414
ANALYTICAL CHEMISTRY, VOL. 51, NO. 3, MARCH 1979
Determination of Organic (Methyl) Mercury in Fish by Graphite Furnace Atomic Absorption Spectrophotometry Gerald T. C. Shum," Harry C. Freeman, and John F. Uthe Department of Fisheries & the Environment, Fisheries and Marine Service, Technology Branch, 1707 Lower Water Street, P.O. Box 550, Halifax, Nova Scotia 63J 257
EXPERIMENTAL
A method to determine organic (methyl) mercury in fish using graphite furnace atomic absorption spectrophotometry is described. Following initial acetone extraction, methylmercury is released from samples by acid cupric bromide, and extracted into toluene. Addition of dithizone to the extract allows the determination of mercury in the furnace. The average recovery of added methylmercuric chloride at 2.00- and 4.00-pg levels is 97.7 f 5 . 5 % . Detection limit is 0.08 pg Hg/g sample. Determination of methylmercury in aqueous sodium thiosulfate after partitioning from toluene, permits an autosampler to be used. Results of the analyses of several fish samples by this technique are compared to those of gas-liquid chromatography and Magos' selective atomic absorption method.
Apparatus. A Perkin-Elmer Model 403 atomic absorption
spectrophotometer equipped with background corrector and a Perkin-Elmer Model HGA 2100 heated graphite furnace were used for all methylmercury determinations. The furnace control settings were: drying a t 100 "C for 40 s, charring at 180 "C for 10 s, and atomizing at 950 "C for 30 s with the purging argon gas interrupted. Total mercury was determined by the semiautoniated method of Armstrong and Uthe ( 1 4 , using a Perkin-Elmer Model 306 atomic absorption spectrophotometer, Technicon autosampler, and proportioning pump. Reagents. All chemicals used were ACS reagent grade and solyents were "distilled-in-glass" grade. Methylmercuric chloride (95+7 ~from ) Ventron Alpha Products was used to prepare 100 ggjmL (as Hg) stock solution in toluene. A second stock of 10.0 gg/mL was prepared weekly by dilution. Both solutions were stored in a refrigerator. Working standard solutions (0.05 to 0.80 wgjmL) were prepared for each run by diluting with freshly prepared 0.107~dithizone in toluene. Extraction solutions were: 0.10 M CuS04.5H20 (aqueous and in 1 N H,SO,) and 3.0 M KBr (in 4 N H2S04). Procedure. The fish samples were homogenized by a Waring blender o r a Brinkmann Polytron unit. Two grams (wet weight) of samples were weighed into 40-mL p>-rex conical centrifuge tubes fitted with Poly-Seal screw caps (A. H. Thomas). The hornogenized sample was shaken vigorously by hand with two 25-mL aliquots of acetone. Each time the mixture was centrifuged for 5 min at 1500 rpm. and the acetone was removed by aspiration. Tilting the tube during the second aspiration facilitated the removal of'the acetone. Five mL of acid bromide and 10 mL of CuSO, (4 iV H2S0,) solution were added to the residue and the mixture was shaken gently by hand. Ten mL of toluene was added next, the tube vias shaken vigorously by hand for 1 min, and then centrifuged for 10 min a t 1500 rpm. The solution was prepared for analysis in the furnace by pipetting 0.5 mL of the toluene extract and 0.5 mL of 0.1070 dithizone into a glass stoppered tube and mixing on a Vortex mixer. Standards were prepared in the same manner, using 0.05-0.40 ppm methylmercury in the dithizone-toluene solution. Twenty to forty jtL of the solution was injected into the furnace for each determination. The peak heights were meaaured and were used to calculate the mercury concentration by the method of standard addition. \Vhen freeze-dried and powdered fish samples were analyzed, an aqueous rather than acidic CuSO, solution was used.
