Anal. Chem. 1990, 62,2408-2412
2400
(10) Faizullah, A. T.; Townshend, A. Anal. Chim. Acta 1985, 767, 225-23 1. (11) Mortatti, J.; Krug, F. J.; Pessenda, L. C. R.; Zagatto, E. A. G.; J0rgensen, S. S. Analyst 1982, 107, 659-663. (12) Alonso, J.: Bartroli, J.; del Valle, M.; Barber, R . Anal. Chim. Acta 1989, 219. 345-350.
(13) Chamsi, A. Y.; Fogg, A. G. Ana/yst 1888, 113, 1723-1727. (14) Hulanicki, A.; Matuszewski, W.; Trojanowicz, M. Anal. Chim. Acta 1987, 194, 119-127.
RECEIVED for review April 16,1990. Accepted June 26,1990.
Bottle-Callbration Static Head Space Method for the Determination of Methane Dissolved in Seawater Kenneth M. Johnson,*J Jeffrey E. Hughes, Percy L. Donaghay, and John McN. Sieburth Graduate School of Oceanography, University of Rhode Island, Narragansett, Rhode Island 02882 Interactions between physical structure and microbial populations play a critical role in controlling the bacterial production and consumption of gases in estuarine and oceanic environments, especially in poorly ventilated water bodies. In the permanently stratified anoxic lower basin of the Pettaquamscutt River in southern RI, fine-scale profiles (1-cm vertical resolution) of the physical structure (salinity, temperature, density, light intensity), particle structure (chlorophyll a fluorescence, light transmission, microbial abundance), and chemical structure (oxygen, methane, hydrogen sulfide, and carbon dioxide) showed an approximately 2 m thick oxic-anoxic transition zone (OATZ) of intense microbial activity usually situated between the 2.5- and 5.5-m depths ( I ) . To determine the association between gases and the chemical, physical, and particle structure would require gas profiles of similar resolution. Current methods for the collection and analysis of gases, however, allow only a small fraction of the 200 or so required samples to be processed in a reasonable time. The primary techniques for gas analysis are dynamic and static head space techniques (2). In dynamic analysis the analytes are stripped from solution with a purging gas (nitrogen, helium, etc.), trapped, and then desorbed or refocused prior to detection. Dynamic analyses for dissolved COz in seawater followed by coulometric detection (3, 4 ) showed quantitative recoveries of C 0 2 with high precision but at a rate too slow for this work (7-10 min sample-'). In static analysis a sealed sample container (usually a serum bottle) is incompletely filled with solution so that the dissolved gases in the liquid phase can partition into the small head space (gas phase) until the partial pressure of the gas is equal in the two phases. The equilibrium gas concentration in the head space can be increased by reducing the pressure in the head space, heating, adding electrolytes, or by keeping the ratio of head space volume (gas phase) to sample volume (liquid phase) small (5-8). The latter technique is simplest in terms of manipulation and calculation but requires tedious gravimetric phase volume determinations for each analysis. In this paper we describe a simple technique for simultaneously calibrating serum bottles for both gas and liquid phases to better than f O . l mL. Once calibrated, the bottles may be reused without recalibration and volume changes due to temperature and salinity can be calculated. These bottles can serve for sample collection, preservation, gas concentration, incubation, and possibly as a reaction vessel in other analytical sequences. With this calibration technique and the static head space method, we have collected up to 100 samples during a
* To whom correspondence should be addressed.
'Present address: Department of Applied Science,Oceanographic and Atmospheric Sciences Division, Brookhaven National Laboratory, Upton, NY 11973.
profiling run, equilibrated them, and analyzed them for CH, by gas chromatography (GC) in a single day.
