Direct determination of mercury in atmospheric particulate matter by graphite plate filtration–electrothermal atomic absorption spectrometry with Zeeman background correction

Spectrochimica Acta Part B 55 Ž2000. 395]402

Analytical note

Direct determination of mercury in atmospheric particulate matter by graphite plate filtration]electrothermal atomic absorption spectrometry with Zeeman background correction Jimmy C. YuU , Bicheng Zhang, Yuen-Kwan Lai Department of Chemistry, The Chinese Uni¨ ersity of Hong Kong, Shatin, New Territories, Hong Kong Received 4 November 1999; accepted 15 February 2000

Abstract A new analytical method has been developed for direct analysis of mercury in atmospheric particulate matter by a graphite plate collection technique coupled with Zeeman electrothermal atomic absorption spectrometry ŽETAAS.. A small porous graphite plate filter was used initially for collecting particulates and subsequently as a platform in ETAAS. A relatively high ashing temperature of 2508C and a low atomization temperature of 9008C could be achieved by using both a PdCl 2-coated pyrolytic graphite tube and a chemical modifier consisting of a 5% ŽNH 4 . 2 S solution. Excellent linearity was maintained over the range of 0]5 ng of Hg. When sampling at a flow rate of 350 ml miny1 for 3 h, this was equivalent to a working concentration range of 0]80 ng my3 of particle-bound mercury in air. Analysis of the NIST SRM 1648, Urban Particulate Matter, gave a recovery of 101.4%, and a precision of 9.0% R.S.D. The detection limit was estimated to be 0.74 ng my3. The accuracy was within "8% when compared with the traditional glass-fiber filtration]acid digestion]cold-vapor atomic absorption spectrometry method. Q 2000 Elsevier Science B.V. All rights reserved. Keywords: Mercury determination; Atmospheric particulate matter; Electrothermal atomic absorption spectrometry; Zeeman background correction; Graphite plate filtration

U

Corresponding author. Tel.: q852-2609-6268; fax: q852-2603-5057. E-mail address: [email protected] ŽJ.C. Yu. 0584-8547r00r$ - see front matter Q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 5 8 4 - 8 5 4 7 Ž 0 0 . 0 0 1 6 6 - X

396

J.C. Yu et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 395]402

1. Introduction

Mercury ŽHg. is a potentially toxic environmental pollutant that is among the most highly bioconcentrated trace metals in the human food chain, and several international committees have targeted Hg for special attention with regard to its emissions, cycling, and health effects w1x. Mercury in atmospheric particulate matter ŽAPM. originates from both natural Žcrustal, aquatic, etc.. and anthropogenic sources Žsmelters, fossil fuel combustion, waste incineration, etc.. w2,3x. Particulate mercury can be a very important form of atmospheric Hg under certain conditions w4x. To understand the cycling and distribution of this element in the environment, accurate determination of particulate mercury is necessary w5x. The most widely used methods for determining Hg are cold-vapor atomic absorption spectrometry ŽCVAAS. and cold-vapor atomic fluorescence spectrometry ŽCVAFS. w6x. Although the coldvapor techniques are convenient for measuring gas-phase Hg, they cannot be directly used for determining Hg in APM. The solid samples must be digested and then the solution is reduced and finally aerated to liberate all elemental Hg into the gas phase w7x. A conventional system for sampling APM consists of a filter pack, and the APM collected is subsequently digested in an acid mixture followed by spectrometric analysis w8x. The major disadvantage of the conventional technique is that it involves several filter and sample handling steps which require ultra-clean sampling techniques and a clean room facility to avoid contamination w9x. Moreover, the time required for this tedious process is usually very long. The limitations and advantages of some common techniques for the determination of particulate mercury have been summarized in a recent review paper w5x. To overcome these limitations, alternative sampling and detection methods have been proposed. For example, Liang et al. w10x deposited APM onto an electrothermal atomizer tube by jet-impaction, and then analyzed by ETAAS or laser-excited AFS. More recently, Lu et al. w11x collected particulate-phase mercury on a quartz