T h e determination of low concentrations of total mercury in biological and environmental samples has become routine with t h e development of t h e cold vapor atomic absorption spectrophotometric method (I). However, it has been shown that the major form of mercury accumulated in most fish and shellfish edible tissue is methylmercury ( 2 ) a n d not the less toxic inorganic form ( 3 ) . Therefore t h e need t o distinguish t h e chemical form of mercury in samples is obvious. Several gas-liquid chromatography based methods for the determination of methylmercury in fish have been published ( 4 - 7 ) , but these are tedious and still subject t o review (8). A selective atomic absorption method t o determine inorganic and methylmercury in biological materials was described by Magos (9),and was modified by Kacprzak and Chvojka (10). Matsunaga and Takashashi determined organic mercury by t h e cold vapor method using solvent extraction and back extraction into a reagent suitable for reduction ( 1 1 ) . T h e rapid development of heated furnace atomization for atomic absorption offered new opportunities in trace metal analysis. Inorganic mercury was first determined by this technique by Ediger (12),who used ammonium sulfide as a stabilizing agent. Alder and Hickman (13) demonstrated the stabilizing effect of hydrochloric acid and hydrogen peroxide on mercury determination in the graphite furnace. T h e work of these earlier researchers on standard solutions demonstrated t h e potential of the graphite furnace for use in developing a method for determining trace quantities of mercury in fish. T h e graphite furnace, unlike the cold vapor technique which could be applied to mercury only, could be used t o determine other trace elements. T h e purpose of this study was t o develop a simple and reliable method that can be used to determine methylmercury in fish tissue by laboratories with atomic absorption capability. In this method, a modified toluene extraction adapted from the method of Uthe e t al. (5) was used, and the methylmercury content in the toluene was directly determined in the graphite furnace, using dithizone as a stabilizing agent. An aqueous system was adapted by extracting the methylmercury from the toluene with sodium thiosulfate a n d determining its concentration in t h e graphite furnace. 0003-2700/7910351-0414$01 0010
RESULTS A N D DISCUSSION S t a b i l i z a t i o n of M e t h y l m e r c u r y i n the F u r n a c e . Methylmercury was stabilized in the furnace by the addition of dithizone in toluene added to a n equal volume of sample extract. It was noted t h a t the atomic absorption of methylmercury in a swordfish extract reaches a maximum arid dithizone added (Figure 1). T h e optimum plateau a t 0.087~ charring temperature t o give the maximum absorption a t atomization was fairly narrow and ranged from 175-200 "C (Figure 2 ) , but t h e signals were reproducible. This low charring temperature caused some buildup of residue in the graphite tube, but this was reduced by prewashing the fatty samples with acetone and by atomizing the samples for 30 s. T h e graphite tube was given a high temperature burn two or three times a day to ensure no significant residue buildup, arid it could be used for 400-500 injections. T h e standard curve C
1979 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 51,
NO. 3,
MARCH 1979
415
Table I. Determination of Methylmercury (MeHg) in Fish and Shellfish (in pg Hg/g) by the Heated Graphite Furnace (Authors' Lab), with Comparative Analyses Data
MeHg in toluene ext.
sample freeze-dried oyster halibut canned tuna scallop muscle lobster tomalley swordfish muscle swordfish muscle swordfish muscle canned tuna 1 canned tuna 2 frozen swordfish liver 1 swordfish liver 2 swordfish liver 3 swordfish muscle swordfish muscle swordfish muscle swordfish muscle Not detectable.
-
50
i
f
E
30
.-I" m m 20 ; X W 0
a
0.06, 0.06 5.49, 5.55 0.78, 0.74 0.06, 0.06 0.05, 0.06 0.40, 0.42 2.52, 2.48
N.D.Q 5.31, 5.68 0.76 2 0.05 ( 6 ) N.D. N.D. 0.43 i 0.06 2.37 i 0.06 3.00 i 0.10 0.40, 0.40 1.90, 1.95
1
2
3
-
0.41, 0.42
-
0.41, 0.42 8.67, 8.25 1.50, 1.62 1.96 i 0.15 (4) 3.98, 4.22
1
2 3 4
9.14, 8.39 1.38, 1.81
-
-
-
1.60, 1.50
0.87, 0.81
/
external lab 2 inorganic Hg by total Magos' Hg
external lab 1 MeHg by total GLC Hg
total Hg 0.18 t 0.02b 5.79 t 0.07 0.92 i 0.03 0.06 i. 0.02 0.74 i 0 02 0.41 * 0.02 2.54 i 0.07 3.34 i 0.06 0.50 I0.01 2.23 * 0.19
0.02 4.55 0.60 0.05 0.09
0.25 4.77 0.80
-
-
0.200 0.034
5.85 0.95
0.64
-
-
-
-
0.11
-
-
2.75
-
0.022 0.109 0.143
3.03
0.42 2.45 3.31
1.05, 1.00 19.1 t 0.72
2.27 t 0.12 2.02, 1.99 4.40, 4.41 1.71, 1.70 0.92 i 0.12
Figures are the mean and standard deviation of triplicate analyses except otherwise indicated. 60
E
-.-
authors' lab MeHg in Na1S10, ext.