EXPERIMENTAL SECTION Gas- and liquid-phase volumes of 50-mL serum bottles (Wheaton, Millville, NJ, No. 223745) were calibrated with distilled water. The empty bottles and a gas impermeable butyl rubber septum stopper (Belco Glass, Vineland, NJ, No. 2048-11800) were weighed to four decimal places to get a total tare weight (WJ. After weighing, the bottles were immersed in a distilled water constant-temperature bath (Model RTE-8, Neslab, Portsmouth, NH) at a calibration temperature of 25 "C. When the bottles came to temperature, a Gilson 5-mL white polypropylene disposable pipet tip (Gilson Medical Electronics, Middleton, WI, No. P-5OOO), hereafter called the calibrator, was forced into the filled serum bottle until the tip contacted the bottle bottom, as shown in Figure 1. The calibrator sealed the bottle, which was then inverted and shaken gently from side to side until water no longer drained from the calibrator. After the calibrator was carefully removed from the upright bottle, leaving both a liquid and gas phase, the bottle was stoppered, dried, and reweighed ( Wz). The liquid-phase volume in milliliters at the calibration temperature (V1,,,) was calculated from the difference (W, - W,) (9) corrected for the buoyancy of air (10). The bottle was reimmersed to displace the gas phase, and the bottle's stopper, pierced with a 20-gauge syringe needle, was reinserted so that the water displaced by the stopper flowed through the needle until the stopper was completely seated. After the needle wai removed, the stoppered bottle was dried and reweighed (W3). The gas-phase volume in milliliters at the calibration temperature (V,,J was calculated from the corrected (IO) weight difference in grams ( W , - W2).The total volume of + the bottle at the calibration temperature (EVet) equals V,,,,. This procedure was repeated three times for each bottle with a different calibrator tip each time. Field samples were collected from a stable donut-shaped floating platform (11)moored in the deepest basin of the Pettaquamscutt River (12) by siphoning through a 1 cm i.d. black polyethylene hose (13) to prevent alterations by pumping. The siphon tubing was attached to an electronic profiler (Sea Bird Electronics, Belvue, WA), and the siphon was maintained by discharging into a sump tank extending 1 m below the water surface in the center of the platform. Gas samples were obtained as in Figure 1. The serum stopper used to seal the bottle was pierced with a 20-gauge hypodermic needle to ensure that the gas phase remained at ambient pressure when the stopper was seated. After seating, it was secured with a 20-mm aluminum seal (Belco Glass, No. 2048-00150 or equivalent) and hand crimper (Wheaton No. 224303). A simple plastic tool shown in Figure 2 was fabricated to aid in the insertion and withdrawal of the calibrator from the serum bottles. By use of the tool, the average time for insertion, withdrawal, and crimping was about 25 s, but insertion and removal can be made without the tool in about the same time if bottles with narrow mouths that cause the calibrator to seal too tightly are discarded. To retard microbial activity, the sample bottles were placed on ice until the liquid and gas phases were equilibrated in a constant-temperature shaking bath at 100 revolutions min-' for at least 12 h.
0003-2700/90/0362-2408$02.50/00 1990 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990
2400
the Henry's law constant and the Bunsen solubility coefficient
(PI KH= &/RT
Flgure 1. Steps in the calibration of both the liquid- and gas-phase volume of serum bottles using a hydrophobic disposable pipet tip: (A) serum bottle and pipet tip; (6) serum bottle filled with seawater sample; (C) pipet tip fully inserted into the bottle: (D) bottle inverted to drain excess sample and create gas phase: (E) gastight sealed sample with known liquid and gas volumes.
=j$j ..... ,:, .,:
Ver.tical d r a i n hole
........
i : /
i
:
H o r i z o n t a l hole f o r the g r i p 1
where KHis the Henry's law constant in mol L-' atm-', p is the Bunsen solubility of CH, in L L-' atm-', 22.356 is the molar volume of CH, in L mol-', R is the gas constant in L atm mol-' K-l, and T is in K. Combining yields a dimensionless quantity Kd = (/3/22.356)RT
(4)
Substituting for Kd in eq 3 gives C1" = Cg((P/22.356)RT + V,/VJ
(5)
From P V = nRT, the concentration of CH, in the gas phase in mol L-' is P / R T where P is the partial pressure of CH, at 1 atm of total pressure after equilibration as measured by gas chromatography. Substituting into eq 5 yields the relationship CI" = (P/RT)((P/22.356)RT + Vg/VJ (6)
_... ,
p = KH(22.356)
B C V
Flgure 2. (A) A 7-cm long plastic tool bored through lengthwise with a 6 mm i.d. drain hole to facilitate the insertion and removal of ( 6 ) calibrator tips into and from serum bottles as shown in Figure 1. (C)
Friction-fitted tool forming a water-tight seal with the calibrator tip. A stainless steel rod (grip) is inserted through horizontally aligned holes bored in both the calibrator tip and tool.
