1982
Anal. Chem. 1984, 56, 1982-1986
Table 111. Recovery of Alkyl p-Hydroxybenzoate in Soy Sauce
alkyl p-hydroxybenzoate
added,”
ethyln-propylsec-propyln-butylsec-butyl-
av recovery,b %
re1 std dev, %
1 1 1
89.3 97.8 94.0
2.5 1.6
1
103.0
1
100.0
mg
2.4
6.0 1.0
’AHB in ethanol was added to the soy sauce. b n = 5.
clear. Normal gas chromatogram peaks are shown in Figure 3. In conclusion, y C D was succesfully used in a cleanup technique for trace analysis of organic compounds in food samples. The present technique may be applicable to the cleanup of various trace organic compounds in food and environmental samples. Registry NO. EtHB, 120-47-8;PHB, 94-13-3;IPHB, 4191-73-5; BHB, 94-26-8; IBHB, 17696-61-6;a-HCH, 319-84-6; 0-HCH, 319-85-7; ./-HCH, 58-89-9; 6-HCH, 319-86-8; p,p’-DDD, 72-54-8; p,p’-DDE, 72-55-9;p,p’-DDT, 50-29-3; EPN, 2104-64-5;THF, 109-99-9;a-CD, 10016-20-3;0-CD, 7585-39-9;T-CD,17465-86-0; di-n-butyl phthalate, 84-74-2;di-n-hexyl phthalate, 84-75-3;di2-ethylhexyl phthalate, 117-81-7;butyl phthalyl butyl glycolate, 85-70-1; nitrofen, 13410-72-5;chloronitrofen, 64047-88-7;oxadiazon, 19666-30-9;aldrin, 309-00-2; dieldrin, 60-57-1; endrin, 72-20-8; dichlorvos, 62-73-7; diazinon, 333-41-5; dimethoate, 6051-5; malathion, 121-75-5;fenitrothion, 122-14-5;parathion, 5638-2; 3,4-benzopyrene,50-32-8;4,5-benzopyrene,192-97-2;sorbic acid, 110-44-1;dehydroacetic acid, 520-45-6;benzoic acid, 65-85-0; ethyl ether, 60-29-7; ethyl alcohol, 64-17-5; dioxane, 123-91-1; trichloroethylene, 79-01-6.
LITERATURE CITED Breslow, R. Science 1982, 218, 532-537. Bender, M. L.; Komiyama, M. “Cyclodextrln Chemistry”; Sprlnger-Verlag: New York, 1978. Uekama, K. Yakugaku Zasshi 1981, 10 1 857-873. Tanaka, M.; Mizobuchl, Y.; Sonoda, T.: Shono, T. Anal. Left., Part A 1981, 14, 281-290. Tanaka, M.; Mizobuchl, Y.; Kuroda, T.; Shono, T. J. Chromatogr. 1981, 279, 108-112. Tanaka, M.; Kawaguchl, Y.; Nakae, M.; Mizobuchi, T.; Shono, T. J. Chromatogr. 1982, 246, 207-214. SmolkovB-Keulemansovi, E. J. Chromatogr. 1982, 251, 17-34. Fujimura, K.; Ueda, T.; Ando, T. Anal. Chem. 1983, 55, 446-450. Freeman, D. H.; Cheung, L. S. Science 1981, 214, 790-792. ~
0
2 4 6 T i m e , minutes
8
Figure 3. Gas chromatograms of alkyl p-hydroxybenzoate released from inclusion complexes in soy sause samples by various solvents: (1) sec-propyl p-hydroxybenzoate, (2) ethyl p-hydroxybenzoate, (3) n-propyl p-hydroxybenzoate, (4) sec-butyl p-hydroxybenzoate, (5) n -butyl p -hydroxybenzoate.
relative to the size of these compounds and 7-CD cavity. Precision and Accuracy. Soy sauce samples fortified with AHB were used to estimate the precision and accuracy of this analytical method using 7-CD (Table 111). The relative standard deviation was acceptable for AHB sample analysis. Another advantage of the use of y C D is that coloring substances in solvent extract of soy sauce cannot make a inclusion complex. The extract of AHB from inclusion complexes is
Kazuyoshi Matsunaga* Masaaki Imanaka Tatsuo Ishida Institute of Environment and Public Health of Okayama Prefecture Uchio, Okayama 701-02, Japan Takuzo Oda Department of Biochemistry Cancer Institute, Okayama University Medical School Okayama 700, Japan
RECEIVED for review October 21,1983. Resubmitted February 16,1984. Accepted May 3, 1984.
