JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 653 Direct Determination of Cadmium and Lead in Geological and Plant Materials by Electrothermal Atomic Absorption Spectrometry Franci DolinQek Janet Stupar and Vinko Vrlieaj Jozef Stefan Institute University of Ljubljana 61 11 1 Ljubljana Yugoslavia A laboratory-assembled atomic absorption spectrometer and graphite cup atomizer were evaluated for direct analysis of solid samples. A number of geological (soil and sediment) and botanical reference materials were analysed for lead and cadmium either by direct introduction (0.02-5 mg) or by a slurry technique (5-200 mg). Integrated absorbance was measured and aqueous standards were used for calibration. An (NH4)2HP04 chemical modifier was necessary in the analysis of botanical samples.The effect of particle size on the accuracy and precision of analytical results was studied. Atomization of small sample aliquots (20-500 pg) by direct ‘weighing in’ sample introduction technique requires fine grinding (1 pm) whereas slurry sample introduction is more tolerant to particle size effects. Keywords Cadmium; lead; geological and botanical reference materials; solid sample introduction; electrother- mal atomic absorption spectrometry Direct determination of trace and minor elements in solid samples of geological and biological origin has long been a challenge to analytical chemists employing atomic spectro- metric methods of analysis. Several advantages of these techniques have been pointed out in the Among the most important are the ability to analyse very small samples and to reduce sample contamination in comparison with conventional solution analysis.A variety of technical approaches have been investigated for the introduction of solid samples into flames plasmas and graphite furnaces for atomization. The most convenient and easiest way of achieving this was to prepare a slurry from the powdered sample. Nebulizing slurries for intro- duction into a flame or plasma however proved to be inefficient producing severe systematic errors in the analy- tical result^.^^^ On the other hand electrothermal atomiza- tion owing to its specific features proved to be the most suitable for direct analysis of solids. It has been extensively employed for the determination of elements having low and medium volatility in a variety of Cadmium and lead have been the most frequently determined elements because of their environmental importance.’ 1-14~19-22924+30933 Solid samples can be introduced into a graphite furnace by a direct ‘weighing in’ technique or by pipetting slurries.Although the latter is preferred there are two major problems involved in these sampling techniques which affect the accuracy and precision of analytical results. The first is sampling error the difference between the content of the analyte of the measured fraction and that of the bulk of the sample which depends on the sample homogeneity and which is further related to the mean particle size and distribution. The sampling error is likely to be critical when using the direct ‘weighing in’ approach where extremely small amounts (20-500 pg) of geological or botanical samples are analysed directly.When employing the slurry sample introduction technique this sampling error can be largely eliminated by generally taking a sample amount of 5-200 mg. The only prerequisite in such a situation is that the slurry should be homogeneous during pipetting. This can be accomplished either by adding a thickening agent or by employing mechanical or ultrasonic agitation. Majidi and H ~ l c o m b e ~ ~ ~ ~ * brought attention to sedimentation and to the volumetric errors both of which might have a significant influence on the accuracy of analytical results. The second major source of error could result from incomplete vaporization of the analyte element in the furnace.Matrices of low volatility and refractory elements are more susceptible to this source of error. As the vaporization of the analyte in the furnace might in some way be related to the particle size of the sample particle size can be assumed to be an important parameter govern- ing the precision and accuracy of analytical results. This has been realized by several workers who have investigated the effects of particle size and distribution on the results obtained by direct analysis of solids using eletrothermal atomic absorption spectrometry (ETAAS). Fuller et aZ.,23 carrying out direct determination of Cr and V in ore samples by slurry ETAAS concluded that acceptable precision and accuracy could be obtained pro- vided the samples are ground to at least a 25 pm size and aqueous standards are used for calibration.It was further shown that sampling error becomes significant when part- icle size increased above this value. Jackson and Newmanls and Hinds et aZ.16 have measured lead and cadmium in different types of soil samples by direct introduction of slurried samples