Electrochemical sensors applied to pollution monitoring: Measurement error and gas ratio bias — A volcano plume case study
Introduction
The commercial development of low-cost, lightweight electrochemical gas sensors (e.g. as manufactured by Alphasense, CityTech, Membrapor), with low power requirements, brings new opportunities within environmental science for in-situ monitoring of atmospheric gases. The incorporation of electrochemical sensors into ‘Multi-Gas’ systems used for volcano monitoring has enabled a range of gases (including H2S, SO2, CO, H2, HCl) to be characterised in volcanic plumes (e.g. see Roberts et al., 2012 and references therein). Temporal variations in volcanic gas composition (as identified from gas ratios) can be used to infer changes in subsurface processes and inform long-term volcano hazard monitoring (e.g. Shinohara et al., 2008, Aiuppa et al., 2009). Measurements of volcanic gas emissions also inform the assessment of downwind plume impacts on the atmosphere and ecosystems. Recently, similar miniature electrochemical sensors have also been deployed in networks to characterise urban pollution (Hasenfratz et al., 2012, Li et al., 2012, Mead et al., 2013), using both fixed stations and mobile sensors (e.g. carried by cyclists, pedestrians and on urban tramways).
A challenge in both urban and volcanic environments is to quantify gas abundance from the sensor current or voltage output. The sensor is exposed to complex mixtures of gases whose abundances may fluctuate rapidly, due to close proximity to the pollution source(s), in-homogenous dilution by turbulent eddies or local wind field variations affecting exposure of the sensor. The sensor itself has a finite response time, typically 10–30 s to reach 90% of signal (see Manufacturer's datasheets (Alphasense.com) and Roberts et al., 2012, Mead et al., 2013). In such a period, gas exposure to a volcanic plume at a given location may vary substantially as dictated by the source dynamics and changing wind field. Physical limitations to the time-response of electrochemical sensing devices include the rate of diffusion of species within the sensor, which can be compounded by absorption and desorption of gases from surfaces within the instrument, e.g. for HCl (Roberts et al., 2012). For miniature electrochemical sensors (~ cm, < 10–20 g), a relatively fast (10–30 s) response time has been achieved through novel sensor design minimising distances between the electrodes, electrolyte and gas diffusion barrier, resulting in very small, low cost and low power sensing devices. Where weight and power requirements allow, electrochemical detector sensitivity can be improved and time-responses reduced to just a few seconds by using portable (~ 2 kg) instruments (Interscan, Inc.) e.g. as deployed by Kelly et al. (2013) for in-situ measurements of H2S and SO2 alongside a UV ozone spectrometer in Redoubt volcano plume.
Determining the measurement uncertainties under complex, heterogeneous plume conditions is non-trivial if gas fluctuations occur on similar timescales to the sensor response, and particularly where cross-sensitivities also need to be subtracted in post-processing: the sensor output is a function of its time-dependent response(s) to its recent exposure to (multiple) gases whose abundances may rapidly fluctuate with time. In other words, gas ratios computed for an unvarying source will show variations (artefacts) related to the exposure time histories and sensor responses. This can be a particular issue where sensors are exposed to intermittent ‘puffs’ of volcanic plume, and if attempts are made to identify rapid fluctuations in source composition. Such measurement errors have not been considered in detail to date in volcanic gas ratios reported from Multi-Gas systems. Here, careful analysis of H2S and SO2 electrochemical sensor measurements of the plume of Miyakejima volcano (Japan), at both crater-rim and downwind locations (where gas abundances fluctuate to contrasting extents), are used to identify these sources of error. The focus is how finite (and contrasting) response times of the sensors can act to introduce measurement uncertainties and bias in the reported gas ratios (e.g. H2S/SO2) under rapidly fluctuating gas concentrations (particularly in cases where interferences are subtracted in data post-processing). Sensor response modelling and a forward ‘instrument model’ are developed to illustrate Multi-Gas instrument behaviour. Improved data analysis approaches are discussed, including sensor response modelling (Roberts et al., 2012) and an integrated data analysis method. The aim is to highlight a source of inaccuracy in Multi-Gas monitoring of volcano emissions and their reported gas ratios (e.g. H2S/SO2, CO2/SO2), and to outline methods for improved quantitative characterisation of complex fluctuating plume environments (e.g. volcanic and urban) by electrochemical sensors.
