Article pubs.acs.org/est
Coupling Automated Radon and Carbon Dioxide Measurements in Coastal Waters Isaac R. Santos,* Damien T. Maher, and Bradley D. Eyre Centre for Coastal Biogeochemistry, School of Environment, Science and Engineering, Southern Cross University, Lismore, NSW, 2480, Australia ABSTRACT: Groundwater discharge could be a major, but as yet poorly constrained, source of carbon dioxide to lakes, wetlands, rivers, estuaries, and coastal waters. We demonstrate how coupled radon (222Rn, a natural groundwater tracer) and pCO2 measurements in water can be easily performed using commercially available gas analysers. Portable, automated radon and pCO2 gas analysers were connected in series and a closed air loop was established with gas equilibration devices (GED). We experimentally assessed the advantages and disadvantages of six GED. Response times shorter than 30 min for 222Rn and 5 min for pCO2 were achieved. Field trials revealed significant positive correlations between 222Rn and pCO2 in estuarine waterways and in a mangrove tidal creek, implying that submarine groundwater discharge was a source of CO2 to surface water. The described system can provide high resolution, high precision concentrations of both radon and pCO2 with nearly no additional effort compared to measuring only one of these gases. Coupling automated 222Rn and pCO2 measurements can provide new insights into how groundwater seepage contributes to aquatic carbon budgets.
■
INTRODUCTION Very little is known about the role of groundwater in delivering carbon to surface waters.1 Dissolved organic and inorganic carbon concentrations in groundwater are often much higher than those in surface waters. Therefore, groundwater seepage may play a significant role in carbon budgets in freshwater and marine ecosystems even if the volumetric contribution is small. In river systems, some of the baseflow seepage is automatically included in downstream river carbon measurements.1 In marine and estuarine systems, however, site specific measurements are needed to resolve the contribution of submarine groundwater discharge (SGD) to carbon budgets. The modern definition of SGD includes both a terrestrial (freshwater) and a marine (recirculated seawater) component.2 While fresh groundwater is a source of “new” water and carbon, recirculated seawater can be a source of “recycled” organic matter respiration products such as nutrients and carbon dioxide.3,4 Recent technological advances (i.e., automation5) have increased our ability to assess groundwater discharge in complex systems using natural tracers such as radon (222Rn, half-life = 3.84 days). Radon is a biogeochemically conservative noble gas and a member of the 238U decay chain.6 Since uranium is present in nearly all sediments and has a half-life of billions of years, any water that remains in contact with sediments for at least several hours acquires a radon signal. As a result, radon is typically 2−4 orders of magnitude higher in groundwater than surface waters. The main advantage of using natural geochemical tracers such as radon is that the water column integrates the signal coming into the system from multiple groundwater and porewater pathways. Small scale variations, which are often not of interest, are smoothed out.7 Radon sources to surface waters other than groundwater discharge are often negligible and can be easily quantified.2 © 2012 American Chemical Society
Much of the carbon dioxide literature in aquatic environments has focused on quantifying exchange rates at the water− air interface8 and on internal cycling processes such as respiration and photosynthesis.9 The few investigations assessing how SGD may influence carbon cycling have relied on grab samples, and focused on either dissolved inorganic (DIC) or organic (DOC) carbon. For example, SGD-derived DIC10,11 and DOC12,13 fluxes were suggested to be significant components of regional carbon budgets. The radon and the carbon dioxide scientific communities have evolved independently in the last several years. Several radon investigations evoke the carbon cycle to justify the need for studying groundwater surface water exchange,3 and several carbon cycle investigations put groundwater discharge in the “to do” list.9,14,15 In addition, much of the groundwater community has focused on modeling studies from a water resources rather than a carbon cycle perspective. In this paper, we contribute to bridging the gap between these communities by demonstrating that automated, high precision, high resolution radon and carbon dioxide measurements can be easily performed simultaneously using portable gas detectors. We report the results of laboratory experiments designed to assess the performance of six gas equilibration devices and the first coupled, automated 222Rn-pCO2 field measurements. Our preliminary observations demonstrate that groundwater discharge may represent a major control on CO2 dynamics in surface water bodies. Received: Revised: Accepted: Published: 7685
May 15, 2012 June 7, 2012 June 14, 2012 June 14, 2012 dx.doi.org/10.1021/es301961b | Environ. Sci. Technol. 2012, 46, 7685−7691
Environmental Science & Technology
■
Article