into the electrother- mal atomizer. It has been shown that significantly more lead is contained in smaller size fractions of the soil in comparison with larger particles and fine grinding (9OOh of particles of less than 11 pm) was found to be essential for complete recovery of these elements from soild samples. However the major source of error was ascribed to incomplete atomization rather than sampling error. The efficiency and reproducibility of pipetting the slurries were affected as the particle size increased beyond 50 pm.Hinds and Jackson1* and Karwowska and Jackson17 have also studied the atomization characteristics of lead from alu- mina matrices and artificial soil samples. It was reported that alumina and clay minerals have the same effect of retarding the vaporization of lead from the soil sample in a comparison with aqueous solutions. Organic matter in the soil however has the opposite effect of shifting the absorbance peak of lead to a shorter appearance time. The use of chemical modifiers and isothermal atomization were strongly suggested for use with direct solid analysis. In addition by careful examination of the absorbance-time profiles the possibility of determing lead fractionation in the soil samples was shown.Olayinka et a1.22 recommended slurry sampling ETAAS for the determination of cadmium in foodstuff samples. The samples were ground to a 44 pm particle size and 0.04-0.2°/6 m/v slurries were prepared by magnetic stirring. The peak height absorbance was mea- sured and calibration was accomplished by a standard654 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 additions technique. Vegetable plant and botanical samples were successfully analysed for cadmium lead copper and other elements by slurry sampling ETAAS. Stephen et af.,20 employing a thickening agent in order to stabilize the slurry found that the precision of measure- ment was not significantly affected by reducing the bulk of the particle size of the powdered samples from 54 to 2 1 pm.Similarly no matrix interference was associated with the change in particle size distribution for slurries of liver samples. Slurries containing 0.08- 10 g of sample per 100 ml were measured with an acceptable precision i.e. 4-8%. Ebdon and Lechotyckiz1 recommended platform atomiza- tion and the use of an (NH4)H2P04 chemical modifier to eliminate matrix interference in the determination of cadmium. Slurries were prepared from amounts of sample between 50 and 500 mg which had been ground to a particle size of less than 20 pm (90% of all particles) in a volume of 25 ml employing magnetic stirring for stabilization. Excellent agreement was obtained between direct analysis and conventional solution atomization employing inte- grated absorbance measurements and aqueous standards for calibration.In contrast to these workers Carri6n et af.24 used 1 mg ml-l slurries of National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1575 Pine Needles with much coarser particles (less than 160 pm) and found good correlation with wet digestion solution analysis. A slurry sample of NIST SRM 1 575 Pine Needles prepared by ultrasonic homogenization served for calibration. Miller-Ihli25~29 carried out an extensive investi- gation of slurry preparation for direct multi-element analy- sis of a variety of biological and botanical RM samples. Slurries were prepared using 5-10 mg of sample in 5 ml of 5% HNOJ containing 0.04% Triton X-100. Relatively coarse particles lying in a narrow range (250-600 pm) yielded better precision than particles typically found in NIST SRMs ( t 2 5 0 or t 4 2 5 pm) which is in contrast to the observations of previous workers.15*19 Ultrasonic probe and vortex mixing have been critically compared. Although the precision of the measurements was found to be comparable the former technique was pre- ferred particularly for elements and samples in which settling of the heavy particles in the slurry might produce a systematic error in the analytical results. This paper deals with the evaluation of the performance of a laboratory- assembled atomic absorption spectrometer and a graphite cup atomizer designed specifically for direct analysis of solid samples. Special emphasis is given to the importance of the size of the sample particles and the distribution in affecting the accuracy and precision of the determination of lead and cadmium in a variety of NIST SRMs.The direct weighing in technique and a slurry sample introduction technique are critically compared. Experimental Apparatus A laboratory-assembled atomic absorption spectrometer and graphite furnace were employed in this work to achieve so-called stabilized temperature platform furnace condi- tions as suggested by Slavin et uf.26 The spectrometer was constructed from partly-modified commercial units and laboratory-made electronic parts. A schematic diagram of the instrument is presented in Fig. 1. The element hollow cathode lamp (1) and deuterium lamp (1) are positioned at right-angles to each other and the emission of both lamps passes through a semitransparent quartz mirror (2) in the same optical path.The lamps operate successively at a frequency of 1 kHz (see Fig. 2) and each lamp is switched on for 0.33 ms every 1 ms. At time intervals when neither lamp is activated the graphite furnace emission signal was measured and subtracted from the lamp signal. The high 1 2 J 11 12 13 . 