- Email: [email protected]
Determination of copper, nickel, and cadmium in sea water by apdc chelate coprecipitation and flameless atomic absorption spectrometry
Anolytica
Chimica Acta,
91 (1977)
189-197
OElsevier Scientific Publishing Company,
Amsterdam - Printed in The Netherlands
DETER&lINATIQN OF COPPER, NICKEL, AN-D CADMIUPcf IN SEA WATER BY APDC CHELATE COPRECIPITATION AND FLAMELESS ATOMIC ABSORPTION SPECTROMETRY
EDWARD A. BOYLE
and JOHN hi. EDMOND
Department of Earth and Ranetary Cambridge. JfA 02139 (U.S.A.)
(Received
Sciences.
dfassachusetts
Institute
of Technology,
1st November 1976)
SUMMARY Copper, nickel, and cadmium can be determined in 100 ml of sea water by coprecipitation with cobalt pyrrolidinedithiocarbamate and graphite atomizer atomic absorption spectrometry. Concentration ranges !ikely to be encountered and estimated (1 o) analytical precisions are l-6 nmol kg-’ (20.1) for copper, 3-12 nmol kg-’ (tO.3) for nickel and 0.0-1.1 nmol kg-’ (~0.1) for cadmium. The technique may be applied to fresh-water samples with slight modification.
This report describes a technique for the determination of copper, nickel, and cadmium in sea water. The metals are concentrated from a lOO-ml sample by cobalt pyrrolidinedithiocarbamate coprecipitation [ 11. The precipitate is redissolved in an organic solvent and digested to simplify the sample matrix. The resulting solution is analyzed by atomic absorption spectrometry with heated graphitz atomization. Extension of the method to freshwater samples is discussed briefly. EXPERIMENTAL
Prevention
of contamination
Quantitative methods for elements at very low concentration levels must preclude contamination of the sample, as well as satisfy accuracy and sensitivity requirements. Elaborate clean-room techniques entail considerable inconvenience in ternIs of cost and availabili@ and do not entirely eliminate introduction of contamination by the analyst. For this reason the method was designed for use in an ordinary laboratory with the precautions outlined below. Reagents are stored in polyethylene or polypropylene bottles and leached in 1 M hydrochloric acid for at least one day. For acid storage, bottles are leached in f3 &I hydrochloric acid at 80°C for 2 h. When not in use, bottles are covered with plastic ba@
All glassware and plastic ware are stored in a covered bath wit.h a leaching solution consisting of 0.1 M hydrochloric acid and 0.1 M nitric acid. Immediately prior to use, the glassware is rinsed with high-purity distilled water and used without drying. The distilled water should be of sufficient purity to give no detectable blank if treated as a sample. Other precautions are that: disposable poIyethylene gloves are used for critical handling steps; samples are allowed minimal exposure to open laboratory air; microliter pipette tips are flushed several times with 0.1 M hydrochloric acid before each use; and silicone rubber stoppers are kept in 2% EDTA solution between each use. Other precautions are mentioned ia the Procedure. Although not used here, a laminar-flow clean bench would help ensure integrity of the sample during manipulation steps and is recommended. Equipment Two special items are illustrated in Fig. 1. The filtration funnel (A) is made by joining a 3.5ml fine sintered glass frit to the bottom of a tall-form 200-ml Pyrex beaker. The redissolution apparatus (B) is made from plexiglass and sihcone rubber stoppers. A frame for securing the filtration funnel into a 250-ml filter flask is necessary; this frame may be constructed from plexiglass with l/S-ino.d. tygon tubing straps with glass hooks. Reagercts Cobalt chloride sclution. Dissolve 0.2 g of ultrapure CoCl* - 6HZ0 (Spex Industries) in 250 ml of distilled water. APDC solution. Purify a 2% ammonium pyrrolidinedithiocarbamate solution in water by repeated extraction with carbon tetrachloride [ 21. Acid-acetone mixture. Prepare a (95 + 5) solution of acetone and 0.1 M hydrochloric acid; dispense from a 250-ml polyethylene squeeze bottle. Nitripperchloric acid mixture. Prepare 6 M nitric acid, redistilled in a vycor glass still, containing 1% (v/v) ultrapure perchloric acid. ,SILICOF:E
II )
A
2oOML TALL FCi7.4 BEAKER
IS F FAITTED
Fig. 1. (A) Filtration
RUS3ER
B TEFLON
SEAKER PLEXlCbaS
FUNSEL
funnel. (B) Dissolution
STOPPER
\ apparatus.