fiber disc held in a miniaturized device and analyzed using a pyrolysisrgold amalgamationrthermal desorptionrCVAFS technique. This paper describes a new analytical method for direct analysis of APM by coupling a small graphite plate sampling technique with ETAAS detection. It is known that porous electrographite plates are ideal for filtration of air particles w12x. Such a small graphite plate can be inserted directly into a graphite atomization tube for analysis. This is why ETAAS is preferable to cold-vapor atomic absorptionrfluorescence or nuclear methods in this study. After filtration of a specific volume of air, the small graphite plate is inserted into a PdCl 2-coated pyrolytic graphite tube to form a platform for atomization. Since the plate is heated primarily by radiational heat transfer from the hot graphite tube, atomization of analyte from the plate surface is delayed relative to that from the surface of the graphite tube wall. Atomization occurs at the steady-state temperature of the vapor phase, which is also at a higher temperature than in a conventional graphite tube. This method does not require tedious sample digestion and thus avoids risks of sample contamination and loss of analytes. In order to minimize the matrix interference in mercury determination by ETAAS, it is necessary to use the highest possible ashing temperature. Therefore, the selection of a suitable matrix modifier to stabilize mercury in the sample is critical. A good modifier should allow the use of a high enough pyrolysis temperature to remove the bulk of matrix species during thermal pre-treatment of the sample without losing any analyte prior to the atomization stage. Another potential problem is whether the vapor-phase Hg in air can be adsorbed onto the graphite plate while APM is collected by the porous graphite plate filter. The measurement result of mercury in APM by Zeeman ETAAS will be incorrect even if the mercury in the vapor-state part is adsorbed. The objective of this work was to achieve the direct determination of trace mercury in APM using a new analytical procedure consisting of an electrographite plate for filtration of APM and then using the same plate for atomization in a PdCl 2-coated pyrolytic

J.C. Yu et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 395]402

graphite tube with chemical modifier and quantitation of mercury by Zeeman ETAAS.

2. Experimental 2.1. Equipment The measurement of analyte absorbance was carried out using a Hitachi Model 8200 atomic absorption spectrophotometry with Zeeman background correction employing a transverse magnetic field. The AAS was equipped with a GA power source unit and an SSC-300 autosampler. An air sampler, model KB-120E ŽQingdao Experimental Institute of Electron Instrument, China., was used for collecting APM in air with a glassfiber membrane filter as a conventional sampling apparatus. The graphite plate filtering device was made of Teflon, and its details are shown in Fig. 1. Graphite plates Ž8 = 3 = 0.5 mm. were fabricated from high-purity porous electrographite ŽShanghai Carbon Works, China, Model SMF-600, with 18, 20 and 24% porosity. in our workshop.

397

Table 1 Instrumental parameters Wavelength Slit width Lamp current PMT voltage Signal mode Temperature control Time constant

253.7 nm 1.30 nm 6.0 mA 520 V BKG correct Optical 0.10 s

The pores are clearly visible on a scanning electron micrograph ŽFig. 2.. Air was drawn through the middle of the plate by means of a rotary vacuum pump, thereby collecting the atmospheric particles. A dual-plate filtering device, constructed the same way as the single-plate device except that it had two graphite plate cavities, was used to determine the collection efficiency of the graphite plates. Hitachi pyrolytic graphite tubes were used. A mercury hollow cathode lamp ŽHitachi. was used as the line source at 253.7 nm. Absorbance signals Žpeak-height mode. were recorded. The instrumental parameters are shown in Table 1. Palladium chloride-coated pyrolytic

Fig. 1. Schematic diagram of the graphite plate filtering device.

398

J.C. Yu et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 395]402

Fig. 2. SEM image of the graphite plate with 18% porosity.

graphite tubes were prepared by submerging the tubes in a saturated palladium chloride solution for 4 h, drying at room temperature, and heating in an oven at 1208C for 1 h. The procedure was repeated once and the tubes were then placed in the graphite furnace and heated according to the temperature programme. 2.2. Reagents A stock standard solution of 1000 mg mly1 mercury in 2% nitric acid ŽBDH. was used. Working standard solutions were obtained through appropriate dilution of the stock standard solution just before use. All chemicals were of analytical reagent grade. A palladiumŽII. chloride-saturated solution was prepared by dissolving a suitable amount of palladium chloride ŽMerck. in 1 M HCl solution, with gentle heating and dropwise addition of conc. HNO3 . Ultra-pure water of 18.3 MV-cm resistivity was obtained directly from a Milli-Q2 water purification system ŽMillipore.. A solution of 5% Žwrv. ammonium sulfide as a

chemical modifier was prepared by dissolving 5.0 g ammonium sulfide ŽMerck. in 100 ml water. 2.3. Procedure Since the proposed direct determination technique did not involve contamination-prone filter sample handling steps or acid digestion, a clean room facility was not required. To prepare absolutely Hg-free graphite plates for air sampling, each filter plate was fired at 26008C in the graphite furnace at least three times before use. The graphite plate was placed in the Teflon sampling device to filter air for a pre-determined time Žtypically 3 h at a flow rate of 350 ml miny1 .. The plate with APM on it was then inserted into the center of a PdCl 2-coated graphite tube. A 5-ml aliquot of 5% ammonium sulfide chemical modifier was injected onto the plate using an autosampler. Table 2 shows the temperature programme for the Zeeman ETAAS measurements. The Hg content in APM was calculated as sug-

J.C. Yu et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 395]402

399

Table 2 Graphite furnace temperature programme for determination of mercury Stage

Temperature Žo C.