18 - 17 16 15 :+ 1 1 %EL Y02H J Fig. 1 Block diagram of the instrument. I Hollow cathode lamps (element and deuterium); 2 semi-transparent quartz mirror; 3 graphite cup atomizer; 4 monochromator SPM-2 (Carl Zeiss); 5 photomultiplier tube R- 106 (Hamamatsu); 6 photomultiplier tube power supply; 7 modified power unit CRA-90 (Varian); 8 power supply feed-back system; 9 temperature measurement sensor; 10 hollow cathode lamp power supply; 1 1 frequency standard (clock); 12 chopper; 13 furnace emission eliminator; 14 electrical filters; 15 zero adjustment unit; 16 logarithmic amplifer; 17 system for signal subtraction; 18 integrator; 19 dual-channel storage oscillo- scope 564 B (Tektronix); and 20 recorder.EL ZH 0.33 ms 0.33 ms m w w H 1 ms ~ I I J I ' Time - Fig. 2 Lamps operation characteristics. EL element hollow cathode lamp and 2H deuterium lamp frequency signal obtained is split by means of precise timers into two separate signals. The first represents the element hollow cathode signal and the second the deuterium lamp signal. By using the two-channel storage oscilloscope ( 19) both signals can be seen at the same time and overlapping of atomic and background signals can easily be observed in transmission mode [see Fig.3(a)]. The transmission signals are converted into absorbance by means of a logarithmic amplifier (16). The deuterium lamp signal is subtracted from the element hollow cathode lamp signal and the difference represents the atomic absorbance signal. A laboratory-made integrator (1 8) is built into the electronic system and enables the absorbance signal to be integrated. Thus absorbance and integrated absorbance signals can be observed at the same time [see Fig. 3(b)]. The time constant of the instrument can be varied in the range 1.1-1 200 ms enabling fast electronic signals to be followed. The photo- multiplier tube ( 5 ) and lamps (1) are powered from locally- made sources (6) and (7) and run under the recommended conditions.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL.6 655 1 1 TvDe B Type c .v- - I 1 I 0 10 20 Fig. 5 Graphite cups in cross-section I I I 0 5 10 Ti me/s Fig. 3 (a) Atomization of the lead in the presence of NaCl matrix. 1 Lead signal; and 2 background signal. (b) Absorbance and integrated absorbance signals. 1 Peak height absorbance; and 2 integrated absorbance 12 Fig. 4 Graphite cup atomizer. 1 Beam collimator; 2 quartz cover; 3 graphite cup; 4 graphite electrode; 5 quartz chimney; 6 watercooled brass block; 7 assembly screw; 8 power supply connection; 9 aluminium base plate; 1 0 poly(tetrafluoroethy1ene) electrical insulator; 1 1 holder and 12 furnace tension adjustment Graphite cup atomizer The graphite atomizer used was laboratory-made and is similar to a mini-Massmann as produced by Varian Techtron.It is shown in schematic form in Fig. 4. The graphite cup (3) is fixed by means of two support electrodes (4) which are clamped between two large water-cooled brass blocks (6) that serve as electrical terminals (8). The distance between the blocks can be varied by means of a screw (12) which enables the graphite cups to be ex- changed. Spring-loaded support electrodes ensure good and constant electrical contact. The chimney ( 5 ) and its cover (2) are made of quartz. Argon flowing through the chimney is used as the protective gas in order to minimize the oxidation of the carbon. The reconstructed Varian CRA-90 unit serves as the power supply to the graphite cup. The atomization cycle was extended up to 20 s the heating rate was made more rapid (maximum heating rate is more than 2000 K s-l depending on the graphite mass) and a constant temperature was achieved for the graphite cup through a built-in temperature feed-back system equipped with fibre-optic cable.Positioned near the graphite cup the fibre-optic cable is fixed in a small stainless-steel tube and connected to a phototransistor. The whole system faces the incandescent graphite cup wall. Another phototransistor fixed just below the stainless-steel tube serves for tempera- ture control of the graphite cup and the signal can be observed by using the storage oscilloscope. The atomization cycle is also partly reconstructed; a cooling step (of 12 s duration) can be applied prior to final atomization.The known temperature overshoot phenomenon normally ob- served when rapid heating rates are used was also largely eliminated. Graphite cups The graphite cups used in this work consisted of two parts an outer cup