JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 68 5 Studies on the Determination of the Metal Content of Airborne Particulates by Furnace Atomization Non-thermal Excitation Spectrometry* Christian Ludke Erwin Hoffmann and Jochen Skole lnstifut fur Spektrochemie und Angewandfe Spektroskopie (ISAS) Laboratorium fur Spektroskopische Methoden der Umweltanalytik (LSMU) Geb. 11. 1 Rudower Chaussee 5 12489 Berlin Germany Quantifying the metal content of airborne particulates is of outstanding relevance because of its potentially toxic and carcinogenic behaviour. The most widely used measuring strategy is based on filtration of particles from ambient air followed by analysis of the loaded filter tissue after wet digestion. In the present work an alternative approach was investigated which is based on drawing the air through the wall of a porous graphite tube in such a way that particulates are collected on the inner surface of the tube.Moreover the graphite tube acts not only as an efficient collector but can also be employed as an electrothermal atomizer when applying furnace atomization non-thermal excitation spectrometry. If combined with an echelle poly- chromator simultaneous multi-element determinations are possible. The amount of metals in air dust collected in this way were determined at the ng mP3 level with limits of detection calculated on the 3s criterion ranging between 0.1 and 1 ng m-3. The precision represented by the relative standard deviation varied from 0.12 to 0.26 depending upon the concentration of metals in the air.Keywords Furnace atomization non-thermal excitation spectrometry; airborne particulates; filters; multi- element analysis The increasing pollution of the atmosphere by natural and anthropogenic sources requires detailed investigation of a large variety of air polluting substances. Therefore measurements of atmospheric trace metals increasingly gain importance as they provide detailed information which can be used to identify sources of pollutants and follow their movements.' In order to study effects of major concern such as deposition mechan- isms or modifications of the climate the composition of particles of different sizes is of great interest. For these appli- cations it is desirable to measure the concentration of the elements in particles fractionated according to size.A pre- condition is that measurement methods having multi-element capability and high sensitivity are available. In most cases the presence of metals in the atmosphere is associated with airborne particulate matter. For the collection of airborne particulates many different sampling devices have already been developed.2 A conventional method is the collec- tion of particulates by pulling air through a filter tissue digestion of the filter and subsequent analysis of the solution preferably by atomic spectrometric techniques. Disadvantages of these methods are the time-consuming procedure and the fact that the chemical preparation of the samples involves sources of error^.^,^ These can result from blank values of the filters loss of certain elements during the digestion of filter material or contamination of the sample during the preparation procedure.A loss of particles is possible if they are smaller than the pore size of the filter. An alternative way of using filters is to filter the air directly through a porous graphite tube or cup which itself acts as the atomizer in an atomic spectrometric detection system. By applying such a sampling technique the disadvantages of filtering methods mentioned above are overcome. Drawing of air through a porous graphite tube followed by electrothermal atomic absorption spectrometry (ETAAS) has been described by Siemer and Wo~driff,~ as well as by Chakrabarti et aL6 Also combinations of electrostatic accumu- lation7 or single-stage impactions with ETAAS has been employed.Broekaertg reported on the direct determination of Pb and Cd by sampling the air through a porous graphite cup followed by excitation in a hot hollow cathode lamp. All