A Membrane Microfluidic Sensor for Conductivity Measurement of Ocean Dissolved Inorganic Carbon

Abstract

Atmospheric CO2 has increased by > 40% since pre-industrial times leading to significant global warming. The ocean absorption of CO2 is saturating, possibly leading to much more serious consequences than predicted. The latest IPCC special report SR15 warns that a 1.5 oC average temperature increase by 2030 is likely and associated CO2 ocean acidification is projected to amplify the adverse effects of warming. The ocean CO2 system is characterized by measuring at least 2 out of 4 canonical “master variables”: total dissolved inorganic carbon (DIC), total alkalinity (AT), partial CO2 (pCO2), and acidity (pH). Understanding this system is of fundamental importance yet CO2 continues to be chronically under-sampled in both space (including depth profiles) and time. Unfortunately, pCO2 and pH are very sensitive to temperature/pressure, hence autonomous measurements of DIC or AT are much preferred but have yet to be realized. Observations are almost exclusively limited to the surface and use large sensors, e.g. spectrophotometry, NDIR and mass spectrometry, attached to expensive research vessels. These large systems are unlikely candidates for miniaturisation. Very large uncertainties therefore still remain in the CO2 budget.The Argo float network, a global array of >3,000 untethered battery-operated floats with satellite data transfer, enables continuous monitoring of ocean salinity and temperature depth profiles. This float infrastructure awaits new autonomous chemical sensor technology but will impose severe constraints on device footprint, volume, power consumption and long-term environmental stability under high pressure cycling (up to 200 atm). Instantaneous measurement of DIC during float ascent or descent is not possible as the CO2 has to be extracted from the seawater first. Hence at each depth, an autonomous device would have to collect and store a seawater sample for subsequent analysis while the float is at park depth (~1500m). This analysis involves several chemical reactions and partial or complete equilibration across a membrane with high precision conductivity measurements. Microfluidic ocean-relevant chemical sensors have received limited attention. Research efforts have concentrated on colorimetric or luminescence techniques at shallow depths.We investigate a potential miniaturised version of a standard bench technique for DIC measurement based on flow injection conductimetric measurements.[1] To achieve a fully functioning DIC profiling system requires microfluidic lab on chip structures with sample volumes ≤ 1uL (to minimise 3 years of reagent payload), precision channel patterning, multi-layer thermoplastic bonding, and liquid-liquid membrane separation of CO2, with incompatible gas-permeable ion blocking materials, suitable for >3-year ocean deployment at high pressures. Recently we demonstrated long-term PDMS membrane bonding within a thermoplastic manifold [2] as well high pressure resilient thermoplastic bonding for multi-layer and multi-channel devices. [3] We have established the principle of precision micro-metering to inject HCl acid into the seawater sample. [4] The detection principle is based on CO2, transferred across the membrane, dissolves in high strength NaOH leading to a reduction in conductivity. [5] Conductivity electrodes are limited in size by sample volume and must operate continuously in a highly corrosive environment (NaOH, pH12). Previously we attempted a capacitively-coupled approach using thin-film protective insulating layer on copper tracks. However, this resulted in severe signal attenuation. Here we report direct metal contact conductivity cells with active volumes between 0.5 mL and 2.0 mL for DIC detection in the seawater range 1900 – 2400 mmol Kg-1. For these volumes, the use of macroscopic wires or plates as electrodes is not feasible and instead, we developed a thin film metallisation process onto PMMA using a sputtered thin film of gold (< 200 nm) on top of a 10 nm adhesion promoting inter-layer of sputtered Ti.After transfer across the membrane, the eluted CO2 in a NaOH carrier, was drawn through a < 2 mL conductivity cell where the change in impedance versus time was measured. Minimum precision values obtained from relative standard deviation were ~ 0.2 % from peak height and 0.5% from area under curve. This compares favourably with precision values of ~ 0.25 % obtained using a large C4D electrophoresis headstage of similar active measurement volume. The required sample and reagent volumes amounted to ~ 500 mL in total due to the incorporation of a planar membrane into a small volume exchange cell. This compares very favourably with reported attempts at conductivity based DIC analysis where total volumes between 5,000 mL and 10,000 mL were required as, in order to achieve the required precision, bulk wire electrodes and 20cm long membrane tubes were used. We show therefore the first microfluidic-based DIC sensor with accuracy, precision and reagent payload that offers a route to global autonomous ocean CO2 depth profiling. 1. P. Hall and R. Aller, Limnol. Oceanogr., 1992, 37, 1113 https://doi.org/10.4319/lo.1992.37.5.11132. M. Tweedie et al, Lab Chip, 2019, 1287 https://doi.org/10.1039/C9LC00123A3. D. Sun et al., Microfluid. Nanofluid., 2015, 913 https://doi.org/10.1007/s10404-015-1620-24. M. Tweedie et al. https://arxiv.org/abs/1909.018455. M. Tweedie et al., https://doi.org/10.26434/chemrxiv.8852087.v1 Figure 1