Determination of Lead in Atmospheric Particulates by Furnace Atomic Absorption Jerome F. Lech,*>’ Duane Siemer, and Ray Woodriff Department of Chemistry, Montana State University, Bozeman, Mont. 5971 5
w By use of flameless atomic absorption techniques, it is possible to determine traces of lead in atmospheric particulates in very small samples. The implementation of flameless techniques is greatly simplified by using porous graphite as a filter medium. Using the Woodriff furnace as the atomization device and cups made of porous graphite, one can take an air sample and insert the cup into the furnace for determination with no other chemical or physical treatment. The sensitivity of the determination based on a minimum detectable recorder deflection of 1% is 2.5 X 1 0 - 1 2 gram/sample. This is equivalent to 0.005 pg/m3 for a 100-cc air sample. When a standard particulate generator was used, the porous graphite compared favorably with Millipore filters. Organic lead, however, was not found to be retained by either type of filter. The techniques described should be readily applicable to other types of flameless devices. Lead is the most widely used of the nonferrous metals. Atmospheric lead comes from manufacturing, use of pesticides, incineration of refuse, and combustion of coal and leaded gasoline. Of these, gasoline combustion is the major source ( I ) . Since 1923 lead alkyls have been added to most gasolines as antiknock compounds. The organic scavengers ethylene dichloride and ethylene dibromide are also added to prevent lead oxide from depositing in the combustion chamber. This results in a discharge to the atmosphere of the mixed chloride and bromide salts of lead (2). Currently ambient concentrations of atmospheric lead are rising a t approximately 5%/year. The toxicological effects of inhaled lead have been documented (3, 4 ) and will not be discussed. On the other hand, a discussion of the relationship of particle size to respiratory retention is pertinent to this study. In developing a filtration method, it is helpful to know the mechanisms and efficiencies of deposition of particles in the respiratory system and of the retention in and clearance from the system. Ideally if atmospheric particulate concentrations are to be related to health hazards, the filtration system should have collection characteristics similar to or the same as the human lung. The U.S.Department of Health, Education and Welfare publication ( 4 ) carries an excellent discussion of this thesis. The size of a particle has a bearing on whether it is deposited and where. Larger particles tend to deposit in the mucous linings of the nasopharyngeal passages. The maximum efficiency of deposition in the alveolar region is at a particle size between 1 and 2 pm. There is minimum efficiency for a size of about 0.4 pm, but the efficiency increases again as particle size decreases. According to the results of the National Air Surveillance cascade impactor network ( 5 ) , 7545% of the lead particulates for six cities were less than 2 pm mass median diameter for the year 1970 and 60-7070 were less than 1 pm. The annual mass median diameter ranged from 0.42-0.69 pm for lead. Lead associated aerosols, however, Present address, Varian Instrument Division, 611 Hansen Way, Palo Alto. Calif. 840
Environmental Science 8, Technology
have a fairly large particle size distribution. The geometric standard deviation from this mass median diameter ranges from 3.61-5.56. By comparison, this value is much greater for lead-associated aerosols than for those of other elements. This is in agreement with the findings of earlier workers who investigated the size distribution of particulate automobile exhausts (2). Although there has been a great deal of consideration given to the possibility of converting to the use of leadfree automotive fuels, industrywide change to unleaded gasoline of the current octane rating would require extensive replacements and additions of refinery equipment (6). Therefore, change would not be immediate and sudden but, a t best, a slow phasing out of leaded fuels is the most that can be expected. In the meantime, monitoring the concentrations of lead in the atmosphere is still a necessary task. For this reason, development of accurate and sensitive methods of determining lead in the atmosphere which are also expedient yet inexpensive are desirable goals. Atomic absorption, particularly coupled with modern flameless devices, is ideally suited for this purpose. Various authors have sought to make use of the advantages of atomic absorption for the determination of metals in air (7-9). More recently, flameless atomization devices have been applied due to the increased sensitivity that results from their use (10-12). A short time ago a new method of filtration using porous graphite was briefly introduced (13), which, when coupled with the atomic absorption furnace of Woodriff et al. (14, 1 5 ) , gives both simplicity of use and great sensitivit y . Here we intend to consider further some of the aspects of the method not investigated in the brief introductory article, such as the accuracy and sensitivity of the filtration medium compared with membrane filters, as well as other advantages of porous graphite filtration.