r--
E E
.E
50-
40-
c
c
.-W
30-
I r 0
0"
jo-
20 -
10 1
,
,
1
,
,
,
,
,
,
-
-
,
0
200
100
300
Furnace c o n t r o l setting
OC
Figure 2. Optimization of charring temperature for swordfish muscle toluene extract. 0.10% dithizone solution was added to an equal volume of extract. Two to three injections of 20 p L solution. Furnace controls were: drying at 100 'C for 40 s, and atomizing at 950 'C for 30 s
was linear from 1 to 6 ng of mercury, but the samples showed a slight suppression of absorption signal (Figure 3). Much lower sensitivity was obtained with pyrolytically coated graphite tubes (Perkin-Elmer),in contrast to the enhancement effect found for the difficult-to-volatilize elements (15). Extraction of Methylmercury. The extraction procedure is an adaptation of the method of Uthe et al. (51, combined with a n acetone prewashing technique of Watts e t al. ( 7 ) . Prewashing with acetone was necessary with oyster and fatty samples such as liver tissue and tuna canned in oil. Nonfatty muscle tissue did not require this pretreatment but, to be consistent and to standardize the method, the prewashing was carried out on all samples. When the extraction was carried out in a glass centrifuge tube under the present conditions, the acid concentration of the aqueous mixture had to be for a more complete extraction, except in increased to 4.0 9the case of freeze-dried and powdered samples, where the original 1.3 N was used to avoid formation of a gel a t the interface. Analysis of Fish Samples. A number of fish samples were analyzed for methylmercury using this procedure. Total mercury content of these samples were also determined for
v ,
/ , 0
, 2.0
I
4.0
6.0
8.0
10.0
ng Methyl mercury ( a s H g ) Figure 3. Standard (-0-)and sample (-04 calibration curves of toluene extract stabilized with dithizone. The sample was tuna canned in oil. 10 pL of standards and 20 pL of spiked samples were used
comparison (Table I, Authors' Lab.). Recovery was determined by adding 2.0e4.00 pg methylmercury to samples after prewashing with acetone, and comparing results with unspiked samples. The average recovery for nine different samples in
416
ANALYTICAL CHEMISTRY, VOL. 51, NO. 3, MARCH 1979
(60
4 140
.-
120
ng Methyl mercury ( a s Hg)
Figure 4. Standard (-0-)and sample (-e-) calibration curves of 0.005 M Na,S,O, extract. The sample was tuna canned in oil. 10 pL of standards was determined, while the sample was determined by overlaying 10 pL of standard solution onto 10 pL of sample in the furnace. Furnace controls were: drying at 100 OC for 40 s, charring at 150 O C for 30 s, and atomizing at 1200 O C for 30 s with purge gas interrupted
*
duplicate was 97.7 5.5%. The detection limit was 0.08 pg/g sample, using 20 pL of sample plus 20 pL spiking standard. T h e calculation was based on an absorbance signal of 3-mm peak height and an average slope of several sample calibration curves. An average of 92.2% of the mercury in muscle samples was methylmercury. This showed t h a t prewashing the samples with acetone did not remove appreciable amounts of tightly bound organic mercury, confirming the results of Watts and co-workers (7). T h e swordfish liver samples, however, contained a much lower percentage of methylmercury t h a n muscle. This was not surprising because the liver had been suggested as one of the sites for demethylation of methylmercury (16, 17). A d a p t a t i o n t o A q u e o u s S y s t e m . In order to allow laboratories to utilize autosamplers (e.g., Perkin-Elmer AS-1) which cannot handle organic solvents, the methylmercury in the toluene extract was partitioned into aqueous sodium thiosulfate, and then determined by the graphite furnace. It was found t h a t a simple one-step partitioning gave excellent recovery. Two to six mL of the toluene extract was pipetted into a 15-mL glass stoppered conical centrifuge tube, and then 2.0 mL of 0.005 M aqueous Na2S203(freshly diluted from 0.05 M stock) was added. T h e mixture was shaken 40 times by hand, and centrifuged to separate the layers. T h e toluene layer on top was aspirated off and the mercury concentration in the aqueous solution was determined by the method of standard additions (Figure 4). The standards for spiking were prepared in toluene and extracted with the thiosulfate solution. T h e standards were linear to about 5.5 ng a t the optimal sample