After equilibrium, the gas phase was analyzed for CH, by injecting 200 pL of the gas phase into a 2.5 m X 1.5 mm i.d. column of stainless steel or Teflon (PFA) packed with Porapak N of 100/120 mesh (Alltech Associates, Deerfield, IL). Before removal of the 200-pL subsamples of the gas phase for analysis, 200 pL of helium was injected into the bottle to keep the total pressure of the gas phase at 1 atm (14). Methane was determined on a Varian Model 1740 dual column gas chromatograph with flame ionization detection (Varian Associates, Palo Alto, CA). The detector block was at 175 "C, injector at 125 "C, and the column isothermal at approximately 55 "C. Injections were made with a Hamilton 500-pL gastight syringe (Reno, NV, No. 1750 Npt-5) equipped with a Chaney adapter. The carrier gas was helium (99.999%) at a flow rate of 20 mL min-'. The output of the detector was integrated on an Infotronics Model CRS 208 digital integrator (Columbia Scientific, Austin, TX). For standards, we used gas mixtures of 100 and 1000 ppm CHI in helium (Alltech Associates, Deerfield, IL, Nos. GO111 and G0112, respectively). These gas mixtures were withdrawn from their containers via an Alltech septum syringe adapter (No. 8810) and injected directly into the chromatographic column. Calibration curves were made by least-squares analysis. To calculate the in situ CH, concentration, we assumed that the initial CHI concentration in the gas phase of the sealed serum bottle was zero. We used the relationships given by Vitenberg et al. (15) for the equilibrium condition C,"V, = c,v, + c,v, (1) where C: is the concentration of CHI in the liquid phase before equilibration, C1is the concentration of CH4in the liquid phase after equilibration,C is the concentration of CH, in the gas phase after equilibration, fil is the volume of the liquid phase, V, is the volume of the gas phase, and CI = KdC, (2) Kd is the partition or distribution coefficient. Substituting and simplifying yield c1" = Cg(Kd + vg/vI) (3) The relationships given by Stumm and Morgan (16)relate Kd to
Equation 6 is used to solve for the liquid-phase CH, concentration before equilibration (mol L-'). Multiplying the molar volume of CHI by the sample density at the equilibration temperature gives C1" in mol kg-'. For this calculation, V,/ Vl is known from the calibration procedure, the partial pressure of methane is measured directly, and the Bunsen solubility can be obtained from the tables of Wiesenburg and Guinasso (17) or calculated according to the Bunsen solubility equation of Weiss (18) In p = Al + A2(100/T) + A , In (T/100) + S ( B , + B,(T/100) + B,(T/100)2) where A,, A,, A,, B1,B2,and B3 are constants and T is in K. Alternatively, the concentration of CH, in the liquid phase before equilibration (C): in mol L-' can be found by calculating the moles of CHI in the liquid phase (mol,,) after equilibration from mol,, = (PB/ 22.356)( VI,^/ 1000) (7) where is the volume of the liquid phase at the equilibration temperature, and the moles of CH, in the gas phase (mol,) from mol, = (P/RT)(V,,e/lOOO) and adding the two CI" = ((mol,,
+ mol,)/
Vl,J1O0O
(8)
(9)
RESULTS AND DISCUSSION To date, 157 bottles have been calibrated a t least three times each. The mean liquid-phase volume was 55.11 f 1.6 mL. The pooled standard deviation, calculated according to Youden (19),for three determinations of was f0.029 mL, and the relative standard deviation (rsd) was 0.053%. For the same bottles, the mean gas-phase volume (V,,,) was 3.95 f 0.06 mL, the pooled standard deviation was f0.043 mL, and the rsd was 1.0%. The mean phase ratio ( V,,ct/ Vl,ct)was approximately 0.07. For the interbottle variation was much greater than the pooled standard deviation of the calibration method (f1.6 and f0.029 mL, respectively), but for V,,, it was not significantly different from the pooled standard deviation of the calibration method (f0.06 and f0.043 mL, respectively) so that the gas-phase volume was nearly constant for these bottles. Repeated calibrations using the same calibrator tip and stoppers (possibly improving precision) were not made because we wanted the precision of the calibration determination to reflect the interchangeability of the tips and stoppers. The calibration method was checked with field samples. Table I shows that for samples collected at 6 OC with a salinity of 30 ppt the mean difference between the calibrated volume of the liquid phase and the liquid-phase volume at 6 "C (V,J was -0.03 mL, which was equivalent to the pooled standard deviation of the calibration procedure (f0.03 mL).
ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990
2410
Table I. Determination of the Difference between the Calibrated Serum Bottle Liquid-Phase Volume ( VI,,,) and the Liquid-Phase Volume (VI,,)of Seawater ( S = 30 ppt) Collected at 6 "C and the Difference between t h e Calibrated Serum Bottle Gas-Phase Volume ( VS,J and the Gas-Phase Volume Determined Empirically ( V S s )When These Samples Were Equilibrated at 25 "C" bottleno. V,,,,mL Vl,ct,mL
diff
Vg,g,mL Vg,ct,mL diff
6 9 13 28 36 61 82 84 89 90 124
54.22 53.99 54.96 54.61 54.31 54.12 54.12 54.17 54.13 54.21 54.01
54.17 54.08 54.98 54.69 54.41 54.10 54.16 54.24 54.12 54.23 54.03
+0.05 -0.09 -0.02 -0.08 -0.10 +0.02 -0.04 -0.07 +0.01 -0.02 -0.02
mean
54.26 10.28
54.29 10.29
-0.03b
3.70 3.56 3.60 3.70 3.73 3.78 3.61 3.57 3.68 3.65 3.71
3.93 3.73 3.80 3.84 3.87 3.92 3.78 3.75 3.87 3.79 3.92
-0.23 -0.17 -0.20 -0.14 -0.14 -0.14 -0.17 -0.18 -0.19 -0.14 -0.21
3.66 10.07
3.83 10.07
-0.17
" T h e density of the sample water was 1.023604 g mL-'. Absolute mean difference = 0.05 mL.
However, Table I shows that after the bottles were warmed to the calibration temperature, there was a mean difference of -0.17 mL between the gas-phase volume ( V,,) determined and the calibrated gas-phase volume (V,,,,). This difference exceeded the pooled standard deviation of the calibration procedure (f0.04 mL) by a factor of 4. The smaller value for V,, compared to V,,, was due to the increased volume of the liquid phase upon warming the sample to the calibration temperature, thus reducing the observed gas-phase volume. Therefore, it was necessary to correct the gas-phase volume for the differences between the calibration, in situ, and equilibration temperatures. The gas-phase volume of a calibrated bottle at any equilibration temperature ( Vg,e) for a sample of any salinity or in situ temperature equals (10) Vg,e = CVct(1 + a(te - tcJ) - VI,ct(ds/de) where de is the density of the sample at the equilibration temperature, d, the density at the in situ temperature, t , the
equilibration temperature, t, the calibration temperature, and a the volume coefficient of expansion of the Wheaton serum bottle glass (1.9 X mL mL-' OC-'). The first term of eq 10 corrects for the expansion of the glass bottle due to cooling or heating, while the second term corrects the liquid-phase volume for salinity and temperature differences. For 33 samples collected at 6 (Table I), 9.6, and 17 "C with salinities of 30, 24.1, and 0, respectively, the mean differences between the gas-phase volumes at an equilibration temperature of 25 "C calculated from eq 10 (V,,,) and the gas-phase volumes actually measured at 25 "C (V,,,) were -0.034, -0.022, and -0.011 mL, respectively. There were 11 samples at each temperature, and the equilibration temperature equaled the calibration temperature. The mean differences never exceeded the pooled standard deviation (f0.04 mL) of the original gas-phase volume determination (V,,,,). However, the difference between V,,, and V,,, increased with salinity. The regression line (y = mx + b ) calculated from these data ( n = 33) was Vg,e.