Experimental Apparatus and Materials. The spectroscopic apparatus and the atomic absorption furnace used have been described previously (15, 16), and for the sake of brevity will not be repeated here. Two types of cups are used for different purposes. The cups used for direct filtration are of the same dimensions as those commonly used in emission spectroscopy for carrier distillation (ASTM No. S3). These cups, however, are made of a type of graphite with a closely controlled porosity. This graphite has been described more fully in an earlier work (13).The cups used for solutions are similar but are 4 mm shorter and are made of a higher density graphite. It is necessary for this graphite to be impervious to relatively concentrated ( 5 M ) nitric acid solutions. When the cup was wetted by these solutions, the reagent blank was quite high, probably due to the fact that the wet outsides of the cup in contact with the titanium cup holder picked up trace contaminants on the surface. This condition was overcome by making cups of Poco grade BX 91, found to be impervious to nitric acid. The cupholder used for air filtration (Figure 1) is machined from Teflon bar stock. This material was ideal for this purpose because it could be soaked in a large variety of solvents for cleaning, and it exhibited no memory ef-
fects from one operation to the next.. This design is an improvement from the one used earlier (13) because it has no threaded sidearm, a source of leaks, and the linear design is easier to insert into a line. Millipore filters used for comparison were type HAWP plain white cellulose ester filters, 13 mm in diameter, with a pore diameter of 0.45 pm. Swinnex 13 adapters were used to hold the Millipore filters during the filtration. Glassware used was soaked overnight in hot 2-3M nitric acid prepared from reagent grade nitric acid and then in hot 2-3M nitric acid prepared from redistilled nitric acid. The latter step was repeated until the values obtained from blank determinations no longer deviated significantly from those determined previously. The apparatus depicted in Figure 2 was constructed to generate standard particulate atmospheres. It consists of a surplus multiple unit tube furnace with a Pyrex or Vycor tube through the center. Air from the house line is bled in a t 50 ml/min through a miniature valve of the type used in small aquaria, and the flow rate is monitored on a Gilmont flow meter. Lead chloride is placed in a ceramic boat run into the hottest zone of the Vycor tube. At the exhaust end is a two-way valve. One is usually connected to the filter apparatus and the other is led through a tube up into the exhaust duct to prevent contamination of the work area. Organic lead standards were generated in an apparatus consisting simply of a sidearm test tube with a one-hole rubber stopper a t the top and a capillary tube down the center. Ten microliters of an appropriate concentration of tetramethyllead (TML) in toluene is pipetted into the bottom of the tube. The sampling apparatus is attached at the sidearm. A capillary is used to increase the linear velocity of the sweep gas (air) as it impinges the TML solution a t the bottom. To ensure that all of the sample has been swept from the tube, a volume of sweep gas is allowed to pass through the tube equal to several times the internal volume after the TML solution can no longer be detected visibly. Reagents. Concentrated nitric acid used was distilled once in Pyrex. Organic lead used for standardization was 80% TML in toluene obtained from Alfa Inorganics. Water used was low-conductivity water distilled twice in Pyrex. TML solutions were diluted with reagent grade toluene from J. T. Baker Co. Procedure. T o determine the optimum furnace temDerature for determining lead, approximately 1 ng of lead is added to the impervious cups as lead nitrate solution. The cups are then placed under an infrared heat lamp and the solutions are evaporated to dryness. The cups are then placed in a desiccator to prevent possible contamination. Each cup is then threaded onto a l/S-in. graphite rod and inserted into the furnace. Absorbance values are measured in triplicate a t various furnace temperatures. Prior to use the cups must be cleaned of possible contaminants, best done by merely inserting the unused cups into the furnace and removing them when the absorbance reads zero (about 1h min). Standard curves are then run in the same manner except that the furnace temperature is held constant and the amount of lead added to the cups is varied. Standard curves were run for each type of graphite cup a t the 217.0-nm lead line. Standards were also run a t the less sensitive 368.3-nm lead line. This line is not commonly used for absorbance measurements; however, a Boltzmann calculation shows that the lower energy level of this line of lead is sufficiently populated a t 1850°C to enable absorbance to occur from this metastable state. Background absorbance of all types of samples was checked a t the nonresonance 220.35-nm lead line and none was present.