FRAME
SLICCNE RU59ER STOFi’ER (PARTIALLY MLLOW 1
Ammonium sulfate soiufion Prepare 3 - 1OW2M ammonium sulfate (reagent grade) in I.5 10-’ M sulfuric acid (reagent grade). Hydrochloric acid (6 M). Prepare from acid redistilled in a vycor glass still. Standards. Prepare nickel and copper standards (20 clmol ml-‘) and cadmium standard (10 pmol ml-‘) in 0.1 h? hydrochloric acid. l
Procedure Sampfes should be acidified to pH 2 by addition of 6 M hydrochloric acid immediately after collection. This procedure minimizes adsorption on to container walls [ 31. Weigh out approximately 100 g of sample into a 200-ml talEform beaker. Add 250 ~1 of cobalt chloride solution and swirl gently to mix thoroughly. Add 500 ~1 of APDC solution, swirl gently to mix thoroughly, and allow the precipitation to proceed for 5 min before filtration. During this waiting period, remove the filtration funnel from the acid bath, rinse with distilled water, and strap into the filtration rig. Apply full vacuum and rinse the interior of the filtration funnel with two portions of 6 M hydrochloric acid followed by two portions of distilled water. Then filter the sample using full’vacuum. Rinse the interior of ‘Lhe funnel with a small portion of distilled water to remove the last traces of sea salt. Release the vacuum slowly by means of a stopcock valve with an in-line air filter. Remove a 5-ml Teflon beaker from the acid bath and rinse with distilled water. Place it into the dissolution apparatus. Rinse the tip of the f3ltiation funnel with acid--acetone solution and place the fwmel into the top of the dissolution apparatus. Rinse through several small portions of acid-acetone mixture by applying a gentle vacuum.. This procedure dissolves the precipitate and washes it into the beaker. Release the vacuum; remove the beaker, and carefully remove the filter funnel from the dissolution apparatus. Remove the traces of the chelate from the tip of the filter funnel by washing with acid--acetone mixture and collecting the wash in the Teflon beaker. Cover a hot plate with a gIass fiber filter. Place the beaker on the hot plate at very low heat (to avoid bumping) and cover with an inverted 20-ml Pyrex beaker; the glass fiber minimizes spattering of condensate running down the inside of the cover. Tilt the beaker by placing one edge on a glass rod, to prevent condensation from falling back into the beaker. Evaporate to dryness, Add 500 ,~l of nitric-perchlorie acid mixture and evaporate to dryness at slightly higher heat. Avoid charring the residual organic material by overheating. After cooling, the beaker may be covered with Parafilm and stored indefinitely. Along with the samples, prepare a matrix for standards by pipetting 250 1_t1 of cobalt chloride solution into each of three 5-ml beakers, adding 10 pi of APDC solution to each, and evaporating and treating with acid mixture as in the preceding paragraph. Determine blanks by running high- . purity water as a sample. The water may be tested for traces of the metals by varying the volume of several blank samples.
14.2
About one hour before the atomic absorption analysis, pipette 2000 ~1 of 0.1 M hydrochloric acid into each sample and standard matrix beaker. While allowing 15 min for the residue to dissolve, prepare a secondary mixed stzndard by diluting 10 ~1 of cadmium standard plus 50 ,ul of copper standard and 100 ~1 of nickel standard to 100.0 ml with distilled water. Mix thoroughly and pipette 50 ~1 into one of the standard matrix beakers and 100 ,ul into another; the third matrix beaker is used for a matrix blank. After the dissolution period, swirl each beaker gently to mix the solution and rinse the sides of the beaker. This solution is used for the copper and nickel determinations. Prepare a diluted so!ution for the cadmium determinations by pipetting 250 ~1 Born each concentrated sample and standard into dry 5-ml Teflon beakers. Add 750 ~1 of ammonium sifate solution and mix by gentle swirling. Cover all beakers with Parafilm and analyze immediately to minimize evaporation. Determinations were made with a Perkin-Elmer HGA 2100 graphite furnace, model 403 atomic absorption spectrophotometer and model 56 chart recorder. Manufacturer’s recommended operating conditions for wavelength, slit and lamp current were followed. Instrumental parameters for the graphite furnace are given in Table 1. Deuterium arc background correction was applied, with the deuterium arc beam aligned to coincide with the hollow-cathode beam at the center of the graphite tube. Optical windows were kept clean by recommended procedures and were masked outside the optical path to minimize light scattering in the optics. The chart recorder was set to 0.5 absorbance full scale with minimum pen damping. 7JJ’pical peak height output is illustrated in Fig. 2. A 100-g sample with 1 nmol kg-’ of each element processed as described above should give peak aksorbances of approximately 0.02 for copper, 0.01 for nickel, and 0.1 for cadmium.