= O.94(Vg,) + 0.25 ( r = 0.93). The coefficient of determination (r2)was 0.87 so that eq 10 accounts for 87% of the observed variation in the measured gas-phase volumes from V,,,v Table I1 shows the precision of the static head space CHI determinations with calibrated serum bottles on samples from the lower basin of the Pettaquamscutt River. These samples include discrete samples siphoned directly into the serum bottle from an apparently constant depth and batch samples siphoned to an aspirator bottle, mixed, and then dispensed in the serum bottles. These data encompassed a 100-fold variation in the concentration of CH, (l0-6-lO4 M). The mean rsd was not significantly different throughout this range, but the mean standard deviation was proportional to the concentration, decreasing from 0.16 pmol kg-' for lo* M samples to 0.0018 pmol kg-' for M samples. The batch samples gave the best precision (1.4%), but none of the sample procedures including the equilibration of portions of the same sample at different temperatures adversely affected precision, and eq 6 gave the same results at the three different equilibration temperatures (2,12, and 22 "C). The overall precision for the method was 1.6%. The mean daily precision of the CH4 standard gases (100 and 1000 ppm in He) analyzed at
Table 11. Precision of the Precalibrated Serum Bottle Static Head Space CH, Analysis for Pettaquamscutt River Samples date 10J 12187 26/8/88 27/8/22 29/9/22
type
depth, m
in-situ T , "C
salinity, PPt
discretea discrete discrete discrete
4.9 8.0 8.0 2.0
9.68 17.50 17.50 19.00
24.10 26.10 26.10 16.00
batchd batch batch batch batch batch batch batch
1.0 2.0 3.0 4.0 5.0 6.0 3.5 3.5
4.50 8.00 8.00 8.00 8.00 9.00 23.00 23.00
14.50 20.11 24.89 25.56 25.83 26.26 11.70 11.70
equilibration, T, "C
n
mean, wmol kg-'
sd
25.0 2.0b 2.0 3.1
5 4 2 3
1.52 5.44 5.42 1.34
0.030 0.400 0.040 0.026
3.5 3.5 3.5 3.5 3.5 3.5 2.0b 2.0
3 3 3 3 3 3 2 4
0.06 0.29 2.28 2.90 2.99 4.07 3.90 3.90
0.001 0.004 0.030 0.077 0.049 0.046 0.014 0.058
meanC 5112188 5/12/88 5/12/88 5/12/88 5/12/88 5/12/88 11/8/89 12/8/89
meanC overall mean
1.9 7.5 0.8 1.9 2.2%
mean' 16-18 8/89
rsd
2.5 1.5 1.3 2.6 1.6 1.1 0.4 1.5 1.4%
discrete discrete discrete discrete
0.75 1.75 2.75 4.80
25.39 25.65 25.49 14.32
8.29 9.03 9.95 24.02
2, 12, 22e
3 3 3 3
0.06 1.04 2.85 6.27
0.002 0.030 0.010 0.150
3.7 3.8 0.3 2.4 1.8% 1.6%
"See text for explanation. bThese samples were equilibrated 4 h. cGeometric mean. dSee text for explanation. eThree serum bottles were filled a t each depth, but each was equilibrated a t a different temperature (either 2, 12, or 22 "C).
ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990
2411
110 1
0.40-I 0.39
-
0.380.37
-
0.36
-
0.35
-
f
> 0
0.34 0 0.33 ~
5
10
15
20
25
H f
I 20
10
0
0
SHAKING TIME (HRS)
Figure 3. Time course for the equilibrium of the liquid and gas phases in calibrated serum bottles at 2 "C, as shown by the sample con-
centration of CH, calculated from the gas-phase CH, concentration. Water was from 3.5 m in the lower anoxic basin of the Pettaquamscutt River, August 11, 1989. The precision (error bars) of the method is f1.6%.