Direct Collection of Lead in the Atmosphere by Impaction for Determination by Electrothermal Atomization Atomic Absorption Spectrometry Sir: The analysis of metal compounds in aerosols (a suspension of solid or liquid particles in a gas) is of ever increasing concern to the environmentalist, industrial hygienist, and analytical chemist. Of the many analytical techniques available (I),atomic spectrometry has many advantages over competitors in terms of sensitivity, precision, detection limits, accuracy, and cost per analysis and has become widely used in the analysis of metal compounds in aerosols (2). A major problem in applying atomic spectrometric techniques (or any other technique) is that a collection stage is required, followed by sample preparation (mostly dissolution) and then analysis.
This can be tedious and time consuming and lead to losses due to contamination and loss of chemical changes due to sampling, storage, and transport. Levels may be so low that a long sampling period is required to obtain a measurable amount with current instrumentation. A further disadvantage is that these methods can be in retrospective and not exclude the possibility of an ingestion of a short, sharp, accidental exposure, which over a period of several hours or even days may be considered not dangerous. There is, therefore, a need for a system which will give a real time analysis of metal compounds in an aerosol.
0003-2700/84/0356-1982$01.50/00 1984 American Chemical Soclety
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
Several workers have developed monitors (3,4) based on flame atomization, but an attractive alternative is an electrothermal atomizer (5) which includes increased sensitivity and no need for large volumes of combustible gases and consequently is less hazardous. In addition, the extra sensitivity of an electrothermal atomizer can be used in a semicontinuous monitoring of metal compounds in the atmosphere. Although the atomizer works on a discontinuous principle, a measurement every few minutes is normally quite acceptable in situations of occupational exposure, in which monitoring is retrospective. For the measurement of metal compounds in an aerosol, a sampling system is required and includes a means of collecting the sample, a device to trap the metal compounds, and a means of measuring the amount of air sampled. The most suitable methods for collection are filtration, electrostatic precipitation, or impaction (2, 6). Filtration is unsuitable for a continuous or semicontinuous basis and electrostatic precipitation is not particularly suitable due to problems of collection efficiency at high flow rates and standardization (7). Torsi and co-workers have described electrostatic precipitation of metal compounds in aerosols for analysis by electrothermal atomization atomic absorption spectrometry (ETAAS) (8-10). Impaction into a liquid collector is unsatisfactory due to problems of unabsorbed particulates, but dry impaction techniques combined with ETAAS offer the possibility of the semicontinuous measurement of metal compounds in aerosols. In this paper an impaction device connected to a commercial ETAAS is described and preliminary results of the determination of lead in the laboratory atmosphere are presented.
THEORY Collection of an Aerosol on a Graphite Tube by Inertial Impaction. Basic Principles. Inertial impaction techniques have found extensive application in the collection and measurement of aerosol particles and have been analyzed both theoretically and experimentally more thoroughly than any other aerosol separation process (11,12). An impactor is an instrument in which an aerosol, issuing from a narrow jet, impinges on a plate or impaction surface, and the aerosol particles are deposited on it because of their inertia. It is usually used when some information on particle size distribution of an aerosol is required. The first instrument was described by May (13)and consisted of four jets and sampling sites. The jets were progressively finer, so that the speed of the aerosol increased and finer particles were impacted on the sides and removed from the aerosol. A size-gradingoccurred which assisted in assessing the samples. This instrument was the forerunner for commercial cascade impactors, most notably the Anderson Impactor (Anderson Samplers, Inc., Atlanta, GA) which has the advantage of being inexpensive and relatively simple to use. Several new and more sophisticated particle size measuring systems are available including a ten-stage piezoelectric cascade impactor that features in situ electronic weighing and gives complete mass concentration and size distribution in minutes (Californian Measurements, Inc., Sierra Madre, CA). In this new collection system there was no need or desire to collect separate particle sizes, but rather to collect the total metal compounds in an aerosol on a graphite tube which was subsequently analyzed by ETAAS. The basic principle is shown in Figure 1 in which an aerosol is passed or drawn through a jet with the output stream directed against the graphite (impaction surface) furnace opposite the injection hole. The impaction plate deflects the flow rate to form an abrupt bend in the streamline (characteristic curve) and particles with sufficient inertia are unable to follow the streamline and impact with the graphite surface at a greater distance from the center. Extremely small particles will remain airborne and flow out of the system. This
AEROSOL F I.OW
1983
PATTE. A N
ALUMINIUM SAMPLING T U B E
T A N T U L U M JET
‘E
\
I
/ P A R T I C L E SIZE NOT COLLECTED
TUBE
\ I M P A C T I O N SURFACE
PARTICLE S I Z E COLLECTED
Figure 1. Schematic cross sectional view of the impaction system.