atomization temperature (Figure 4). However, the sample calibration curves were linear to a wider range, for example, canned tuna as shown in the same figure. T h e extraction of methylmercury from toluene into thiosulfate was complete in most cases (Table I). T h e average recovery of 14 extractions of 5 different types of samples was 102 k 5.17'~. Sample extracts that were low in methylmercury, for example, oyster and lobster tomalley (Table I) were concentrated two to three times by the partitioning, thus increasing the sensitivity of the method. C h e c k S a m p l e S t u d y . Six of the freeze-dried samples were sent to a n external laboratory for methylmercury de-
termination by the gas-liquid chromatographic method of Uthe et al. ( 5 ) , and also for total mercury determination. Except for the halibut sample, the results (Table I, External Lab I) were in good agreement. T h e low percentage of methylmercury in oyster and lobster tomalley found by us was verified. T h e halibut results indicated a difference in the quantitation of mercury b u t confirmed t h a t most of the mercury in this fish was methylmercury. T h e same halibut, tuna, and three freeze-dried swordfish samples were analyzed in a third laboratory using a modified Magos' (9) selective reduction atomic absorption method. Total and inorganic mercury in the fish were determined and organic mercury was calculated by difference (Table I, External Lab 2 ) . The mercury results obtained by this laboratory were almost identical to ours, and thus substantiated our finding, that these samples had very little inorganic mercury. CONCLUSIONS The furnace atomic absorption method for the analysis of methylmercury in fish presented here, does not distinguish the different forms of organic mercury, b u t is based on the findings t h a t with rare exception (18), the organic form of mercury found in fish has been methylmercury ( 2 ) . When in doubt, the species of organic mercury can be identified by a thin-layer chromatographic technique similar to the one described by Westoo ( 4 ) or T a t t o n and Wagstaff (19). T h e present method enables laboratories engaged in trace metal analyses to carry out methylmercury determinations in fish without having to install separate gas-liquid chromatographic equipment which is costly to operate and maintain (20). In addition, use of an autosampler allows methylmercury determination a t a much faster rate than gas-liquid chromatographic methodology. ACKNOWLEDGMENT T h e authors express their gratitude to A. Lutz of the Freshwater Institute, Fisheries and Oceans Canada, and A. S. Hall of the National Marine Fisheries, U.S.Department of Commerce for their cooperation in the comparative studies. LITERATURE CITED W . R. Hatch and W . L. Ott, Anal. Chem., 40, 2085 (1968). G. Westoo, "Chemical Fallout", M. W. Miller and G. G. Berg, Eds., Charles C Thomas, Springfield, Ill., 1969, p 75. P. L. Bidstrup, "Toxicity of Mercury and Its Compounds", Elsevier, New York, 1964. G. Westoo. Acta Chem. Scand., 20, 2131 (1966). J . F. Uthe, J . Solomon, and B. Grift, J . Assoc. Offic. Anal. Chem., 55, 583 (1972). M. L. Schafer, U. Rhea, J. T. Peeler, C. H. Hamilton, and J. E. Campbell, J . Agric. Food Chem., 23, 1079 (1975). J. 0. Watts, K. W. Boyer, A. Cortez, and E. R. Elkins, Jr., J . Assoc. Offic. Anal. Chem.. 59. 1226 11976). A. J. Malanos'ki, K. Helrich, S. 'Williams, B. J. Prima, and S. W . Butler, J . Assoc. Offic. Anal. Chem., 61, 397 (1978). L. Magos, Analyst(London), 96, 847 (1971). J . L. Kacprzak and R. Chvojka, J . Assoc. Offic. Anal. Chem., 59, 153 (1976). K. Matsunaga and S. Takahaski, Anal. Chim. Acta, 87, 487 (1976). R. D. Ediger, At. Absorpf. Newsl.. 14, 127 (1975). J. F. Alder and D. A. Hickrnan, Anal. Chem., 49, 336 (1977). F. A. J. Armstrong and J . F. Uthe, At. Absorpt. Newsl., 10, 101 (1971). R. E. Sturgeon and C. L . Chakrabarti. Anal. Chem., 49, 90 (1977). W. D. Burrows and P. A. Krenkel. Environ. Sci. Techno/..7 , 1127 (1973). K . R. Olson, K. S. Squibb, and R. J. Consins, J . Fish. Res. Board Can., 35, 381 (1978). S. Yarnanaka and K. Ueda. Bull. Environ. Contam. Toxicol., 14, 409 (1975). J. O'G. Tatton and P. J. Wagstaffe. J . Chromatogr., 44, 284 (1969). J . F. Uthe and F. A. J. Armstrong, Toxicol. Environ. Chem. Rev., 2, 45 (1974).
RECEI~TD for review September 5 , 1978. Accepted November 29, 1978.
