Chapter 3 Sampling methodology and analysis of samples
3.7 Well head sampling
3.1 Introduction to DOC sampling
When sampling groundwater for DOC, or any other constituent, the need to obtain accurate data is absolute. The problems of obtaining DOC samples representative of the groundwater in the aquifer have been widely discussed (Sharp et al., 1993; Turrell, 1994; Tupas, 1994, Sugimura and Suzuki, 1988; Wilhams et a l, 1993). During the sampling and storage of water new surfaces, pressures, temperatures, light levels and contamination can alter the DOC concentrations on a time scale of hours (Sharp et al.,
1993). As much as 20% of the DOC in unpreserved samples may disappear in less than one day (Kirchman et al., 1991), so groundwater samples must be preserved as soon as possible after sampling, in order to minimise change.
In this study, a sampling strategy was used that allowed the collection of water samples that were representative of the DOC concentration in the groundwater, and that prevented the DOC concentration from changing before analysis was complete. Following initial studies at University College London (McArthur, 1995, pers. comm.; Turrell, 1994) a sampling strategy was developed by the author, specifically to overcome the problem of obtaining low DOC field blanks and ensuring sample integrity at low (<1 mg/l) DOC concentrations. Field blanks are blanks filled in the field with ultra-pure water from a Winchester, and laboratory blanks are blanks filled in the laboratory with ultra-pure water.
For this sampling strategy development several aspects of sampling were examined: • sample containers: how the nature of the container influenced DOC concentrations; • sample collection: how the physical act of sampling influenced DOC concentrations; • cleanliness: what standards were necessary to avoid contaminating samples;
• filtration: how filtration influenced DOC concentrations;
• acidification and sparging: how bacterial alteration can be minimised by removing a source of metabolite COo.
3.2 Sample containers
Containers used to store DOC samples must be rendered free from any contaminant organics (Sharp et al., 1993). Samples were therefore collected in glass containers, as they are more easily cleaned of organics (Sharp et al., 1993). Results from this thesis suggest that organics leaching from plastic containers contaminate samples with DOC (see Chapter 3.5.1). Glass containers of 50 ml volume, made by Shimadzu, were used to collect samples because this minimised sample manipulation; vials are placed in the Shimadzu 5000 auto-sampler for analysis, rather than being transferred to other sample bottles, a process that can lead to contamination (Turrell, 1994). Shimadzu glass was preferred to Pyrex glass, as the latter had been shown to result in DOC in samples increasing with time (McArthur, 1993. Pers. comm.).
The Shimadzu vials were cleaned for 24 hours with chromic acid, a strong oxidising agent, rather than baking as the latter method might activate the surface of the glass and lead to the sorption or organics onto the container walls (Sharp et al., 1993). Following cleaning, vials were rinsed with ultra-pure water and filled with 1% sodium metabisulphite, a strong reducing agent and bacterial poison, which prevents bacterial growth. The vials were sealed with aluminium foil, kept in place with a Teflon cap.
3.3 Sample Collection
At the well head, the Shimadzu vials were emptied of the sodium metabisulphite and placed immediately under the stream of water from the borehole sampling tap, or directly in the spring, allowing the vial to overflow for approximately 15 seconds. This ensured that, while the vial was adequately rinsed, it was not open to the atmosphere, which is a potential source of contamination (Sharp et al., 1993; see Chapter 3.6.2). Vials were then sealed with aluminium foil, held in place with a Teflon cap. The aluminium foil that was used to cap the vials was cut into pieces of the appropriate size in the laboratory and stored in an acid cleaned jar until use in the field. The foil was handled at all times using surgical gloves and tongs. Samples were collected in triplicate in order to assess sample reproducibility.
3.3.1 Cleanliness
Because of the low DOC concentrations measured, a rigorously clean sampling strategy had to be employed. Fingerprints contain high concentrations of organic compounds (Sharp et a i, 1993); a fingerprint on the inside of a Shimadzu vial can contribute 1 mg/l DOC (Turrell, 1994), so powder-free, surgical gloves were worn at all times during the sampling, acidification and sparging procedures, to prevent contamination from skin. DOC samples from boreholes were taken directly from the sampling tap or outlet valve and DOC samples from springs were taken as close to the spring source as possible. Sampling taps were washed with sodium metabisulphite and then allowed to run for about five minutes prior to sampling. The boreholes were either pumping constantly or were purged for approximately 15 minutes prior to sampling.
