Determination of natural 32P and 33P in rainwater, marine particles and plankton by low-level beta counting

Nuclear Instruments and Methods in Physics Research A 338 (1994) 560-567 North-Holland

NUCLEAR INSTRUMENTS & METHODS IN PHYSICS RESEARCH Section A

33P

Determination of natural 32 P and in rainwater, marine particles and plankton by low-level beta counting N.A. Waser *°', A.P. Fleer, T.R. Hammar, K.O. Buesseler, M.P. Bacon Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA

(Received 6 July 1993) Methods were developed for determining the natural levels of 32 P and 33 P in samples of rainwater, marine particulate matter and plankton . Preconcentration from rainwater consists of the extraction of radiophosphorus on alumina. Filtration of large volumes of seawater is required to obtain sufficient quantities of suspended particulate matter. The radiochemical scheme, which consists of a series of precipitations and a column separation, results in a pure source containing the two beta emitters . The activities of 32 P and 33 P in the source are determined separately by an external absorber method on low-level beta counters . The yields are monitored by the addition of stable phosphate in rainwater samples and by the natural levels of stable phosphate in marine particulate matter and plankton . The methods allow the specific activities and the activity ratio to be determined with a 10% precision. 1. Introduction A wide variety of radioisotopes are produced by cosmic rays in the atmosphere, in the oceans and in rocks. These radiotracers have been applied to problems in geochemistry, geology, atmospheric chemistry and oceanography . In atmospheric chemistry, studies of 3ZP (t 1 1 2 = 14 .28 d, Emax =1 .71 MeV) and 33p (t112 = 25 .3 d, E maX = 1.71 MeV) can provide information on the residence time of tropospheric aerosols . In the upper ocean, they allow in situ rates of biological uptake of phosphorus and turnover times of P to be determined [1-5]. Very low counting backgrounds and, more importantly, high activities are essential to the detection of 32P and 33 P by low-level ß counting . The background can be reduced by the use of an anticoincidence system surrounding the primary detector . High activities can be obtained by preconcentration in the field. Preconcentration of the activity in rainwater is achieved here by concentrating the activity of the dissolved phase on alumina, a specific adsorbent of phosphate, in a system similar in principle to the one used by Silker [6]. Large-volume filtration is utilized to collect suspended particulate matter . To separate 3Z P and 33 P a nuclear technique must be used . The two isotopes are distinguished by their half-lives and by the very distinct * Corresponding author . t Present address: Department of Oceanography, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada .

energy spectra of the electrons emitted in their decay. Their half-lives are not different enough to enable their separation by decay-curve analysis. In contrast the absorption characteristics of the two isotopes differ enough that an external absorber of a given thickness will absorb all the radiation from 33 P but only a small fraction of the radiation from 32 P. The sampling protocols, radiochemical procedure, and counting procedure are explained in this paper. Specific activities of 32p and 33 P and activity ratio 33p/32p in rainwater collected at Woods Hole and in plankton samples collected off Bermuda at the Bermuda Atlantic TimeSeries station (31°50'N, 64°10'W) are presented. 2. Sampling protocols Rain samples were obtained with a collector (4 mZ) situated on the roof of the Clark Building in Woods Hole . The sampler enabled the collection of the 10 to 60 I of rainwater necessary to determine radiophosphorus. The rainwater was spiked with 100 wmoles of stable KH Z P04, which served as carrier and enabled the yield of the extraction and purification procedures to be determined . The sample was left to equilibrate for 4-6 h and was then pumped (with a peristaltic pump) through a plexiglas unit containing about 7 g of alumina (activated alumina, 98% powder, Matheson Coleman and Dell) packed between two porous polyethylene filters (3 .5 cm in diameter). The extraction efficiency of P04 was better than 98% for a volume of 60 1 of rainwater and flow rates of 5 1/min

0168-9002/94/$07.00 © 1994 - Elsevier Science B.V . All rights reserved SSDI 0168-9002(93)E0844-1