Section snippets
Electrochemical sensing of pollutant gases
An overview of the use of miniature electrochemical sensors to monitor volcanic plume gases (SO2, H2S, HCl, CO, H2) and urban pollution (CO, NO, NO2) is given by Roberts et al. (2012) and Mead et al. (2013), the former also detailing use of electrochemical sensors within in-situ (Multi-Gas) monitoring systems in volcanology (e.g. Aiuppa et al., 2005a; Shinohara, 2005, Witt et al., 2008b). Issues concerning the application of electrochemical and other miniature sensors to environmental
Materials and methods
The ‘Multi-Gas’ instrument used in this study incorporates Alphasense electrochemical sensors, SO2-AF, H2S-A1, and NO2-A1, as described by Roberts et al. (2012). Gas is drawn over the sensors by a miniature pump, a filter (hepa-vent) prevents particles > 0.3 μm entering the instrument and each sensor also includes its own dust filter. Minimal use of tubing limits potential adsorption of gases within the instrument. Calibrations were performed in October 2007, shortly after the fieldwork, exposing
Discrepancies in H2S/SO2 gas ratios detected at Miyakejima volcano
The 1 Hz electrochemical sensor measurements of SO2 and H2S in Miyakejima volcano plume, as detected both downwind and in the near-source (crater-rim) plume, are illustrated in Fig. 1 as scatter plots of H2S versus SO2 mixing ratios in the downwind and crater-rim plume. H2S is derived from either the NO2-A1 sensor (with no interference), or the H2S-A1 sensor (where the interference from SO2 has been subtracted). Robust regression is used to determine H2S/SO2, which yields very similar gradient
Implications for Multi-Gas volcano monitoring, and wider electrochemical sensor applications
Through detailed investigation of Multi-Gas measurements at Miyakejima volcano, this study has shown how errors can manifest in the determination of gas ratios from in-situ sensors with non-identical response times, when sensors are exposed to fluctuating plume gas concentrations (e.g. at the crater-rim), and particularly when interferences need to be subtracted in the data analysis. A key question is to what extent this source of measurement error may influence reported volcanic gas ratios
Conclusions
There is a growing interest in the use of electrochemical sensors within environmental science as a low-cost, low-power and highly portable gas measurement technique to detect and characterise sources of and exposure to pollution. Such sensors are commonly deployed alongside other in-situ detectors (e.g. infra-red sensors for CO2, H2O) for characterisation of gas mixtures such as volcanic emissions. Here we show how the non-instantaneous time response of electrochemical sensors can lead to
Acknowledgements
We are particularly grateful to Darryl Dawson of Alphasense Ltd who provided sensor calibrations and measurement advice. TJR acknowledges a NERC studentship (2006-9) during which the instrument was deployed, and further funding from LABEX VOLTAIRE (VOLatils – Terre Atmosphère Interactions – Ressources et Environnement) ANR-10-LABX-100-01 (2011-20). TJR is grateful to C.F. Braban, P.T. Griffiths, R.A. Freshwater, R.A. Cox and R.L. Jones for advice during instrument development, and we thank E.
References (30)
- et al.
The 2007 eruption of Stromboli volcano: insights from real-time measurements of the volcanic gas plume CO2/SO2 ratio
J. Volcanol. Geotherm. Res.
(2009) - et al.
A model of degassing for Stromboli volcano
Earth Planet. Sci. Lett.
(2010) - et al.
Magma storage, transport and degassing during the 2008–10 summit eruption at Kīlauea Volcano, Hawai'I
Geochim. Cosmochim. Acta
(2013) - et al.
Rapid chemical evolution of tropospheric volcanic emissions from Redoubt Volcano, Alaska, based on observations of ozone and halogen-containing gases
J. Volcanol. Geotherm. Res.
(2013) - et al.
The use of electrochemical sensors for monitoring urban air quality in low-cost, high-density networks
Atmos. Environ.
(2013) - et al.
Electrochemical sensing of volcanic gases
Chem. Geol.
(2012) - et al.
