Lab-on-Chip Measurement of Nitrate and Nitrite for In Situ Analysis of

Jul 26, 2012 - Here we present the first of a new generation of microfluidic chemical analysis systems with sufficient analytical performance and robu...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/est

Lab-on-Chip Measurement of Nitrate and Nitrite for In Situ Analysis of Natural Waters Alexander D. Beaton,†,∥,* Christopher L. Cardwell,† Rupert S. Thomas,‡ Vincent J. Sieben,† François-Eric Legiret,†,§ Edward M. Waugh,† Peter J. Statham,§ Matthew C. Mowlem,† and Hywel Morgan‡ †

National Oceanography Centre, Southampton SO14 3ZH, United Kingdom The Nano Group, University of Southampton, Southampton SO17 1BJ, United Kingdom § Ocean and Earth Science, University of Southampton, Southampton SO14 3ZH, United Kingdom ∥ Bristol Glaciology Centre, University of Bristol, University Road, Bristol BS8 1SS, United Kingdom ‡

ABSTRACT: Microfluidic technology permits the miniaturization of chemical analytical methods that are traditionally undertaken using benchtop equipment in the laboratory environment. When applied to environmental monitoring, these “lab-onchip” systems could allow high-performance chemical analysis methods to be performed in situ over distributed sensor networks with large numbers of measurement nodes. Here we present the first of a new generation of microfluidic chemical analysis systems with sufficient analytical performance and robustness for deployment in natural waters. The system detects nitrate and nitrite (up to 350 μM, 21.7 mg/L as NO3−) with a limit of detection (LOD) of 0.025 μM for nitrate (0.0016 mg/L as NO3−) and 0.02 μM for nitrite (0.00092 mg/L as NO2−). This performance is suitable for almost all natural waters (apart from the oligotrophic open ocean), and the device was deployed in an estuarine environment (Southampton Water) to monitor nitrate+nitrite concentrations in waters of varying salinity. The system was able to track changes in the nitrate−salinity relationship of estuarine waters due to increased river flow after a period of high rainfall. Laboratory characterization and deployment data are presented, demonstrating the ability of the system to acquire data with high temporal resolution.

■

INTRODUCTION High-resolution measurements of nitrate and nitrite are essential for our understanding of biogeochemical nitrogen cycling in aquatic systems. Determination of nutrient levels in natural waters is traditionally performed via manual sample collection and subsequent laboratory analysis. This method is time and labor intensive, and also carries the inherent risks of sample contamination and degradation during transfer and storage. To eliminate these issues, as well as facilitate high temporal and spatial resolution monitoring, several groups have proposed and developed instrumentation to perform near continuous nutrient analysis in situ,1−7 in a range of aquatic environments. Despite these developments, large physical size and/or excessive power and reagent consumption (and in some cases poor levels of detection) limit the ways in which these systems can be deployed. Technologies that enable on-site measurements, including recent advances in the development of oceanographic platforms (e.g., Argo floats,8 autonomous underwater vehicles (AUVs)9 and oceanic gliders10) have provided several opportunities for the widespread deployment of in situ sensor networks for longterm monitoring. Although these platform technologies exist © 2012 American Chemical Society

and are currently in use for physical metrology, the lack of miniaturized in situ chemical analyzers means that measurements on moving platforms have largely, but not exclusively (e.g., see refs 9, 11), been restricted to conductivity, temperature, and dissolved oxygen. While electrochemical techniques,12 capillary electrophoresis,13 chromatography,14 and UV absorption15 methods have been applied to nitrate analysis of natural waters, spectrophotometric methods using color forming reagents have proven to be the most stable, sensitive, and selective, and have formed the basis of several in situ nutrient analyzers for environmental analysis (e.g., refs 1−7). Some in situ wet chemical systems have been made available commercially (e.g., NAS-3x16 and the SubChemPak Analyzer17). Direct UV absorption methods have also widely been used in situ (SUV-6,15 ISUS,18 and ProPS19), and provide high-frequency data (1 Hz) without the need for reagents. While several good data sets have been achieved (e.g., Received: Revised: Accepted: Published: 9548

February 2, 2012 July 14, 2012 July 26, 2012 July 26, 2012 dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

Figure 1. (A) Fluidic path diagram indicating the three syringes, fifteen valves, and three absorption cells. Inputs for standard solutions are indicated as STD. Hydrochloric acid (6%) is denoted as HCl. (B) CAD drawing of the microfluidic chip with fluidic connections labeled.

typically 2 orders of magnitude lower than nitrate. Almost all spectrophotometric methods for nitrate determination in natural waters require its reduction to the more reactive nitrite prior to detection.42 A variety of nitrate reduction methods have been demonstrated, including reduction by hydrazine,43 copperized cadmium,44 zinc,45 and nitrate reductase,46 and irradiation by ultraviolet light.47 Methods using hydrazine, nitrate reductase, and UV light require long wait periods for the reduction to occur, while a zinc reduction column recently demonstrated for estuarine waters by Ellis et al.48 was replaced daily due to degradation. Although cadmium exhibits high toxicity, reduction by copperized cadmium represents the most well established method and is capable of achieving reduction efficiencies approaching 100%14 without requiring long wait periods. Of the spectrophotometric methods available (see 14, 49), the Griess assay (diazotization with sulphanilamide and subsequent coupling with N-(1-naphthyl)-ethylenediamine dihydrochloride (NED) to form an intensely colored azo dye) is established as the mainstay of colorimetric nitrite analysis and is not subject to interferences in oxygenated seawater. We have developed a field-deployable platform for automated in situ colorimetric nitrate analysis using the Griess assay, which is the first of a new generation of miniaturized field analysers based on microfluidic technology. The system is highly configurable (up to six blank corrected measurements per hour) and has low power consumption (1.5 W). The system (without reagents and power supply) is approximately 100 mm (diameter) by 200 mm (length). The use of on-chip sequential absorption cells (2.5 mm and 25 mm) that use colored poly(methylmethacrylate) (PMMA) to exclude background light50 results in a high-sensitivity system with a dynamic range (0.025 to 350 μM) suitable for deployment in a wide variety of natural waters.