these * Presented at the XXVIII Colloquium Spectroscopicurn Internationale (CSI) York UK June 29-July 4 1993. direct sampling methods except the last one have only single- element detection capability. The aim of the present work is to demonstrate a new method for the sensitive multi-element determination of trace metal contents in airborne particulates. In a simple sampling device air was pulled through the wall of a thermally cleaned porous graphite tube so that particulates were collected on the inner surface of the tube. After inserting the tube into the furnace atomization non-thermal excitation spectrometry (FANES) Source'O transient emission spectrometric signals were gener- ated by using an optimized working programme.For regis- tration of simultaneous spectrometric signals an tchelle polychromator was used." This method promises quick and sensitive multi-element analysis of collected particulates with- out any sample pre-treatment. Besides it is to be expected that the low limits of detection will allow short sampling times and analysis of the airborne particulates which are fractionated according to size. Experimental Sample Collection Sampling of airborne particulates was carried out by using a special porous graphite tube as filter. These graphite tubes were manufactured in-house from a block of pure graphite (28 mm long 6 mm i.d.and 1 mm thick). They had neither an injection hole nor any coating. According to the carbon products manufacturer (Ringsdorff-Werke Bonn Germany) the porosity was higher than that normally used for furnaces in analytical applications. The lower hardness of the graphite used is the reason for the shorter life of the tubes. The lifetime of a tube is about 60 firings. Before being used as a sampler each graphite tube was cleaned by three electrothermal heatings in the FANES source to 2600°C for 2 s. After this the furnace blanks were read in the normal heating programme (see Table 1). Cleaned tubes were stored separately in small polypropene containers before being used. For use as a filter the thermally cleaned graphite tube was held in a plastic vessel which was connected to a vacuum pump and had an opening for the entry of air.Inside the plastic vessel the graphite tube was fixed gas-tight on its ends686 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 Table 1 Working conditions for FANES source Element set 1 Ni 352.5 nm Fe 248.3 nm Cr 425.4 nm (:a 393.4 nm Sr 460.7 nm Mg 279.6 nm Temperature/ Ramp rate/ Step time/ Argon pressure/ Discharge current/ C "C s - I S h Pa mA 200 5 47 35 No power 53 800 200 5 2200 1600 3 35 No power 12 1013 50 50 50 1013 0 0 0 90 0 Element set 2 Na 586.9 nm Pb 405.8 nm Mn 403.1 nm Cu 324.8 nm T1 276.8 nm Cd 228.8 nm Temperature/ Ramp rate/ Step time/ Argon pressure/ Discharge current/ "C "C s - ' S hPa mA 200 5 47 35 No power 53 300 75 5 I800 lo00 4 35 No power 11 1013 20 20 20 1013 0 0 0 90 0 so that the air had to pass through the tube wall from inside to outside [see Fig.l(b)]. The system for size fractionated sampling of airborne par- ticulates consists of a combination of a standard sampling head for filter tissues of 50mm in diameter followed by the porous graphite tube in the plastic vessel. The standard sampling head was equipped with a poly( tetrafluoroethylene) 2 d I Fig. 1 Air sampling device (LI) scheme of set-up; and (h) detail A. 1 sampler; 2 magnetic valve; 3 timer; 4 pump; 5 porous graphite tube; 6 rubber seal; 7 plastic vessel; and 8 standard sampling head (PTFE) membrane filter of 5 pm pore size (Membrane Filter Type TE 38 Schleicher & Schiill Dassel Germany). Particles which have passed through the PTFE filter are collected in the graphite tube mounted behind it.An identical graphite tube without a pre-filter was used to collect the total air dust matter in the same sampling time. Six replicate samples were taken with the air sampling device as shown in Fig. 1. The sampling conditions were as follows. Samples (0.5-1.5 m3 of air) were taken at the site of the laboratory buildings in