ANALYTICAL CHEMISTRY, VOL. 51, NO. 3, MARCH 1979
Determination of Organic (Methyl) Mercury in Fish by Graphite Furnace Atomic Absorption Spectrophotometry Gerald T. C. Shum," Harry C. Freeman, and John F. Uthe Department of Fisheries & the Environment, Fisheries and Marine Service, Technology Branch, 1707 Lower Water Street, P.O. Box 550, Halifax, Nova Scotia 63J 257
EXPERIMENTAL
A method to determine organic (methyl) mercury in fish using graphite furnace atomic absorption spectrophotometry is described. Following initial acetone extraction, methylmercury is released from samples by acid cupric bromide, and extracted into toluene. Addition of dithizone to the extract allows the determination of mercury in the furnace. The average recovery of added methylmercuric chloride at 2.00- and 4.00-pg levels is 97.7 f 5 . 5 % . Detection limit is 0.08 pg Hg/g sample. Determination of methylmercury in aqueous sodium thiosulfate after partitioning from toluene, permits an autosampler to be used. Results of the analyses of several fish samples by this technique are compared to those of gas-liquid chromatography and Magos' selective atomic absorption method.
Apparatus. A Perkin-Elmer Model 403 atomic absorption
spectrophotometer equipped with background corrector and a Perkin-Elmer Model HGA 2100 heated graphite furnace were used for all methylmercury determinations. The furnace control settings were: drying a t 100 "C for 40 s, charring at 180 "C for 10 s, and atomizing at 950 "C for 30 s with the purging argon gas interrupted. Total mercury was determined by the semiautoniated method of Armstrong and Uthe ( 1 4 , using a Perkin-Elmer Model 306 atomic absorption spectrophotometer, Technicon autosampler, and proportioning pump. Reagents. All chemicals used were ACS reagent grade and solyents were "distilled-in-glass" grade. Methylmercuric chloride (95+7 ~from ) Ventron Alpha Products was used to prepare 100 ggjmL (as Hg) stock solution in toluene. A second stock of 10.0 gg/mL was prepared weekly by dilution. Both solutions were stored in a refrigerator. Working standard solutions (0.05 to 0.80 wgjmL) were prepared for each run by diluting with freshly prepared 0.107~dithizone in toluene. Extraction solutions were: 0.10 M CuS04.5H20 (aqueous and in 1 N H,SO,) and 3.0 M KBr (in 4 N H2S04). Procedure. The fish samples were homogenized by a Waring blender o r a Brinkmann Polytron unit. Two grams (wet weight) of samples were weighed into 40-mL p>-rex conical centrifuge tubes fitted with Poly-Seal screw caps (A. H. Thomas). The hornogenized sample was shaken vigorously by hand with two 25-mL aliquots of acetone. Each time the mixture was centrifuged for 5 min at 1500 rpm. and the acetone was removed by aspiration. Tilting the tube during the second aspiration facilitated the removal of'the acetone. Five mL of acid bromide and 10 mL of CuSO, (4 iV H2S0,) solution were added to the residue and the mixture was shaken gently by hand. Ten mL of toluene was added next, the tube vias shaken vigorously by hand for 1 min, and then centrifuged for 10 min a t 1500 rpm. The solution was prepared for analysis in the furnace by pipetting 0.5 mL of the toluene extract and 0.5 mL of 0.1070 dithizone into a glass stoppered tube and mixing on a Vortex mixer. Standards were prepared in the same manner, using 0.05-0.40 ppm methylmercury in the dithizone-toluene solution. Twenty to forty jtL of the solution was injected into the furnace for each determination. The peak heights were meaaured and were used to calculate the mercury concentration by the method of standard addition. \Vhen freeze-dried and powdered fish samples were analyzed, an aqueous rather than acidic CuSO, solution was used.
