OR/17/011 Method
| White, D, Williams, P J, Civil, W, and Lapworth, D J. 2017. A field based method for pre-concentration of micro organics using solid phase extraction. British Geological Survey Internal Report, OR/17/011. |
Sample site collection
The objective of the initial trial was to test how stable a selection of compounds were on the cartridges and if the compounds detected in a conventional bottled water sample reflected those sorbed onto the cartridge. Sample sites OSP, East Hanney and East Shefford, were all surface waters receiving input from sewage treatment works of differing sizes, two of these were in rural areas and the other in an urban area. The groundwater sample from Boxford was from a previously well characterised borehole in a rural area. These sites were selected to find a wide range of contaminants.
Sample handling precautions
All through the process, from sample collection to pre-concentration onto the cartridges all precautions were taken to reduce the possibility of contaminating the samples with personal care products including insect repellents, sun screens, medicated ointments, as well as vehicle fuel, grease or any other possible contaminant.
Samples were collected in clean glass bottles supplied by the National Laboratory Services (NLS) which were inspected for cleanliness prior to use. No bottles were re-used. Bottles were stored away from sources of contamination with the lids tightly closed. Bottles were labelled using only the labels provided by NLS and no permanent maker was used.
Sample collection methods
For the field trial, samples from surface waters were either collected directly from the stream or by using a pump where it was either too hazardous or difficult to do so. A Solinst 410 peristaltic pump was used together with pump tubing and rigid HDPE tubing that had previously been washed in a solution of Virkon and rinsed with deionised water. During the initial field trial each site had dedicated tubing to prevent carry over.
For the groundwater sample a Waterra 3-stage pump was used to purge the borehole then samples were taken with the Solinst peristaltic pump with dedicated, Virkon washed and deionised water rinsed tubing. Each surface water site sampled during the initial study had dedicated, washed pump and HDPE tubing stored in separate plastic bags before and after use.
All samples were refrigerated at 4°C on return to the laboratory and left to stand at least over night for any particulate matter to settle.
Sample method blanks
To test the likely inputs of contaminants from different stages of the sampling process a blank was taken.
Sample bottle blank. Ultra-pure water direct from the dispenser was run into an NLS glass bottle to test NLS bottles for contamination. This was run through the system and onto the cartridge, replicating the procedure for other samples.
Sample processing set up
Pre-conditioned, numbered and dated sorbent Oasis® HLB cartridges were supplied by NLS in sealed Corning centristar centrifuge tubes to protect the cartridges and stop them drying out. Nitrile gloves were worn during the handling of the cartridges, tubing and samples to avoid contamination. As the system was run in a laboratory, the whole system was put in a fume cupboard to stop possible contamination from volatiles within the laboratory environment.
A 12-channel Ismatec high precision multichannel dispenser was used to drip the sample onto the pre-treated cartridge using new red-red (1.14 mm internal diameter) or yellow-yellow (1.42 mm internal diameter) double-bridge Tygon© auto analyser pump tubes. Additional new Tygon ©tubing was attached from the influent end of the pump tube into the sample bottle, carefully making sure the input was above the bottom of the bottle to avoid disturbance of settled particulate matter.
| Product code | Description | Use |
| ACF00007-CP | Tygon© Lab Tubing, Non-DEHP, 1/8"ID x 1/4"OD | Extension tube |
| WZ-96449-48 | Ismatec SC0430 2-Stop, E-LFL Tubing, 2.79 mm (purple-white) | Effluent |
| WZ-96449-30 | Ismatec SC0421 2-Stop, E-LFL Tubing, 1.14 mm (red-red) | Influent |
| WZ-96449-34 | Ismatec 2-Stop, E-LFL Tubing, 1.42 mm (yellow-yellow) | Influent |
The effluent from the cartridge was pumped into an effluent container using purple-white (2.79 mm internal diameter) double bridge Tygon© auto analyser pump tubes. The effluent tubes were larger than the influent tubes to allow for the resistance of the sample passing through the resin.
Tygon© tubing was used for any extensions needed between the pump tubes and sample bottle, cartridge or effluent containers. Where needed, 0.5 cm length of purple-white Tygon© were pushed over the end of the smaller diameter pump tubes to act as seals and size changers before connecting the larger diameter Tygon© tubing. All influent tubing was changed between each sample and the effluent tubing was changed as needed.