of 8 mm 0. d. 1 mm wall thickness and 9-12 mm in height; and a small inner cup of 5.5 mm 0. d. 0.5 mm wall thickness and 5-8 mm in height. The latter can either be placed on the bottom of the outer cup (type A Fig. 5 ) or hung on the edge of the hole made through the bottom of the outer cup (type B and type C Fig. 5). The inner cups are only heated by thermal conductivity and radiation from the outer cup and hence act as a platform. Hanging cups are particularly suitable for larger sample amounts and for solid samples that form carbonaceous residues or increase in volume during atomization (blood etc.).All inner cups with the exception of type C are easily inserted and removed from the outer cup by means of specially designed tweezers. Inner cup type C is similar to type B and has a graphite ring placed in the lower part which is fixed by self- friction (a push-fit) hence a much slower heating rate is obtained. The heating rate can also be varied by the height of the inner cup and thus a number of different cups similar to type B can be made for any particular analytical application. Up to 400 measurements can be carried out using a single graphite cup for the determination of cadmium and lead. Graphite cup temperature. The outer cup temperature control was performed by means of a phototransistor connected via an electronic circuit to the storage oscillos- cope or to the plotter.The corresponding voltages obtained were calibrated for temperature (in K) using an optical pyrometer (Pyrowerke). The pyrometer was focused at the inner wall surface of the outer cup. Slight variations of this system caused by unstable electrical contacts between the656 3000 4 I JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1991 VOL. 4 ) Fig. 6 Heating characteristics of the outer cup and different inner cups. 1 Cup type A; 2 cup type B; 3 cup type C; and 4 outer cup 1 I J C J 0 2.5 5 2.5 5 1 . 0 L l Q ;;OECl; n & 0.5 1500 K 0 2 0 0 .o Fj 1500 :Fi 0.4 0.5 0.2 0 0 5 10 0 5 10 Time/s Fig. 7 Absorption profiles of cadmium in different graphite cups (a) and (b) type A; (c) and (d) type B (5 mm); and (e) and 0 type B (8 mm).Variation of absorbance and integrated absorbance with the heating rate. Sample 5 pl 0.025 pg ~ m - ~ . T=temperature profile support electrodes and cup can occur and periodic calibra- tion is recommended. As the rate of entry of the sample into the analytical volume of the cup depends on the heating rate of the inner cup the temperature-time functions for the inner cups presented in Fig. 6 were measured at certain selected outer cup temperatures. It can be seen clearly from this figure that the temperature-time functions of the inner cups investigated are phase delayed which is a character- istic feature of a platform. Moreover? at equilibrium? only the cup type A temperature reaches the outer cup tempera- ture whereas the equilibrium temperatures of cup types B and C are much lower (the temperature difference being approximately 600 K for these two types). Graphite cup characteristics.In order to investigate the performance of the graphite cup atomizer? cadmium and lead solutions were atomized using graphite cups of types A and B ( 5 and 8 mm in height). Aqueous solutions were used instead of solid samples so as to achieve better reproducibil- ity of results. Measurements were performed at a constant outer cup temperature while the heating rate was varied. Results presented in Figs. 7 and 8 quantitatively illustrate the extent of the delay in appearance and broadening of the I I I I I 0 2.5 5 0 2.5 5 s 0 5 10 1900 K e 0.5 0.5 0 % - 0 5 10 j C 1 .o El LoL]- 1900 K * 0.5 0.5 0 0 0 5 10 Time/$ 0 5 10 Fig.8 Absorption profiles of lead in different graphite cups (a) and (b) type A; (c) and (d) type B (5 mm); and (e) and u> type B (8 mm). Variation of absorbance and integrated absorbance with the heating rate. Sample 5 pl 1 pg ~ m - ~ . T=temperature profile 0 10 Time/s 20 Fig. 9 Absorption profiles of cadmium in various graphite cups. Sample NBS SRM 1645 River Sediment; 0.02 mg. 1 Cup type A wall atomization (without inner cup); 2 cup type B; 3 cup type C; and 4 outside cup temperature 2100 K absorbance-time profiles with decreasing heating rates. All of these phenomena were even more pronounced when cup type C was used for atomization. Despite these differences in peak shapes the integrated absorbances were almost constant when the same masses of cadmium or lead were atomized.The absorption peaks always appeared in the interval where the outside cup temperature was constant. Varying the heating rate of the outer cup and/or changing the type of inner cup influenced the rate of entry of cadmium or lead atoms into the analytical volume signifi- cantly? and hence extended the useful range for the determination of these elements over two orders of magni- tude without changing the analytical line. As an example determination of cadmium in National Bureau of Stan- dards (NBS now NIST) SRM 1645 River Sediment containing 10.2 pgg-l of cadmium is shown in Fig. 9. A 0.02 mg portion of the sample was atomized using an inner cup of type C without obtaining an excessively high peak absorbance value.