Anal. Chem. 1981, 53, 1035-1038
terference effects, other than those previously discussed, for the levels encountered in unprocessed foods. The results in Tables VI and VI1 illustrate the performance of the TNS over a range of trace concentrations and in a variety of crop matrices. These preliminary results for unprocessed foods and the data presented in Tables I11 and IV justify further study of the tandem nebulization system as a means for providing improved simultaneous multielement analysis.
LITERATURE CITED (1) Wolnik, K. A.; Kuennen, R. W.; Fricke, F. L. In “Developments In Atomic Plasma Spectrochemical Analyses”; Barnes, R., Ed.; Heyden and Son Inc.: Philadelphia, PA, in press. (2) Munter, R. C. ICf Inf. News/. 1979, 5 , 368.
1035
(3) Mertz, W. I n “Ultratrace Metal Analysis in Biological Sciences and (4) (5) (8) (7) (8)
(9)
(10)
Environment”; Risby, T. H., Ed.; American Chemical Society: Washington, DC, 1979; Chapter 1. Robbins, W. 6.; Caruso, J. A. Anal. Chem. 1979, 51, 889A. Thompson, M.; Pahiavanpour, B.; Walton, S. J.; Kirkbright, G. F. Analyst (London) 1978, 103, 568. Nishibe, J., presented at Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlantic City, NJ, 1980, Abstract No. 428. Fed. Regist. 1979, 44 (148), 44940. Thompson, M.; Pahlavanpour, B. Anal. Chim. Acta 1979, 709, 251. Hahn, M. H.; Kuennen, R. W.; Caruso, J. A.; Fricke, F. L. J. Agrlc. Food Chem., in press. Thompson, M.; Pahlavanpour. 6.; Watson, S. J.; Kirkbright, G. F. Ana/ysf (London) 1978, 103, 705.
RECEIVED for review December 12, 1980.
Accepted March 20,
1981.
Determination of Lead in Air by Electrothermal Atomic Spectrometry with Electrostatic Accumulation Furnace Glancarlo Torsi, * Elio Desimoni, Francesco Palmlsano, and Luigia Sabbatlni Istltuto di Chimica Analitica, Universiti, Via Amendola 173, 70126 Sari, Italy
An apparatus is descrlbed by which the elemental analysis of atmospheric particulate matter can be performed by combining “in situ” electrostatic accumulation with electrothermal atomic absorption spectrometry. The calibration of such an apparatus can be accomplished simply by uslng aqueous standard solutions of the analyzed metal. Sampling volumes are 100-300 cm3 with sampling times of 30-90 s. Detection limits are on the order of g.
In a previous paper (1) Torsi and Desimoni described the basic principle of a new technique named “electrostatic accumulation furnace for electrothermal atomic spectrometry” (EAFEAS) by which the elemental analysis of atmospheric particulate matter could be easily performed by combining the electrostatic precipitator technique with electrothermal atomic absorption spectrometry (EAAS). The novelty of the technique stands in using the electrothermalfurnace (a simple graphite cylinder) both as particulate matter precipitator and as atomization device. Such a combination gives a technique with high sensitivity, great selectivity, rapidity, and absence of contamination. EAAS is characterized by detection limits in the order of g and the average urban levels of a metal in air particulate matter range between 0.1 and 0.01 pg/m3 (2). This suggests that an EAFEAS analysis can be performed by sampling just 10-100 cm3of air which can mean a total time of analysis of only a few minutes. Such volumes are small compared with the volumes collected with usual techniques (3-5)and consequently the results obtained have different meaning since the value obtained by conventional methods represents an integral value over a large period of time while the EAFEAS results may be considered a quite instantaneous value, the sampling time being only a few minutes. This can be of great importance when prompt action must be taken or large samples are not available.