TABLE
1
Instrumental Element
CU
parameters Sample volume (PI) 50
for HGA
2100
Gas flow continuous (cm’ min-‘)
Dry cycle
Char cycle
30
20 5 130°C
25 s 850°C
2400°C 12s 2600° C
Ni
100
30
35s 110°C
20 1150°C
Cd
50
20
25s 110°C
20s 500°C
=For Cd, this was followed
by cleaning for 5 s at 26OO’C.
Atomization cycle
10s
5s 1600°C’
193
COPPER
NICKEL
CADMIUM
Fig. 2. Representative peak height output for copper, nickel, and cadmium. Du$icate injections are ixzdiczted by brackets. Peak height expected for 1 nmol kg-’ (Cu. Ni) and 0.1 nmol kg-’ (Cd) in a 100-g sample is indicated by bar.
All samples and standards should be injec’ted in duplicate, or more if duplicates disagree significantly. Standards spanning the sample absorbance range are run before and aft&r each series of about five samples and a drift correction is applied. The volumes and standard concentrations are chosen to remain within the linear range, which should be checked regularly. Experience suggests that the calibration plots are reliably linear below an absorbance of 0.15. The cobalt chloride should be sufficiently pure that the standard matrix blank gives a peak height below 5 - lo-* absorbantxz units. Where a small makix blank is observed, this value is subtracted from the peak heights of the standards. With some instruments a small (
RESULTS
Two standard addition plots for a deep Sargasso Sea water sample are shown in Fig. 3. Expsrience from a large number of such plots indicates that copper and nickel recoveries are essentially quantitative, whereas cadmium recovery is about 10% lower. Table 2 gives the results of duplicate analyses on water samples taken from a profile at a station south of New Zealand (GEGSECS station 293). The pooled standard deviations are: copper, 20.13 nmol kg-l; nickel, kO.28 nmol kg-‘; and cadmium, 20.10 nmol kg’.
c;?..Y.xzx.xl. nru,.J“ Fig. 3. Standard additions plot for a sea-water sample. TABLE
Best-fit lines are indicated_
2
Replicate sea-x-cater analysis (Results are given as nmol kg-‘.)
Wver
Nickel
Cadmium
216
1.28 1.14
4.5 4.7
0.31 0.38
916
2.48 2.17
5.8 6.2
0.71 6.73
1733
2.25 2.19
7.2 7.1
0.87 -
2326
2.85 2.63 2.74
7-S 7.0 7.1
1.07 1.12 1.29
4926
4.55 4.08
8.2 7.8
0.98 1.26
Sample depth (m)
A comparison
of copper analyses by this technique
with fiame atomic
absorption analyses after a similar coprecipitation technique [I, 43 is given in Table 3. Individual resulis from the two methods agree within the estimated errors. The systematic difference is due to propagation of the standardizaticn and blank errors and is within the stated precision. A sisificant systematic error unique to either flame or flameless analysis is ruled out. The accuracy of the anaiyses is estimated to be similar to the precision. The results of the standard additions tests sue&est an accuracy better than 10%
195
TABLE 3 Comparison of flame and flameless atomic absorption copper
analyses of sea-water samples for
Sample NO-~
Flameless (nmol kg-‘)
Flame (nmol kg-‘)
Differenceb
284 290 292 293 294
1.80 1.99 1.35 1.50 1.39
1.80 1.87 0.98 1.33 1.28
0.00 0.12 0.37 0.17 0.11
=See ref. 4 for description of samples. bThe average difference is 0.15. which is consistent with the estimated precisions of 0.13 (flameless) and 0.15 (flame).