Five samples could be run at once with the 12 channel Ismatic pump but not all samples ran at the same speed. Due to particulate matter collecting onto the resins the influent pump speed was reduced during pumping or specific sample lines were interrupted to stop loss of sample.
Effluent from the cartridge was collected in a bottle and when approximately 500 ml had successfully gone through the cartridge, pumping was stopped and the effluent tubing run dry. The resulting effluent was poured into a measuring cylinder, the volume was noted and the cartridge was put back in the centrifuge container. The centrifuge container was labelled in pencil with the sample name, the volume of effluent and the date of processing. An additional NLS sample ID label was attached to the centrifuge tube and the corresponding label, together with the sample information was recorded on the NLS sample log sheet.
During the initial set-up both the influent and effluent pump tubes were controlled by the same dispensing pump. This caused a problem when needing to reduce the speed of the influent when the cartridges became blocked as the effluent was reduced at the same time. When a second peristaltic pump was available the system was modified so that the influent and effluent were controlled by separate pumps. This made the system easier to control especially with the high particulate concentrations found in some of the surface water samples.
Evidence from the initial trial
To check the repeatability of the process across different concentrations and for different compounds samples were taken from sites likely to have different contaminant loads. A groundwater sample was taken from a previously characterised borehole know to have a fewer contaminants at low concentrations. Several surface water samples were taken from downstream of sewage treatment works of differing sizes from the small local works to the main area works. A laboratory blank was also taken.
To check the repeatability and stability of the samples, 5 water samples were collected into NLS bottles at each site and on returning to the laboratory 4 of the samples were split between 2 pre-conditioned sorbent Oasis® HLB cartridges. This yielded 1 water sample and 8 cartridges for each sample site. The water samples and sorbed samples were sent to NLS for semi-quantative, analysis using a multi residue LCMS scan with the following timetable for each sample site:
| Week | Analysis |
| 1 | Water sample, cartridge 1 and 2 |
| 2 | Cartridge 3 and 4 |
| 3 | Cartridge 5 and 6 |
| 4 | Cartridge 7 and 8 |
No compounds were detected in any of the blank samples; from this we could conclude that the Tygon tubing, NLS bottles and lab method used was not contributing contamination to the samples. Appendix 1 - Trial results gives details of the results for the bottled water and sorbed cartridge replicate duplicate samples from the trial.
Bar charts (Figure 1) show the average concentration of all the cartridges and bottled water sample for the pumped groundwater and the surface water sample OSP which had the greatest contaminant load, the error bar is one standard deviation. Full data is given in Appendix 1 - Trial results.
Figure 2 shows the variability in reported selected contaminants and concentrations with time between the duplicate, replicate samples as well as the un-processed water sample from the highly impacted OSP site. Compounds were selected due to their relatively high concentrations in the sample. As can be seen, some compounds look very stable at the reported concentrations, while there is slight variation in others. Detection limits for Atenolol, Codeine and Lidocaine is 0.001 µg/l, there is no reported LOD for Diclofenac.
The results from each cartridge was compared to results from the original water sample analysed soon after sampling. This is shown for some of the compounds from one of the sites.
Thames surface water field trial, UK
After the initial success of the technique, the trial was expanded and samples were taken along the length of the River Thames. The River Thames rises as groundwater fed streams in the Cotswolds, and flows through rural areas, towns and cities before flowing through London and out to sea. There are a series of locks on the river which restrict the tidal range to below Teddington lock and through most of London. Surface waters in the Thames are impacted by rural (agriculture, small sewage treatments works), urban (industry, larger sewage treatments works) and saline conditions, this was seen to be a good test of the field procedure. The Thames was in flood during parts of the study with the river containing higher particulate load.
A total of 34 samples were taken along the length of the Thames from the source to the sea in January and February 2016. Most samples in this project were taken direct from the bank without the use of a pump, however, the Solinst peristaltic pump was used where it was too hazardous to sample in this manner. The tidal part of the Thames (consisting of 15 samples) was sampled from a boat using the Solinst pump with the inlet 1 m below the water surface. When the pump was used all tubing was rinsed thoroughly with sample water prior to sampling. Duplicate samples were processed and additional blanks were run during this study to test the field procedures.
- Sample bottle blank. Ultra-pure water direct from the dispenser was run into an NLS glass bottle to test NLS bottles for contamination. This was run through the system and onto the cartridge, replicating the procedure for other samples.
