Understanding the dynamic response of Durafet-based sensors: A case study from the Murderkill Estuary-Delaware Bay system (Delaware, USA)

Abstract

The use of Durafet-based sensors has proliferated in recent years, but their performance in estuarine waters (salinity < 20) where rapid changes in temperature and salinity are frequently observed requires further scrutiny. Here, the responses of the Honeywell Durafet and its internal (pHINT) and external (pHEXT) reference electrodes integrated into a SeapHOx sensor at the confluence of the Murderkill Estuary and Delaware Bay (Delaware, USA) were assessed over extensive ranges of temperature (1.34–32.27°C), salinity (1.17–29.82), and rates of temperature (dT/dt; −1.46 to +1.53°C (0.5 h)−1) and salinity (dSalt/dt; −3.55 to +11.09 (0.5 h)−1) change. Empirical analyses indicated dynamic errors in the temperature and salinity responses of the internal and external reference electrodes, respectively, driven by tidal mixing were introduced into our pH time-series. These dynamic errors drove large anomalies between pHINT and pHEXT (denoted ΔpHINT−EXT) that reached >±0.8 pH in winter when the lowest temperatures and maximum tidal salinity variability occurred and >±0.15 pH in summer when the highest temperatures and minimum tidal salinity variability occurred. The ΔpHINT−EXT anomalies demonstrated a clear linear relationship with dSalt/dt thereby making dSalt/dt the strongest limiting factor of reference electrode response in our application. A dynamic sensor response correction for the external reference electrode (solid-state chloiride ion-selective electrode, Cl-ISE) was also developed and applied in the voltage domain. This correction reduced winter and summer ΔpHINT−EXT anomaly ranges by >40% and 68.7%, respectively. Summer anomalies were notably reduced to <±0.04 pH across all measurements. Further, this correction also removed the first-order salinity dependence of these anomalies. Consequently, dynamic errors in reference electrode response cannot be ignored and must be considered in future experimental designs. Further work to better understand the dynamic temperature and salinity responses of both reference electrodes is underway. Ultimately, we hope this work will stimulate further discussion around the role and treatment of large ΔpHINT−EXT anomalies as a part of future data quality control and data reporting as well as the dynamic errors in reference electrode response that drive them in the context of Sensor Best Practices.