Active Sampling Device for Determining Pollutants in Surface and Pore Water – the In Situ Sampler for Biphasic Water Monitoring

We designed and evaluated an active sampling device, using as analytical targets a family of pesticides purported to contribute to honeybee colony collapse disorder. Simultaneous sampling of bulk water and pore water was accomplished using a low-flow, multi-channel pump to deliver water to an array of solid-phase extraction cartridges. Analytes were separated using either liquid or gas chromatography, and analysis was performed using tandem mass spectrometry (MS/MS). Achieved recoveries of fipronil and degradates in water spiked to nominal concentrations of 0.1, 1, and 10 ng/L ranged from 77 ± 12 to 110 ± 18%. Method detection limits (MDLs) were as low as 0.040–0.8 ng/L. Extraction and quantitation of total fiproles at a wastewater-receiving wetland yielded concentrations in surface water and pore water ranging from 9.9 ± 4.6 to 18.1 ± 4.6 ng/L and 9.1 ± 3.0 to 12.6 ± 2.1 ng/L, respectively. Detected concentrations were statistically indistinguishable from those determined by conventional, more laborious techniques (p > 0.2 for the three most abundant fiproles). Aside from offering time-averaged sampling capabilities for two phases simultaneously with picogram-per-liter MDLs, the novel methodology eliminates the need for water and sediment transport via in situ solid phase extraction.


GC-MS/MS calibration in hexane. 26
Pump performance. The IS2B peristaltic pump was calibrated prior to each analysis. Two 27 replicate benchtop tests were performed to ascertain the precision of the pump. Results are 28 shown in supplementary table S1. Multiple trials with this device indicated that one type of 29 tubing (PharMed) provided greater consistency in pump performance than did others (e.g., 30 Viton), probably due to the tendency of the latter to deform permanently when pinched by the 31 pump rollers. During the recovery tests, the ISMATEC control unit was set to deliver 200 mL at 32 a pump rate of 140 µL/min / channel to each of the six channels in two consecutive runs (n = 12). 33 Sample collection. Sediment field blanks were collected from locations about 50 yards from the 34 edge of the wetland. The sediment was not impacted by the wastewater effluent, and was S3 therefore used as a quality control. Water field blanks were DI water samples transported from 36 Arizona State University to the wetland, and transferred there into ashed media bottles. 37 Analytics. Calibration standard response accuracy had to be within 20% of expected values. 38 Level 1 QA/QC for quantitation of fiproles was performed using lab control spikes. The absolute 39 recovery of spiked mass was compared to "clean" calibration standards in 1:1 acetonitrile:water 40 (for LC-MS/MS analysis) or 100% hexane (for GC-MS/MS analysis), and these results are 41 displayed in supplementary table S2. Unspiked equipment blanks were used as controls, and the 42 method of quantitation required subtraction of the equipment blank signal from that of the spiked 43

samples. 44
Recovery tests. Water laden with dissolved organic carbon (DOC) was generated by adding 100 45 mg potassium citrate to 3 L of 18.2 MΩ (Milli-Q) water. The water was spiked to 300 ppm (v/v) 46 with Kathon CG/ICP biocide and stored at room temperature in ashed amber media bottles. 2000 47 mL was transferred to a 2-L ashed media bottle and was spiked with 20 ng (nominal 48 concentration 10 ng/L) of the fipronil parent compound, along with the sulfide, sulfone, and 49 amide degradates. A separate 2-L sample of water was spiked with 2 ng (nominal concentration 50 1 ng/L) with fipronil-desulfinyl. Both samples were extracted in separate tests as described 51