- Peristaltic pump method blank. A solution of Virkon was run through the Solinst peristaltic pump tubing and HDPE tubing (plus PE connector), the tubing was run dry followed by ultra-pure water to rinse. A glass beaker was used. A sample of ultra-pure was then pumped into an NLS bottle through the clean tubing to test the tubing for contamination.
- Post-sampling peristaltic pump blank. After sampling the tidal Thames (15 samples) one litre of Ultra-pure water was used to rinse the Solinst peristaltic pump tubing, PE connectors and HDPE rigid tubing. After this the procedure in the field was followed, the sampling beaker was rinsed three times and then filled with 600ml of water before filling a clean NLS bottle. This was to test for carry over between samples as we were unable to wash the tubing between samples.
| Blank sample description | Compound detected (LOD µg/l) | Concentration (µg/l) |
| Peristaltic pump method blank | no compounds detected | |
| Post-sampling peristaltic pump blank (post-tidal Thames survey) | Mepronil (0.001) Carboxin (0.001) |
0.0024 0.014 |
| Post-sampling peristaltic pump blank (post source to Oxford sampling) | Tramadol (0.001) | 0.0001 |
| Laboratory blanks | no compounds detected |
A total of 84 compounds were detected above the LDO in the samples from the Thames; 38 pesticides, 37 pharmaceuticals, 6 surfactants, a fire retardant, sucralose and cocaine.
Thames study conclusions
The post-sampling blanks highlight the importance of cleaning the peristaltic sampling tubing correctly between sample sites and has led to the data being blank corrected for possible carry over.
No problems were encountered when processing sea-water influenced samples during the field solid phase extraction however, processing samples with higher particulate matter was more difficult. Where higher particulate matter was seen, samples needed to be left to settle at least over night, the input had to be above the bottom of the sample bottle and the influent line pump speed had to be reduced or stopped more often as the cartridges became blocked. This meant that the process took longer.
Processing of the cartridges at NLS was more difficult for some parts of the Thames due to the amount of contaminants found in the samples. After correcting for detection limit and blank contaminants, between 10 and 62 compounds were reported in the Thames samples.
Field-trial in Varanasi, India
The city of Varanasi, India, is situated on the banks of the Ganges River and has a population of over 1 million. The city obtains a significant proportion of its drinking water from the sedimentary aquifer system beneath the city. Municipal drinking water supply is from groundwater sources distributed across the city as well as surface water from the River Ganges. Private groundwater supply is common across the city. A campaign of fieldwork to characterise the groundwater quality of drinking water sources within the aquifer system included sampling and broad screening for micro-organic contaminants.
A total of 29 groundwater sites were sampled from paired shallow (<50 mbgl) and deep sites (>100 mbgl) to profile depth changes in environmental tracers as well as spatial variation. Broad screening for micro-organics was carried out by the UK National Laboratory Service following solid phase extraction of groundwater samples in the field. A blank sample was taken out in a sample bottle from the UK and processed in the field to quantify procedural contamination.
Varanasi preliminary results
The field blank was found to be contaminated with 7 substances, two analgesics and 4 insecticides and a pesticide. With the exception of dichlorvos and trinexepac these the concentrations found in the blank sample were <0.015 mg/L.
| CAS#, Substance (Common name) | Group | Concentration (LOD), mg/L |
| Cas# 103-90-2 Acetaminophen (Paracetemol) | Analgesic | 0.012 (0.005) |
| Cas# 63-25-2 Carbaryl | Insecticide | 0.0079 (0.001) |
| Cas# 62-73-7 Dichlorvos | Insecticide | 0.82 (0.005) |
| Cas# 138261-41-3 Imidacloprid | Insecticide | 0.0057 (0.001) |
| Cas# 52-68-6 Trichlorfon (Metrifonate) | Insecticide | 0.0029 (0.001) |
| Cas# 143294-89-7 Trinexepac | Pesticide | 0.18 (0.1) |
| Cas# 15687-27-1 Ibuprofen | Analgesic | 0.014 (0.001) |
For the 30 samples analysed the most frequently detected groups included pesticides, pharmaceuticals and perfluorinated alkylated substances (PFAS). One the bank data had been screened out a total of forty separate compounds were considered true detections in the study, 22 pesticides, 14 pharmaceuticals, 3 PFAS and sucralose. Although mostly detected in low ng/L concentrations in groundwater there are a broad range of compounds that are consistently detected in shallow groundwater and surface waters including frequent detection of sulphonamides.