9, 11, 20, 21), direct UV absorption systems suffer from lower performance than most wet chemical systems (ISUS V3 precision: ± 0.5 μM, accuracy ±2 μM22), interferences,23 drift,24 and high power consumption (6.5 W for ISUS V322). The application of microfluidics (lab-on-chip technology) to in situ colorimetric nutrient analysis provides several significant advantages over traditional “macro-scale” systems. These include (a) the potential for reduced reagent and power consumption (enabling long-term deployment), (b) reduced size, facilitating deployment on a range of oceanographic platforms (e.g., Argo floats and gliders), and (c) reduced cost (through potential mass-fabrication of microfluidic chips (e.g., 25), making distributed sensor networks a possibility (see 26, 27). An early application of microfluidics to nitrite analysis was demonstrated by Greenway et al.,28 and this was extended to nitrate analysis through work by Petsul et al.29 Since then, spectrophotometric,30−32 chromatographic,33 and electrochemical 34 microfluidic systems for nitrate and/or nitrite analysis have been developed. Despite these advances, reports of field deployments and in situ analysis using microfluidic nutrient analysers are less widespread. Cleary et al.35 described the deployment of a microfluidic phosphate system (with 200-μm channel widths) at numerous locations,35−38 although the LOD of 3.16 μM would make it unsuitable for low-nutrient natural waters. Our previous work described the deployment of a microfluidic analyzer for nitrite (featuring 150-μm wide channels).32 In 1994, Jannasch et al. described an osmotically pumped nitrate analyzer featuring a microconduit flow manifold that was capable of long-term nitrate analysis with a detection limit of 0.1 μM.39 Despite successful deployments, slow response time and a long calibration period (8 h)40 make the sensor more appropriate for long-term, low-resolution measurements on stationary platforms (e.g., 41). Nitrate concentrations in natural waters range from subnanomolar in the oligotrophic open ocean to several hundred micromolar in high-nutrient rivers where agricultural runoff contributes significantly, while in oxic waters nitrite is

■

MATERIALS AND METHODS The microfluidic platform (shown in Figure 1) is based on a circular block of PMMA with a diameter of 100 mm. The 9549

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

Figure 2. Calibration curves with associated error bars (± 1σ) based on three repeats. (A) Calibration data for the long (25 mm) cell for nitrate standards up to 100 μM. Points up to 20 μM are fitted with a linear regression line. (B) Calibration data for the short (2.5 mm) cell for nitrate standards, which remains linear up to 100 μM. (C) Calibration data for the 25-mm cell for low nitrate standards. (D) Calibration data for the 25-mm cell for low nitrite standards. A flow rate of 300 μL/min (per syringe) was used for data in graphs A and B, and a flow rate of 150 μL/min (per syringe) was used for the data in graphs C and D (to maximize reduction efficiency at low concentrations).

the LED and photodiode 14.5 mm apart and using a 10.15-mm long “light tube” to transmit light between the LED and the cell. The light tube is formed by milling into the tinted PMMA substrate and is filled with clear, high refractive index Norland Optical Adhesive 68. For nitrite analysis, fluid is routed through a reference cell then mixed with Griess reagent before passing through a 0.25m-long serpentine mixing channel. Absorption is then determined in the two sequential measurement cells, which are separated by a milled groove to prevent cross-talk. For nitrate analysis, fluid is combined with the imidazole buffer, passed through a 0.46-m serpentine mixer then through an offchip cadmium tube (SEAL analytical, UK) before passing through the reference cell and mixing with the Griess reagent. An off-chip cadmium tube was used during prototyping, but could be fabricated as part of the chip in future (e.g., 39). The cadmium tube can be conditioned periodically52 to recover reduction efficiency by passing 5 mM copper sulfate solution (2 min), 6% HCl (4 min), and the copper sulfate solution again (6 min), which are pumped through syringe 2 (in place of the buffer solution) and selected using dedicated valves. To reduce its internal volume, the cadmium tube was shortened to 160 mm and compressed to a thickness of 1.5 mm so that it had internal volume of 80 μL. Griess reagent and nitrite standards were prepared as described previously.31 Nitrate stock solution (100 mM) was prepared by dissolving 10.11 g of potassium nitrate in 1000 mL of Milli-Q water. Serial dilution with Milli-Q (initial testing) or artificial seawater (further testing and deployment) was performed to make the various working standards. For the second (longer) deployment, 0.1% chloroform was added to preserve standards. The imidazole buffer solution was created