N.A . Waser et al. I Nucl. Instr. and Meth . in Phys. Res. A 338 (1994) 560-567 or less . This efficiency was confirmed by laboratory experiments. In the laboratory, 50-140 1 of distilled water and filtered seawater (0 .2 wm) from Vineyard Sound were spiked with stable phosphate . Phosphate was extracted on alumina with efficiencies higher than 98% in the case of distilled water . In seawater, the efficiency was lower and variable . Different size fractions of particles were sampled . Particles were collected with 293 mm diameter Millipore" filters (3 or 8 wm mesh size) . These samples were collected with a simple filtration system which consisted of either an impeller pump or a diaphragm pump connected to a 293 mm filter holder . The pump was placed upstream of the filter holder. For surface samples a vacuum hose was lowered to a depth of 3-5 m over the side of the ship . For deep samples the vacuum hose was attached to a weight on the hydrowire, and the rest of the hose was allowed to float

561

freely at the surface . With this system we could pump 1000 1 at 10 to 45 psi in about 1 h. The filter clogged fast and was changed as soon as the flow rate started to decrease, which was about every hour. The filters were immediately frozen to avoid release of phosphorus compounds [7] . Particles larger than 67, 150, 300 or 500 ltm were harvested with plankton nets of the indicated mesh size. The particles were sampled at the surface and at depth . Samples were recovered from the cod end, gravity filtered on WhatmanTM filters and immediately frozen .

3 . Radiochemical procedure The A1 2 0 3 is removed from the filter holders and slurried in 1N NaOH . NaOH (1N) removes adsorbed P

Fig. 1 . Analytical scheme for the purification of radiophosphorus . Reagent 1 refers to a solution of ammonium molybdate. Reagent 2 refers to a solution containing MgCl and NH 4CI.

562

N.A . Waser et al. /Nuct. Instr. and Meth. in Phys. Res. A 338 (1994) 560-567

from A' 203, but large volumes are necessary for good recoveries (> 50%). For rain samples, 0.8 to 1 1 of 1N NaOH is typically used and subsequently boiled down to about 200 ml . The alkaline solution is then poured gradually into a 6N HN03 solution . The pH of the solution is kept below 2 to avoid formation of massive amounts of aluminum hydroxide colloids . The frozen MilliporeTM filters and plankton samples are refluxed for a few hours in concentrated HN03 until the solution is clear. If refractory particles are present, the solution is filtered . The solution is then taken down to a small volume and diluted to about 1N HN03 with distilled water. An aliquot of the solution is analysed for phosphate (i .e ., the yield monitor) by the classic molybdenum blue method [8,9]. From this point on, the procedure is the same for all the different types of sample. The chemical purification of phosphorus is based on a series of specific phosphate precipitations, which have been described in the classic chemistry and radiopurification methods [10-12] and have been developed recently for applications to oceanographic problems [4]. The flow chart of the purification scheme is presented in Fig. 1 . The sample solution in HN03 is heated, and an excess of ammonium molybdate is added. This solution is heated to about 30-40°C and stirred until a yellow precipitate of ammonium phosphomolybdate (NH4)3 P04(Mo0 3 ) 12 appears. The ammonium phosphomolybdate precipitation is a preliminary separation for reducing the amount of heavy metals present, particularly Fe, Co, Ni, Cr, Ti and Zr [11] . While stirring is maintained, heating is stopped and the precipitation process is allowed to continue . If heating continues too long (at temperatures higher than 50°C), molybdic acid will start to precipitate, and the precipitate will be contaminated by Si, As and V [10] . After about 1 h the precipitate is vacuum filtered through a Millipore TM HA filter (0 .45 wm, 47 mm in diameter) and washed with 1N HNO3 . Occasionally the precipitate is so fine that a 0.2 wm MilliporeTM filter has to be used . The precipitate is dissolved in NH 40H and the solution acidified to 1N HN03. Ammonium phosphomolybdate is precipitated a second time and redissolved with ammonia. The pH of the solution is then lowered to about 7 with HCI. The solution is cooled in an ice bath, and a reagent containing MgCl 2 and NH 4Cl is added to the cold solution . Drops of NH 40H are added while stirring, leading to the formation of a white crystalline precipitate of ammonium magnesium phosphate. An excess of concentrated NH 40H is added. The NH 4 MgP0 4 - 6H 20 precipitate in solution is stirred for 15 min and cooled for half an hour in an ice bath . This step is not a separation, since there are a large number of interferences included with the precipitate . The advantage of the NH 4 MgP0 4 - 6H 20 precipitate is that it can be dissolved in HCl while (NH 4 )3PO4