3.4 Filtration
DOC samples collected for this study were not filtered, and therefore the DOC concentration in the sample is a measure of the total non-volatile organic carbon, rather than the non-particulate DOC. As particulate DOC usually comprises <5% of the total organic carbon in groundwater it contributes very little to the overall DOC concentration in unfiltered groundwater samples (Sharp et al., 1993).
Filtration removes particulate organic matter (molecules with a diameter >0.45 |im) and sterilises samples; bacteria alter the DOC concentration in a sample by metabolising the DOC (see Chapter 2). There are approximately 100-1000 bacteria per ml of groundwater (Chapelle, 1993) and each bacterium has a mass in the order of 10'^ mg (Schlegal, 1985), so bacterial mass does not form a significant part of the DOC concentration in unfiltered groundwater samples. Bacterial growth and metabolism of DOC can be prevented in unfiltered samples, by acidification and sparging of the sample which renders all the bacteria in the sample inactive (see Chapter 3.5).
During filtration DOC concentrations may be altered, through its sorption onto the filter, through contamination by the filtering equipment, or from the release of DOC from or
the filter (Sharp et a l, 1993; Norrman, 1993; Turrell, 1994). This was another reason for not filtering DOC samples.
3. 5 Sample preservation
3.5.1 Acidification and sparging
Samples for DOC analysis were acidified to pH 3.0 in the field and sparged with nitrogen gas for 15 minutes in order to remove dissolved inorganic carbon (DIC) and to preserve samples. A low pH minimises heterotrophic activity but converts bicarbonate (HCO3 ) to carbon dioxide, which promotes autotrophic bacterial activity. Sparging with nitrogen gas removes this carbon dioxide. The downside of sparging after acidification is that it removes any acid-volatile DOC. Until a method is developed for DOC analysis in the presence of DIC, this problem cannot be overcome, although the volatile organic carbon fraction in groundwaters is usually less than 0.05 mg/l (Thurman, 1985).
Acidification was done with 1.6 M sulphuric acid from a glass syringe with a hypodermic needle. Acid was added through the aluminium foil of the sample vial during sparging in order to minimise exposure of the sample to the atmosphere and to maintain positive pressure in the vial during addition, further minimising the potential for contamination. Sparging was done by hypodermic needles inserted through the aluminium foil caps, using nitrogen from a portable nitrogen cylinder.
Contaminant DOC might be expected to be added to samples via acidification and sparging. Trials were therefore conducted, both in the field and in the laboratory (Table 3.1), to compare DOC in blanks that were acidified and sparged, with DOC in blanks that were neither acidified nor sparged. Also tested were blanks acidified with a Hach titrator and blanks acidified with a glass syringe (Table 3.1).
Table 3.1. DOC in field and laboratory blanks untreated, and treated in various ways. All concentrations in mg/l.
LA
Field blanks Laboratory blanks
Untreated Acidified (Hach) and
sparged
Acidified (syringe) and sparged
Untreated Acidified (Hach)
0.043 0.081 0.035 0.050 0.042
0.071 0.055 0.041 0.033 0.070
0 .0 1 1 0.131 0.048 0.049 0.041
0.053 0.068 0.036 0.024 0.041
Acidification of ultra-pure water laboratory blanks using a Hach titrator did not introduce any significant source of DOC contamination (Table 3.1), but acidification of ultra-pure water blanks in the field using a Hach titrator introduced slight contamination (Table 3.1). Acidification and sparging of ultra-pure water blanks in the field using acid from a glass syringe gave lower blanks with a lower range (Table 3.1). This suggests that acidifying blanks using a syringe gives more reproducible results than acidifying blanks using a Hach titrator.
3.5.2 Freezing
Sparging and acidifying groundwater samples is effective in preserving DOC samples, but is time consuming (15 minutes per sample), and is impractical for remote areas due to the need for bottled nitrogen. Freezing may preserve DOC samples but Sugimura and Suzuki (1988) noted a 15-20% loss of DOC in freezing. However, they gave no details of sample handling nor freezing method, nor why the DOC loss occurred. Sharp et al. (1993) suggested that quick freezing may prevent this loss.
In order to test this hypothesis, and so develop a more practical preservation methodology, field trials were carried out to determine the effectiveness of freezing. Shimadzu vials proved fragile during freezing in dry ice, whether the vials were full or half full, so alternative containers were therefore sought. Plastic vials were tried, although previously avoided because of the risk of contamination from plasticisers. Using ultra-pure water and freezing using dry ice (-78 "C), comparisons were made of laboratory and field blanks after freezing in acid washed, 30 ml low density polyethylene (LDPE) vials and freezing in acid washed 50 ml, polypropylene vials. The results are discussed below.