microchip was machined in 5.0-mm-thick tinted PMMA (Plexiglas GS 7F60, Rohm, Darmstadt, Germany) by micromilling (LPKF Protomat S100 micromill). A solvent vapor bonding procedure was used to polish the channel surfaces and to bond the two halves.51 The chip incorporates a fluidic manifold that permits selection of one of four standards (nitrate or nitrite), the sample, and a blank (Milli-Q or artificial seawater). Fluid control is achieved using 15 microinert solenoid valves (LFNA1250325H, Lee Products Ltd., UK) which are mounted directly to the chip. A custom-designed pumping architecture drives fluid through three 2.78-mm internal diameter titanium syringes (syringe 1 for sample/standard, syringe 2 for the buffer solution, and syringe 3 for the Griess reagent). Viton o-rings (Polymax Ltd., UK) are used as moving seals on the pump plungers, which are driven by a Size 11 stepper-motor-based linear actuator (Haydon Kerk, USA). Hall-effect sensors are used for syringe pump feedback and control. Green 525-nm LEDs (HLMP-CM14-Z30DD, Avago Technologies, CA, USA) and TAOS TSLG257-LF photodiodes (TX, USA) were fixed directly to the chip31 using Norland Optical Adhesive 68 (NJ, USA). All on-chip microchannels are 150 μm wide and 300 μm deep, except the optical absorption cells which are 300 μm wide and 300 μm deep. The chip contains three absorption cells: a 25-mm reference cell, a 25-mm measurement cell (for concentrations below 30 μM), and a 2.5-mm measurement cell (for concentrations above 30 μM). The use of dark PMMA reduces the amount of LED background light (i.e., that which did not pass through the fluid in the absorption cell) reaching the photodiode. For the 2.5-cm cell, additional background light rejection is achieved while maximizing the light passing through the cell by spacing 9550

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

Figure 3. Flow-rate (per syringe) vs reduction ratio of the cadmium tube for 100 μM standards with error bars (± 1σ). This was verified with 1 μM and 40 μM standards and found to be concentration independent.

by adding 3.4 g of imidazole (Fisher Scientific, UK) and 1 mL of 10 mM copper sulfate solution to 1000 mL of MQ water and adjusting the pH to 7.8 with concentrated hydrochloric acid. All fluids (including waste) were stored in Flexboy bags (Sartorius Stedim, UK). The system was housed in a darkened water-tight acrylic tube terminated with acetal plastic end-caps and sealed with o-rings. On the bench, the system was automated using a program written in LabVIEW and data were collected using a National Instruments PCI 6289 card. When used in situ the system was controlled using a custom electronics package consisting of eight 18-bit analogue-to-digital inputs (for Hall-effect sensors and photodiodes), four constant-current LED drivers, a stepper motor driver for the syringe pump, a temperature sensor, a real time clock, and 16 spike and hold circuits for the valve solenoids. A microcontroller was programmed to perform all low level operations, with the final data being stored on a 2 GB flash memory card (filtered to 1 Hz). For the operations described here, data were also streamed over an RS-232 connection and monitored in real time. The system was calibrated in the lab by sequentially passing known nitrate and nitrite standards (0.5, 1, 5, 10, 15, 20, 30, 50, 70, and 100 μM) made using artificial seawater (ASW). Absorption was calculated by comparing the optical intensity measured by the photodiode (after a 120 s reaction wait time) for each standard with that of the blank measurement (ASW) using the Beer−Lambert law. Before each measurement, the pump was programmed to flush the system five times with the new sample or standard to be measured, which was necessary to prevent carryover between measurements. For its first in situ deployment the microfluidic analyzer was submerged (to approximately 0.5 m) in Southampton Water for 70 h in April 2010. In this case the deployment period was limited by the volume of reagents that could be contained within the housing. A flow rate of 300 μL/m per syringe was chosen to optimize sampling frequency. The system was deployed with an 80 μM nitrate standard, an 80 μM nitrite standard (in order to monitor reduction efficiency), and an ASW blank. A 0.45-μm pore size Millex HP inline filter (Millipore, USA) was placed on the sample inlet, and salinity and temperature were recorded by an Ocean Seven 320+ CTD probe (Idronaut, Italy) placed next to the sensor. In May 2012 the system was deployed for a period of 26 days. This time, reagent bags were suspended above the outside of the sensor

housing so that they could be accessed and changed weekly without terminating the deployment. During both deployments bottle samples were collected from just below the surface, filtered through a 0.45-μm membrane, frozen, and analyzed on a QUAATRO high performance SFA system (SEAL Analytical Ltd., UK).

■

RESULTS AND DISCUSSION Figure 2 shows calibration curves generated for each of the two measurement cells. At nitrate concentrations higher than 30 μM the 25-mm cell no longer shows a linear relationship between absorption and nitrate concentration (due to nonlinearity of the Beer−Lambert law at high absorption values). For measurements above 30 μM the 2.5-mm cell can be used (which is linear up to 350 μM). Calibration curves using low concentration nitrate and nitrite standards (0.05, 0.1, 015, 0.2, and 0.25 μM) are shown in Figure 2C and D. All data are in triplicate and error bars show the standard deviation. Limit of detection (LOD) is normally defined as three times the standard deviation of the blank measurement.53 In the case of this analyzer, where each sample or standard measurement has an associated blank from which absorption is calculated, LOD is governed by point-to-point variations in the absorption measurements of known standards. To estimate the limit of detection, ten subsequent blank-corrected absorption measurements of 0.05 μM nitrate and nitrite standards were performed. The LOD was taken as three times the standard deviation of these sequential absorption measurements (taken over a 3-hour period). This results in LODs of 0.025 μM for nitrate and 0.02 μM for nitrite. The efficiency of a cadmium reductor in reducing nitrate to nitrite is dependent on several factors (e.g., residence time (i.e., flow rate), temperature, salinity, and pH54,55). To determine the optimum flow rate for reduction, equimolar nitrite and nitrate standards were passed sequentially through the cadmium tube at seven different flow rates, and the ratio between the measured absorption values for each set of nitrate and nitrite measurements was used to calculate the measured reduction ratio as a percentage (see Figure 3). The flow rate for optimum reduction (where the ratio between nitrate and nitrite measurements is 100%) was found to be 150 μL/min per syringe. This produces a total flow rate of 300 μL/min through the cadmium tube and 450 μL/min through the measurement 9551