the same time was 2.5 and 1.8%, respectively. Daily precision ranged from 0.3 to 5.0% for three to seven standard gas replicates per day. The identical results for both standard and sample analyses were consistent with syringe-based techniques. Better precision would require the implementation of the recommendations given by Weiss (20) for the gas chromatographic determination of CHI (thermostated gas sampling valves, etc.). Regarding sensitivity, inspection of eq 5 illustrates the effect of reducing the ratio Vg/Vl on the concentration of CHI in the gas phase. A decrease in Vg/VI increases the gas-phase concentration (CJ, assuming C: remains constant. The mean phase ratio ( Vg/VI) of 0.07 for the serum bottles in this work gave a 10-fold enrichment of CH, in the gas phase compared to equilibrations of the same samples with equal liquid- and gas-phase volumes. This increased sensitivity reduced the size of the subsample (50-200 pL) needed for gas chromatographic analysis and made in possible to resample without disturbing the phase equilibrium as long as the total pressure in the bottle was kept constant. For waters undersaturated or in equilibrium with respect to atmospheric CHI concentrations, eqs 6 and 9 hold, but for gas chromatographic detection, larger subsamples would be needed (2000 pL). The time needed for the equilibration of CH4 at 2 "C was determined as shown in Figure 3. Because the calibrated bottles were to be used in subsequent incubation experiments, we equilibrated and stored our samples at 2 "C to retard microbial activity instead of risking contamination with a chemical preservative. In Figure 3, the sample CHI concentration calculated from the measured gas phase CH, concentration was unchanged after 2 h of shaking (0.388 pmol kg-' compared to 0.390 and 0.389 pmol kg-' at 4 and 24 h, respectively). During this experiment, three bottles were equilibrated (shaken) for 24 h and then stored at approximately 2 "C for 78 h before analysis. The mean (0.389 pmol kg-') and standard deviations (f0.008 pmol kg-l) for these three samples were identical with the 24-h result. Table I1 shows that for samples taken on August 26,1988, from a depth of 8.0 m, and run on August 26 and 27 equilibration was complete after 4 h even though these samples had a 10-fold higher CH, concentration. The 4-h samples exhibited a lower precision than the bottles analyzed at 24 h. Figure 4 shows the recovery when a volume of a Matheson-certified mixture of 1 % CHI in He or a volume of pure CHI (99.0%) was added to sealed head space bottles by using the sample syringes previously described. The highest con-
1 2
4
10
8
6
12
14
METHANE CONCENTRATION
Figure 4. Recovery of CH, from samples analyzed by the static head space method as a percent of the apparent CH, concentration of these samples. The apparent concentration (pg/kg) was obtained by cai-
culation for the gas addition samples. T "C 10
13
16
19
22
25
28
DEPTH
6 10
100
1000
10 00
log PCH4 (PPW
Figure 5. Vertical distribution of the log of the partial pressure of CH, (pCH, in ppm) and temperature through the oxidized and reduced layers of waters on ewer side of the oxic-anoxic interface (Eh = 0 = dotted line at 3.7 m) in the lower basin of the Pettaquamscutt River estuary. The profile was made on August 15, 1989, and the CH, samples were equilibrated at 2 "C and analyzed on August 16. The error bars are f3.2% of the measured partial pressure, which is twice the mean precision of the analysis (1.6%).
centration of CHI was prepared by adding CH, to head space bottles containing sulfide-free Pettaquamscutt River water from a depth of 2.0 m and a salinity of 16.0 ppt, while the other concentrations were obtained by adding CH, to head space bottles containing distilled water. After addition, the bottles were equilibrated at 2 "C and then analyzed after 24 h. Three bottles were prepared for each gas addition experiment, and the overall precision of these analyses was 2.9%, which is consistent with the results shown in Table 11. The mean recovery for the additions was 92.7 % of the calculated value, and there was no relationship between concentration and the
Anal. Chem. WOO, 62, 2412-2414
2412
percent recovery over the concentration range tested (0.2-15.0 wmol kg-'). A representative CHI profile obtained in the lower basin of the Pettaquamscutt River is presented in Figure 5. It depicts temperature and the log of the partial pressure of CHI (pCH,) in ppm above and below the oxic-anoxic interface. Given the magnitude and variation (39-4200 ppm) of the in situ CH4 concentrations in the lower basin, the precision of the method (