonstrate the representa-
Probabilistic risk model for assessing hydrogen fuel
tiveness of the sample collected.
contamination effects in automotive FC systems. Int J
This review highlights the similarities and differences be- Hydrogen Energy 2018;43(9):4143e59.
tween current sampling strategies. From a hydrogen fuel [12] Shabani B, Hafttananian M, Khamani S, Ramiar A, Ranjbar A.
perspective, it is currently difficult to evaluate the impact or Poisoning of proton exchange membrane fuel cells by
the difference it would have on the hydrogen fuel quality contaminants and impurities: review of mechanisms,
analysis results. Therefore, comparative sampling studies are effects, and mitigation strategies. J Power Sources
2019;427:21e48.
needed to support the standardization of hydrogen fuel
[13] Beurey C, Gozlan B, Carre  M, Bacquart T, Morris A, Moore N,
sampling. Arrhenius K, Meuzelaar H, Persijn S, Rojo A, Murugan A.
As highlighted in this review, the implementation of online Review and survey of methods for analysis of impurities in
analysers for a selection of species would be extremely valu- hydrogen for fuel cell vehicles according to ISO 14687:2019.
able as part of the sampling strategy and to validate the Front Energy Res 2020;8:391.
representativeness of the sampling strategy. [14] Murugan A, de Huu M, Bacquart T, van Wijk J, Arrhenius K, te
Ronde I, Hemfrey D. Measurement challenges for hydrogen
vehicles. Int J Hydrogen Energy 2019;44:19326e33.
[15] H2 quality - retail stations [Online]. Available, https://www.
Declaration of competing interest nrel.gov/hydrogen/assets/images/cdp-retail-infr-25.
20210203.jpg.
The authors declare that they have no known competing [16] Aarhaug TA, Kjos OS, Ferber A, Hsu JP, Bacquart T. Mapping
financial interests or personal relationships that could have of hydrogen fuel quality in Europe. Front Energy Res 2020
appeared to influence the work reported in this paper. Nov 11.
[17] Aarhaug TA, Kjos O, Bacquart T, Valter V, Optenhostert T.
Assessment of hydrogen quality dispensed for hydrogen
refuelling stations in Europe. Int J Hydrogen Energy
Acknowledgements 2020;46:29501e11.
[18] Bacquart T, Moore N, Storms W, Chramosta N, Morris A,
Murugan A, Gozlan B, Lescornez Y, Fe rat S, Pinte G, Carre
 M.
The Joint Research Project «Metrology for hydrogen vehicles 2»
Hydrogen fuel quality for transport e first sampling and
is supported the European Metrology Programme for Innova- analysis comparison in Europe on hydrogen refuelling
tion and Research. The EMPIR initiative is co-funded by the station (70 MPa) according to ISO 14687 and EN 17124. Fuel
European Union's Horizon 2020 research and innovation pro- Commun 2021;6(100008).
gramme and the EMPIR Participating States. [19] Bacquart T, Moore N, Hart N, Morris A, Aarhaug TA, Kjos O,
Aupretre F, Colas T, Haloua F, Gozlan B, Murugan A.
Hydrogen quality sampling at the hydrogen refuelling
references station e lessons learnt on sampling at the production and at
the nozzle. Int J Hydrogen Energy 2020;45(8):5565e76.
[20] ASTM D7606 - 17. Standard practice for sampling of high
pressure hydrogen and related fuel cell feed Gases. West
[1] Hydrogen Europe's position paper on the sustainable and
Conshohocken, PA, USA: ASTM International; 2017.
smart mobility strategy. Hydrogen Europe; 2020. Hydrogen
[21] ISO 19880-1:2020. Gaseous hydrogen d fuelling stations d
Europe_SSMS_Paper_final.pdf.
Part 1: general requirements. Geneva, Switzerland:
[2] Hydrogen Roadmap Europe. Hydrogen roadmap
International Standardization Organisation; 2020.
Europe_Report.pdf (europa.eu). Hydrogen Europe; 2019.
[22] Airborne Labs launches new hydrogen fuel sampling units
[3] Lin R-H, Ye Z-Z, Wu B-D. A review of hydrogen station
[Online]. Available, https://www.h2-view.com/story/
location models. Int J Hydrogen Energy 2020;45:20176e83.
airborne-labs-launches-new-hydrogen-fuel-sampling-units/
[4] Kim H, Kim B-I, Thiel D. Exact algorithms for incremental
. [Accessed 6 April 2021].
deployment of hydrogen refuelling stations. 2021. Post-Print
[23] SUMI-FINE [Online]. Available, http://benkankikoh.com/en/
hal-03284509, HAL.
cylinder/sumi-fine.html. [Accessed 28 January 2021].
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 3 4 8 3 9 e3 4 8 5 3 34853

[24] SPECTRA-SEAL ® calibration gas mixtures [Online]. at low nanomole per mole levels in high-pressure
Available, http://hiq.linde-gas.com/en/images/SPECTRA% aluminium cylinders for ambient measurements. Metrologia
20SEAL%20Information%20Sheet_tcm899-92259.pdf. 2018;55(2).