EXPERIMENTAL SECTION Approach. We have used a RAD7 radon detector and a Licor 7000 CO2 detector because they are robust, durable, precise, and stable. In addition, they can be powered with 12 V batteries and run in a continuous mode. These characteristics make these detectors ideal for unattended field deployments. The RAD7 uses an electric field with a silicon semiconductor detector to attract the positively charged polonium daughters (218Po+ and 214Po+) which are then counted as a measure of the radon concentration in air. The full radioactive equilibrium between 218Po (half-life=3.1 min) and 222Rn takes about 15 min. The Licor is a differential nondispersive infrared gas analyzer that measures CO2 based on the difference in the adsorption of infrared radiation between two sampling cells (a reference cell with 0 ppm CO2 and the measurement cell). A portable computer is connected to the Licor to log data. While these detectors were the most suitable for this study, several other gas analysers could be used depending on user needs. In addition, gases such as methane could also be detected in the air stream. Water was continuously pumped into a gas equilibration device (GED) that equilibrates a stream of flowing water with a stream of air which is being recirculated through gas detectors in a closed loop (Figure 1). The air is pumped at 1 L min−1
simultaneously before using automated, portable analysers. Previous work on radon and CO2 relied mostly on grab samples and focused on soil gas exchange processes18,19 rather than groundwater as a source of carbon to surface waters. Assessing Gas Equilibration Devices. The major challenge in using the described system is to extract gases from water in an effective manner. We have experimentally evaluated the performance of six commercially available GEDs. Two batches of 200 L of water were used to assess how quick each device equilibrates gases between the air and water phase. The first batch of water consisted of tap water equilibrated with the atmosphere with air bubblers (i.e., low radon, low CO2 waters). The second batch of water consisted of groundwater highly enriched in both pCO2 (∼12 000 μatm) and 222Rn (∼105 dpm L−1). Each experiment was started using the low concentration water, then the water pump was switched to the high concentration water until equilibrium was reached, and finally the pump was returned to the low concentration water until concentrations stabilized again. All experiments were conducted in the laboratory at 22 °C. The same water pump (800 GPH submersible pump) was used for all experiments. However, each GED constricts the water flow in a different manner resulting in different flow rates. We have also experimentally assessed the influence of Drierite on CO2 response times by deploying two Licor 7000 in series; one just upstream of the Drierite, and another just downstream of the Drierite column. The first GED assessed was a shower head exchanger. It is a clear acrylic tube (10 cm diameter; 25 cm long) which had water flowing through it continuously at 2.8 L min−1. A nozzle (“shower head”) aspirates the water into fine droplets that maximize gas equilibration. The water exits the device over an inverted cup that isolates the closed air loop from the ambient air.16,17 The second device was a marble exchanger originally used for measuring CO2 in turbid waters20 that consists of an acrylic tube (50 cm long, 10 cm diameter) filled up with inert glass marbles. The marbles increase the surface area available for exchange and reduce the air volume. Water was pumped to the top of the exchanger at 5.4 L min−1, flowed along the marble pathway, and then exited the exchanger through a hole on the bottom. A closed air circuit was established against the water flow.20 The third device was a passive hydrophobic tubular membrane (Accurel PP V8/2 HF by Membrana GmbH, Germany). The inner diameter of the tubing was 5.5 mm and the wall thickness was 1.55 mm. Diffusion through the membrane pores ensures equilibrium between the gas phase inside the membrane and the water phase.21 We used a 2 m long membrane coil mounted on an open wire frame, but longer/shorter coils can be used. This membrane is unique in that it can be suspended in the water and no water pumping is required, decreasing power requirements for long-term deployments.22 In the laboratory, the membrane was suspended in the different batches of water. Water was recirculated around the membrane to simulate river conditions and to avoid the development of a stagnant film that could slow down equilibration. Other gas permeable materials such as silicone tubing can be used in a similar manner.23 The fourth, fifth and sixth devices were different sizes of Membrane Contactors (Liqui-Cel). These contactors are polypropylene hollow-fiber degassing cartridges with a sponge-like structure. The water in the cartridge flows inside
Figure 1. Schematic layout of the coupled 222 Rn and pCO2 measurement system. Water is continually pumped into a gas equilibration device. A closed air loop is developed between the gas detectors and the gas equilibration device. Air is pumped at 1 L min−1 using the RAD7 internal pump. Temperature, pressure, and conductivity loggers inside the GED allow the calculation of gas concentrations in the source water.