for the determination of added Cu, Ni and Cd. MO suitable reference standard sea-water samples are available for comparison, however. The limiting factor in the precision is not the atomic absorption determination; this is indicated by a comparison of the pooied precision of replicate injections with the overall precision (Table 4). For copper and nickel the limiting factor is probably quantitative transfer of the precipitate to the final solution. For cadmium the additional error is probably associated with the slightly lower (and therefore potentially more variable) precipitation efficiency. Improvements in precision require improved sample handling procedures rather than instrumentation. DISCUSSION The method described resulted from various optimization experiments. Preliminary work with cadmium is described below to illustrate the methodology. Cadmium is used as the example because it exhibited the largest variations with respect to changes in analytical conditions. Initial work showed that cobalt interfered with the determination of cadmium. Injection of a standard into a cobalt chloride matrix resulted in rapidly decreasing peak absorbance on successive injections, with a char TABLE 4 Comparisonof precision for replicate samples with replicate injections Element
Pooled standard deviations (nmol kg-‘) Samples
n
Injections
n
Cu
0.13
6
0.04
25
&! Y’
0.10 0.28
5 6
0.03 0.10
22 19
temperature of 300°C and an atomization temperature of 1600°C. When this series of injections was followed by a high-temperature (2600°C) tube cleaning cycle for 10 s, the sensitivity was restored to its initial level. Apparently, cobalt coats the graphite surface and retards reduction of cadmium to the atomic state. The cleaning cycle removes cobalt and allows reduction to proceed unhindered. Similar coatin, = effects have been used to minimize carbide formation [ 51. In addition to this cumulative coating effect, cobalt also affects the peak height of cadmium in a cleaned tube (Fig. 4 A, B). Use of a higher atomization temperature does not eliminate this effect (Fig. 4C). Such errors can be eliminated by using standards containing the same cobalt concentration as the samples. Sensitivity is enhanced by minimizing the amount of cobalt used for coprecipitation and by dilution of the final concentrate. Finally, addition of ammonium sulfate to the matrix allows the use of a higher charring temperature [S] and minimizes interferences. The coprecipitation procedure is applicable to fresh water samples, with the following modifications. Organic material dissolved in fresh water often precipitates on acidification and clogs the fritted glass discs. To avoid this problem, a filter rig with disposable acid-leached glass, fiber filters should be substituted for the filtration funnels. For some highly colored water samples, the organic material inhibits precipitation at the reagent levels recommended above. If this effect is observed, the cobalt and APDC levels should be increased until the chelate does not pass through the filter. The recovery efficiency should be monitored for each different type of sample. Coixlusions APDC chelate coprecipitation coupled with flameless atomic absorption provides a simple and precise method for the determination of nanomol kg-’ levels of copper, nickel, and cadmium in sea water. With practice, the method is not overly time-consuming. It is reasonabie to expect to complete sample
F:g. 4. (A) Effect of lo-‘-’ hi Co on a cadmium standard cwve. (B) Effect of cobalt concentration on cadmium peak height at 1600°C atomiztion temperature. (C) Effect of atomization temperature on cobalt interference, [Co] = lo-‘-’ M, [Cd] = 115 nmol-‘.
197
concentration in Iess than 20 min, digestion in about 4 h, and sample preparation in another hour; atomic absorption time should average about 5 min per element. Excellent results have been obtained on the distribution of nickel and cadmium in the ocean by this technique [ 7,8]. REFERENCES 1 E. A Boyle and J. M. Edmond, in T. R. P. Gibb (Ed.), Analytical Methods in Oceanography, Advances in Chemistry Series. No. 147, American Chemical Society, Washington D.C., 1975, p_ 44. 2 W. Slavin, Atomic Absorption Spectroscopy, InterJcience-Wiley, New York, 1968, p. 75. 3 D. E. Robertson, Anal. Chim Acta, 42 (1968) 533. 4 E k Boyle and J. M. Edmond. Nature (London), 253 (1975) 107. 5 J. I-I. Runnels, R. Merryfield and Il. B. Fisher, Anal. Chem.. 47 (1975) 1258. 6 R D. Ediger, At. Absorpt. Newel.. 14 (1975) 127. 7 F. R. Sclater, E. A Boyie and J. &L Edmond, Earth Planet. Sci. Lett., 31 (1976) 129. 8 E. A_ Boyle, F. R Sclater and J. M. Edmond. Nature (London), 263 (1976) 42.
