A critical issue for the long-term monitoring of atmospheric trace gases is precision and accuracy of the measurement systems employed. Both measuring and preparing reference gas mixtures for trace gases are challenging due to, for example, adsorption and desorption of the substances of interest on surfaces; this is particularly critical at low amount fractions and/or for reactive gases. Therefore, to ensure the best preparation and measurement conditions for trace gases in very low amount fractions, usage of coated materials is in demand in gas metrology and atmospheric measurement communities. This study focuses on testing potential adsorption and desorption effects for different materials or coatings that are currently used or that may be relevant in the future for the measurements of greenhouse gases. For this study we used small volume chambers designed to be used for adsorption studies. Various materials with or without coatings were loaded into the small cylinder to test their adsorption and desorption behavior. We used the aluminum cylinder as the measurement chamber and glass, aluminum, copper, brass, steel and three different commercially available coatings as test materials. Inserting the test materials into the measurement chamber doubles the available geometric area for the surface processes. The presented experiments were designed to investigate the pressure dependency of adsorption up to 15 bar and its temperature dependency up to 80
Long-term atmospheric monitoring of trace gases requires great attention to precision and accuracy. In order to achieve a high level of compatibility for data obtained at different sites and/or at different times, the World Meteorological Organization (WMO) has recommended compatibility goals for measurements of trace gases within its Global Atmosphere Watch (GAW) Programme (2016). These challenging limits can be achieved not only by regular calibration with standard gases of known composition, but also by limiting any cause of amount fraction alteration. During their relatively long lifetime, on the order of decades, standard gas cylinders may not be stable due to diffusion, leakage, regulator effects, gravimetric fractionation and surface processes
Key results of the abovementioned studies point out that the adsorption behavior is pressure- and temperature-dependent. All mentioned studies used larger volume (10, 29.5 or 50 L) cylinders, which were already in use as standard cylinders. Their approach on filling varied from compressing natural air
In this study, we aim at distinguishing these effects among various materials under controlled conditions in a previously characterized measurement chamber
The affinity of adsorption and desorption deviates largely for different species on various surfaces. Some coatings may provide inert, corrosion-resistant, or hydrophobic surfaces and enable usage of metals instead of polymers with ambiguous outgassing effects. According to the current literature, surface losses are critical especially for more reactive gases during the preparation of the standards. In the gas metrology community, this issue has already been investigated in more detail, i.e., for species such as ammonia using test tubes with various coatings
This study contributes to the limited literature on the discussion of surface effects of different materials for the species
In order to understand the adsorption and desorption behavior of various materials, high-pressure (up to 130 bar) and small-volume (5 L) cylinders of aluminum and steel were designed. These cylinders served as measurement chambers in which various test materials can be inserted. Since the aluminum cylinder showed smaller effects with respect to surface effects in the previous study
The fillings were done using compressed air from high-pressure 50 L aluminum cylinders (LUX3586 and LUX3575). These two cylinders are called the mother cylinders and their air content the mother mixture from here on. A mother cylinder was directly connected to a small expansion volume (0.5 L) made of stainless steel (316L-HDF4-500 from Swagelok). In addition to the mother mixture, another mixture of comparable content and from a cylinder of comparable material and equipment to the mother cylinder was measured to check the stability of the measurement device. This mixture (from cylinder LUX3579) is referred to as the working gas. All three cylinders were filled by Carbagas, Switzerland, with compressed air according to their own protocol. The filling history of the cylinders is known only to the extent that the cylinders were filled with compressed air only. In order to test for higher amount fractions of CO, the mother mixtures were spiked: a known amount of pure CO gas was injected into a known volume (60 mL) and was pushed into the sample cylinder using another compressed air mixture as carrier gas. For example, after spiking the mother mixture, the composition of LUX3575 was 428.59
Schematic of the experimental setup. The aluminum cylinder is placed in the oven (denoted by the red box). The cylinder is filled through the expansion volume from the mother cylinder. At the outlet of the cylinder, the dashed lines show the three possible paths into the analyzer: through the rotary valve, direct tubing or mass flow controller (MFC). The equipment related to the cleaning procedure is denoted in blue.
Material loadings into the cylinder were conducted as follows: glass pieces were inserted in order to avoid sharp metal–metal contact points between the sample pieces and the cylinder inner surface. These consisted of a ladder and two rod-shaped glass pieces (Fig.