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

Figure 4. Estuarine data from 70-h test deployment in Southampton Water. A. Absorption values for the sample, 80 μM nitrate standard, and 80 μM nitrite standard as a function of time, with water height (i.e., the tide) shown as a green line (data from SOTONMET, Associated British Ports). B. Absorption data converted to concentration using on-board standard measurements. Three clear outliers (visible in Figure 4A) were removed. Also plotted are manually collected bottle samples. C. Temperature and salinity. D. Mixing diagram plotting each nitrate+nitrite measurement against salinity (and linear regression line).

cells (when the flow of the sample, buffer, and Griess reagent is combined). Reduction ratios over 100% at flow rates lower than 150 μL/ min indicate that over-reduction of nitrite (to hydroxylamine and ammonia) is occurring (see 55), resulting in an undermeasurement of the true nitrite value. While low flow rates that lead to over reduction will give erroneous measurements and should be avoided, the use of on-board calibration standards means that accurate measurements can still be made at flow rates that are higher than the optimum. For example, using a flow rate higher than 150 μL/min would increase the sampling frequency (by pumping faster) at the expense of color development (i.e., sensitivity). Figure 4 shows data from the first in situ deployment in Southampton Water, during which the system performed 284 blank-corrected nitrate plus nitrite measurements (142 estuarine water samples and 142−80 μM standard measurements) and 19 blank-corrected 80 μM nitrite standard measurements. Absorption measurements taken by the 2.5mm cell for the samples and each of the standards are shown in Figure 4A and absorption values converted into concentration using each corresponding measurement of the on-board nitrate standard are shown in Figure 4B. Three clear outliers have been removed for the concentration plots, but included in for absorption plots (these were possibly caused by bubbles or debris residing in the flow cell for short periods). Outliers identified in the raw data by anomalously high or low intensity measurements for either blank or standard are therefore easy to flag up and exclude. Figure 4C shows salinity and temperature plots for the deployment period.

During the 70-h test deployment, nitrate+nitrite concentrations measured by the sensor ranged from 34 to 102 μM, and were influenced by the tide. The standards showed an average reading of 0.114 absorption units (AU) and a standard deviation of 0.010. The downward trend in the absorption values of the nitrate standard (estimated as 1.2 × 10−4 AU per hour) is likely the result of a gradual reduction in the efficiency of the cadmium tube. After the deployment, the efficiency of the cadmium tube was recovered in situ using HCl and copper sulfate solutions. Southampton Water is a tidal estuary which receives nutrient rich freshwater inputs from two main rivers (the Test and the Itchen) with an annual discharge equivalent to 1.54 × 106 m3/ day. In addition Southampton Water receives a wastewater input of 0.1 × 106 m3/day.56 Nitrate and nitrite concentrations are affected by the tide, which exhibits a double high water with peaks approximately 2 hours apart. Nitrate concentrations are also affected by changes in runoff caused by precipitation events. Nitrate is reported to behave conservatively57 in Southampton Water,56 except during the summer months when nitrate levels have been reported to drop below detection limits at salinities greater than 34 (i.e., not within our measurement range) as a result of biological production. Nitrite also generally behaves conservatively, although it can be affected by wastewater inputs (sewage plants) at point sources along the estuary. Figure 4D shows a mixing diagram demonstrating the relationship between salinity and measured nitrate+nitrite concentrations for the 70-h test deployment. Assuming there are no additional processes occurring in the estuary that result in the addition or removal of nitrate (i.e., 9552

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

Figure 5. Estuarine data from 26-day deployment in Southampton Water. (A) Processed nitrate+nitrite data and bottle samples (red crosses). Ten clear outliers were removed from this data set. (B) Salinity. (C) Temperature. (D) Daily discharge data for the River Test (Environment Agency) and daily rainfall data for Southampton (taken from www.southamptonweather.co.uk). (E) Water height, SOTONMET, Associated British Ports.

Figure 6. Mixing diagram containing all measurements from the 26-day deployment separated into four sections representing days 1−10 (before any major rainfall, black diagonal crosses), days 11−17 (first large rainfall, blue crosses), days 18−21 (second large rainfall, pink circles) and days 22−26 (minimal rainfall, green squares). Corresponding regression lines show the evolution of the nitrate−salinity relationship due to changing endmember concentration.

conservative mixing57), the riverine end-member nitrate concentration can be estimated using linear regression. The linear regression (NOx = −13(± 1)S + 461(± 17), where S = salinity and NOx = nitrate+nitrite concentration) predicts an end-member riverine nitrate concentration of 461 μM (± 17 μM). This agrees favorably with previous measurements for this time of year (the SONUS report56,58).