3.5.2.1 Freezing using LDPE
The field blanks collected in LDPE bottles were thawed at room temperature, transferred from the Nalgene bottles to the Shimadzu vials and then analysed (Table 3.2).
Table 3.2. DOC using LDPE Nalgene bottles.
DOC (m g /l) in field blanks, a cid ified and
sparged
DOC (m g /l) in field blan k s, frozen
0.035 0.053 0.017 0.066 0.081 0.080 0.027 0.034 0.038 0.058 M ean 0.040 ±0.049 M e a n 0.058 ±0.034
The samples were transferred in a laminar flow hood, containing an activated carbon filter, which provides a contamination free environment. There was no significant increase in the DOC concentration in blanks transferred between vials (Table 3.3).
Table 3.3. Results of trails to determine whether the transferring of samples between
DOC (mg/l) in lab blank not transferred between
vials
DOC (mg/l) in lab blank transferred with hood off
DOC (mg/l) in lab blank transferred with hood on
0.006 0.003 0.029
0.006 0 .0 2 1 0 .0 1 0
0.033 0.078 0.030
0.140 0.016 0.015
0.006 0.033 0 .0 1 0
Mean 0.013 ±0.049 Mean 0.030 ±0.034 Mean 0.019 ±0.020
3.52.2 Freezing using PP
PP centrifuge tubes were tested as they could fit directly into the Shimadzu 5000 TOC autosampler and could be used to collect DOC samples. Twenty laboratory blanks were taken in PP centrifuge tubes and five vials were analysed for DOC immediately, five vials
analysed after 1 hour, and 10 vials placed in a freezer and analysed for DOC 24 hours later, after thawing by two separate methods (natural thawing at room temperature, hot thawing under hot water) (Table 3.4).
Table 3.4. Resu ts of freezing trials using polypropy ene (PP) centrifuge tubes
DOC (m g/l) in blank in glass vials DOC (m g/l) in PP blank analysed im m ediately D OC (m g/l) in PP blank after 1 hour D O C (m g /l) in PP blank, frozen, and
hot thawed'
D O C (m g/l) in PP blank, frozen, and
thawed^ 0 . 0 1 2 0.049 0.061 0.024 0.055 0.009 0.024 0.015 1.54 0.104 0.005 0.037 0.014 1.57 0.107 0.024 0.078 0.040 0.391 0.027 0 . 0 1 2 0.024 0.022 0.205 0.103 Mean 0.012 ±0.014 Mean 0.042 ±0.045 M ean 0 .0 3 0 ±0.040 M ean 0.746 ±1.50 Mean 0.079 ±0.073
= the blanks were frozen in the PP vials and then thawed quickly by running the bottom half of the vial under hot running water, ensuring that the top of the vial did not come into contact with the hot water. The thawing took approximately 4 minutes.
" = the blanks were frozen in the PP vials and then thawed at room temperature. The thawing took approximately three hours.
There was a considerable difference between the DOC in the blanks in the glass vials and the DOC in blanks in PP tubes frozen and then hot thawed, although the hot thawed blanks had a very large range. These results show that freezing blanks in PP gives elevated DOC concentrations and bad reproducibility, suggesting that during the freezing and thawing process some breakdown of the plastic occurs and leaches into the blank.
It is concluded that, until more trials can be carried out to determine which plastics can withstand the freezing process and can be used for DOC collection, that DOC samples be preserved by acidification and sparging.
3.5.3 Storage
Acidification and sparging preserves DOC samples (Turrell, 1994); nevertheless they were kept cool and in the dark whenever possible. DOC samples which have been
acidified and sp a rg e d can be s to re d at ro o m tem p eratu re, for up to o n e m o n th , with no significant cha nge in the D O C co n c e n tra tio n (Fig. 3.1 and Fig. 3.2) (T u rrell, 1994; M c A r th u r , 1995. pers. c o m m .). 2.9 2.74 2.54 Error bar 2.3 40 60 80 100 120 140 160
storage time (hours)
Figure 3.1. Chart showing the results of storing acidified and sparged samples for up to six days. The error bar was calculated from the mean deviation about the mean of 28 DOC samples (after Turrell, 1993).
1.3 1 '-2- E O Q LlJ E rro r i b a r I 10 s to r a g e tim e (d a y s)
Figure 3.2. Chart showing the results of storing acidified and sparged samples for up to twenty days. The error bar was calculated from the mean deviation about the mean of 28 DOC samples (McArthur, 1993. Unpublished data).