ji&-l Figure 1 . Teflon cupholder
I
I
0 REFCREYCE ETHOD
\
I
I CEFAM C BOAT
8
GRAPH TE F LTEF ADAPTER
ALTOTRANSFORMER
Figure 2. Particulate generator
Volume 8, Number 9, September 1974
841
The filtering procedure for porous graphite cups is relatively simple. After a set of cups has been cleaned by insertion into the furnace and has cooled down, one of them is placed in the Teflon holder. A syringe is attached to one end of the holder with a piece of surgical tubing in such a way that, on drawing out the plunger, air is forced to flow through the cup from the inside to the outside. The result is the deposition of the particulates on the inner surface of the cup. The volume of air sampled can be adjusted so that a sufficient amount of deposited lead gives significant absorbance. Commonly 100 cm3 are sufficient. The lead in the filtrate can then be determined without further pretreatment by inserting the cup into the furnace and reading the resulting absorbance. For membrane filters, on the other hand, after the air sample is drawn through the filter, the filters must undergo digestion to decompose the filter material, thus eliminating background absorbance. Wet digestion is preferred over dry ashing since the filters vaporize rather vigorously during the dry ashing procedure with a possible loss of sample. After the filtration step, the membrane filter is removed from the holder and placed in the bottom of a graduated centrifuge cone. Concentrated distilled nitric acid (300 p1) is then added. The centrifuge cone is placed in a boiling water bath, and the filter is digested until the solution ceases to fume and appears nearly colorless. The solution is then diluted to 1 ml, and 100 pl aliquots are placed in the impervious cups, and the absorbance is determined as described earlier.
Results and Discussion Figure 3 shows the variation of absorbance with temperature for gram of Pb a t the 217.0-nm line. Approximately the same curve is obtained with different furnaces and different optical arrangements. From this curve it can be seen that the best operating temperature for lead is 1845°C. If standard curves are run as close to this temperature as possible, then the temperature for running samples can vary as much as *25"C without losing accuracy. The standard curves for lead shown in Figure 4 are a t the usual 217.0-nm analytical lines. Since the shapes of the two types of cups differ, as well as the densities and heat capacities of the two different graphites used, it is necessary to run standard curves for each. The sensitivities for the two are similar (2.5 X 10-l2 gram). Both curves were run under the same conditions-i.e., photomultiplier 850 V, hollow cathode 6 ma, entrance slit 150 pm, exit slit 120 pm, reciprocal linear dispersion 11 A/mm. By use of the porous graphite cups, several 100 cm3 air samples were taken throughout the day on the Montana State University campus approximately one week before and one week after the beginning of the school term. Because it takes only a few minutes to collect a sample, it is possible to determine short-term variations in atmospheric concentrations. In Figure 5 are comparative plots of the time variations in concentration on these two days. Weather conditions were similar on both days. The effect of student population on the local concentrations of lead in the air can be seen. Not only is the average concentration greater on September 27, but the fluctuations are also greater with sharp increases occurring during the between-class periods. Figure 6 shows the results of samples taken along Interstate 90 north of Bozeman, Mont., one day before and one week after the official opening of that section of highway. Care was taken to select a location as far removed from local traffic centers as possible. Even though vacation traffic was already a t an end, there is a noticeable difference in the atmospheric concentrations on the two days. 842
Environmental Science & Technology
E
1.2
16
I9
I8
17
20
Temperature x
Figure 3. Temperature curve for lead
10-
8-
A 7b r
i
!/?
,
!
,
!
7
8
1-
I 1
4 5 6 G r a m s X 10"
2
9
10
Figure 4. Lead standard curves at 21 7 0 nm A Porous graphite 0 Impervious graphite
0 c4
1
PC GAINES HALL
....- - - -. . S.Df.
IO
S e w 27
1
, 9
,
1
0
i
i
1
2
1
$
3
4
5
Figure 5. Results of samples taken on the Montana State U n i versity campus
c
3-1
:~ c.