When the system was redeployed at the same location in Southampton Water on May 25, 2012 it operated continuously for more than 26 days and performed 5616 individual measurements (1872 blank measurements, 1872 80-μM standard measurements, and 1872 estuarine water measurements). The processed nitrate+nitrite time series, salinity data, and temperature data are presented in Figure 5. Rainfall data from a weather station in Southampton (indicative of rainfall 9553

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

While the system described here is intended to prove the concept of high-performance in situ microfluidic nutrient analysis, simple fluidic modifications could further reduce reagent consumption and increase temporal response in future design iterations. For example, including an extra valve would allow Griess reagent to be returned to the storage bag on all but the final stroke of each flush cycle, reducing reagent consumption 5-fold. The implementation of a multiplex stop flow regimeMSF (recently demonstrated by Ogilvie et al.65)could allow multiple analyses to occur simultaneously, increasing the temporal response while retaining the high performance of stop flow measurements. While nitrite standards are unstable due to oxidation, nitrate standards can be preserved for several months using a biocide (e.g., 0.1% chloroform66). Griess reagent must be protected from light, but will eventually (after several weeks) develop color and increase its background absorption. Griess reagent can be preserved for longer (e.g., several months40) by separating the sulphanilamide and NED components and mixing them in situ, a task that could be performed on a microfluidic chip. The use of rapid prototyping tools for chip fabrication (such as micromilling used here) allows fluidic design iterations to proceed relatively quickly. While microfluidic nitrate and nitrite analysis using lab-onchip technology has previously been demonstrated on the bench,29 this is the first report of an in situ stand-alone system deployed in the field. The system performs high-resolution measurements of nitrate plus nitrite, as well as just nitrite, and has a performance (LOD 0.025 μM for nitrate, 0.02 μM for nitrite) superior to other in situ nitrate systems (e.g., UV absorption). The small size and low power consumption make it suitable for integration into a range of oceanographic platforms including AUVs and buoy systems. In the future, the integration of on-chip pumps and valves will allow further reductions in size and fluidic dead volumes. The deployment of this new device in an estuarine system begins to demonstrate the potential of lab-on-chip nutrient analysis for inclusion in large-scale networks for ocean observation and monitoring of other natural water systems.

over the river catchment area) and discharge data for the River Test (one of the two main rivers draining into Southampton Water) are also presented in Figure 5 and allow us to link changes in the nitrate+nitrite and salinity time series to changes in runoff caused by precipitation events. Figure 5D shows how a significant rainstorm event on day 17 was followed by a marked increase in river discharge. Figure 5B shows that this was accompanied by an influx of fresh water (decrease in salinity) which lasted for approximately 5 days. The salinity deviation was accompanied by a corresponding large increase in nitrate+nitrite levels measured by the lab-on-chip analyzer (Figure 5A), with nitrate+nitrate concentration peaking at 188 μM on day 22. The tidal data in Figure 5E show that this occurred on a neap tide, during which the flushing time for the estuary has been estimated at 76 h (compared to 26 h on a spring tide59). Deviations in estuarine salinity and nitrate concentrations following large rainfall events are well documented (e.g., 60, 61), and increases in riverine nutrient flux due to storm events have been shown to produce favorable conditions for phytoplankton growth.62,63 While measurements by the labon-chip analyzer show an increase in nitrate concentrations which correspond to the influx of freshwater following a large rainfall event, the mixing diagram in Figure 6 also shows how the nitrate−salinity relationship evolved in response to changes in the riverine end-member concentration. To aid interpretation the time series has been separated into four sections which are plotted with different markers on the mixing diagram. The initial section (days 1−10) represents a period of relatively constant river discharge (before any major rainfall), and the mixing diagram predicts a riverine end-member concentration of 577 ± 28 μM. After the first period of rainfall (days 11−17) the predicted riverine end-member drops to 477 ± 19 μM, and then to 336 ± 15 μM (days 18−21) after the second large rainfall (and influx of high-nitrate, low-salinity water). For the remainder of the time series (days 22−26) the predicted endmember increases again to an average of 451 ± 20 μM. This behavior is consistent with dilution of end-member waters due to increased river discharge associated with rainfall events, and is well documented elsewhere (e.g., 64). The data presented here show the clear advantages of high-resolution in situ monitoring over low-resolution sampling, which could struggle to decouple short-term variability due to semidiurnal tidal signals from other influencing factors such as episodic rainfall events. When operated continuously, the sensor consumed 0.088 L of Griess reagent, 0.088 L of buffer solution, and 0.029 L of standard solution per 24-h period. Each individual measurement required 300 s (which includes 180 s for the five flush cycles and a 120 s wait time for the color-forming reaction to occur), allowing a maximum of 6 blank-corrected measurements per hour. In the deployments described here, any drift due to changes in the reduction efficiency of the cadmium tube or temperature effects on the color forming reaction was minimized by analyzing regular calibration standards. The filter was not changed throughout the 70-h deployment, during which it filtered 0.084 L of estuarine water. For the 26day deployment, the filter was changed weekly (when the reagent bags were swapped) as a precautionary measure. Previous testing in Southampton Water showed the Millex HP filter capable of filtering 2.5 L of estuarine water before becoming blocked, equivalent to 12 weeks of continuous deployment at similar suspended particular matter loadings.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; phone: +442380599689. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We gratefully acknowledge the following research funding: EPSRC/NERC EP/E016774/1, NERC Oceans 2025 Theme 8.1, EPSRC EP/D057620 (studentship to A.B.), and EU FP7 SENSEnet (PITN-GA-2009-237868).