Berlin- Adlershof at a height of 3 m above the ground and at a distance of > 1.5 m from the buildings for a period of 2.5-4 h. A mechanical rotary pump suitable to maintain a pressure difference of 650 mbar (1 bar= lo5 Pa) at the sampling tube was used. Instrumentation To analyse the metal content of particulates collected in the graphite tube it was inserted into the FANES source.There the tube serves both as an electrothermal atomizer and as a cathode for the hollow cathode discharge. The computer assisted work programme of FANES involves temperature and time settings a pump-down step filling argon up to a low pressure level and switching the discharge voltage. Atoms generated by electrothermal atomization from the particulates collected were excited by impaction with the 4 r GraDhite tube Anode 1 Collimator Galvanometer Quartz Slits u k J mirror scanner prism I -e Pipette tip - - - - - Furnace ready for inserting the loaded graphite tube Fig. 2 Scheme of the experimental arrangementJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 687 discharge electrons. To measure the emission signal the FANES source was operated in conjunction with the Cchelle polychromator in a tetrahedral mounting designed by Falk et On the focal plane of the echelle polychromator about 120 optical fibres were fixed at the wavelength position of the most sensitive analytical lines.By connecting the appropriate optical fibres to the built-in photomultiplier tubes (PMTs) six elements can be measured simultaneously. A change in these connections allows the measurement of several sets of elements. For this study two sets of elements with six elements in each were chosen for simultaneous determination. With regard to their ashing and atomization temperatures the elements of each set were chosen so as to fit the compromise conditions best. The analytical signals decrease by < 10% when compro- mise rather than optimized conditions are used.For each set of elements separate tubes were taken for sampling and analy- sis. The working conditions of the FANES source (listed in Table 1) were adapted to the sets of elements also listed in the table. Analytical lines were selected on the basis of their providing the best detection limits as indicated by previous FANES reports.">' An oscillating quartz refractor plate 3 mm thick mounted on a galvanometer scanner behind the entrance slit of the polychromator (slit-width 150 pm) permits background correc- tion by wavelength modulation. A three-step square wave modulation was employed and the modulation frequency was 36Hz. All operational parameters of FANES as well as data acquisition and wavelength modulation were controlled by appropriate software installed in a PC.A schematic diagram of the experimental arrangement is given in Fig. 2. Both the FANES source and the Cchelle polychromator were manufac- tured in-house. Reagents and Calibration Reference solutions were prepared from inductively compled plasma (ICP) multi-element standard solution IV (Merck 11355.01) in 0.5 mol dmP3 HNO (sub-boiling quality). Neither the distilled water nor the HNO used contained measurable concentrations of the elements of interest here. These reagents were distilled sub-boiling in-house and the purity confirmed by FANES and ICP mass spectrometry measurements. For calibration a sample aliquot of 20 mm3 of reference solution was injected manually with a micropipette into the graphite tube still outside the furnace as seen in Fig.2. After this the tube was kept in a horizontal position and carefully inserted into the FANES source. The furnace was then closed and the 6 7 8 9 Fig. 3 Schematic sectioning of the digestion vessel 1 PTFE cap; 2 spring; 3 glassy carbon plate; 4 PTFE band; 5 graphite tube; 6 PTFE beaker; 7 glassy carbon vessel; 8 PTFE distance piece; 9 1 cm3 HNO f 3 cm3 HF concentrated heating programme started. The reproducibility of repeated injections of reference solutions in the uncoated tube ranges between 0.07 and 0.10 (relative standard deviation calculated from 12 measurements). For