(Mo0 3)12 cannot, thus facilitating the subsequent column separation . The NH 4MgPO 4 - 6H 2 0 precipitate is vacuum filtered onto a Millipore HA filter (0 .45 wm, 47 mm in diameter), washed with dilute NH 40H, and dissolved in 9N HCl. A cation exchange column of AG-50W-X8 Cl, 100-200 mesh (analytical grade resin from Biorad Laboratory) is conditioned with 3 volumes of 9N HCl. The solution is loaded on the column and 1 volume of 9N HCl is passed through to rinse. The cation exchange resin allows separation of Ca, K, Fe, Al, V, W, Zn, Zr and Ti, which will interfere with the final NH 4 MgPO4 - 6H 2 0 precipitation [10-12]. Concentrated NH 40H is added to make the pH neutral, and the solution is put again in an ice bath prior to the last precipitation. NH 4MgP04 - 6H 2 0 is precipitated as described previously and vacuum filtered on a preweighed Millipore HA filter (0 .45 wm, 25 mm in diameter). The area of the precipitate is 2.75 cm 2. The precipitate is hygroscopic, and it is thus dried carefully until the weight is constant . It is then mounted face down on a film of Mylar TM (thickness = 0.9 mg/cmZ). The film of Mylar is supported by a white Delrin TM ring . A silver planchet (140 mg/cm2 ) is placed on top of the filter, and the source is sealed with tape to ensure no change in the weight of the precipitate . 4. Counting procedure 4.1. Separation procedure

The source is counted on an anticoincidence lowlevel ß counter similar in design to that of Lal and Schink [13] . Typically, the background of the counters ranges from 0.25 to 0.45 ± 0.02 cpm. A nuclear method is used to separate the two radioisotopes of P based on the difference in energy of the beta particles emitted in their decay. The intensity of a source I as a function of the external absorber thickness x is simply approximated by : I =I0 exp[-(ln 2Xx/x12)], where Io is the intensity of the source with no external absorber [14] . The activity of 32p is decreased by a factor of 2 when an aluminum foil of 72 ± 2 Mg/CM2 is placed between the source and the detector (Fig . 2) . The half-thickness of absorption is lower than the estimate of 84 Mg/CM2 of Libby [14] .The estimated mass absorption coefficient A(~u = In 2/x1 2) of 9.7 ± 0.3 cm 2/g is comparable to the mass absorption of 10 .8 cm 2/g calculated from the empirical formula: A = 22/Em 33 for 0.5 < Emax _< 6 MeV [15] . The difference observed in half-thicknesses produces a change in the calculated absorbed intensity of 1 .5 to 2.4% for the range of absorber used, i.e ., 10-20 mg/cm 2. In the case of 33 P, a foil of only 4.3 ± 0.1 Mg/CM2 blocks 50% of the R radiation (Fig . 3) . Thus from the absorption curves of 32 P and 33p one

N.A . Waser et al. /Nucl. Instr. and Meth. in Phys. Res. A 338 (1994) 560-567

563

can choose a thickness of aluminum that will block essentially all the radiation from 33 P and decrease the radiation from 32 P by a known small amount . The thickness of aluminum is chosen so that the estimated activity of 33 P in the source would be decreased to less than 0 .02 cpm when the source is counted with the absorber . In order to separate 32 P and 33 P the samples are counted repeatedly in time with and without external absorber . The time-dependent curve generated without absorber represents the total activity of the sample . The curve with absorber represents the activity of 32 P, decreased slightly due to the absorption of 0 particles in the absorbing material . The initial net count rates 32 No and 33 N o of 32 P and 33 P (i .e., the net count rates at the time the sample was first counted) are then determined by the difference of the two time-dependent curves corrected for external absorption . The equations describing the total gross count rates without absorber Nt and with absorber Na at a given time are the following :

ïsU m N

U z"