Ramp time Žs.

Hold time Žs.

Internal flow of purge gas Žml miny1 .

Dry Ash Atomize Clean out Cool down

80]140 250 900

40 30

0 0 10 4 5

200 200 0 200 200

gested by Chakrabarti et al. w13x: C x Ž ng r m3 . s 10 3 wŽ C2 y C1 .Ž A x y A1 . r Ž A 2 y A1 . q C1 x V Ft where F is the rate of filtration in milliliters per minute; t is the sampling time in minutes; C1 and C2 are the concentrations in nanograms per milliliter of the aqueous standards; A x , A1 and A 2 are the peak height absorbances due to the unknown mercury content of the sample collected, and of the first and the second standard solution, respectively Ž A 2 value should be bigger than A x value and A1 value should be smaller than A x value.; V is the volume in microlitres of the aqueous standard injected on the plate.

3. Results and discussion 3.1. Selection of chemical modifiers Previous studies have shown that loss of mer-

cury would occur during the drying and ashing cycles, but this effect could be minimized with matrix modification w14,15x. We tested six chemical modifiers and the results are summarized in Table 3. The best chemical modifier was the 5% ammonium sulfide solution. When used together with a PdCl 2-coated pyrolytic graphite tube, an ashing temperature of up to 2508C was possible. Even with an ordinary pyrolytic graphite tube, an ashing temperature as high as 2258C could be used. The effect of the amount of ammonium sulfide added on the absorbance of mercury was also investigated. Fig. 3 shows that 250 mg of ŽNH 4 . 2 S provides a maximum absorbance of 0.121 for 5 ng mercury at 2508C ashing temperature. The role of ammonium sulfide as a chemical modifier could be attributed to the formation of stable mercuryŽII. sulfide at the ashing temperature w16x. The palladium chloride-coated tubes could also preserve mercury by the formation of an intermetallic compound Žor alloy. between mercury and palladium w17,18x. Fig. 4 shows the atomization curve of 5 ng Hg with 5 ml of 5%

Table 3 The highest thermally stable temperature tolerated at ashing stage with different chemical modifiers for 5 ng Hg Chemical modifiers

Peak-height absorbance

R.S.D.% Ž n s 3.

Tolerated ashing temperature Žo C.

250 mg ŽNH4 .2 Sa 250 mg ŽNH4 .2 S 15 mg Na2 S2 O3 250 mg ŽNH2 .2 CSa 25 mg ŽNH4 .2 Sq 5 mg MgŽNO3 .2 7.5 mg Pd q 5 mg MgŽNO3 .2 6.9 mg Pd q 4.6 mg MgŽNO3 .2 q 0.38 mg Se

0.121 0.120 0.115 0.058 0.116 0.008 0.008

2.31 2.64 3.54 6.51 2.33 9.78 14.81

250 225 150 200 200 180 180

a

Using PdCl 2 -coated pyrolytic graphite tube.

400

J.C. Yu et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 395]402 Table 4 Collection efficiency test for mercury in APM using graphite plates with different porosities

Fig. 3. Effect of the amount of ŽNH 4 . 2 S on ashing temperature.

ŽNH 4 . 2 S solution over a temperature range of 900]20008C. It shows that complete atomization of mercury can be achieved at 9008C. Such a low atomization temperature should minimize interferences from the matrix species in APM. 3.2. Collection efficiency of particle-bound mercury Since particle-bound mercury accounts for only a small fraction of the total Hg in ambient air w4x, the influence of vapor-phase mercury on the measurement should be investigated. Lech et al. w19x found that porous graphite failed to adsorb mercury vapor unless it was electroplated with a thin layer of gold. Our result also showed that adsorption of mercury vapor on graphite surface was negligible, and it should not affect the measurements. The collection efficiency of the porous graphite plate filter was tested using a dual-plate filtering device. The porosity of the second Žbottom. plate was 18% while plates of different porosities were used as the first Žtop. plate. The amount of mercury found in the first plate over

Fig. 4. Effect of atomization temperature on absorbance.