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL.6 657 Procedures Reagents and standard solutions Doubly distilled de-ionized water was used for preparing the slurries and for diluting stock standard solutions ( 1000 mg dm-3). Standard stock solutions were prepared by dissolving the appropriate salts or metals (all of analytical- reagent grade). An aqueous solution of (NH4)2HP04 (25 g dm-3) was used for chemical modification. Solid sample and slurry preparation Several powdered solid samples obtained from NIST and from the International Atomic Energy Agency (IAEA Vienna) were used in this work. Prior to analysis all of the samples were dried for 4 h at 80 "C. A part of each sample was ground in a vibrational mill (Retsch Model MM-2) equipped with either a stainless-steel or zirconia grinding jar and balls.In contrast to previous work,29 no significant lead or cadmium contamination was found after grinding silica. The samples were stored in polyethylene containers. After each grinding operation the device was carefully cleaned in order to prevent contamination of the following sample. Slurries were prepared by diluting 5-200 mg of powdered sample in 1-25 ml of doubly distilled water containing 0.04% Triton X- 100. If necessary a correspond- ing amount of (NHJ2HP04 was added directly to the slurry. Analytical procedure The samples were weighed either directly into the inner graphite cup or into special glass containers using Mettler Model AE- 1 63 and UM-3 microbalances. When weighed directly the inner cups were easily inserted into the outer cup and removed after completion of the analysis.Samples from the containers were introduced into the inner graphite cups using a small glass funnel. By using this sample introduction technique with samples weighing less than 30 pg the results showed that 98-102% of the sample entered the cup. Typically 0.02-5 mg of sample were taken for a single measurement. Slurries were placed directly into the inner graphite cups using a micropipette (1 0-20 pl). Prior to sampling the slurry was homogenized using an ultrasonic device with a 3 mm diameter titanium probe (Cole-Par- mer ultrasonic homogenizer Model 47 10). The instrumen- tal parameters employed in the determination of cadmium and lead are given in Table 1. The analytical signals were observed using the dual-channel storage oscilloscope (Tek- tronix Model 564B) and integrated absorbance was mea- sured.The mean (3 and standard deviation (a) were calculated from at least 11 independent measurements of each sample using either the weighing in technique or the slurry introduction procedure. Results and Discussion Particle Size Distribution Measurements In direct analysis of solids by ETAAS the particle size distribution of the sample has been shown to be an important parameter influencing the quality of the re- sults. 15~16s20f3*27*29 The accuracy and precision of measure- ments can be affected either by inadequate homogeneity or poor sample representation and/or incomplete recovery of the analyte from the matrix during the atomization process.An investigation was therefore conducted to assess the magnitude of these effects in the determination of cadmium and lead in a variety of samples employing both the direct weighing in and the slurry sample introduction techniques. For that purpose original and ground RMs were taken and particle size distribution measurements performed on both Table 1 Instrument settings and graphite cup parameters used in the determination of cadmium and lead in aqueous solutions and different solid materials Parameter Cd Pb Wavelengthhm Spectral band pass/nm Lamp current/mA Graphite cup type Argon flow rate/ml min-' Drying- TemperatureK Time/s Ashing- TernperatureK Time/s Cooling- TemperatureK Time/s TemperatureK Time/s Lamp current/mA Atomization- Hydrogen background corrector 228.8 0.6 5 A,B,C 3000 380-420 10-90 600-900 5-60 295 10 1400-2300 2-10 15-25 * Modifier added 125 pg of (NH4)2HP04.283.3 0.5 6 A,B,C 3000 380-420 10-90 600- 1 OOO* 5-60 295 10 1600-2300 2-12 15-25 sets of samples using a low-power helium-neon laser diffractional type granulometer Model HR 850 Cilas Alcatel. Some characteristic values such as the median (&) 90% diameter (Ow) and 10% diameter (Dlo) were taken from the measured distribution graphs and are summarized in Table 2. It is evident from these results that some of the original RMs in particular those of plant tissue were relatively coarse having broad particle size distribution. It should however be borne in mind that these materials were made for conventional analytical procedures where a minimum of 100-250 mg of sample are required for the analysis.In general additional grinding