(1) 32N +33N, 33 = 32 No exp(-32At), 32 N3213 + 33N ß 32,8 (2) Na = where 32N and 33 N are the net count rates of 3Z P and 33P; 3zß is the fraction of 32 P transmitted through the absorber, i.e ., 328 = exp(-ln 2x/71 .6); x is the thickness of the absorber utilized ; 33 /3 is the fraction of 33 P transmitted through the absorber, i.e ., 338 = exp(-In 2x/4 .3); and 32 A is the decay constant of 32 P. In practice the samples are counted so that each count in time has an 8% or less counting error, both with the external absorber and without. Typically, the thickness of the absorber ranges from 10 to 20 mg/cm 2. Na (i.e., 32 N 3zß + 33 N 3313) is usually approximated by 32N 3213 . The two curves N,(t) and Na(t) are then fitted by least squares. The source is counted repeatedly over 20 to 60 days . The general equations used for the fit are: Nt

Absorber thickness

(mg/cm2)

Fig. 2. Net count rate of 32P as a function of the external absorber thickness.

=

N, =a exp(-at), Na =b exp(- 32 At) .

Absorber thickness (mg/=2)

Fig. 3. Net count rate of 33 P as a function of the external absorber thickness.

(3) (4)

Nt is always approximated by a simple exponential decrease, because the decay curve Nt does not show evidence for the presence of two radionuclides (see Figs . 5 and 6). This is because the half-lives of 32 P and 33P are so close to one another. The values of 32No and 33 No were determined by the intercepts, a and b, of the two curves with the y axis . At this point the term 33 N0 33,6 was computed to verify that it was indeed much smaller than 32No 328 as approximated in Eq . (2). 33N o 3313 was found to be < 0.02 cpm, which corresponds to the error on the background of the 0 counters . For some samples of extremely low total activity, thin absorbers have to be used to be able to separate the two radionuclides. The residual activity of

564

N.A . Waser et al. /Nucl. Instr. and Meth. i n Phys. Res. A 338 (1994) 560-567

33p in the absorber curve (33N 33ß > 0.02 cpm) can be corrected for by subtracting the estimated activity of 33P from the curve with absorber . The correction is iterated until the calculated values 32N o and 33No converge to constant values . The initial net count rates, 32 N' and 33N O, are converted into absolute activities 32Ao and 33A0 by taking into account the counting efficiency, e, and the yield of the procedure, y. The absolute activity at the time of collection is : A(('=(Nu/yE) exp(At), where t is the time elapsed between collection and initial counting time, and A is the decay constant . The activity ratio 33p/32p (R) is R = 33A0/ 32AO =33N0 32C / 32 N0 33 6 . The counting efficiency, e, for a given radioisotope varies with the thickness of the source according to the following relationship [14] : e/EO= (1 -exp(-A d))l(w d), (7) where A is the self-absorption coefficient in cm 2/Mg; d is the thickness of the source in mg/cm2; and e and e() are the counting efficiencies for a source of thickness d and for an infinitely thin source, respectively . Typically d varied from 2 to 12 mg/cm2. The thickness of the silver planchet used for backing material was about 140 mg/cm2, which is well beyond the saturation thickness of 33 P (12 mg/cm2) and very close to the saturation thickness of 32 P (160 mg/cm2) [16] . 4.2. Calibration Phosphorus-32 is commercially available from DuPont in a solution of KH 2P04 . This 32p source requires standardization. Verification of the purity of the tracer was obtained by determining the half-life and the half-thickness of absorption as described previously (Fig. 2). The half-life and the half-thickness were, within the errors, equal to the expected values . The solution is standardized by counting a thin source of 32 P, evaporated on a stainless steel planchet on a 21r gas-flow proportional counter (Nuclear Measurements Corporation, model PCC-11T) . The counter had been calibrated by the manufacturer with beta-emitting sources of different E., and was recalibrated at WHOI with a standard of (47Pm (E maX = 0.224 MeV; Amersham). The calibration agreed with that of the manufacturer and a 72% counting efficiency was determined for 32p on the 2 ,rr gas-flow proportional counter. The standardization of the 32 P in KH 2 P04 solution from DuPont being done, the calibration of the lowlevel ß counters could then be undertaken . For this purpose, a known amount of 32p was precipitated as NH 4 MgP04 - 6H 20 with varying amounts of stable