all elements studied linear cali- bration curves were determined. Two different methods were used to verify the accuracy of the calibration. (i) Isoformation by high-pressure vapour-phase digestion inside the graphite tube loaded with (a) airborne particulates and (b) dried reference solution.For this purpose a PTFE beaker containing the graphite tube was placed on top of a distance piece in the glassy carbon digestion vessel with only the distance piece standing in a mixture of concen- trated acids (1 cm3 HNO and 3 cm3 HF). A schematic diagram showing the sectioning of the digestion vessel is shown in Fig. 3. Six such glassy carbon vessels each with a spring-fixed lid were placed in a high pressure asher.I6 During the digestion procedure the temperature was maintained at 90°C and the pressure at 9 MPa for 1 h and 170 "C and 10 MPa respectively for another hour. After the gas-phase digestion the graphite tubes were analysed by FANES.(ii) Particulates collected at the inner surface of the tube were dissolved in 5rnoldm- HN03. For this purpose a small PTFE beaker containing the graphite tube was filled with HNO (1.5 cm3) till the tube was fully covered. After 8 h the graphite tube was removed and the remaining acid was analysed by electrothermal vaporization ICP-MSl7 according to the standard additions method. Results and Discussion Calibration The results of the different calibration procedures listed in Table 2 show acceptable agreement within the different methods. The results missing for Cd TI and Cu with the dissolution procedure were caused by diluting the analytes below the concentration which could be readily measured. Differences in calibration procedures checked by statistical tests were found to be random events.Although for Mn and Ni larger variations were observed a significance test at the Table2 Comparison of trace metal contents based on different calibration procedures; n = 5 Element Cd Pb Mg T1 c u Mn Ni Calibration with reference solution/ ng m-3 6.1 3t_ 0.4 212f 16 174+ 15 1.4f0.3 16+ 1 89+ 16 2 7 f 3 After isoformation/ ng m-3 6.6 f0.5 230+ 18 170f 14 2.5 f 0.5 15+ 1 64+6 36+7 After dissolution ng rn- 260 3t_ 70 120 f 40 - - - 100 f 30 50+7 Table 3 Metals contents determined in airborne particulates; n = 6 Element Ni Fe Cr Ca Sr Mg Cd Na TI Pb Mn c u Particulates unfractionated/ ng m-3 12$1 781 f 62 13+1 5200 f 420 27+2 580 & 29 7.2 2 514f 133 45$7 365 t- 72 174 +_ 21 64+_ 12 Particulates less than 5 pm/ ng m- 8.6 f 0.6 20+3 3 0.3 41 f 9 0.5$0.1 2.9 -t 0.3 1.9 -t 0.7 11.4f2 5.8 f 0.9 7.8 & 1 0.710.1 2.7 f0.9 LOD/ ng m-3 0.4 2 0.3 3 0.2 0.3 0.2 0.3 1 0.8 0.1 0.6688 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL.9 Table 4 Composition of metal contents in airborne particulates measured by various techniques Sampling method Sample Sample Measuring volume/m3 pre-'treatment method Filtering by quartz 60-70 Wet digestion ETAAS * Filtering by porous 1.5 None FANES fibre filter tissues ICP-AESt graphite tube Metal content/ng m-3 Element Cd Cr Ni Sr c u Pb Fe Ca Ref. 184 2.21-6.45 8.51-25.17 4.3- 10.1 13.4-29.2 26.0-363.3 89-49 1 1097-625 1 2216-4598 Ref. 195 0.99-3.04 7.0-16.2 7.3-11.5 - 33.0-110 '73-303 - - This work (n = 9) 5.8 & 0.7 15-t 1.2 15 f 1.5 26f3 45f5 228 f 40 633 2 59 4520 2 500 * ETAAS used for Cd and Ni.t ICP-AES used for Cr Ni Sr Cu Pb Fe and Ca. 5 Mean values for the year. 99% level could not confirm their difference. The difference in TI values cannot be seen as real either as the measurements were made near the detection limit where higher errors are more likely. Measured Concentration in Air Measured concentrations of metal content in airborne particu- lates are given in the first column of Table 3. Flow rates through the graphite tubes range between 140 and 200 dm3 h-' and air volumes sampled range between 1.1 and 1.5 m3. The approximate sampling time of 8 h gave a concentration value in the worst case (Ni= 12 ng m-3) a factor 30 higher than its limit of detection (LOD). Provided that a reliable determi- nation is still possible at 