Chimica Acta,
91 (1977)
189-197
OElsevier Scientific Publishing Company,
Amsterdam - Printed in The Netherlands
DETER&lINATIQN OF COPPER, NICKEL, AN-D CADMIUPcf IN SEA WATER BY APDC CHELATE COPRECIPITATION AND FLAMELESS ATOMIC ABSORPTION SPECTROMETRY
EDWARD A. BOYLE
and JOHN hi. EDMOND
Department of Earth and Ranetary Cambridge. JfA 02139 (U.S.A.)
(Received
Sciences.
dfassachusetts
Institute
of Technology,
1st November 1976)
SUMMARY Copper, nickel, and cadmium can be determined in 100 ml of sea water by coprecipitation with cobalt pyrrolidinedithiocarbamate and graphite atomizer atomic absorption spectrometry. Concentration ranges !ikely to be encountered and estimated (1 o) analytical precisions are l-6 nmol kg-’ (20.1) for copper, 3-12 nmol kg-’ (tO.3) for nickel and 0.0-1.1 nmol kg-’ (~0.1) for cadmium. The technique may be applied to fresh-water samples with slight modification.
This report describes a technique for the determination of copper, nickel, and cadmium in sea water. The metals are concentrated from a lOO-ml sample by cobalt pyrrolidinedithiocarbamate coprecipitation [ 11. The precipitate is redissolved in an organic solvent and digested to simplify the sample matrix. The resulting solution is analyzed by atomic absorption spectrometry with heated graphitz atomization. Extension of the method to freshwater samples is discussed briefly. EXPERIMENTAL
Prevention
of contamination
Quantitative methods for elements at very low concentration levels must preclude contamination of the sample, as well as satisfy accuracy and sensitivity requirements. Elaborate clean-room techniques entail considerable inconvenience in ternIs of cost and availabili@ and do not entirely eliminate introduction of contamination by the analyst. For this reason the method was designed for use in an ordinary laboratory with the precautions outlined below. Reagents are stored in polyethylene or polypropylene bottles and leached in 1 M hydrochloric acid for at least one day. For acid storage, bottles are leached in f3 &I hydrochloric acid at 80°C for 2 h. When not in use, bottles are covered with plastic ba@
All glassware and plastic ware are stored in a covered bath wit.h a leaching solution consisting of 0.1 M hydrochloric acid and 0.1 M nitric acid. Immediately prior to use, the glassware is rinsed with high-purity distilled water and used without drying. The distilled water should be of sufficient purity to give no detectable blank if treated as a sample. Other precautions are that: disposable poIyethylene gloves are used for critical handling steps; samples are allowed minimal exposure to open laboratory air; microliter pipette tips are flushed several times with 0.1 M hydrochloric acid before each use; and silicone rubber stoppers are kept in 2% EDTA solution between each use. Other precautions are mentioned ia the Procedure. Although not used here, a laminar-flow clean bench would help ensure integrity of the sample during manipulation steps and is recommended. Equipment Two special items are illustrated in Fig. 1. The filtration funnel (A) is made by joining a 3.5ml fine sintered glass frit to the bottom of a tall-form 200-ml Pyrex beaker. The redissolution apparatus (B) is made from plexiglass and sihcone rubber stoppers. A frame for securing the filtration funnel into a 250-ml filter flask is necessary; this frame may be constructed from plexiglass with l/S-ino.d. tygon tubing straps with glass hooks. Reagercts Cobalt chloride sclution. Dissolve 0.2 g of ultrapure CoCl* - 6HZ0 (Spex Industries) in 250 ml of distilled water. APDC solution. Purify a 2% ammonium pyrrolidinedithiocarbamate solution in water by repeated extraction with carbon tetrachloride [ 21. Acid-acetone mixture. Prepare a (95 + 5) solution of acetone and 0.1 M hydrochloric acid; dispense from a 250-ml polyethylene squeeze bottle. Nitripperchloric acid mixture. Prepare 6 M nitric acid, redistilled in a vycor glass still, containing 1% (v/v) ultrapure perchloric acid. ,SILICOF:E
II )
A
2oOML TALL FCi7.4 BEAKER
IS F FAITTED
Fig. 1. (A) Filtration
RUS3ER
B TEFLON
SEAKER PLEXlCbaS
FUNSEL
funnel. (B) Dissolution
STOPPER
\ apparatus.