T h e determination of low concentrations of total mercury in biological and environmental samples has become routine with t h e development of t h e cold vapor atomic absorption spectrophotometric method (I). However, it has been shown that the major form of mercury accumulated in most fish and shellfish edible tissue is methylmercury ( 2 ) a n d not the less toxic inorganic form ( 3 ) . Therefore t h e need t o distinguish t h e chemical form of mercury in samples is obvious. Several gas-liquid chromatography based methods for the determination of methylmercury in fish have been published ( 4 - 7 ) , but these are tedious and still subject t o review (8). A selective atomic absorption method t o determine inorganic and methylmercury in biological materials was described by Magos (9),and was modified by Kacprzak and Chvojka (10). Matsunaga and Takashashi determined organic mercury by t h e cold vapor method using solvent extraction and back extraction into a reagent suitable for reduction ( 1 1 ) . T h e rapid development of heated furnace atomization for atomic absorption offered new opportunities in trace metal analysis. Inorganic mercury was first determined by this technique by Ediger (12),who used ammonium sulfide as a stabilizing agent. Alder and Hickman (13) demonstrated the stabilizing effect of hydrochloric acid and hydrogen peroxide on mercury determination in the graphite furnace. T h e work of these earlier researchers on standard solutions demonstrated t h e potential of the graphite furnace for use in developing a method for determining trace quantities of mercury in fish. T h e graphite furnace, unlike the cold vapor technique which could be applied to mercury only, could be used t o determine other trace elements. T h e purpose of this study was t o develop a simple and reliable method that can be used to determine methylmercury in fish tissue by laboratories with atomic absorption capability. In this method, a modified toluene extraction adapted from the method of Uthe e t al. (5) was used, and the methylmercury content in the toluene was directly determined in the graphite furnace, using dithizone as a stabilizing agent. An aqueous system was adapted by extracting the methylmercury from the toluene with sodium thiosulfate a n d determining its concentration in t h e graphite furnace. 0003-2700/7910351-0414$01 0010
RESULTS A N D DISCUSSION S t a b i l i z a t i o n of M e t h y l m e r c u r y i n the F u r n a c e . Methylmercury was stabilized in the furnace by the addition of dithizone in toluene added to a n equal volume of sample extract. It was noted t h a t the atomic absorption of methylmercury in a swordfish extract reaches a maximum arid dithizone added (Figure 1). T h e optimum plateau a t 0.087~ charring temperature t o give the maximum absorption a t atomization was fairly narrow and ranged from 175-200 "C (Figure 2 ) , but t h e signals were reproducible. This low charring temperature caused some buildup of residue in the graphite tube, but this was reduced by prewashing the fatty samples with acetone and by atomizing the samples for 30 s. T h e graphite tube was given a high temperature burn two or three times a day to ensure no significant residue buildup, arid it could be used for 400-500 injections. T h e standard curve C
1979 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 51,
NO. 3,
MARCH 1979
415
Table I. Determination of Methylmercury (MeHg) in Fish and Shellfish (in pg Hg/g) by the Heated Graphite Furnace (Authors' Lab), with Comparative Analyses Data
MeHg in toluene ext.
sample freeze-dried oyster halibut canned tuna scallop muscle lobster tomalley swordfish muscle swordfish muscle swordfish muscle canned tuna 1 canned tuna 2 frozen swordfish liver 1 swordfish liver 2 swordfish liver 3 swordfish muscle swordfish muscle swordfish muscle swordfish muscle Not detectable.
-
50
i
f
E
30
.-I" m m 20 ; X W 0
a
0.06, 0.06 5.49, 5.55 0.78, 0.74 0.06, 0.06 0.05, 0.06 0.40, 0.42 2.52, 2.48
N.D.Q 5.31, 5.68 0.76 2 0.05 ( 6 ) N.D. N.D. 0.43 i 0.06 2.37 i 0.06 3.00 i 0.10 0.40, 0.40 1.90, 1.95
1
2
3
-
0.41, 0.42
-
0.41, 0.42 8.67, 8.25 1.50, 1.62 1.96 i 0.15 (4) 3.98, 4.22
1
2 3 4
9.14, 8.39 1.38, 1.81
-
-
-
1.60, 1.50
0.87, 0.81
/
external lab 2 inorganic Hg by total Magos' Hg
external lab 1 MeHg by total GLC Hg
total Hg 0.18 t 0.02b 5.79 t 0.07 0.92 i 0.03 0.06 i. 0.02 0.74 i 0 02 0.41 * 0.02 2.54 i 0.07 3.34 i 0.06 0.50 I0.01 2.23 * 0.19
0.02 4.55 0.60 0.05 0.09
0.25 4.77 0.80
-
-
0.200 0.034
5.85 0.95
0.64
-
-
-
-
0.11
-
-
2.75
-
0.022 0.109 0.143
3.03
0.42 2.45 3.31
1.05, 1.00 19.1 t 0.72
2.27 t 0.12 2.02, 1.99 4.40, 4.41 1.71, 1.70 0.92 i 0.12
Figures are the mean and standard deviation of triplicate analyses except otherwise indicated. 60
E
-.-
authors' lab MeHg in Na1S10, ext.