■

REFERENCES

(1) Johnson, K.; Sakamoto-Arnold, C.; Beehler, C. Continuous determination of nitrate concentrations in situ. Deep-Sea Res., Part A 1989, 36, 1407−1413. (2) Vuillemin, R.; Le Roux, D.; Dorval, P.; Bucas, K.; Sudreau, J. P.; Hamon, M.; Le Gall, C.; Sarradin, P. M. CHEMINI: A new in situ CHEmical MINIaturized analyzer. Deep-Sea Res., Part I 2009, 56, 1391−1399. (3) Daniel, A.; Birot, D.; Blain, S.; Treguer, P.; Leilde, B.; Menut, E. A submersible flow-injection analyser for the in-situ determination of nitrite and nitrate in coastal waters. Mar. Chem. 1995, 51, 67−77.

9554

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

using an in situ ultraviolet spectrophotometer. Limnol. Oceanogr.: Methods 2009, 7, 132−143. (24) Christensen, J. P.; Melling, H. Correcting Nitrate Profiles Measured by the In Situ Ultraviolet Spectrophotometer in Arctic Ocean Waters. Open Oceanogr. J. 2009, 59−66. (25) Li, H.-F.; Lin, J.-M.; Su, R.-G.; Cai, Z. W.; Uchiyama, K. A polymeric master replication technology for mass fabrication of poly(dimethylsiloxane) microfluidic devices. Electrophoresis 2005, 26, 1825−33. (26) Johnson, K. S.; Needoba, J. A.; Riser, S. C.; Showers, W. J. Chemical sensor networks for the aquatic environment. Chem. Rev. 2007, 107, 623−40. (27) Diamond, D.; Coyle, S.; Scarmagnani, S.; Hayes, J. Wireless sensor networks and chemo-/biosensing. Chem. Rev. 2008, 108, 652− 79. (28) Greenway, G.; Petsul, P. H.; Haswell, S. J. Characterisation of a micro-total analytical system for the determination of nitrite with spectrophotometric detection. Anal. Chim. Acta 1999, 387, 1−10. (29) Petsul, P. H.; Greenway, G. M.; Haswell, S. J. The development of an on-chip micro-flow injection analysis of nitrate with a cadmium reductor. Anal. Chim. Acta 2001, 428, 155−161. (30) Baeza, M.; del, M.; Ibanez-Garcia, N.; Baucells, J.; Bartrolí, J.; Alonso, J. Microflow injection system based on a multicommutation technique for nitrite determination in wastewaters. Analyst 2006, 131, 1109−15. (31) Sieben, V. J.; Floquet, C. F. A.; Ogilvie, I. R. G.; Mowlem, M. C.; Morgan, H. Microfluidic colourimetric chemical analysis system: Application to nitrite detection. Anal. Methods 2010, 2, 484. (32) Beaton, A. D.; Sieben, V. J.; Floquet, C. F. A.; Waugh, E. M.; Abi Kaed Bey, S.; Ogilvie, I. R. G.; Mowlem, M. C.; Morgan, H. An automated microfluidic colourimetric sensor applied in situ to determine nitrite concentration. Sens. Actuators, B 2011, 156, 1009− 1014. (33) Murrihy, J.; Breadmore, M.; Tan, A.; Mcenery, M.; Alderman, J.; Omathuna, C.; Oneill, A.; Obrien, P.; Advoldvic, N.; Haddad, P. Ion chromatography on-chip. J. Chromatogr., A 2001, 924, 233−238. (34) Shiddiky, M. J. a; Won, M.-S.; Shim, Y.-B. Simultaneous analysis of nitrate and nitrite in a microfluidic device with a Cu-complexmodified electrode. Electrophoresis 2006, 27, 4545−54. (35) Cleary, J.; Slater, C.; McGraw, C.; Diamond, D. An Autonomous Microfluidic Sensor for Phosphate: On-Site Analysis of Treated Wastewater. IEEE Sens. J. 2008, 8, 508−515. (36) Slater, C.; Cleary, J.; Lau, K.-T.; Snakenborg, D.; Corcoran, B.; Kutter, J. P.; Diamond, D. Validation of a fully autonomous phosphate analyser based on a microfluidic lab-on-a-chip. Water Sci. Technol. 2010, 61, 1811−8. (37) Slater, C.; Cleary, J.; McGraw, C. M.; Yerazunis, W. S.; Lau, K.T.; Diamond, D. Autonomous field-deployable device for the measurement of phosphate in natural water. Proc. SPIE 2007, 67550L.1−67550L.8. (38) McGraw, C. M.; Stitzel, S. E.; Cleary, J.; Slater, C.; Diamond, D. Autonomous microfluidic system for phosphate detection. Talanta 2007, 71, 1180−5. (39) Jannasch, H. W.; Johnson, K. S.; Sakamoto, C. M. Submersible, Osmotically Pumped Analyzer for Continuous Determination of Nitrate in situ. Anal. Chem. 1994, 66, 3352−3361. (40) Dickey, T.; Frye, D.; Jannasch, H.; Boyle, E.; Manov, D.; Sigurdson, D.; Mcneil, J.; Stramska, M.; Michaels, A.; Nelson, N. Initial results from the Bermuda Testbed Mooring program. Deep-Sea Res., Part I 1998, 45, 771−794. (41) McNeil, J. D.; Jannasch, H. W.; Dickey, T.; McGillicuddy, D.; Brzezinski, M.; Sakamoto, C. M. New chemical, bio-optical and physical observations of upper ocean response to the passage of a mesoscale eddy off Bermuda. J. Geophys. Res. 1999, 104, 15537− 15548. (42) Patey, M.; Rijkenberg, M.; Statham, P.; Stinchcombe, M.; Achterberg, E.; Mowlem, M. Determination of nitrate and phosphate in seawater at nanomolar concentrations. TrAC, Trends Anal. Chem. 2008, 27, 169−182.