Analysis times in conventional techniques can be reduced by analyzing directly (3) portions of the filtering membrane on which the sample has been collected, but in this case, tedious handling of the filter and problems associated with high blank signals cannot be avoided. However, in the course of EAFEAS measurements, some difficulties were encountered concerning the standardization of the technique. Since gaseous standards with known constant load of particulate matter and particle size distribution are not available, the following calibration procedure was followed (a) The air was filtered through a 0.8-pm millipore membrane and lead in the residue determined by anodic stripping voltammetry (ASV). Since this procedure did not correspond to a 100% recovery (1) subsequent steps were as follows. (b) Lead was determined by EAFEAS in an air sample not subjected to membrane filtration. (c) Lead was determined by EAFEAS in the same air sample after membrane filtration. (d) Analyses of samples from (b) and (c) were carried out in triplicate and averaged. (e) The difference between the two averages was assumed to correspond to voltammetric estimate. Such a procedure is, however, too tedious for routine use of the proposed method. Therefore a new apparatus has been devised in order to check the possibility of capturing all the particles present in the sampled gas and of calibrating such an apparatus by the standard technique of using aqueous solutions of the investigated metal. The present paper describes the experimental apparatus, the influence on measurements of different parameters (such as potential, flow rate, etc.), and the calibration technique. An example of the performance of the new apparatus is also reported, by monitoring the variations in lead content in air particulate matter over a certain lapse of time.
EXPERIMENTAL SECTION Apparatus. All the absorption measurements were performed with a Perkin-Elmer 460 spectrometer with a 0.2-nm slit width. The output was stored on a Gould Advance 4000 storage oscil-
0003-2700/81/0353-1035$01.25/00 1981 American Chemical Society
1036
ANALYTICAL CHEMISTRY, VOL. 53, NO. 7, JUNE 1981
(4 a i r , bag
ET
to bubbler
X k 2
Flgure 2. Schematic flow of air or N, in the cell during (a) deaeration, (b) air sampling, and (c) air sampling from plastic bag container. b
2
Flgure 1. Schematic view of the cell (for a complete description see text).
loscope with option 4001 from which a hard copy of the transienb could be obtained, when necessary, with the aid of a strip-chart recorder. The lamp used was a Perkin-Elmer Intensitron operating at 283.2 nm. Although not strictly necessary, D, background correction was used. The atomizer power supply was a 230/20 V transformer connected to a “Variac” tension variator and was manually switched on and off. No accurate temperature control was used. Since the peak height and the peak area methods gave similar results in the calibration with 1or 2 pL of Pb2+solutionsand since (see below) the shape and position of the curve were the same starting from either solution or particulate matter, only peak heights were measured. Electrostatic Accumulation Furnace Device. The apparatus is schematically shown in Figure 1. On a Plexiglass rectangular base, a, which can be fastened to the spectrometer flame mount, are connected two water-cooled brass slabs, b, via two screws, c1 and c2 The screws serve also as electrical connection, d. The graphite tube, e, is held in place by two hollow graphite screws, f, one for each slab. The left slab is also connected to a metallic cylinder, g, and the end sealed by a quartz window, h. A connection for external gas introduction tubes, i, is also provided. On the right side of the base there is a slide,j, which can be moved on the guide-rail, k, by a knob, 1. A scale (not shown in the figure) measures displacements. Through a special mount on the slide, a tungsten wire, ml (M3N8grade, Pierce Inorganics), or a capiUary can be fastened so that they can be inserted into the graphite tube. A spring-loadedmechanism rotates the mount out of the optical path when the mount is located on the extreme right position. A BNC connector,m2, is also fixed at the base for the high-voltage lead to the wire. A Plexiglass cover, n, permits controlled atmosphere operations and can be raised, when necessary. It has a quartz window, 0,and two connections for external tubes, p1 and p,. Chemicals. The Pb2+standard solutions used for calibration procedure were prepared just before analysis from a 0.1 N HN03 stock solution containing 10 ppm of Pb(N03), (CarloErba reagent