First plate porosity

24%

20%

18%

Second plate porosity

18%

18%

18%

Collection efficiency of the 1st plate

69%

92%

99%

the total amount of mercury in the two plates was taken as the collection efficiency of the first plate. Table 4 shows the results of the collection efficiency test for mercury in APM using the graphite plate filters. The test results indicated that the collection efficiency of the graphite plate filter with porosity of 18% for mercury in APM was 99% when flow rate and collection time were 350 ml miny1 and 3 h, respectively. 3.3. Precision and detection limit The precision of the method was evaluated by replicate determinations for the aqueous solution of 5 ng mercury standard with 5 ml of 5% ammonium sulfide chemical modifier. The relative standard deviation ŽR.S.D.. for 11 replicate determinations was 5.0% for 5 ng of mercury. The analytical calibration curve Žpeak-height mode. for mercury with the ammonium sulfide chemical modifier shows an excellent linear response in the concentration range of 0]5 ng Hg. The regression equation for the calibration curve was y s 1.7= 10y2 xy 2.6= 10y2 with a correlation coefficient of 0.9950 Ž n s 7., where xs the analyte mass Žng. and y s peak absorbance. When sampling at a flow rate of 350 ml miny1 for 3 h, this is equivalent to a working concentration range of 0]80 ng my3 of particle-bound mercury in air. Table 5 shows the analytical results of three aqueous standard solutions prepared by independent digestion of SRM No. 1648 Urban Particulate Matter. The average recovery Ž n s 3. of mercury was 101.4% when compared with literature values w20x. A precision of 9.0% and detection limit of 0.74 ng my3 were obtained from 11 replicate measurements for each of the samples.

J.C. Yu et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 395]402

401

Table 5 Determination of mercury in three samples prepared by independent digestion of NIST SRM 1648 with 5% ŽNH 4 . 2 S modifier Sample

Theoretical wHgxa Žng mly1 .

Found wHgx Žng mly1 .

Recovery Ž% .

R.S.D. Ž% .

Detection limit estimated Ž3s . Žng my3 .

1 2 3

17.32 9.67 17.60

16.67 10.50 17.49

96.25 108.6 99.38

8.9 9.5 8.7

0.85 0.74 0.64

101.4

9.0

0.74

Average a

The theoretical concentrations were calculated from the literature value of mercury in SRM No. 1648 and the amount of the standard used in preparing the three samples.

Detection limits were estimated from 3s of the blank samples Ž; 0.047 ng., assuming sampling at a flow rate of 350 ml miny1 for 3 h Žtotal volume of air s 0.063 my3 ..

The average concentration of particulate-bound Hg in our laboratories was 2.2 ng my3 . The accuracy was better than "8% when compared with the traditional glass-fiber filter collection]CVAAS technique.

3.4. Interference study Zeeman background correction is an effective technique for minimizing matrix interference associated with direct solid sampling AAS w21x. We found that the use of Zeeman background correction caused a slight drop in sensitivity but it greatly enhanced the reproducibility. Interferences by various substances in the determination of mercury by Zeeman ETAAS were studied by spiking different amounts of the substances into Hg sample solutions. Eighteen major components of APM were selected as foreign ions in the determination of 5 ng Hg and 250 mg ŽNH 4 . 2 S chemical modifier. The results are shown in Table 6. No significant interference was found since the relative error ŽR.E.. was no more than "5%. 3.5. Accuracy and results of laboratory measurements The accuracy of the proposed method was compared with a traditional procedure of glass-fiber filtration]acid digestion followed by cold-vapor atomic absorption spectrometry determination w22,23x. Simultaneous samplings of APM were carried out in our laboratories. Graphite plates of 18% porosity were used and the flow rate was maintained at 350 ml miny1 for 3 h. Table 7 shows the results of the parallel measurements.

4. Conclusions Mercury associated with atmospheric particulate matter can be determined with good accuracy and precision using a graphite plate sampling Table 6 Interference tests of foreign ions in 5 ng Hg Foreign ions

Amount of foreign ions added Žng.

Amount of Hg found Žng.

R.E. Ž%.

Al3q AsŽIII. Ba2q Cd2q Cr3q Cu2q Fe3q Kq Mg2q Mn2q Naq Ni2q Pb2q SbŽIII. SeŽIV. Zn2q Cly SO4 2y

500 5 50 5 25 12.5 5 500 50 1000 50 10 500 250 250 500 50 1000

5.13 4.86 4.78 4.83 4.75 5.16 4.86 4.97 4.83 5.02 5.25 4.97 4.85 4.97 5.19 4.92 5.25 5.02

q2.6 y2.8 y4.4 y3.4 y5.0 q3.2 y2.8 y0.6 y3.4 q0.4 q5.0 y0.6 y3.1 y0.7 q3.8 y3.6 q5.0 q0.4

402

J.C. Yu et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 395]402

Table 7 Comparison of the developed method and the traditional glass]fibre filter]CVAAS technique Sample No.