Gas and aerosol emissions from Villarrica volcano, Chile
J. Volcanol. Geotherm. Res.
(2011) A new technique to estimate volcanic gas composition: plume measurements with a portable multi-sensor system
J. Volcanol. Geotherm. Res.
(2005)- et al.
Evolution of volcanic gas composition during repeated culmination of volcanic activity at Kuchinoerabujima volcano, Japan
J. Volcanol. Geotherm. Res.
(2011) - et al.
Degassing of CO2, SO2, and H2S associated with the 2009 eruption of Redoubt Volcano, Alaska
J. Volcanol. Geotherm. Res.
(2013)
Fumarole compositions and mercury emissions from the Tatun Volcanic Field, Taiwan: results from multi-component gas analyser, portable mercury spectrometer and direct sampling techniques
J. Volcanol. Geotherm. Res.
Foundations of Analog and Digital Circuits
Chemical mapping of a fumarolic field: La Fossa Crater, Vulcano Island (Aeolian Islands, Italy)
Geophys. Res. Lett.
Rates of carbon dioxide plume degassing from Mt Etna volcano
J. Geophys. Res.
Real-time simultaneous detection of volcanic Hg and SO2 at La Fossa Crater, volcano (Aeolian Islands, Sicily)
Geophys. Res. Lett.
Cited by (22)
W-CeO<inf>2</inf> nanospheres gas sensor array for accurate and selective H<inf>2</inf>S detection in exhaled breath
2024, Chemical Engineering JournalGeochemical monitoring of volcanoes and the mitigation of volcanic gas hazards
2020, Forecasting and Planning for Volcanic Hazards, Risks, and DisastersBlue-light photoelectrochemical sensor based on nickel tetra-amined phthalocyanine-graphene oxide covalent compound for ultrasensitive detection of erythromycin
2018, Biosensors and BioelectronicsCitation Excerpt :Further, several researchers are investigating electrochemical sensors owing to their advantages of high selectivity and sensitivity, small size, and low cost (Felix and Angnes, 2017; Guo, 2017a; Guo and Ma, 2017). These sensors have found wide use in bioanalysis, medical diagnoses, and pollutant monitoring (Guo, 2016, 2017b; Roberts et al., 2014). While PEC analysis shares the advantages of electrochemical analysis, it is more precise, faster, and non-contact in nature, as the sensing phenomenon is based on the photoelectric activity of the material employed, and the analysis is performed using light as the excitation signal and the photocurrent generated as the detection signal (Zhang et al., 2016; Zhang and Zhao, 2013; Zhou et al., 2015).
Ratiocalc: Software for processing data from multicomponent volcanic gas analyzers
2015, Computers and GeosciencesCitation Excerpt :Electrochemical and spectroscopic sensors exhibit different response times (e.g., T90~10 s for CO2 Gascard NG and ~20 s for SO2 Citytechnology at 20 °C) due to the different kinetics of the processes, and these change with the gas concentration (and redox reactions and infrared absorption). Evidence from this difference is shown by Roberts et al. (2014) and manifests as a common phase shift between two signals and different levels of noise. Ratiocalc can be used to apply a fast Savitzky–Golay filter (Savitzky and Golay, 1964) to the signals for the purpose of smoothing the noisiest data and to find the phase shift by cross-correlating the two signals.
New field-based agricultural biomass burning trace gas, PM<inf>2.5</inf>, and black carbon emission ratios and factors measured in situ at crop residue fires in Eastern China
2014, Atmospheric EnvironmentCitation Excerpt :To minimise measurement impacts from the combination of lag times and the rapidly changing trace gas mixing ratios experienced during in situ smoke plume measurements (Reid et al., 2005), the outputs from the gas sensors were averaged over 20s for trace gas emission ratio (ER) derivation. This averaging period was further increased to 120s when comparing the gas and aerosol measurements, and such averaging is commonly used in plume investigations when relying on different instruments to target diverse species whose concentrations will later be used to derive emissions ratios (e.g. Reid and Hobbs (1998); Roberts et al., 2014). Fieldwork took place in Eastern China in June 2014, coinciding with the main harvest and burning time in this part of China (Fig. 1a), which the MODIS Active Fire product (Giglio et al., 2003) confirms (Fig. 1b).