3.6 Blanks
3.6.1 Trip blanks
The effects of storage of blanks in cleaned Shimadzu vials was assessed. DOC concentrations in lab blanks analysed for DOC immediately, were compared with DOC concentrations in trip blanks (blanks made in the laboratory, taken into the field and returned unopened) stored in cleaned Shimadzu vials for four days (Table 3.5). The results suggest that storage in cleaned vials does not contribute any significant DOC contamination.
Table 3.5. Results of laboratory blank storage trials.
DOC (mg/L) lab blanks DOC (mg/L) in lab blanks stored for 4 days
0.056 0.043 0.065 0.036 0 .0 1 0 0.063 0.036 0.041 Mean 0.042 ±0.049 Mean 0.046 ±0.024 3.6.2 Field blanks
The problems associated with obtaining true field blanks are considerable and have resulted in DOC field blanks up to 0.5 mg/1, i.e. higher DOC concentrations than groundwater samples (Turrell, 1994). The ultra-pure water used for field blanks when sampling the Lower Greensand aquifer in Sussex and Surrey (see Chapter 4), was made using an Elga water purification unit which gave laboratory blanks of <0.05 mg/1. This water purification unit was replaced with an Elga Maxima Ultra-pure Water unit which can give DOC laboratory blanks of <0.02 mg/1, and the field blanks on all other sampling trips were made from ultra-pure water from the new unit.
Field blanks are taken by rinsing vials with ultra-pure water three times prior to filhng, and field samples are taken by holding the vial under a constant steam of water from the sampling tap. Some field blanks contained more DOC than samples, and as laboratory blanks made from ultra-pure water were much lower than field blanks (0 .0 1 to 0.1 mg/1), it was concluded that the method of collecting field blanks was at fault. To test this hypothesis, field blanks were collected by exposing the vials to the atmosphere for varying times in between successive rinsings, and DOC contamination in the field blanks assessed as a function of time (Table 3.6).
Table 3.6. Results of the field trails on contamination from exposure to the air
Time of exposure' DOC (mg/1) in the field blank
o' 0.043
1 0.136
10 (1326
30 0.115
300 0 .1 2 0
= the time the vial was left open to the air between each of three rinses. “ = the vial was filled under a constant stream of water from the glass tank.
It was found that taking field blanks under a constant stream of water gave DOC concentrations approaching the detection limit of the Shimadzu 5000 and lower than blanks collected by successive rinsings (Table 3.5). It is concluded that the factors that contribute to high blanks do not affect the samples, which were taken using a different methodology, and therefore, DOC measurements are reported without blank corrections.
3.7 Well head sampling
Dissolved oxygen, pH, temperature, alkalinity and conductivity were measured at each borehole or spring site. Dissolved oxygen measurements were made using a pHOX 962 Oxygen Meter. At springs, where a flow cell could not be used, the oxygen meter was placed directly in the spring, although this method may have resulted in slightly elevated measurements of dissolved oxygen in these springs. The 100% saturation calibration
was checked at each site and the meter recalibrated if neeessciry. The 0% calibration was checked at the start of each sampling trip, and recalibrated if necessary. The 100% calibration and 0% calibration - did not need recalibration unless the meter was switched off for more than two weeks.
Temperature measurements were made using the temperature probe in the pHOX meter and pH was measured using a Whatman PHA 260 pH meter. The pH meter was corrected for temperature, and calibrated at each site using buffer solutions of pH 7.00 and pH 4.00. The pH measurements were made in a flow cell.
Alkalinity was measured in the field by titrating the sample with 1.6 or 0.16 Normal sulphuric acid, using a Hach digital titrator. Titrations were monitored by pH electrode and end points determined from the inflection point of the titration curve. Two or three titrations were carried out to ensure a precise measurement, usually within two Hach digital titrator units (1/800 ml).
Electrical conductivity was measured using a WTW LF318 conductivity meter placed in a flow cell. The meter took approximately ten minutes to equilibrate to the correct temperature. It was calibrated using 0.005 M potassium chloride solution, 706 p,S/cm at
25"C.
Samples for cation and anion analysis were collected in 30 ml, acid washed, LDPE plastic Nalgene bottles. Cation samples were acidified with three drops of 50% hydrochloric acid, anion samples were not acidified. Two blanks were taken at each site, using ultra-pure water stored in Winchesters, acidifying one with 50% hydrochloric acid.
Sampling records of all sites are shown in Appendix 6.