8 times the LOD a total analysis times of about 2 h is sufficient as the measuring time by FANES is only 2min and to change tubes takes only a few seconds.The comparison with measurements of metals in airborne particulates carried out by established methods shows satisfac- tory agreement as seen in Table4. The comparative values were measured in 198818 and 199119 on selected sites in Berlin by the Association for Technical Inspection. Values for 1993 were not yet available at the time of writing but they should not differ widely from previous measurements. Size-fractionated Sampling The porous graphite tube collectors mounted behind the standard filter head were also analysed after loading them into the FANES source.To compare the metal content in total air dust with the metal content in particulates of less than 5 pm in size the values in both cases are summarized in Table 3. The LODs given in the last column are based on a 3s estimation of 15 furnace blanks related to a sample volume of 1 m3 of air. In most cases the LODs are one order of magnitude lower than the measured contents in particulates less than 5 pm in size. The elements with the highest contents in the air e.g. Ca Mg Nay Fe and Pb exist for the most part as particles greater than 5 pm in size. Elements with contents below 100 ng mV3 such as Ni Cry Cd T1 and Cu were generaly found as particles of less than 5 pm in size. This result is presented graphically in Fig. 4. 80 60 A + 2 40 E a 20 0 Fig. 4 71.7 26.3 Ni Ccl Cr TI Cu Sr Fe Na Ca Mg Mn Pb Element Percentage of given elements contained in particulates of less than 5 pm size Conclusion Collection of particulates by filtration of air through the wall of a porous graphite tube followed by simultaneous multi- element detection using a FANES-Cchelle system has proved to be a useful tool for quantifying the metal content of airborne dust. The procedure is quick and easy does not need any chemical preparatory steps and can be calibrated with refer- ence solutions.The detection power of the procedure presented was high enough to analyse the fraction of particulates smaller than 5 pm contained in 1 m3 of air. The financial support by the Senatsverwaltung fur Wissenschaft und Forschung des Landes Berlin and the Bundesministerium fur Forschung und Technologie is gratefully acknowledged.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL.9 689 1 2 3 4 5 6 7 8 9 10 11 References Gordon G. E. in Air Pollutants and Their ESfects on the Terrestrial Ecosystem ed. Legge A. H. and Krupa S. V. Wiley New York 1986 138. Klockow D. Fresenius’ Z . Anal. Chem. 1987 326 5. Tolg G. and Tschopel P. Anal. Sci. 1987 3 199. Tschopel P. Kotz L. Schulz W. Veber M. and Tolg G. Fresenius’ Z . Anal. Chem. 1980 302 1. Siemer D. D. and Woodriff R. Spectrochim. Acta Part B 1974 29 269. Chakrabarti C. L. He X. Wu S. and Schroeder W. H. Spectrochim. Acta Part B 1987 42 1127. Torsi G. and Bergamini D. Ann. Chim. 1982 79 45. Lian Z.-w. Wei G.-t. Irwin R. L. Walton A. P. Michel R. G. and Sneddon J. Anal. Chem. 1990,62 1452. Broekaert J. A. C. Bull. Soc. Chim. Belg. 1976 85 755. Falk H. Hoffmann E. and Liidke C. Prog. Anal. At. Spectrosc. 1988 11 417. Schmidt K. P. Becker-ROB H. and Florek S. Spectrochim. Acta Part B 1990 45 1203. 12 13 14 15 16 17 18 19 Falk H. Becker-Rol3 H. Florek S. Hoffmann E. Ludke C. and Tischendorf R. in 4. Coll. Atomspektrom. Spurenanal. ed. Welz B. Perkin-Elmer Uberlingen 1987. Falk H. Hoffmann E. and Liidke C. Spectrochim. Acta Part B 1984 39 283. Naumann B. Knull B. Kerstan F. and Opfermann J. J. Anal. At. Spectrorn. 1988 3 1121. Baxter D. C. Nichol R. Littlejohn D. Liidke C. Skole J. and Hoffmann E. J. Anal. At. Spectrorn. 1992 7 727. Knapp G. Intern. J. Environ. Anal. Chem. 1985 22 71. Hoffmann E. Ludke C. and Scholze H. J. Anal. At. Spectrum. submitted for publication. Technical Report No. D-89,485 ed. TUV Berlin e.V. Berlin 1989 170. Technical Report No. U-92/279 ed. TUV Berlin-Brandenburg Berlin 1992 58. Paper 3/07135I Received December 2 1993 Accepted February 22 1994