r--
E E
.E
50-
40-
c
c
.-W
30-
I r 0
0"
jo-
20 -
10 1
,
,
1
,
,
,
,
,
,
-
-
,
0
200
100
300
Furnace c o n t r o l setting
OC
Figure 2. Optimization of charring temperature for swordfish muscle toluene extract. 0.10% dithizone solution was added to an equal volume of extract. Two to three injections of 20 p L solution. Furnace controls were: drying at 100 'C for 40 s, and atomizing at 950 'C for 30 s
was linear from 1 to 6 ng of mercury, but the samples showed a slight suppression of absorption signal (Figure 3). Much lower sensitivity was obtained with pyrolytically coated graphite tubes (Perkin-Elmer),in contrast to the enhancement effect found for the difficult-to-volatilize elements (15). Extraction of Methylmercury. The extraction procedure is an adaptation of the method of Uthe et al. (51, combined with a n acetone prewashing technique of Watts e t al. ( 7 ) . Prewashing with acetone was necessary with oyster and fatty samples such as liver tissue and tuna canned in oil. Nonfatty muscle tissue did not require this pretreatment but, to be consistent and to standardize the method, the prewashing was carried out on all samples. When the extraction was carried out in a glass centrifuge tube under the present conditions, the acid concentration of the aqueous mixture had to be for a more complete extraction, except in increased to 4.0 9the case of freeze-dried and powdered samples, where the original 1.3 N was used to avoid formation of a gel a t the interface. Analysis of Fish Samples. A number of fish samples were analyzed for methylmercury using this procedure. Total mercury content of these samples were also determined for
v ,
/ , 0
, 2.0
I
4.0
6.0
8.0
10.0
ng Methyl mercury ( a s H g ) Figure 3. Standard (-0-)and sample (-04 calibration curves of toluene extract stabilized with dithizone. The sample was tuna canned in oil. 10 pL of standards and 20 pL of spiked samples were used
comparison (Table I, Authors' Lab.). Recovery was determined by adding 2.0e4.00 pg methylmercury to samples after prewashing with acetone, and comparing results with unspiked samples. The average recovery for nine different samples in
416
ANALYTICAL CHEMISTRY, VOL. 51, NO. 3, MARCH 1979
(60
4 140
.-
120
ng Methyl mercury ( a s Hg)
Figure 4. Standard (-0-)and sample (-e-) calibration curves of 0.005 M Na,S,O, extract. The sample was tuna canned in oil. 10 pL of standards was determined, while the sample was determined by overlaying 10 pL of standard solution onto 10 pL of sample in the furnace. Furnace controls were: drying at 100 OC for 40 s, charring at 150 O C for 30 s, and atomizing at 1200 O C for 30 s with purge gas interrupted
*
duplicate was 97.7 5.5%. The detection limit was 0.08 pg/g sample, using 20 pL of sample plus 20 pL spiking standard. T h e calculation was based on an absorbance signal of 3-mm peak height and an average slope of several sample calibration curves. An average of 92.2% of the mercury in muscle samples was methylmercury. This showed t h a t prewashing the samples with acetone did not remove appreciable amounts of tightly bound organic mercury, confirming the results of Watts and co-workers (7). T h e swordfish liver samples, however, contained a much lower percentage of methylmercury t h a n muscle. This was not surprising because the liver had been suggested as one of the sites for demethylation of methylmercury (16, 17). A d a p t a t i o n t o A q u e o u s S y s t e m . In order to allow laboratories to utilize autosamplers (e.g., Perkin-Elmer AS-1) which cannot handle organic solvents, the methylmercury in the toluene extract was partitioned into aqueous sodium thiosulfate, and then determined by the graphite furnace. It was found t h a t a simple one-step partitioning gave excellent recovery. Two to six mL of the toluene extract was pipetted into a 15-mL glass stoppered conical centrifuge tube, and then 2.0 mL of 0.005 M aqueous Na2S203(freshly diluted from 0.05 M stock) was added. T h e mixture was shaken 40 times by hand, and centrifuged to separate the layers. T h e toluene layer on top was aspirated off and the mercury concentration in the aqueous solution was determined by the method of standard additions (Figure 4). The standards for spiking were prepared in toluene and extracted with the thiosulfate solution. T h e standards were linear to about 5.5 ng a t the optimal sample