substantially reduced the mean particle size of all of these materials with the exception of the IAEA Soil 7. (This soil already existed in fine powder with a median diameter of 9 pm.) In addition to this the distribution of particle sizes obtained after the grinding stage was much narrower. Analysis of Solids by Direct Weighing In Sample Introduc- tion Lead and cadmium were determined by the introduction of weighed amounts of solid samples directly into the graphite Table 2 Particle size distribution data of original and ground reference material samples Sample Diameterlpm NBS 1633a IAEA Soil 7 NBS 1645 NBS 1571 NIST 1572 NBS 1573 Coal Fly Ash River Sediment Orchard Leaves Citrus Leaves Tomato Leaves Original sample Finely ground sample 14 46 2 9 31 2 13 55 1 38 189 6 78 265 7 88 350 6 2 8 0.6 7 29 0.7 4 24 1 13 38 3 13 37 3 11 40 3658 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL.6 Table 3 Determination of cadmium in reference materials by ETMS using the direct weighing in method of solid sample introduction Result found (Xk o)/pg g-l Sample NBS 1633a IAEA Soil 7 NBS 1645 NBS 1571 NIST 1572 NBS 1573 Coal Fly Ash River Sediment Orchard Leaves Citrus Leaves Tomato Leaves * Value not certified. Certified value Amount of (Xf aypg 8-1 sample/mg 1 .Of 0. 15 0.6-1.3 - 0.3-1.5 10.2k 1.5 0.05-0.15 0.1 1 kO.01 1-3 0.03 f 0.0 1 2-5 3* 0.8- 1.2 Original sample Finely ground sample 0.82 k 0.2 1 1 .oo k0. 12 5.9k 1.5 6.70k0.87 0.13 k 0.04 0.73 k 0.07 0.97 k 0.12 0.09 k 0.03 0.02 1 k 0.004 0.033 & 0.007 2.0 k 0.2 2.3 k O .l Table 4 Determination of lead in reference materials by ETAAS using direct weighing in method of solid sample introduction Result found Q k a)/pg g-I Sample NBS 1633a IAEA Soil 7 NBS 1645 NBS 1571 NIST 1572 NBS 1573 Coal Fly Ash River Sediment Orchard Leaves Citrus Leaves Tomato Leaves Certified value (Xf a)/= g-1 72.4 f 0.4 55-7 1 714k28 4 5 f 3 13.3 f 2.4 6.3 f 0.3 Amount of sampleimg 0.1-0.3 0.1-0.5 0.02-0.03 0.25-0.5 0.5-1.2 1-1.5 Original sample Finely ground sample 56.2 f 2.2 69.4 f 8.3 67.3 k 4.5 63.9 k 7.0 800 f 280 806 k 201 32k5 36k2 10.3 f 2.2 12.1 k0.8 4.2 k 0.8 6.5k 1 cups. Different inner cups (see Fig. 5 ) were employed for particular samples depending on the type of sample and concentration of the analyte present.For example the analysis of NBS SRM 1645 River Sediment of relatively high lead and cadmium contents required the use of inner cup type C facilitating a slow release of analyte elements during the atomization process. Although the amount of sample used for a single measurement was in the range 20-150 pg. Analysis of biological and plant materials produced a high background coinciding with the atomic absorption peak. Overlapping of these signals was largely reduced by the use of the (NH4)2HP04 chemical modser which is illustrated in Fig. 10. Therefore the use of a chemical modifier was obligatory for the determination of cadmium and lead in these samples if accurate results are to be obtained employing aqueous standards for calibration.Foaming of some of the biological and plant samples was observed during the atomization process which might result in sample loss and poor precision of measurement. This problem was solved by directing a small stream of argon in a direction perpendicular to the bottom of the inner cup during the drying and ashing step. After each measurement the residue if any was also blown-out of the cup by means of the stream of argon. The results of the determination of cadmium and lead for some of the selected RMs by direct introduction of weighed amounts of sample into the graphite atomizer are summarized in Tables 3 and 4. Sample aliquots used for the determination of these elements were in the milligram range with the exception of NBS SRM 1645 River Sediment IAEA Soil 7 and NBS SRM 1633a Coal Fly Ash in the determination of lead where sub-milligram amounts were analysed.The original materials produced results that were pre- dominantly lower than the certified values. Additional grinding of these samples improved the accuracy of results with the exception of the soil sample where particle size distribution was not substantially changed. Plant material samples ground to a particle size of t 4 4 pm should yield acceptable results provided sample aliquots of 0.5-2 mg are atomized. The same would probably apply for soil and geochemical samples ground to particle sizes of less than 10 pm in diameter.16 Sample homogeneity can definitely be a dominant factor in the production of erroneous results when very small sample aliquots (20-500 pg) are atomized.As can be deduced from the data for NBS SRM 1633a Coal Fly Ash (see Table 2) although reducing the particle size (Ow) from 46 to 8 pm did not totally eliminate error in the result for lead. It would therefore appear that sub-micro- metre particle size is required for geological and related samples if accurate results are to be obtained by direct weighing in sample introduction. Analysis of Solids by Slurry Sample Introduction The convenience of sampling slurries in the direct analysis of powdered samples was first pointed out by Fuller et uLZ3 Sample amounts in the range 5-10 000 mg were typically ~ s e d ~ ~ ~ ~ ~ to prepare slurries which largely eliminated the problem of how well the specimen represents the sample.The homogeneity of the slurry during pipetting of aliquots of 10-50 pl for atomization was maintained by the addition of a thickening agent,*’ magnetic stirring16 or ultrasonic homogenizati~n.