I N T E R S T A T E 90
sept
t I
z , 21
Results of highway samples
Figure 6.
76A
k
5:
/ '-1
b 4-
32-
~
1
2
3
Figure 7.
4 5 6 Grams X IO8
7
8
9
10
Standard curves at 368.3 n m
A Porous graphite. 0 Impervious graphite
On the average, the lead concentrations on the latter date were twice as high; however, it should be pointed out that the anomaly a t 4:40 is due to the traffic from the construction workers. T o determine the validity of these results, it was felt that the porous graphite cups used should be characterized with respect to precision and accuracy. For this, the previously described apparatus shown in Figure 2 was used. It was later found that compressed air from the house line could be used in place of the nitrogen tank with no additional blank absorbance. A rheostat was also connected in series with the tube furnace to enable finer adjustment of the furnace temperature. A cardboard and glass shield was constructed around the furnace and the tube to serve a s a convection shield since drafts would rather suddenly change the system temperature by a few degrees. A regulated power supply was also necessary because locally there are substantial voltage fluctuations. With these modifications, the temperature a t the center of the tube within the furnace could be held constant to within fl/q°C for several hours. With lead bromide in the ceramic boat, a temperature of 381"C, and a sampling time of 1 min a t a flow rate of 50 ml/min, 18 samples were taken. The average amount of lead collected was 9.7 x 10-10 gram and the relative standard deviation was 10.070. When nitrogen instead of air was used, several samples of the particulate containing carrier gas, after being filtered through the porous graphite, were injected directly into the furnace. This could be done by fitting a shortened
side arm with a rubber septum and injecting both unfiltered and filtered gas into the furnace through the septum. Filtration of the particulates by the porous graphite proved to be quantitative. Photomicrographs of the particles generated are shown in Figure 8, a and b. These particles were collected by placing microscope slides in the bottom of the Pyrex tube of Figure 2 and lowering the flow rate to 5-10 ml/min to allow the condensation particles to settle. Figure 8a is taken 8 cm from the ceramic boat and Figure 8b, 23 cm from the boat. The particles range in size from 0.1-1.5 pm. This is very similar to the situation typically found in air samples ( 5 ) . Since activated carbon has been successfully used by others to adsorb organic lead quantitatively (17), the adsorptive properties of our porous graphite cups for organic leads were investigated. When lo-' gram of lead as TML in toluene was pipetted into the organic lead evaporator monitoring the less sensitive 368.3-nm line, there was no absorbance above blank values detected for several determinations, for either the graphite cups or the Millipore filters. To determine if there is some saturation point below which organic leads are quantitatively adsorbed, the same procedure was repeated using gram of lead as T M L and monitoring the more sensitive 217.0-nm line. Again, no organic lead was retained by the porous graphite cups. To determine the accuracy for filtration of particulates by porous graphite, a comparison with membrane filters was attempted since an absolute air particulate standard would be a difficult task. Several filters from four different lots of 13-mm Millipore filters had an average of about 9 ng/filter, with relative standard deviations for the different lots ranging from 18-67%. The results are shown in Table I. This is unsuitable for comparison a t the 217.0-nm line, so a pair of standard curves was made a t the 368.3nm line (Figure 7) having a lower sensitivity-i.e., 1 X 10-9 gram. This enabled a comparison between the two filtration methods, because the Millipore filter blank was insignificant a t this line. By use of the particulate generator of Figure 2, samples were taken of the effluent particulate containing gases alternating between porous graphite filters and Millipore filters so that any drifts in the rate of evolution of lead would be seen in both filters. It turned out, however, that no such drifts or trends were observed. The results of this comparison are listed in Table 11. Since the Millipore filters are digested and only an aliquot used for analysis, particulates from the generator are collected for a longer period of time. To facilitate comparison, then, the number of grams collected per minute is calculated. The values obtained were similar a t 95% confidence, using the student's t analysis. The value obtained for t was 2.00 as opposed to a rejection criteria of 2.08 for that confidence level. An F value of 4.1 shows that the precision of porous graphite filtration is greater than that for filtration using Millipore filters. In both cases, one value was rejected a t 95% confidence as a result of a rejection quotient calculation.