atomization temperature (Figure 4). However, the sample calibration curves were linear to a wider range, for example, canned tuna as shown in the same figure. T h e extraction of methylmercury from toluene into thiosulfate was complete in most cases (Table I). T h e average recovery of 14 extractions of 5 different types of samples was 102 k 5.17'~. Sample extracts that were low in methylmercury, for example, oyster and lobster tomalley (Table I) were concentrated two to three times by the partitioning, thus increasing the sensitivity of the method. C h e c k S a m p l e S t u d y . Six of the freeze-dried samples were sent to a n external laboratory for methylmercury de-
termination by the gas-liquid chromatographic method of Uthe et al. ( 5 ) , and also for total mercury determination. Except for the halibut sample, the results (Table I, External Lab I) were in good agreement. T h e low percentage of methylmercury in oyster and lobster tomalley found by us was verified. T h e halibut results indicated a difference in the quantitation of mercury b u t confirmed t h a t most of the mercury in this fish was methylmercury. T h e same halibut, tuna, and three freeze-dried swordfish samples were analyzed in a third laboratory using a modified Magos' (9) selective reduction atomic absorption method. Total and inorganic mercury in the fish were determined and organic mercury was calculated by difference (Table I, External Lab 2 ) . The mercury results obtained by this laboratory were almost identical to ours, and thus substantiated our finding, that these samples had very little inorganic mercury. CONCLUSIONS The furnace atomic absorption method for the analysis of methylmercury in fish presented here, does not distinguish the different forms of organic mercury, b u t is based on the findings t h a t with rare exception (18), the organic form of mercury found in fish has been methylmercury ( 2 ) . When in doubt, the species of organic mercury can be identified by a thin-layer chromatographic technique similar to the one described by Westoo ( 4 ) or T a t t o n and Wagstaff (19). T h e present method enables laboratories engaged in trace metal analyses to carry out methylmercury determinations in fish without having to install separate gas-liquid chromatographic equipment which is costly to operate and maintain (20). In addition, use of an autosampler allows methylmercury determination a t a much faster rate than gas-liquid chromatographic methodology. ACKNOWLEDGMENT T h e authors express their gratitude to A. Lutz of the Freshwater Institute, Fisheries and Oceans Canada, and A. S. Hall of the National Marine Fisheries, U.S.Department of Commerce for their cooperation in the comparative studies. LITERATURE CITED W . R. Hatch and W . L. Ott, Anal. Chem., 40, 2085 (1968). G. Westoo, "Chemical Fallout", M. W. Miller and G. G. Berg, Eds., Charles C Thomas, Springfield, Ill., 1969, p 75. P. L. Bidstrup, "Toxicity of Mercury and Its Compounds", Elsevier, New York, 1964. G. Westoo. Acta Chem. Scand., 20, 2131 (1966). J . F. Uthe, J . Solomon, and B. Grift, J . Assoc. Offic. Anal. Chem., 55, 583 (1972). M. L. Schafer, U. Rhea, J. T. Peeler, C. H. Hamilton, and J. E. Campbell, J . Agric. Food Chem., 23, 1079 (1975). J. 0. Watts, K. W. Boyer, A. Cortez, and E. R. Elkins, Jr., J . Assoc. Offic. Anal. Chem.. 59. 1226 11976). A. J. Malanos'ki, K. Helrich, S. 'Williams, B. J. Prima, and S. W . Butler, J . Assoc. Offic. Anal. Chem., 61, 397 (1978). L. Magos, Analyst(London), 96, 847 (1971). J . L. Kacprzak and R. Chvojka, J . Assoc. Offic. Anal. Chem., 59, 153 (1976). K. Matsunaga and S. Takahaski, Anal. Chim. Acta, 87, 487 (1976). R. D. Ediger, At. Absorpf. Newsl.. 14, 127 (1975). J. F. Alder and D. A. Hickrnan, Anal. Chem., 49, 336 (1977). F. A. J. Armstrong and J . F. Uthe, At. Absorpt. Newsl., 10, 101 (1971). R. E. Sturgeon and C. L . Chakrabarti. Anal. Chem., 49, 90 (1977). W. D. Burrows and P. A. Krenkel. Environ. Sci. Techno/..7 , 1127 (1973). K . R. Olson, K. S. Squibb, and R. J. Consins, J . Fish. Res. Board Can., 35, 381 (1978). S. Yarnanaka and K. Ueda. Bull. Environ. Contam. Toxicol., 14, 409 (1975). J. O'G. Tatton and P. J. Wagstaffe. J . Chromatogr., 44, 284 (1969). J . F. Uthe and F. A. J. Armstrong, Toxicol. Environ. Chem. Rev., 2, 45 (1974).
RECEI~TD for review September 5 , 1978. Accepted November 29, 1978.