~~*~~-~~ The last was found to be superior particularly for slurries containing large heavy particles.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 659 Table 5 Determination of cadmium in reference materials by ETAAS using slurry sample introduction Result found (Xf a)lpg g-I Sample (Xf a)/* g-1 mg ml-l sample Slurry Certified value concentration/ Original sample Finely ground NBS 1633a Coal Fly Ash IAEA Soil 7 NBS 1645 River Sediment NBS 1571 NIST 1572 NBS 1573 Orchard Leaves Citrus Leaves Tomato Leaves *Value not certified.1 . O f 0.15 2013 10013 - 2012 10013 10.2f 1.5 1015 100125 0.1 1 fO.01 20012 0.03 f 0.0 1 9512 3* 5/ 1 2515 10015 1.1 f O . 1 1.0 f 0.1 1.3 f 0.1 1.2 f 0.1 7.9 f 0.5 8 . 0 ~ 1.0 0.08 f 0.06 - 2.7 f 0.6 3.2 f 1.5 2.5 kO.5 1 .Of 0.1 1 .Of 0.1 1.2 f 0.1 1.1 20.1 8.9 -t 0.6 9.6 f 0.8 0.12 -t 0.01 0.029 f 0.003 3.0f 0.3 2.7 f 0.4 2.6 -t 0.2 Table 6 Determination of lead in reference materials by ETAAS using slurry sample introduction Result found (X-t a)lpg g-' Sample ( 2 a)/* g-' mg ml-l sample Slurry Certified value concentration/ Original sample Finely ground NBS 1633a Coal Fly Ash IAEA Soil 7 NBS 1645 River Sediment NBS 1571 Orchard Leaves NIST 1572 Citrus Leaves NBS 1573 Tomato Leaves 72.4 f 0.4 20/2 55-7 1 2015 200/15 100110 714f28 512 80120 4 5 f 3 1012 100110 13.3 k 2.4 2012 200/2 6.3 f 0.3 2012 10015 69.0 f 7.6 7 1.8 f 6.4 59.8 f 7.2 52.6 f 5.8 704f 134 7 4 t h 112 54.1 f 11.0 47.7 f 7.6 15.9 f 4.5 6.2 k 1.3 6.3 f 2.5 14.0f 1.3 67.4 f 6.0 71.3k4.3 57.6 f 5.2 51.8f 1.0 774-t 108 748 -t 60 52.7 k 5.3 48.2 f 1.9 13.6-tO.5 13.6 f 0.5 6.5 f 1.1 6.5 5 0.4 The effect of particle size on the quality of the analytical results has been discussed by several workers but opinions were fairly different.Fuller e? ~ 1 1 . ~ ~ analysing ore and silicate rock samples set a limit of 25 pm above which sampling was the major problem. Hinds e? a1.,16 on the other hand reported poor recovery of cadmium and lead from larger particles in the atomization of soil surries. In order to achieve good results these samples should be ground until 90% of particles are less than 1 1 pm in diameter.Ebdon and LechotyckP accurately determined cadmium in biological and botanical reference materials which had been ground until 90% of the particles were less than 20 pm. Stephen e? all9 reported that stable slumes of vegetable and protein samples can be obtained if the majority of the particles are less than 50 pm in size. Contrary to these workers Miller- Ihliz5 found better precision for coarse (250-600 pm) but more uniform NBS SRMs of mixed diet when measuring eight different elements. In the present study a variety of geological and plant material samples were slurried for the determination of cadmium and lead. The mass of the sample and the particle size distribution were the variable parameters. Nine consecutive measurements of 10 pl aliquots were made. The results are summarized in Tables 5 and 6.In comparison with the direct weighing in sample introduction 1-3 orders of magnitude greater sample amounts were used in the preparation of slurries which should almost eliminate the problem of sample representa- tion. Indeed all of the values but one obtained for the six different RMs were closely matched with the certified values. The variation of sample mass in the range 10-200 mg did not influence the accuracy of the results even when plant material powders of fairly non-uniform particle size distribution (see Table 2) were atomized. Fine grinding of these samples which significantly reduces both the mean particle size and distribution did improve the precision of measurement.The same effect was observed for the determination of lead in the soil sediment and coal fly ash samples. However the particle size in these samples with the exception of the coal fly ash was only slightly affected by additional grinding. It could therefore be anticipated that particle size distribution would have little effect on the accuracy and precision of results when slurries of biological botanical or food samples were atomized. Although a slight improvement in either accuracy or precision for slurries of geological soil or sediment samples was obtained grinding to a particle size less than 30 pm15 is recommended for these660 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 100 50 - E D C .