FRAME
SLICCNE RU59ER STOFi’ER (PARTIALLY MLLOW 1
Ammonium sulfate soiufion Prepare 3 - 1OW2M ammonium sulfate (reagent grade) in I.5 10-’ M sulfuric acid (reagent grade). Hydrochloric acid (6 M). Prepare from acid redistilled in a vycor glass still. Standards. Prepare nickel and copper standards (20 clmol ml-‘) and cadmium standard (10 pmol ml-‘) in 0.1 h? hydrochloric acid. l
Procedure Sampfes should be acidified to pH 2 by addition of 6 M hydrochloric acid immediately after collection. This procedure minimizes adsorption on to container walls [ 31. Weigh out approximately 100 g of sample into a 200-ml talEform beaker. Add 250 ~1 of cobalt chloride solution and swirl gently to mix thoroughly. Add 500 ~1 of APDC solution, swirl gently to mix thoroughly, and allow the precipitation to proceed for 5 min before filtration. During this waiting period, remove the filtration funnel from the acid bath, rinse with distilled water, and strap into the filtration rig. Apply full vacuum and rinse the interior of the filtration funnel with two portions of 6 M hydrochloric acid followed by two portions of distilled water. Then filter the sample using full’vacuum. Rinse the interior of ‘Lhe funnel with a small portion of distilled water to remove the last traces of sea salt. Release the vacuum slowly by means of a stopcock valve with an in-line air filter. Remove a 5-ml Teflon beaker from the acid bath and rinse with distilled water. Place it into the dissolution apparatus. Rinse the tip of the f3ltiation funnel with acid--acetone solution and place the fwmel into the top of the dissolution apparatus. Rinse through several small portions of acid-acetone mixture by applying a gentle vacuum.. This procedure dissolves the precipitate and washes it into the beaker. Release the vacuum; remove the beaker, and carefully remove the filter funnel from the dissolution apparatus. Remove the traces of the chelate from the tip of the filter funnel by washing with acid--acetone mixture and collecting the wash in the Teflon beaker. Cover a hot plate with a gIass fiber filter. Place the beaker on the hot plate at very low heat (to avoid bumping) and cover with an inverted 20-ml Pyrex beaker; the glass fiber minimizes spattering of condensate running down the inside of the cover. Tilt the beaker by placing one edge on a glass rod, to prevent condensation from falling back into the beaker. Evaporate to dryness, Add 500 ,~l of nitric-perchlorie acid mixture and evaporate to dryness at slightly higher heat. Avoid charring the residual organic material by overheating. After cooling, the beaker may be covered with Parafilm and stored indefinitely. Along with the samples, prepare a matrix for standards by pipetting 250 1_t1 of cobalt chloride solution into each of three 5-ml beakers, adding 10 pi of APDC solution to each, and evaporating and treating with acid mixture as in the preceding paragraph. Determine blanks by running high- . purity water as a sample. The water may be tested for traces of the metals by varying the volume of several blank samples.
14.2
About one hour before the atomic absorption analysis, pipette 2000 ~1 of 0.1 M hydrochloric acid into each sample and standard matrix beaker. While allowing 15 min for the residue to dissolve, prepare a secondary mixed stzndard by diluting 10 ~1 of cadmium standard plus 50 ,ul of copper standard and 100 ~1 of nickel standard to 100.0 ml with distilled water. Mix thoroughly and pipette 50 ~1 into one of the standard matrix beakers and 100 ,ul into another; the third matrix beaker is used for a matrix blank. After the dissolution period, swirl each beaker gently to mix the solution and rinse the sides of the beaker. This solution is used for the copper and nickel determinations. Prepare a diluted so!ution for the cadmium determinations by pipetting 250 ~1 Born each concentrated sample and standard into dry 5-ml Teflon beakers. Add 750 ~1 of ammonium sifate solution and mix by gentle swirling. Cover all beakers with Parafilm and analyze immediately to minimize evaporation. Determinations were made with a Perkin-Elmer HGA 2100 graphite furnace, model 403 atomic absorption spectrophotometer and model 56 chart recorder. Manufacturer’s recommended operating conditions for wavelength, slit and lamp current were followed. Instrumental parameters for the graphite furnace are given in Table 1. Deuterium arc background correction was applied, with the deuterium arc beam aligned to coincide with the hollow-cathode beam at the center of the graphite tube. Optical windows were kept clean by recommended procedures and were masked outside the optical path to minimize light scattering in the optics. The chart recorder was set to 0.5 absorbance full scale with minimum pen damping. 7JJ’pical peak height output is illustrated in Fig. 2. A 100-g sample with 1 nmol kg-’ of each element processed as described above should give peak aksorbances of approximately 0.02 for copper, 0.01 for nickel, and 0.1 for cadmium.