P0 4. The sources were dried to constant weight and then sealed and counted. The counting efficiency for 32p was roughly constant in the range of thicknesses encountered in all the samples. The efficiency for an infinitely thin source, Eo, was calculated by evaporating drops of a standard solution of 32 P on an area of 2.75 cm 2, equal in size to the area of the precipitate, centered on the Mylar film . A MilliporeTM filter and a silver planchet were placed on top of the source . The efficiency for an infinitely thin source and for all thicknesses from 2 to 8 Mg/CM2 was 49% . This efficiency is slightly smaller than the 55% efficiency of 234Th (and 234 Pa) on the same system [171. A standard of )47Pm from Amersham is used as an analogue of 33 P. The )47Pm standard is evaporated on a silver planchet, and its absorption curve is generated with aluminum absorbers. The measured half-thickness of absorption is 4.4 ± 0.1 mg/cm2, within experimental error identical to the value of 4.3 ± 0.1 Mg/CM2 for 33 P. A known activity of 147Pm is evaporated on a Mylar film . A filter and a silver planchet are put on top of the mylar film . The ratio of the net count rate to the activity of this standard is the counting efficiency of 33 P, 33E(), for an infinitely thin source . To calculate 33 e for all thicknesses, a carrier free unknown activity of 33P (in KH 2P0 4 purchased from DuPont) is precipitated as NH 4MgP04 - 6H 20 with increasing amounts of stable P04. A curve was generated of the net count

0 .96 0 .941 0 .921 0 .901 0 .88

.84 0 .82 0 .80 0 .78 0 .76' 0 .74 0 .72 0 .70 0 .68 0 .66 1 .0

1 .5

2 .0

2 .5

3 .0

Source thickness

3 .5

4 .0

4 .5

(mg/=2)

Fig. 4. Relative counting efficiency of 33 P as a function of source thickness.

565

N.A . Waser et al. / Nucl. Instr. and Meth. in Phys. Res. A 338 (1994) 560-567 rate as a function of thickness. The net count rate for 0 33 t47Pm standard was assumed to be E.. The allowed the value of EO (i .e ., 26%) to be determined . The rest of the curve could then be calibrated . The curve did not exhibit any maximum as observed previously for weak beta emitters [18], possibly because (1) the maximum occurs at < 1 mg/cm Z or (2) self-scattering is small in NH 4 MgPO4 - 6H 2 0 . Therefore the efficiencies might be offset by up to 3% when Eq . (3) is used to calculate the efficiency of 33 P . Fig . 4 shows the counting efficiency, 33E, as a function of sample thickness . The counting efficiencies for an infinitely thin source, 33 E0, were the same within 1% for all four counters used in this study. The relative efficiency curve was fitted by the function given in Eq . (3) . The exponent u of the curve was determined on each detector and varied between 0 .145 to 0.160 and averaged 0 .150 ± 0 .01 cm'/mg . These values are identical to each other within the error of the regression . The error on the absolute efficiency is estimated to be less than 2% for an error of 0 .01 cm Z /mg on A . The counter efficiency e (expressed in %) for a source thickness d (expressed in mg/cm 2 ) is calculated from the experimental relationship :

mg/cmZ

E = 26(1- exp(-0 .150d))/(0 .150d) .

(8)

4.3. Yield The chemical yield is calculated as the ratio of the amount of stable P in the source relative to the amount of stable P initially present. The method for determination of the yield varies according to the sample type . For rainwater samples 100 wmoles KH 2 PO4 are added to each sample . The amount of P0 4 present in the final NH 4 MgPO4 - 6H 2 0 precipitate is determined colorimetrically with the classic molybdenum blue method [8,9] . For plankton and suspended particulate samples, the initial amount of PO, is measured colorimetrically after digestion with nitric acid. When the amount of phosphate exceeds 100 wmoles, the sample is split so that thin sources can be obtained to minimize self-absorption of 33P in the source . Experience showed that 50-100 wmoles of stable P in the source optimized the total net count rate and minimized self-absorption of 33P for samples collected in the Sargasso Sea . Usually, for plankton samples obtained from 15 to 20 min tows, the initial amount of stable P exceeds 100 wmoles and the sample is split . For suspended particles, the samples typically have 30-60 wmoles of P0 4 initially and are not split. 4 .4. Errors The errors on the absolute initial activities 32A° and 33A0 are computed for the start time of the first count .