- 0 2.5 Time/s 5.0 Fig. 10 Effect of chemical modifier on cadmium absorption profile.Graphite cup type B (8 mm). Sample IAEA A2 Dried Animal Blood (2.16 mg). (a) Without chemical modifier and (b) chemical modifier added. 1 Cd absorbance; and 2 background absorbance materials in order to achieve a more uniform evaporation of elements from the matrix. Conclusions An atomic absorpition spectrometer capable of measuring transient signals was assembled and a graphite cup atom- izer suitable for direct atomization of solid samples was designed. A variety of inner cups acting as a platform could be incorporated in the graphite atomizer thus enabling the rate of entry of the analyte into the analytical volume to be varied substantially. This facilitates the extension of the useful analytical range over two orders of magnitude without changing the analytical line.No matrix condensa- tion problems often experienced with tube atomizers were observed. Different geochemical (soil and sediment) and biological (plant) RMs were analysed and direct weighing in and slurry sample introduction techniques compared. Aqueous standards were used for calibration and the addition of an (NHJ2HP04 chemical modifier was found to be necessary in the analysis of botanical samples. Slurry sample introduction was found to be superior to the direct weighing in technique for the following reasons (i) better accuracy and precision were obtained enabling easier and more rapid sample handling and offering the possibility of dilution at high analyte concentration; and (ii) the particle size and the distribution of the sample was found to be extremely important in the atomization of small sample aliquots (20-500 pg) when using direct weighing in sample introduction.A sub-micrometre particle size would prob- ably be required for geological samples in order to achieve acceptable accuracy and precision. In contrast the results obtained by slurry sample introduction using sample amounts that were 1-3 orders of magnitude larger were less susceptible to particle size effects but the effects of particle size should not be totally ignored particularly when geochemical samples are analysed. Professor S. A. Katz of the Rutgers University New Jersey USA and Dr. W. Frech of the Department of Analytical Chemistry University of Umei Sweden are thanked for the fruitful discussion and valuable suggestions in preparing this manuscript.The supply of graphite material by B. Hutsch of Ringsdorff-Werke is also gratefully acknow- ledged. References 1 Langmyhr F. J. Talanta 1977 24 277. 2 Slavin W. and Manning D. C. Prog. Anal. At. Spectrosc. 1982 5 243. 3 Langmyhr F. J. and Wibetoe G. Prog. Anal. At. Sprectrosc. 1983 8 193. 4 Kurfurst U. Fresenius Z. Anal. Chem. 1987 328 3 16. 5 Ajlec R. and Stupar J. Spectrochim. Acta Part B 1983,38,7. 6 Stupar J. and Ajlec R. paper presented at The Third International Colloquium on Solid Sampling with Spectros- copic Methods Wetzlar October 10th- 12th 1988. 7 Chakrabarti C. L. Wan C. C. and Li W. C. Spectrochim. Acta Part B 1980 35 93. 8 Rygh G. and Jackson K. W. J. Anal. At. Spectrom.1987,2 397. 9 Atsuya I. Itoh K. and Akatsuka K. Fresenius Z. Anal. Chem. 1987,328 338. 10 Fleckenstein J. Fresenius Z. Anal. Chem. 1987 328 396. 11 Vollkopf U. Grobenski U. Tamm R. and Welz B. Analyst 1985 110 573. 12 Strubel G. Rzepka-Glinder V. Grobecker K. H. Fresenius 2. Anal. Chem. 1987 328 382. 13 Lucker E. Rosopulo A. and Kreuzer W. Fresenius Z. Anal. Chem. 1987,328 370. 14 Homer E. and Kurfurst U. Fresenius 2. Anal. Chem. 1987 328 386. 15 Jackson K. W. and Newman A. P. Analyst 1983 108 261. 16 Hinds M. W. Jackson K. W. and Newman A. P. Analyst 1985 110 947. 17 Karwowska R. and Jackson K. W. J. Anal. At. Spectrom. 1987 2 125. 18 Hinds M. W. and Jackson K. W. J. Anal. At. Spectrom. 1987 2 441. 19 Stephen S. C. Littlejohn D. and Ottaway J. M. Analyst 1985,110 1147. 20 Stephen S. C. Ottaway J. M. and Littlejohn D. Fresenius Z. Anal. Chem. 1987 328 346. 21 Ebdon L. and Lechotycki A. Microchem. J. 1987 36 207. 22 Olayinka K. H. Haswell S. J. and Grzeskowiak R. J. Anal. At. Spectrom. 1986 1 297. 23 Fuller C. W. Hutton R. C. and Preston B. Analyst 1981 106 913. 24 Carridn N. de Benzo Z. A. Moreno B. Fernhndez A. Eljuri E. J. and Flores D. J. Anal. At. Spectrom. 1988 3 479. 25 Miller-Ihli N. J. J. Anal. At. Spectrom. 1988 3 73. 26 Slavin W. Manning D. C. and Carnrick G. R. At. Spectrosc. 1981 2(5) 137. 27 Majidi V. and Holcombe J. A. Spectrochim. Acta Part By 1990 45 753. 28 Holcombe J. A. and Majidi V. J. Anal. At. Spectrom. 1989 4 423. 29 Miller-Ihli N. J. Fresenius J. Anal. Chem. 1990 337 271. 30 Epstein M. S. Carnrick G. W. Slavin W. and Miller-Ihli N. J. Anal. Chem. 1989 61 1414. 31 Miller-Ihli N. J. J. Anal. At. Spectrorn. 1989 4 295. 32 Hoenig M. and Van Hoeyweghen P. Anal. Chem. 1986,58 26 14. 33 Baxter D. C. and Frech W. Fresenius J. Anal. Chem. 1990 337 253. Paper I /0025 7K Received January 18th 1991 Accepted July Ist 1991