TABLE
1
Instrumental Element
CU
parameters Sample volume (PI) 50
for HGA
2100
Gas flow continuous (cm’ min-‘)
Dry cycle
Char cycle
30
20 5 130°C
25 s 850°C
2400°C 12s 2600° C
Ni
100
30
35s 110°C
20 1150°C
Cd
50
20
25s 110°C
20s 500°C
=For Cd, this was followed
by cleaning for 5 s at 26OO’C.
Atomization cycle
10s
5s 1600°C’
193
COPPER
NICKEL
CADMIUM
Fig. 2. Representative peak height output for copper, nickel, and cadmium. Du$icate injections are ixzdiczted by brackets. Peak height expected for 1 nmol kg-’ (Cu. Ni) and 0.1 nmol kg-’ (Cd) in a 100-g sample is indicated by bar.
All samples and standards should be injec’ted in duplicate, or more if duplicates disagree significantly. Standards spanning the sample absorbance range are run before and aft&r each series of about five samples and a drift correction is applied. The volumes and standard concentrations are chosen to remain within the linear range, which should be checked regularly. Experience suggests that the calibration plots are reliably linear below an absorbance of 0.15. The cobalt chloride should be sufficiently pure that the standard matrix blank gives a peak height below 5 - lo-* absorbantxz units. Where a small makix blank is observed, this value is subtracted from the peak heights of the standards. With some instruments a small (
RESULTS
Two standard addition plots for a deep Sargasso Sea water sample are shown in Fig. 3. Expsrience from a large number of such plots indicates that copper and nickel recoveries are essentially quantitative, whereas cadmium recovery is about 10% lower. Table 2 gives the results of duplicate analyses on water samples taken from a profile at a station south of New Zealand (GEGSECS station 293). The pooled standard deviations are: copper, 20.13 nmol kg-l; nickel, kO.28 nmol kg-‘; and cadmium, 20.10 nmol kg’.
c;?..Y.xzx.xl. nru,.J“ Fig. 3. Standard additions plot for a sea-water sample. TABLE
Best-fit lines are indicated_
2
Replicate sea-x-cater analysis (Results are given as nmol kg-‘.)
Wver
Nickel
Cadmium
216
1.28 1.14
4.5 4.7
0.31 0.38
916
2.48 2.17
5.8 6.2
0.71 6.73
1733
2.25 2.19
7.2 7.1
0.87 -
2326
2.85 2.63 2.74
7-S 7.0 7.1
1.07 1.12 1.29
4926
4.55 4.08
8.2 7.8
0.98 1.26
Sample depth (m)
A comparison
of copper analyses by this technique
with fiame atomic
absorption analyses after a similar coprecipitation technique [I, 43 is given in Table 3. Individual resulis from the two methods agree within the estimated errors. The systematic difference is due to propagation of the standardizaticn and blank errors and is within the stated precision. A sisificant systematic error unique to either flame or flameless analysis is ruled out. The accuracy of the anaiyses is estimated to be similar to the precision. The results of the standard additions tests sue&est an accuracy better than 10%
195
TABLE 3 Comparison of flame and flameless atomic absorption copper
analyses of sea-water samples for
Sample NO-~
Flameless (nmol kg-‘)
Flame (nmol kg-‘)
Differenceb
284 290 292 293 294
1.80 1.99 1.35 1.50 1.39
1.80 1.87 0.98 1.33 1.28
0.00 0.12 0.37 0.17 0.11
=See ref. 4 for description of samples. bThe average difference is 0.15. which is consistent with the estimated precisions of 0.13 (flameless) and 0.15 (flame).