grade). A set of cleaned Pyrex glassware (one for each concentration) was used for preparing standard solutions. Care was taken to avoid interchanging of containers in order to eliminate memory effects. Nitrogen UPP grade, prepurified over a BASF 0, absorber was used as purging gas. Calibration. The calibration of the apparatus was obtained by depositing 1pL of Pb2+standard solution inside the graphite furnace. This was accomplished with a 1-pLcapillary bent at one end and inserted in a holder fastened to the mount of the slide in place of the tungsten wire (see the descriptionof the apparatus). The fiiing of the capillary,whose bent end was treated with liquid wax to facilitate the detachment of the solution droplets (6),was accomplished by using a “Drummond cup” like device. The volume of the drop was calculated by several weighings. It was filled with the desired solution by suction from a Teflon cup brought manually to the tip of the capillary. Then the capillary
was moved inside the graphite tube at the desired point and, after the solution was expelled, pulled out of the furnace and the optical path. The usual steps in EAAS were then followed (drying 90 “C, ashing = 400 “C, and atomizing 2000 “C). The calibration curve obtained with this procedure was linear up to 3 X g. A detection limit (S/N = 3) of 8 X g was found. EAFEAS Measurements Procedure. The deposition of Pb by electrostatic precipitation was accomplished by inserting the tungsten wire in the desired position inside the furnace. While a constant flow rate of air was maintained through the furnace, high voltage was applied between the tungsten wire (negative pole) and the graphite rod. When large volumes were required, the air was drawn through the furnace by suction with a pump and the sampled volume calculated by use of flow meter readings and collection times. More often, since the sampled volumes were very small, volumes were determined more precisely by connecting the EAFEAS device to a closed, calibrated vessel from which water was drained at a controlled rate. When the sampling was over, the high-voltage source was disconnected and the wire was pulled out of the furnace and of the light beam. A gas-purge step to remove residual air in the EAFEAS device was done by flowing oxygen-free nitrogen for 3-5 min. Diagrams of the flow system during the sampling and purge step are shown in Figure 2. The usual procedure for EAAS (see “calibration procedure” paragraph) was then followed omitting the drying step.
RESULTS AND DISCUSSION Electrostatic Precipitation. An electrostatic precipitator is a device which can capture small solid or liquid particles by charging them and making them impinge, by a suitable electric field, against a conducting surface to which they transfer their charge remaining at the same time captured. The charge acquired by a particle is a function of many parameters including field strength, residence time, dimension, and surface chemical composition. In a cylindrical configuration, like the one used here, the charged particle is subjected to two forces. The first one, due to the gas flow in which the particle is entrained, produces an axial velocity V,; the second force, due to the electrical field acting on the charged particle, produces a radial velocity V,. While it is easy to calculate an average V , from the section of the cylinder and from the flow rate, it is rather difficult to calculate V, since many of the parameters cited above are difficult to evaluate with any degree of accuracy. A survey of the difficulties involved in this kind of calculation is offered, for instance, in ref 6 and 7. In our case, the situation is still more complicated since the precipitator has dimensions generally not encountered in normal precipitators designed for large volumes and heavy loads. While our low sample load satisfies the condition for using ideal limiting equations, actual “ideal behavior” is probably not realized since (a) the central wire is very big compared to the external cylinder whose dimensions must be necessarily small to permit its use as a furnace in EAAS measurements and (b) the “corona discharge” is confined to the tapered tip of the tungsten wire which results in a poorly
ANALYTICAL CHEMISTRY, VOL. 53, NO. 7, JUNE 1981
1037
X
I 15
20
O ’ l
-5
0
5
10
V (kvolts
Influence of potential on peak height at two dlfferent average air velocities: (0)140 cm s-‘;( X ) 12.5 cm s-’.
d (mn] (reference d i s t a m l
Flgure 3.
i J 05
L
I
100
$0
FLOW
RATE
I 1BO
(.my’s)
Flgure 4. Influence of velocity on peak height: (0)normally tapered wire; ( 0 )blunt wire.