Nt (t) and Na(t) are fitted with a least-squares method and the errors aN,o and QN o are given by the nonweighted least squares fit (samples typically are counted long enough to accumulate 300-500 counts for low count rate samples and > 1000 counts for rain samples) . The errors Q32 N o and Q33N o are deduced from 32 propagation theory . Since the backgrounds and /3 are well known (the errors associated with these quantities are small), Q32 N o and Q33 N o simplify to : ((T3z No) 0`33

N'

z

= 1/ 32 '6QN o, z

(10 )

The details of the calculation are given in Waser (1993) . The errors Q32A o, o-33A o, and oR are then easily derived by applying propagation theory. In practice, for the rain samples the errors associated with the counting statistics are small (usually < 2%) and thus the errors on the initial activities, 32AO and 33A ° , and the ratio, R, are dominated by the errors in the counting efficiencies . For plankton samples, the net count rates are much lower in general and the errors are due to both counting statistics and counting efficiencies . 5 . Results and discussion The net count rates with and without absorber are shown in Fig. 5 for a rain sample collected at Woods Hole and in Fig . 6 for a plankton tow sample collected off Bermuda . The figures show typical count rates e .OT 6 .01 5 .0y 4 .0-

Total

3 .o-

NN

2 .0

é U z

1.

o. o .e

0 .70 . 6'

0 .5- _

Absorber 23 mg/=2

0 .4' 0 .3+ 0

25 Days

30

Fig . 5 . Net count rate with and without aluminum absorber, Na and N, respectively, of a rain sample collected at Woods Hole on January 16, 1991 .

566

N.A . Waser et al. /Nucl. Instr. and Meth . in Phys . Res. A 338 (1994) 560-567 ited by the accuracy of the determination of the counting efficiencies . The precision of the measurements is

3 . OT

2 .t

limited by the precision of the counting efficiencies, the yield and the counting statistics for the low activity

~ Total

samples. The reproducibility of the measurements (specific activities and ratio) is found to be estimated within the 10-13% precision. The specific activities of

1 .s

3ZP in plankton are compared to previous measure-

ments and are found to be higher by factors of 10 to 100 (Table 1). This is explained by the differences in the amount of stable phosphate in seawater in the

different

ocean provinces, the Sargasso Sea off Bermuda being an especially phosphate-poor region. The specific activities of 3Z P and 33 P and the activity

0 .4t 0 .3t

ratio 33 P/ 32p in rainwater are in good agreement with the previous measurements done both at Bombay and Westwood, New Jersey (Table 1).

Absorber 20 .4 mg/=2

4

6

8

10

12

14

16

18

20

22

Days

Fig. 6. Net count rate with and without aluminum absorber, N and N respectively, of a plankton tow sample collected off Bermuda on May 5, 1991 .

observed for rain samples (20 to 60 1) ,and plankton tows (15 to 20 min). Suspended particulate matter

samples filtered from 2000 to 5000 1 of seawater show

very low count rates, similar in magnitude to the plankton tow samples. The specific activities of 3Z P and 33 P and the activity ratio 33 P/ 32p in rain samples and in triplicate samples from a plankton tow are presented in Table 1. The accuracy of the measurements is lim-

6. Conclusion The method described allows the measurement of natural levels of 32 P and 33 P in rainwater (10 to 60 1), suspended particulate matter (filtered from 2000 to 5000 l of seawater), and plankton (15 to 20 min tows

off Bermuda). The individual activities and the activity ratio in such samples can be determined to within

roughly 10%. The successful measurement of the weak emitter, 33 P, in particulate phases was due to the low amount of stable P in the samples collected off Bermuda. In ocean provinces containing ten times more stable P, self-absorption of the low energy ß particles

Table 1 Specific activities of 3Z P and 33 P and activity ratio 33 P/ 32p in rain samples collected at Woods Hole and in triplicates of a plankton tow (300 [Lm mesh net) collected off Bermuda. For comparison previous measurements are listed in the lower part of the table Sample type Plankton tow triplicates Bermuda Rainwater Woods Hole