with the RAD7 internal pump. Since an air moisture content of less than 10% is required for the RAD7, an air drying unit is implemented in the air-loop. We use Drierite desiccant as suggested by the RAD7 manufacturer. While CO2 may be absorbed by Drierite if the air residence time is long, the 1 L min−1 air pump rate avoided any major Drierite interference. Nafion tubes can be used to extend the Drierite life, but these were avoided as they are not a required component of the system. A water shutoff valve in the air stream just downstream of the GED was used to avoid accidentally pumping water into the detector. A one-way air valve was connected in the tubing just upstream of the GED to avoid water reversing into the air stream. A vent (long thin tube) could also be used to avoid a pressure build up inside the GED. However, we have avoided such a vent as it could result in ambient air getting inside the GED. Temperature and pressure sensors are deployed inside the GED to allow the calculation of gas partial pressures in the sample water as a function of observed air concentrations. This basic approach has been independently developed for 222Rn and CO2,16,17 but these dissolved gases had not been measured 7686
dx.doi.org/10.1021/es301961b | Environ. Sci. Technol. 2012, 46, 7685−7691
Environmental Science & Technology
Article
Figure 2. Experimental response times of 222Rn and pCO2 using different GEDs. The blank areas in each plot represent the time when the pump was kept in low concentration tap water equilibrated with the atmosphere (pCO2 = 390 μatm, 222Rn < 0.1 dpm L−1). The gray area represents when the pump was shifted to a groundwater container highly enriched in pCO2 (∼12 000 μatm) and 222Rn (105 dpm L−1). The concentrations are shown as ratio between individual observations and the groundwater concentration. The response times (t90%) were estimated by fitting polynomial curves to the data. Analytical uncertainties are often smaller than the symbol size.
micropores in contact with a counter-current air flow.24,25 The fourth GED (hereafter referred to as small Liqui-Cel) was a 17.80 cm long cartridge with a membrane area of 0.18 m2 at a water flow rate of 1.2 L min−1 (MiniModule part number G543). The fifth GED (hereafter referred to as medium LiquiCel) was a 26.60 cm long cartridge with a membrane area of 0.99 m2 and a water flow rate of 2.0 L min−1 (MiniModule part number G541). The sixth GED (hereafter referred to as large Liqui-Cel) was a 28.18 cm long cartridge with a membrane area of 1.4 m2 and a water flow of 8 L min−1 (Extra flow part number G420). Regardless of the GED, after some time the 222Rn and CO2 in the closed air loop reach equilibrium with the concentration in the source water. Calculations were made to determine the dissolved gas concentration in water using well established protocols. The radon solubility was calculated as a function of temperature and salinity,26 whereas pCO2 was calculated from the dry molar fraction of CO2 in the gas stream as a function of temperature, salinity, and the pressure in the GED relative to the ambient pressure.16,27
achieving equilibrium between the gas and water phase for both gases (Figure 2). However, each GED had a different response time. The response time was defined as the time taken for gas concentrations to reach 90% (t90%) of its final concentrations when changing the water pump from low to high concentrations, and again from high to low. Overall, the marble exchanger and the large Liqui-Cel had the shortest response time for both gases most likely as a result of the relatively high water flow rates (i.e., 8 and 5.4 L min−1, respectively) supported by these GEDs. The passive membrane had by far the longest response time. This membrane is unique in that gases have to diffuse across the porous membrane wall. The radon response time was at least twice as long as the CO2 response time, reflecting the known low radon diffusivity.28 We emphasize that these experiments represent an extreme test to our system since we used two batches of water with extremely different gas concentrations (i.e., tap water equilibrated with the atmosphere versus groundwater). Gas concentrations in natural water bodies are unlikely to change so abruptly. The response time for 222Rn was always >15 min longer than the pCO2 response time, reflecting the time taken for a radioactive equilibrium between 222Rn and its daughter 218Po which is detected by the RAD7. The response time for the different GEDs will depend on a number of factors such as (1)
■
RESULTS AND DISCUSSION Advantages and Disadvantages of GEDs. Our laboratory experiments revealed that the six GED were effective in 7687
dx.doi.org/10.1021/es301961b | Environ. Sci. Technol. 2012, 46, 7685−7691
Environmental Science & Technology
Article
Table 1. Summary of Key Issues to Consider When Choosing the Ideal GED for Different Applications shower head marble passive membrane small liqui-cel medium liqui-cel large liqui-cel
response time
water pumping
potential for clogging
potential for biofouling
example of application
short shortest very long long short shortest
yes yes no yes yes yes
low low low very high high high
low high very high high high high
survey and times series in coastal waters survey in turbid waters unattended time series in remote rivers and lakes clean groundwater surveys in clear waters surveys in clear waters
the water flow rate, (2) the air flow rate, (3) the volume of the air loop, and (4) the efficiency of gas transfer between the water and air phase.5,29 Higher water flows will presumably deliver gases faster to the system while higher airflows will promote mixing and delivery to the detector.30 While manipulating all the variables influencing the response time of the different GEDs is out of the scope of this article, previous work using the RAD7 and a shower head exchanger demonstrated that a water flow of 17 L min−1 and an airflow rate of 3 L min−1 can shorten the radon response from