defined charging time for the particles. For all the reasons mentioned above a pragmatic approach has been followed trying to verify experimentally the characteristics of our device. Influence of Potential and Flow Rate. The efficiency of an electrostatic precipitator may be increased, in theory, by increasing the potential or decreasing the flow rate. Both actions produce a relative increase in VI and the number of elementary charges on the particles. The influences of these parameters on the EAAS signal strength are shown in Figures 3 and 4. To properly evaluate the effects of potential and flow rate variations, it is necessary to operate with a gas of known constant load of particles. Since, as mentioned above, such a gas is not available, it was decided to use air stored in a plastic collapsable bag connected to the air inlet of our EAFEAS device (see Figure 2c). As already found (6))the air in the bag showed a decay in the particles load which can be followed, rather easily, by monitoring the peak height for lead a t regular intervals of time. In this way the relative particulate content could be calculated at any time by interpolation and a measure in different experimentalconditions can be normalized to it as done in Figures 3-5. From Figures 3 and 4 (upper curves) it can be seen that a large region is reached, in which VI and V , variations have no influence on the signal of lead. These results can be easily explained if we assume that all the particles containing lead are captured in a small section of the graphite tube. However the lower curve of Figure 3, which is obtained in the same condition of the upper curve but at a lower flow rate, indicates that at low axial velocity there is a lowering of the precipitator efficiency as the potential increases. This anomalous result,
Peak height as function of the position of the drop (X) or of the wire tip at two different velocities: (A) 25 cm s-l; (0)125 cm s-‘. The air flow is from right to left. Distances are measured along the longitudinalaxis of the carbon tube with respect to its middle point. Positive values are on the right. which should be present at high values of V,, is difficult to explain at the present time. Our tentative explanation is that reentrainment (i.e., the bouncing back of a particle in the gas stream after discharge) due to a very high relative VI occurs when low flow rates and high potentials are used. Both curves in Figure 3 also show that potential values higher than 1.3 kV are necessary to obtain significant particle deposition. The finding may be accounted for simply on considering that the “corona discharge” can be produced only above a well-defined electric field (6). Plots similar to that reported in Figure 3 are obtained if the current flowing between the wire and the cylinder is reported on the abscissa instead of applied potential. Of course current increases indefinitely up to a potential when sparks begin to appear. The potential at which appreciable conduction occurs (current values in the order of microamps) is, of course, a function mainly of the cylinder and wire dimensions and to a lesser extent of the form of the wire tip. The upper curve of Figure 4, obtained with a “normally tapered wire”, shows that the lead signal is practically constant for a large range of values of the axial velocity V, so that even a qualitative estimation of the influence of this parameter on the precipitator efficiency is not possible under these experimental conditions. For these reasons measurements have been made with a wire to which the tip has been cut in order to reduce its efficiency in ionizing the particles (see lower curve in Figure 4). As expected the lead signal increases with decreasing V, even if a “saturation level” is never reached in the actual experimental conditions. Comparison of Data Obtained with Pb Solutions and Pb from Air Particles. In order to strengthen our assumption that all the particles were captured in a small section of the graphite tube and to demonstrate the possibility of a calibration with aqueous Pb2+solution, it was decided (a) to examine the response when 1 or 2 p L of P b solution was deposited in different points along the graphite tube axis, (b) to examine the response when lead from particulate matter was deposited in different points along the graphite tube axis, (c) to compare the results in (a) and (b), and (d) to compare the shape of the signal obtained for lead deposited from aqueous solutions and particulate matter. It is well-known that in EAAS the height of the peak is rather sensitive to the point where the solution has been deposited. The apparatus used in this work is somewhat different from a commercial furnace since (i) there is no central hole for sample injection, (ii) there is no symmetry in the heating rate since part of the furnace is partially shielded from air circulation, and (iii) the expansion of the gas during the Flgure 5.
1038
ANALYTICAL CHEMISTRY, VOL. 53, NO. 7, JUNE 1981
i
out 01 range
0 . 2 i ~w v~e a1 0.1
-71
A u lwrve b ’
w
0
z a m
1 sec
E
0
8
12
16
2,o
v)
m
a
T I M E I hours)
Figure 7. Two series of measurements for monitoring lead content in the air particulate matter in the authors’ laboratory: (- - -) Jan 24, 1980; (-)Jan 25, 1980.