(4) Le Bris, N.; Sarradin, P.-M.; Birot, D.; Alayse-Danet, A.-M. A new chemical analyzer for in situ measurement of nitrate and total sulfide over hydrothermal vent biological communities. Mar. Chem. 2000, 72, 1−15. (5) Thouron, D.; Vuillemin, R.; Philippon, X.; Lourenço, A.; Provost, C.; Cruzado, A.; Garçon, V. An autonomous nutrient analyzer for oceanic long-term in situ biogeochemical monitoring. Anal. Chem. 2003, 75, 2601−9. (6) Clinch, J. R.; Worsfold, P. J. An automated spectrophotometric field monitor for water quality parameters: Determination of nitrate. Anal. Chim. Acta 1987, 200, 523−531. (7) Adornato, L. R.; Kaltenbacher, E. A.; Greenhow, D. R.; Byrne, R. H. High-resolution in situ analysis of nitrate and phosphate in the oligotrophic ocean. Environ. Sci. Technol. 2007, 41, 4045−52. (8) Gould, J.; Roemmich, D.; Wijffels, S.; Freeland, H.; Ignaszewsky, M.; Jianping, X.; Pouliquen, S.; Desaubies, Y.; Send, U.; Radhakrishnan, K.; Takeuchi, K.; Kim, K.; Danchenkov, M.; Sutton, P.; King, B.; Owens, B.; Riser, S. Argo Profiling Floats Bring New Era of In Situ Ocean Observations. Trans. Am. Geophys. Union 2004, 85. (9) Johnson, K. S.; Needoba, J. A. Mapping the spatial variability of plankton metabolism using nitrate and oxygen sensors on an autonomous underwater vehicle. Limnol. Oceanogr. 2008, 53, 2237− 2250. (10) Webb, D. C.; Simonetti, P. J.; Jones, C. P. SLOCUM: An Underwater Glider Propelled by Environmental Energy. IEEE J. Oceanic Eng. 2001, 26, 447−452. (11) Johnson, K. S.; Riser, S. C.; Karl, D. M. Nitrate supply from deep to near-surface waters of the North Pacific subtropical gyre. Nature 2010, 465, 1062−5. (12) Davis, J.; Moorcroft, M. J.; Wilkins, S. J.; Compton, R. G.; Cardosi, M. F. Electrochemical detection of nitrate and nitrite at a copper modified electrode. Analyst 2000, 125, 737−742. (13) Kaniansky, D.; Zelensky, I.; Hybenova, A.; Onuska, F. I. Determination of chloride, nitrate, sulfate, nitrite, fluoride, and phosphate by on-line coupled capillary isotachophoresis-capillary zone electrophoresis with conductivity detection. Anal. Chem. 66, 4258-4264. (14) Moorcroft, M. J.; Davis, J.; Compton, R. G. Detection and determination of nitrate and nitrite: a review. Talanta 2001, 54, 785− 803. (15) Finch, M. S.; Hydes, D. J.; Clayson, C. H.; Weigl, B.; Dakin, J.; Gwilliam, P. A low power ultra violet spectrophotometer for measurement of nitrate in seawater: introduction, calibration and initial sea trials. Anal. Chim. Acta 1998, 377, 167−177. (16) EnviroTech LLC Website; http://envirotechinstruments.com/ files/datasheets/nas3x.pdf. (17) Hanson, A. K. Chemical Analyzer for Mapping Coastal Nutrient Distributions in Real Time. Oceans 2000 IEEE - Where Marine Science and Technology Meet 2000, 1975−1982. (18) Johnson, K.; Coletti, L. In situ ultraviolet spectrophotometry for high resolution and long-term monitoring of nitrate, bromide and bisulfide in the ocean. Deep-Sea Res., Part I 2002, 49, 1291−1305. (19) Zielinski, O.; Fiedler, B.; Heuermann, R.; Kortzinger, A.; Kopiske, E.; Meinecke, G.; Munderloh, K. A new nitrate continuous observation sensor for autonomous sub-surface applications: Technical design and first results. OCEANS 2007 - Europe 2007, 1−4. (20) Pidcock, R.; Srokosz, M.; Allen, J.; Hartman, M.; Painter, S.; Mowlem, M.; Hydes, D.; Martin, A. A Novel Integration of an Ultraviolet Nitrate Sensor On Board a Towed Vehicle for Mapping Open-Ocean Submesoscale Nitrate Variability. J. Atmos. Oceanic Tech. 2010, 27, 1410−1416. (21) Sandford, R. C.; Exenberger, A.; Worsfold, P. J. Nitrogen Cycling in Natural Waters using In Situ, Reagentless UV Spectrophotometry with Simultaneous Determination of Nitrate and Nitrite. Environ. Sci. Technol. 2007, 41, 8420−8425. (22) Satlantic Incorporated. MBARI-ISUS V3 Operation Manual, 2010. (23) Sakamoto, C. M.; Johnson, K. S.; Coletti, L. J. Improved algorithm for the computation of nitrate concentrations in seawater 9555