Conclusion The results obtained seem to indicate that using porous graphite as a filter medium for air particulates compares favorably with Millipore filters of the type generally used for this purpose. Other authors (18) have shown that there is some variation between different portions of a filter when large filters are cut into portions and each portion is analyzed separately. We have found that there is some variation between filters when the whole filter is used. At least some of this variation can be attributed to the inVolume 8 , Number 9, September 1974
843
cles resulting from t h e particulate generatoi right. 23 Cm
from boat
Acknowledgment
The authors wish to thank Varian Techtron for assistance given for this study. The authors are also indebted to Don Fritts of the Veterinary Research Lab, Montana State University, for the microscopic work. Literature Cited (1) Hall, S. K., Enuiron. Sci. Technol., 6,31-5 (1972).
(2) Hirschler, D. A,, Gilbert, L. F., Lamb, F. W., Niebylski, L. M..Ind. Ens. Chem, , 49. 1131-4211957) ~~, ~ .~ ,....,. . ( 3 ) Kehoe,R. A,, J.Air Pollut. Contr. Ass., 19,690-700 (1969). Y
kilrrarion Method
Porous graphite
NO. of samples Glmin X 108 Std dev Re1 std dev Population mean a t 95% confidence
14 3.16 0.25 8% p = 3.16 & 0.15
Millipore filter
9 2.84 0.51 18% I" = 2.84 i 0.39
~
~
~~
(4) U S . Department of Health, Education, and Welfare, "Air Quality Criteria for Particulate Matter," National Air Pollution Control Administration Publication No. AP~49.1969. (5) Lee, R. E. Jr., Goranson, S. S., Enrione, R. E., Morgan, G. B., Enuiron. Sei. Teehnol., 6,1025-30 (1972). (6) American Chemical Society, Washington, D.C.. "Cleaning Our Environment, the Chemical Basis forAction," 1969. (7) Thilliez, G., Anal Chem , 39,427-32 (1967). ... ._ . " . @) Dai"e 's, n.ti., ", . IVIOGLO, n., C n m o , u. IVI., finuaon. s a I ecnnol., 4, 318-22 (1970). ( 91 Hwan ~. ,. ... g, J. Y . , Feldman, F. J., Appl. Speetrose., 24, 371-4 (1970). (10) Loftin, H. P., Christian, C. M., Robinson, J. W., Spectrosc. Lett., 3.,161-74 (1970). I. T n.-.l. 7, Xr..:.-m..L.*&..._~ (11) Matourm, u . , D I U U L ~ . . R . , I L ~ ~ M ICCIMUII I ~ ALULIUC ~*L:~~~. sorption Application Notes, Bull. No. IO-CRA-12, 1-6 (1972). (12) Omang, S. H.,Anal. Chim. Acto, 55,439-41 (1971). (13) Woodriff,R., Lech, J. F.,Anal. Chem., 44,1323-5 (1972). (14) Woodriff, R., Stone, R. W., Held, A. M., Appl. Speetros., 22, 408-11 (1968I . (15) Woodriff,R., Shrader, D., Anal. Chem., 43,1918-20 (1971). (16) Woodriff, R., Culver, B., Shrader, D., Super, A. B., ibid., 45, 230-4 (1973) (17) Snyder, L,.J., ibid., 39,591-5 (1967). .-. T H dtrnnr l._ll,l.l., ~ ~ r. (18) Pierce, J. n -., ..-,., --., __I..__I ", m. - -"1 ~~ (1971).
- ..
. I
-1
-
-
-
~
creased handling necessary when using these membrane filters,'especially when we consider the low levels with which we are working. On the other hand, if we are working with less sensitive devices or methods which require higher concentration and therefore larger volume air sampling, these considerations are of less imp,xtance. Furthermore, the blank levels in the filters can be reduced with pretreatment. It is not difficult ito envision other interesting applicar & o n r nf +ha tions of porous graphit0 6ltrst;nn ...l.ll.-.o.~,n a r i o l l ~;n *.I1 favorahle oronerties t h a t e7anhit.e has for different. t ~ n e s
....
_y~II
."..,... ..~..
___
Receiued f o r review Julr 9, 1973 Accepted May 24, 1974. Work
i