- 1
I-
T I ME Flgure 6. Typlcal example of atomic absorption signals of lead obtained with particulate matter and aqueous solution: (curve a) 1 pL of Pb standard solution; (curve b) atmospheric particulate collected by electrostaticprecipitation (sampling time 2 min; sampled volume 250 cm3; air stream velocity 40 cm s-‘;applied potential 1.8 kV).
atomization step creates a drift of the gas from left to right due to different paths to the bubbler (see Figure 2). The effects of such a complex pattern of heating rates and residence time of vaporized atoms in the optical path are reflected by the results in Figure 5 where the peak heights for lead signal are reported as a function of the distance from the center of the graphite tube at which the Pb solution has been deposited. For a comparison with the results from the collected particle measurements, the peak heights are normalized to the highest value. The deposition of the gas-suspended particles at different points along the graphite tube has been obtained by varying the position of the wire tip. Since the gas flow is from left to right it is obvious that most of the particles could be captured only on the right side on the wire tip. The experimental verification that the “capture section” of the particles is small and located precisely near the wire tip is important since it is well-known that the shape and position of the absorption curve is greatly dependent on the point where the sample is deposited. It must be recalled that VI is not the same for all particles (6) which implies that, even in the best conditions, the particles should be deposited at different points along the graphite tube. Consequently the higher the VI/ V , ratio the smaller should be the “capture section” regardless of the value of VI. From Figure 5 it can be seen that the experimental points obtained at low flow rate (i.e., at high VI/ V , values) follow quite closely those obtained with Pb solution. Both the shift to the right and the slower increase with the distance observed for points obtained at high flow rates (see Figure 5) are in agreement with the reasoning presented above. A rough calculation from the shift of these curves at 0.5 relative peak height, assuming linear drift of the particles and constant electric field, gives a VI value around 100 cm/s. Calculated and experimentallyverified values of VI range between 16 and 170 cm/s (6). The high VI found in our experimental conditions could justify the reentrainment process invoked to explain the results of the lower curve of Figure 3. Moreover the closeness on the X axis of the points obtained with both
Pb solutions or electrostatic precipitation reinforced the idea that (at least at high relative values of VI)the distribution of the electrostatically precipitated particles along the graphite tube is not much broader than the spread of Pb solution. This is further strengthened by the finding that the shape of the atomic absorption signals of Pb, obtained with 1 or 2 p L of solutions and those obtained in the reference experimental conditions from electrostatic precipitation of air-suspended particles, are practically the same (see Figure 6). The experimental findings are not unexpected since in our furnace there is no central hole for sample injection and the ratio between the furnace length and its inside diameter is rather high. The potentialities of the proposed technique can be inferred from Figure 7 which shows the fluctuations of the Pb content in the air-suspended particulate matter of the authors’ lab as monitored by the EAFEAS method. The preliminary studies presented suggest the feasibility of using aqueous Pb standards to generate a calibration curve for the EAFEAS apparatus. Unfortunately, results obtained in our lab indicate that, a t least for Pb, a conventional technique like filter collection followed by ASV does not have sufficient precision or accuracy to unequivocally confirm the quantitative utility of EAFEAS. Specific techniques for counting and locating particles inside the furnace are needed before definite conclusions can be drawn. Work is in progress in this direction.
ACKNOWLEDGMENT V. Sacchetti and G. Cosmai of the mechanical shop are dutifully acknowledged.
LITERATURE CITED (1) Torsi, 0.; Desimoni, E. Anal. Len. 1979, 12 (A 13), 1361-1366. (2) Cooper, I. A. Repoft-BNWL-SA, Battelle Northwest Lab.: Richland, WA 1973, 4690. (3) Pachuta, D. G.; Love, L. J. C. Anal. Cbem. 1960, 52, 444-448. (4) Gileadi, P.; Adams. F. Anal. Chim. Acta 1976, 96, 229-241. (5) Walsh, P. R.; Fasching, S. L.; Duce, R. A. And. Chem. 1978, 48, 10 12-1 0 14. (6) Rose, H. E.; Wood, A. J. “An Introduction to Electrostatic Precipitation in Theory and Practice”; Constable and Co. Ltd.: London, 1956; Chapters 3-5,; (7) Nasser, E. Fundamentals of Gaseous Ionization and Plasma Electronics”; Wley-Interscience: New York, 1971; Chapter 11.
RECEIVED for review September 25,1980. Accepted February 2,1981. Work carried out with the financial assistance of the Italian National Research Council (C.N.R.,Rome) and of the Nuovo Pignone (Bari).