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556

Environmental Science & Technology

Article

(43) Mullin, J. B.; Riley, J. P. The Spectrophotometric Determination of Nitrite In Natural Waters With Particular Reference To Sea-Water. Anal. Chim. Acta 1955, 12. (44) Wood, E. D.; Armstrong, F. A. J.; Richards, F. A. Determination Of Nitrate In Sea Water By Cadmium-Copper Reduction To Nitrite. J. Mar. Biol. Assoc. U.K. 1967, 47, 23−31. (45) Lambert, J. L.; Zitomer, F. Differential Colorimetric Determination of Nitrite and Nitrate Ions. Anal. Chem. 1960, 32, 1684−1686. (46) Patton, C. J.; Fischer, A. E.; Campbell, W. H.; Campbell, E. R. Corn leaf nitrate reductase–a nontoxic alternative to cadmium for photometric nitrate determinations in water samples by air-segmented continuous-flow analysis. Environ. Sci. Technol. 2002, 36, 729−35. (47) Takeda, K.; Fujiwara, K. Determination of nitrate in natural waters with the photo-induced conversion of nitrate to nitrite. Anal. Chim. Acta 1993, 276, 25−32. (48) Ellis, P. S.; Shabani, A. M. H.; Gentle, B. S.; McKelvie, I. D. Field measurement of nitrate in marine and estuarine waters with a flow analysis system utilizing on-line zinc reduction. Talanta 2011, 84, 98− 103. (49) Dutt, J.; Davis, J. Current strategies in nitrite detection and their application to field analysis. J. Environ. Monit. 2002, 4, 465−471. (50) Floquet, C. F. A.; Sieben, V. J.; Milani, A.; Joly, E. P.; Ogilvie, I. R. G.; Morgan, H.; Mowlem, M. C. Nanomolar detection with high sensitivity microfluidic absorption cells manufactured in tinted PMMA for chemical analysis. Talanta 2010, 84, 235−239. (51) Ogilvie, I. R. G.; Sieben, V. J.; Floquet, C. F. a; Zmijan, R.; Mowlem, M. C.; Morgan, H. Reduction of surface roughness for optical quality microfluidic devices in PMMA and COC. J. Micromech. Microeng. 2010, 20, 065016. (52) Bran+Luebbe. QuAAtro Applications Method No. Q-068-05 Rev. Two (Nitrate and Nitrite in Water and Seawater). (53) Long, G. L.; Winefordner, J. D. Limit of Detection: A Closer Look at the IUPAC Definition. Anal. Chem. 1983, 55, 712−714. (54) Nydahl, F. On the optimum conditions for the reduction of nitrate to nitrite by cadmium. Talanta 1976, 23, 349−57. (55) Gal, C.; Frenzel, W.; Muller, J. Re-Examination of the Cadmium Reduction Method and Optimisation of Conditions for the Determination of Nitrate by Flow Injection Analysis. Microchim. Acta 2004, 146, 155−164. (56) Hydes, D. Nutrients in the Solent. In Solent Science - A Review; Collins, N.; Ansell, K., Eds.; Elsevier, 2000; pp 135−148. (57) Morris, A. W. Estuarine chemistry and general survey strategy. In Practical Estuarine Chemistry: A Handbook; Head, P. C., Ed.; Cambridge University Press, 1985; pp 1−60. (58) Hydes, D. J.; Wright, P. N. SONUS: The SOthern NUtrients Study 1995−1997; Southampton Oceanography Centre Report No. 7; Southampton Oceanography Centre: Southampton, U.K., 1999; 130 pp, http://eprints.soton.ac.uk/19118/1/socr007.pdf (accessed August 17, 2012). (59) Wright, P. N.; Hydes, D. J.; Lauria, M. L. O. U.; Sharples, J.; Purdie, D. A. Data Buoy Measurements of Phytoplankton Dynamics in Southampton Water, UK, a Temperate Latitude Estuary with High Nutrient Inputs. Dtsch. Hydrogr. Z. 1997, 49, 203−211. (60) Hubertz, E. D.; Cahoon, L. B. Short-term Variability of Water Quality Parameters in Two Shallow Estuaries of North Carolina. Estuaries 1999, 22, 814−823. (61) Mallin, M.; Paerl, H.; Rudek, J.; Bates, P. Regulation of estuarine primary production by watershed rainfall and river flow. Mar. Ecol.: Prog. Ser. 1993, 93, 199−203. (62) Garcia-Soto, C.; de Madariaga, I.; Villate, F.; Orive, E. Day-today variability in the plankton community of a coastal shallow embayment in response to changes in river runoff and water turbulence. Estuarine, Coastal Shelf Sci. 1990, 31, 217−229. (63) Malej, A.; Mozetie, P.; Malatit, V.; Terzic, S.; Ahel, M. Phytoplankton responses to freshwater inputs in a small semi-enclosed gulf (Gulf of Trieste, Adriatic Sea). Mar. Ecol.: Prog. Ser. 1995, 120, 111−121.

(64) Balls, P. W.; Macdonald, a.; Pugh, K. B.; Edwards, a. C. Rainfall events and their influence on nutrient distributions in the Ythan Estuary (Scotland). Estuarine, Coastal Shelf Sci. 1997, 44, 73−81. (65) Ogilvie, I. R. G.; Sieben, V. J.; Mowlem, M. C.; Morgan, H. Temporal Optimization of Microfluidic Colorimetric Sensors by Use of Multiplexed Stop-Flow Architecture. Anal. Chem. 2011, 83, 4814− 4821. (66) McCormack, T.; David, A. R. J.; Worsfold, P. J.; Howland, R. Flow injection determination of nitrate in estuarine and coastal waters. Anal. Proc. incl. Anal. Commun. 1994, 31, 81.

9556

dx.doi.org/10.1021/es300419u | Environ. Sci. Technol. 2012, 46, 9548−9556