1 2 3

Mercury content in APM in air Žng my3 . Graphite plate]ETAAS

Glass-fiber filter]CVAAS

Accuracy R.E.

0.66 1.96 3.93

0.70 2.13 3.78

y5.7% y8.0% q4.0%

device followed by direct analysis of the plate by Zeeman ETAAS. A relatively high ashing temperature of 2508C can be obtained when PdCl 2coated pyrolytic graphite tubes and a chemical modifier consisting of a 5% ammonium sulfide solution are used. With graphite plates of 18% porosity, the collection efficiency of particlebound mercury was 99%. When sampling at a flow rate of 350 ml miny1 for 3 h, the detection limit was estimated to be 0.74 ng my3 . Linear absorbance response was obtained for a concentration range of 0]80 ng my3 particle-bound mercury in air. Compared to traditional techniques, the proposed method is less time-consuming and it is a better time-resolved procedure. With minor modifications, this method should also allow the direct determination of other trace elements in atmospheric particulate matter.

Acknowledgements The authors are grateful to the Research Grants Council of Hong Kong, Research Project CUHK4187r97P, for financial support. References w1x S.E. Lindberg, J.L. Price, J. Air Waste Manage. Assoc. 49 Ž1999. 520]532. w2x I. Olmez, M.R. Ames, G. Gullu, Environ. Sci. Technol. 32 Ž1998. 3048]3054. w3x R.P. Mason, W.F. Fitzgerald, F.M.M. Morel, Geochim. Cosmochim. Acta 58 Ž1994. 3191]3198. w4x J.Y. Lu, W.H. Schroeder, Talanta 49 Ž1999. 15]24.

w5x J.Y. Lu, W.H. Schroeder, Water Air Soil Pollut. 112 Ž1999. 279]295. w6x H. Morita, H. Tanaka, S. Smimomura, Spectrochim. Acta Part B 50 Ž1995. 69]84. w7x W.L. Clevenger, B.W. Smith, J.D. Winefordner, Crit. Rev. Anal. Chem. 27 Ž1997. 1]26. w8x J.L. Guentzel, W.M. Landing, G.A. Gill, C.D. Pollman, Water Air Soil Pollut. 80 Ž1995. 393]402. w9x G.J. Keeler, M.E. Hoyer, C. Lamborg, Measurements of atmospheric mercury in the Great Lakes Basin, in: C.J. Watras, J. Huckabee ŽEds.., Mercury Pollution } Integration and Synthesis, Lewis Publishers, Boca Raton, FL, 1994, pp. 231]250. w10x Z. Liang, G.T. Wei, R.L. Irwin, A.P. Walton, R.G. Michel, J. Sneddon, Anal. Chem. 62 Ž1990. 1452]1457. w11x J.Y. Lu, W.H. Schroeder, T. Berg, J. Munthe, D. Schneeberger, F. Schaedlich, Anal. Chem. 70 Ž1998. 2403]2408. w12x C.L. Chakrabarti, X. He, S. Wu, W.H. Schroeder, Spectrochim. Acta Part B 42 Ž1987. 1227]1233. w13x C.L. Chakrabarti, B. Marchand, V. Vandernoot, J. Walker, W.H. Schroeder, Spectrochim. Acta Part B 51 Ž1996. 155]163. w14x A. Volynsky, S. Tikhomirov, A. Elagin, Analyst 116 Ž1991. 145]148. w15x G.F. Kirkbright, H.-C. Shan, R.D. Snook, At. Spectrosc. 1 Ž1980. 85]89. w16x M. Filippelli, Analyst 109 Ž1984. 515]517. w17x X.-P. Yan, Z.-M. Ni, Anal. Chim. Acta 272 Ž1993. 105]114. w18x J.F. Alder, D.A. Hickman, Anal. Chem. 49 Ž1977. 336]339. w19x J.F. Lech, D.D. Siemer, R. Woodriff, Spectrochim. Acta Part B 28 Ž1973. 435]439. w20x S.M. Talebi, Int. J. Environ. Anal. Chem. 72 Ž1998. 1]9. w21x G. Strubel, V. Rzepka-Glinder, K.H. Grobecker, K. Jarrar, Fresenius J. Anal. Chem. 337 Ž1990. 316]319. w22x Guidelines for PM-10 Sampling and Analysis Applicable to Recepter Modelling, USEPA, EPA-452rR-94-009. w23x NIOSH, Manual of Analytical Methods, 4th ed., Method No. P & CAM 173, 1998.