[Accessed 6 April 2021]. [36] Bacquart T, Perkins M, Ferracci V, Martin NA, Resner K,
[25] HyCoRa [Online]. Available, http://hycora.eu. [Accessed 28 Ward MK, Cassidy N, Hook JB, Brewer PJ, Irvine JT,
January 2021]. Connor PA, Murugan A. Production and stability of low
[26] Arrhenius K, Büker O, Bartlett S, Aarhaug T, Corte s D. Good amount fraction of formaldehyde in hydrogen gas standards.
Practice Guide on effective sampling and transportation of Int J Hydrogen Energy 2018;43(13):6711e22.
hydrogen from the refuelling station as required by [37] Meuzelaar H, Persijn S, van Wijk J, Bacquart T, Bartlett S,
ISO14687. MetroHyVe, EU project; 2018. p. 2017e20. Murugan A. A2.3.1: review of the available passivation
[27] Fueling protocols for light duty gaseous hydrogen surface treatments for gas cylinders. MetroHyVe, Eu project; 2019.
vehicles J2601_201407. Warrendale, PA, USA: SAE p. 2017e20.
International; 2014. [38] Formaldehyde calibration standards [Online]. Available,
[28] The Performax® cylinder passivation: a brand new approach https://www.scottecatalog.com/PDFFiles.nsf/
for aluminium cylinders [Online]. Available, https://www. 7295b1c05fa9e8e38525667c00412c82/
gasanalysisevent.com/en/speakers/d-09/. [Accessed 6 April 17b5c97fc9211d1785256de a0055e09d/$FILE/
2021]. FormaldehydeMixes1122.1.pdf. [Accessed 6 May 2019].
[29] Arrhenius K, Bohle n H, Bartlett S, Meuzelaar H, Persijn S, van [39] van Wijk J, Bossaert D, Ferracci V, Amico di Meane E,
Wijk J. A4.4.5: good practice on the suitability of vessels and Martin N. D1.1.7 - report on the stability study of NH3
gas cylinders for sampling hydrogen as required by mixtures in cylinders. MetNH3 EU project; 2017.
ISO14687. MetroHyVe EU project; 2020. p. 2017e20. [40] Lototskyy M, Davids MW, Swanepoel D, Louw G, Klochko Y,
[30] Satar E, Nyfeler P, Pascal C, Niederhauser B. Towards an Smith F, Haji F, Tolj I, Chidziva S, Pasupathi S, Linkov V.
understanding of surface effects: testing of various materials Hydrogen refuelling station with integrated metalhydride
in a small volume measurement chamber and its relevance compressor: layout features and experience of three-year
for atmospheric trace gas analysi. Atmos Measure Tech operation. Int J Hydrogen Energy 2020;45:5419e25.
2020;13(1):119e30. [41] Carre M. How to ensure H2 quality without increasing H2
[31] Morris A, Bacquart T, Allen ND, Moore N, Ward MK, analysis cost?. In: "Hydrogen project" workshop, France;
Bartlett S, Murugan A, Storms W. Challenges in hydrogen 2018.
fuel sampling due to contaminant behaviour in different gas [42] Meuzelaar H, Liu J, Persijn S, van Wijk J, ven der
cylinders. Int J Hydrogen Energy 2021. In press. Veen AM. Trace level analysis of reactive ISO 14687
[32] Bartlett S, Bacquart T, Murugan A, van Wijk J, Meuzelaar H, impurities in hydrogen fuel using laser-based
Persijn S. Deliverable 4: report on the results of the spectroscopic detection methods. Int J Hydrogen Energy
interlaboratory comparison for offline hydrogen purity 2020;45(58):34024e36.
analysis with conclusions on the participant agreement and [43] Bacquart T, Murugan A, Carre  M, Gozlan B, Aupretre F,
recommendations for future improvements. MetroHyVe EU Haloua F, Aarhaug TA. Probability of occurrence of ISO
project; 2020. p. 2017e20. 14687-2 contaminants in hydrogen: principles and examples
[33] Barone G, Smith D, Higgins M. "Characterizing the from steam methane reforming and electrolysis (water and
performance of surface modifications that enhance chlor-alkali) production processes model. Int J Hydrogen
sensitivity, reliability, reproducbility and accuracy of Energy 2018;43(26):11872e83.
analytical instruments," SilcoTek [Online]. Available, https:// [44] Bacquart T, Arrhenius K, Persijn S, Rojo A, Aupretre F,
www.silcotek.com/hubfs/TI%2011-21-17%20Enhancing% Gozlan B, Moore N, Morris A, Fischer A, Murugan A,
20Reliability%20and%20Accuracy%20of%20Analytical% Bartlett S, Doucet G, Laridant F, Gernot E, Fernandez TE,
20Instruments.pdf. [Accessed 6 April 2021]. Gomez C, Carre  M, De Reals G, Haloua F. Hydrogen fuel
[34] A chromatographic view of a SilcoNert®-treated sample flow quality from two main production processes: steam
path [Online]. Available, https://theanalyticalscientist.com/ methane reforming and proton exchange membrane water
fileadmin/tas/issues/App_Notes/07217-silcotek-app-note- electrolysis. J Power Sources 2019;227170:444.
supplied.pdf. [Accessed 6 April 2021]. [45] Doc. 211/17. Hydrogen vent systems for customer
[35] Kim ME, Kang JH, Kim YD, Lee DS, Lee S. Development of applications: European industrial Gases association.
accurate dimethyl sulphide primary standard gas mixtures European Industrial Gases Association (EIGA); 2017.