Reference

5 May 1991 5May1991 5 May 1991

this work

27 Dec 1990 16 Jan 1991 21 Apr 1991

this work

Plankton tow South California Bight Bedford Basin Rainwater Bombay Bombay Bombay Westwood, New Jersey

Lal and Lee [1] Lal et al . [2] Lai et al. [l8] Goel et al . [19] Lal et al . [20] Walton and Fried [21]

32P/P

33P/P

32P

33P

33p/ 32P

[mBq/mg P] a

[mBq/mg P] a

[mBq/1]

[mBq/1]

[Bq/Bq]

13 .2+1 .2 11 .8±1 .2 11 .2+1 .2

17 .5+2.2 18 .3±1 .8 15 .0+2.0

1.33+0.17 0.17 1 .55+0.22 1 .34+0.19 13 .7+11.5 .5 6.8+0.7 15 .0±1 .7

16 .3+1,7 8.5+1 .0 12 .3±1.7

0.09 1 .21+0.09 1 .24+0.11 0.82±0.13

4.7-15.8

0.40-0.56

0.13-17 0.33-40 11-39 1-25 6-63 8-48

There is roughly 1 mg P per 1 g of dry plankton off Bermuda. NB : 1 o error is reported for each sample .

N.A . Waser et al. /Nucl. Instr. and Meth . in Phys. Res. A 338 (1994) 560-567 emitted by 33p in the source would prevent determination of 33 P by the present method and counter design . References [1] D. Lal and T. Lee, Nature 333 (1988) 752. [2] D. Lai, Y. Chung, T. Platt and T. Lee, Limnol . Oceanogr . 33 (1988) 1559 . [3] T. Lee, E. Barg and D. Lai, Limnol . Oceanogr. 36 (1991) 1044 . [4] T. Lee, E. Barg and D. Lai, Anal . Chim. Acta 260 (1992) 113. [5] N.A . Waser, Ph.D . Thesis (WHOI/MIT, 1993). [6] W.B . Silker, R.W . Perkins and H.G . Rieck, Ocean Engng. 2 (1971) 49 . [7] R.W. Collier and J.M. Edmond, Prog . Occanogr . 13 (1984) 113. [8] J. Murphy and J .P. Riley, Anal . Chim . Acta 27 (1962) 31 . [9] F. Koroleff, in : Methods of Seawater Analysis, eds. K. Grasshoff, M. Ehrhard and K. Kremling (Verlag Chemie, Weinheim, 1983) p. 125. [10] W.F . Hillebrand, G.E.F. Lundell, H.A. Bright and J.I .

[11] [12]

[13] [14] [15] [16] [17] [18] [19] [20] [21]

567

Hoffman, in : Applied Inorganic Analysis (Wiley, 1959) p. 694 . W.T. Mullins and G.W. Leddicotte, The Radiochemistry of Phosphorus (NAS-NS-3056, Washington DC, 1962). T.P . Whaley and L.W. Ferrara, in : Environmental Phosphorus Handbook, eds. E.J . Griffith, A. Beeton, J.M . Spencer and D.T. Mitchell (Wiley, New York, 1972) p. 313. D. Lai and D.S . Schink, Rev . Sci. Instr. 31 (1960) 395. W.F. Libby, Anal . Chem . 19 (1947) 2. W.J . Price (ed.), Nuclear Radiation Detection (McGrawHill, New York, 1958) p. 18 . G.R . Choppin and J. Rydberg, (eds .), Nuclear Chemistry Theory and Applications (Pergamon, New York, 1960). A.P . Fleer, in : Marine Particles: Analysis and Characterization, eds. D.C . Hurd and D. W. Spencer (Geophysical Monograph 63, Washington DC, 1991) p. 227. D. Lai, N. Narasappaya and P.K. Zutshi, Nucl . Phys ., 3 (1957) 69. P.S . Goel, N. Narasappaya, C. Prabhakara, Rama Thor and P.K. Zutshi, Tellus XI (1959) 91 . D. Lai, Rama Thor and P .K . Zutshi, J. Geophys. Res. 65 (1960) 669. A. Walton and R. Fried, J. Geophys. Res. 67 (1962) 5335 .