below. 52
For bench top extraction, the sampler was assembled with two 3-channel PTFE manifolds for 53 water inlet, and six 1-mL SDB-L SPE cartridges (25 mg of resin), conditioned and rinsed with 54 acetonitrile and LCMS grade water, respectively. Both IS2B inlet tubes were placed into the 55 spiked lab-created water with the IS2B control unit set to deliver 200 mL at 140 µL/min/channel. 56 The effluent tubes from the SPE cartridge were each placed into separate weighed 1000 mL S4 media bottles. At the end of the pumping period, the SPE cartridges were rinsed with 1 mL 58 LCMS water, and eluted with 1 mL of acetonitrile, followed by 1 mL of 1:1 hexane:acetone. The 59 serial eluates from each channel were combined, divided into two 1 mL aliquots, evaporated 60 under nitrogen, and one set of aliquots was reconstituted to 1 mL of acetonitrile (ACN), while 61 the other was reconstituted to 1 mL hexane. The resulting ACN solutions were diluted by 50% 62 with water, and the ACN/ After the recovery test, the pump calibration was assessed by comparing the set volume on the 71 control unit with the volumes collected in the effluent capture bottles. The volumes were 72 determined by dividing the mass difference between the empty and full bottles by the density of 73

water. 74
A similar procedure was used to determine the recovery efficiency using an AutoTrace 280 by 75 Dionex (Sunnyvale, CA). Because the AutoTrace was equipped with 3 mL rather thatn 1 mL 76 adaptors, it was loaded with 500 mg/3mL SDB cartridges (8 replicates total), which were 77 conditioned as described above. Benchtop comparisons indicated that the 25 mg and 500 mg 78 resin beds performed similarly, but 500 mg cartridges were on hand for this study. 200 mL of 79 spiked DOC-laden water with 1 ng/L of targets was loaded onto each cartridge at 1 mL/min, and S5 eluted serially with 2 mL of acetonitrile and 2 mL of hexane:acetone (1:1) at 1 mL/min. The 81 eluates were commingled and blown down to dryness under nitrogen before being reconstituted 82 to 2 mL of acetonitrile. These samples were split for GC-MS/MS analysis and LC-MS/MS 83 analysis. LC samples were diluted by 50% with LCMS grade water prior to analysis. GC 84 samples were solvent switched to hexane prior to analysis. 85

Method Detection Limit.
A sample of lab-generated water (as described above) was used to 86 determine the baseline signal for each analyte. Nine replicate samples were generated, and two 87 were subsequently omitted, resulting in six degrees of freedom. The method detection limit 88 (MDL) was calculated as described by the Environmental Protection Agency. 1 This method was 89 used to determine the MDL using both the AutoTrace and IS2B preconcentration devices. Since 90 the IS2B and AutoTrace each have six channels, the process was run twice: once with three 91 spiked replicates and three unspiked controls, and once with six spiked replicates. A student's t-92 value (99% confidence interval) of 3.14 was used, and was multiplied by the standard deviation 93 of 7 replicates. The calculated MDLs were checked against the following criteria: 94

Instruments and analysis
Bellefonte, PA). The mobile phase consisted of 40% acetonitrile and 60% water flowing at a rate 133 of 400 µL/min with a total runtime of 12 min, with a gradient profile of 10% ACN/min starting 134 at t = 1.00 min. Analytes were introduced into the mass spectrometer using an electrospray 135 ionization probe operating in negative mode, and multiple reaction monitoring (MRM) was used 136 for qualitative analysis. Optimized conditions for the ionization and fragmentation of the 137 analytes are specified below. Quantitation was performed using a 5 point calibration curve in 1:1 138 acetonitrile:water. GC mass spectrometric analysis was performed using an Agilent 7890 gas 139 chromatograph coupled to an Agilent 7000 triple quadrupole mass spectrometer (Agilent 140 Technologies, Santa Clara, CA) operating in positive mode, and MRM was used for qualitative 141 analysis. Absolute recovery of all compounds was performed by using 4-or 5-point calibration 142 curves and subtracting the concentration in the unspiked matrices from those of the spiked 143 matrices. Equipment blanks using 18.2 MΩ (Milli-Q) water were run prior to all deployments, 144 and grab sample controls included field blanks of Milli-Q water.