Since the cylinder was exposed to outside air in between loadings of different materials, a specific cleaning procedure was applied to eliminate water vapor. The measurement chamber was first pumped down to 0.05 mbar using a dry piston vacuum pump (EcoDry M15 from Leybold), then filled with 2 bar of
Figure
An overview of data included in this study. The pressure values indicate the pressure in the small cylinder at the beginning of each replicate run.
For the pressure dependency experiments, data analysis was based on
In order to investigate the temperature dependency, the cylinder was placed into a climate cabinet (Angelantoni ACS Challenge 600) at the Swiss Federal Institute of Metrology (METAS). The temperature of the cabinet was set to
Temperature cycle set at the climate cabinet.
In Fig.
Box plots for all materials for the species
Amount fraction difference relative to the start of the experiment for
In order to highlight the changes during the emptying of the measurement chamber, we show differences of the measured amount fractions from the initial amount fraction (
For
Since CO and
Based on the results of the pressure tests, the temperature experiments were conducted within a pressure range for which no pressure effect should occur, with the exception of Dursan®. In order to graphically distinguish the temperature effect on various materials, data were split into four different groups (Fig.
Figure
Temperature experiments grouped according to temperature response.
Figure
The DLC loading clearly showed a different temperature response compared to all other materials, especially with regard to the variability of its replicates.
Moreover, it should be noted that, during the set of the measurements presented in this study, the aluminum cylinder experienced the temperature cycle 30 times. This presumably resulted in a change of the background effect over the course of the presented analysis in the range of 0.04
The presented setup enabled the investigation of surface effects under extreme conditions which favored adsorption and desorption. Compared to common usage in the atmospheric measurement and gas metrology communities, our study has differed in cylinder size, geometric-surface-to-volume ratios, and pressure and temperature ranges. Previous studies
In addition to the properties of the materials, pressure and temperature play a role on surface effects. The following assumption lies behind the pressure experiments: if the material has adsorbed a significant amount of gas while filling the cylinder, this should be desorbed towards the end of the experiments controlled by desorption. The onset of the desorption for all tested materials except Dursan® and partly DLC was observed well below atmospheric pressures.
Increasing temperature is expected to facilitate desorption by providing the required energy to desorb the gas molecules from the surface and mix into the gas phase. On the contrary, cooling the cylinder and its content favor adsorption, and it is expected that this results in a decrease in the measured amount fraction.
Testing pieces cut from the aluminum and steel cylinders commonly used in the community would be a valuable addition to enable direct comparison between the commonly used cylinder materials and the produced material blocks at low pressures and high temperatures. Moreover, in order to observe significant surface effects, materials of very high surface areas can be inserted into the measurement chamber. Some ideas would be using thin metal plates, metal spheres or metal pieces resulting from manufacturing processes (e.g., metal chips).
We have presented the pressure- and temperature-dependent response of the species
These experiments show that all coatings do not necessarily enable more passive surfaces and might result in enhancements when exposed to pressure and temperature changes. Materials currently used by the atmospheric measurement community for storing standard gases are well suited under 80
Data can be obtained upon request. Please contact Markus Leuenberger (markus.leuenberger@climate.unibe.ch).
Amount fraction differences compared to the start of the experiment for CO and
ML and ES designed the project. ES performed all measurements and wrote the draft of the paper. CP and BN helped with the selection of gases to be studied and provided access to the laboratories and technical units of METAS. PN helped set up the measurement system and solved technical problems. All were actively involved with the final version of the paper.
The authors declare that they have no conflict of interest.
This project is supported by a research contract (F-5232.30052) between the Swiss Federal Institute of Metrology (METAS) and the University of Bern as well as the SNF project Klima- und Umweltphysik: Isotope im Erdklimasystem (icoCEP)(SNF-200020_172550). The authors would like to thank the workshop of the University of Bern for the production of the cylinders, as well as METAS Gas Analysis Laboratory and METAS workshop for their technical support during this work.
This research has been supported by the Federal Institute of Metrology METAS (grant no. F-5232.30052) as well as Isotope im Erdklimasystem (icoCEP; SNF-200020_172550).
This paper was edited by Brigitte Buchmann and reviewed by three anonymous referees.