for the determination of added Cu, Ni and Cd. MO suitable reference standard sea-water samples are available for comparison, however. The limiting factor in the precision is not the atomic absorption determination; this is indicated by a comparison of the pooied precision of replicate injections with the overall precision (Table 4). For copper and nickel the limiting factor is probably quantitative transfer of the precipitate to the final solution. For cadmium the additional error is probably associated with the slightly lower (and therefore potentially more variable) precipitation efficiency. Improvements in precision require improved sample handling procedures rather than instrumentation. DISCUSSION The method described resulted from various optimization experiments. Preliminary work with cadmium is described below to illustrate the methodology. Cadmium is used as the example because it exhibited the largest variations with respect to changes in analytical conditions. Initial work showed that cobalt interfered with the determination of cadmium. Injection of a standard into a cobalt chloride matrix resulted in rapidly decreasing peak absorbance on successive injections, with a char TABLE 4 Comparisonof precision for replicate samples with replicate injections Element
Pooled standard deviations (nmol kg-‘) Samples
n
Injections
n
Cu
0.13
6
0.04
25
&! Y’
0.10 0.28
5 6
0.03 0.10
22 19
temperature of 300°C and an atomization temperature of 1600°C. When this series of injections was followed by a high-temperature (2600°C) tube cleaning cycle for 10 s, the sensitivity was restored to its initial level. Apparently, cobalt coats the graphite surface and retards reduction of cadmium to the atomic state. The cleaning cycle removes cobalt and allows reduction to proceed unhindered. Similar coatin, = effects have been used to minimize carbide formation [ 51. In addition to this cumulative coating effect, cobalt also affects the peak height of cadmium in a cleaned tube (Fig. 4 A, B). Use of a higher atomization temperature does not eliminate this effect (Fig. 4C). Such errors can be eliminated by using standards containing the same cobalt concentration as the samples. Sensitivity is enhanced by minimizing the amount of cobalt used for coprecipitation and by dilution of the final concentrate. Finally, addition of ammonium sulfate to the matrix allows the use of a higher charring temperature [S] and minimizes interferences. The coprecipitation procedure is applicable to fresh water samples, with the following modifications. Organic material dissolved in fresh water often precipitates on acidification and clogs the fritted glass discs. To avoid this problem, a filter rig with disposable acid-leached glass, fiber filters should be substituted for the filtration funnels. For some highly colored water samples, the organic material inhibits precipitation at the reagent levels recommended above. If this effect is observed, the cobalt and APDC levels should be increased until the chelate does not pass through the filter. The recovery efficiency should be monitored for each different type of sample. Coixlusions APDC chelate coprecipitation coupled with flameless atomic absorption provides a simple and precise method for the determination of nanomol kg-’ levels of copper, nickel, and cadmium in sea water. With practice, the method is not overly time-consuming. It is reasonabie to expect to complete sample
F:g. 4. (A) Effect of lo-‘-’ hi Co on a cadmium standard cwve. (B) Effect of cobalt concentration on cadmium peak height at 1600°C atomiztion temperature. (C) Effect of atomization temperature on cobalt interference, [Co] = lo-‘-’ M, [Cd] = 115 nmol-‘.
197
concentration in Iess than 20 min, digestion in about 4 h, and sample preparation in another hour; atomic absorption time should average about 5 min per element. Excellent results have been obtained on the distribution of nickel and cadmium in the ocean by this technique [ 7,8]. REFERENCES 1 E. A Boyle and J. M. Edmond, in T. R. P. Gibb (Ed.), Analytical Methods in Oceanography, Advances in Chemistry Series. No. 147, American Chemical Society, Washington D.C., 1975, p_ 44. 2 W. Slavin, Atomic Absorption Spectroscopy, InterJcience-Wiley, New York, 1968, p. 75. 3 D. E. Robertson, Anal. Chim Acta, 42 (1968) 533. 4 E k Boyle and J. M. Edmond. Nature (London), 253 (1975) 107. 5 J. I-I. Runnels, R. Merryfield and Il. B. Fisher, Anal. Chem.. 47 (1975) 1258. 6 R D. Ediger, At. Absorpt. Newel.. 14 (1975) 127. 7 F. R. Sclater, E. A Boyie and J. &L Edmond, Earth Planet. Sci. Lett., 31 (1976) 129. 8 E. A_ Boyle, F. R Sclater and J. M. Edmond. Nature (London), 263 (1976) 42.