collection system will therefore separate the metal compounds into sizes, particles larger than a certain aerodynamic size are removed from the aerosol by impaction on the graphite surface and those smaller remain airborne and pass through the collection system. Calculation of Impaction Efficiency. Airborne particles vary widely in chemical composition, size, shape, homogeneity, and concentration (2). It is desirable for this new collection system to efficiently collect a wide range of particle sizes, but more importantly, to ensure small particle size collection because it is accepted that small particles are more dangerous than larger ones because they are carried more efficiently into the lungs. The efficiency of the impactor system will be defiied by its characteristic c w e which shows the dependence of the fraction of metal compounds in the aerosol deposited on their diameter. The simple geometry of this impaction system will lead to very complex particle motion which will be strongly dependent on particle size. If the particle motion is governed by Stoke’s law, the parameter which governs the efficiency of the impactor is the Stoke’s number (Stk) or impaction parameter which depends not only on the particle size but also the average flow velocity (0and diameter (D)of the jet
where L is the stopping distance of the particle, p is the density of the particle, I.L is the viscosity of the medium, dp is the diameter of the particle, and Cc is the Cunningham correction factor which takes care of the fact that the relative velocity of the aerosol at the surface of a sphere is not zero for small particles which approach the mean path of particles. At dp 2 5 pm and atmospheric pressure it can be assumed that Cc = 1. 7 is the relaxation time which is the product of particle mass (m)and mobility (P) and is a useful quantity for the analysis of a complex particle motion and is the time for a particle to adjust or relax its velocity to a new condition of forces.
The Reynolds number characterizes aerosol flow through the jet (flow Reynolds number) or around an aerosol particle (particle Reynolds number) and provides information on whether the flow is laminar (Re < 2000) or turbulent (Re > 2000) and is proportional to the ratio of inertial forces to frictional forces acting on each element in the aerosol.
pUD
P UdP
c1
IJ
Re = -(flow Re) or
-(particle Re)
(3)
1984
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11,
SEPTEMBER 1984
Table I. Theoretical Calculations of Stoke Number, Particle Size, and Reynolds Number jet diameter, mm
dist from jet to graphite surf, mm
0.5 0.5 0.5 0.5 0.5 0.5
0.5 0.5 5.0 5.0 8.0 8.0 0.5 0.5 5.0 5.0 8.0 8.0 0.5 0.5 5.0 8.0 8.0 8.0
1.0 1.0 1.0 1.0 1.0 1.0 1.5 1.5 1.5 1.5 1.5 1.5
av flow velocity, ms-’
flow rate,
Lmid 15 1 15
1272.9 84.9 1272.9 84.9 1272.9 84.9 318.3 21.2 318.3 21.2 318.3 21.2 141.5 9.4 141.5 9.4 141.5 9.4
1 15
1 15
1 15 1 15
1 15 1 15
1 15 1
Stoke 2 2 20 20 32 32
1 1 10 10 16 16 0.67 0.67 6.67 6.67 10.67 10.67
particle size, 10% m
Reynolds no. (flow)
Reynolds no. (particle)
0.20 0.80 0.65 2.52 0.82 3.18 0.40 1.60 1.30 5.87 1.79 6.40 0.62 2.39 1.95 7.55 2.47 9.55
43 406 2 895 43 406 2 895 43 406 2 895 21 708 1446 21 708 1446 21 708 1446 14 476 962 14 476 962 14 476 962
4 268 11383 13 870 35 858 17 497 45 249 21 339 85 355 69 351 313 147 95 410 341 121 14 704 3 765 46 245 11895 58 577 15 045
I
TANTULUM JET
0 0
2
3
GRAPHITE I U B E
4
AERODYNAMICDI AMETE R p I
0
I
0.47
OUTLET(2) T O VACUUM PUMC
I
0.94
m
Figure 3. The impactor system consisting of an aluminum rod and plate for attachment to an Instrumentation Laboratory 655 temperature
Flgure 2. Typical theoretical Impactor efficiency curve.
controlled furnace.
Collection efficiency curves are often plotted in the general form of collection efficiency vs. the square root of the Stokes number, Stk, which is directly proportional to particle size. A typical theoretical impactor efficiency curve is shown in Figure 2. The average flow velocity (0at the jet can be calculated from the flow rate divided by the cross sectional area of the jet. Design Characteristics of the Impaction System. Several authors (14, 15) have made theoretical calculations of impactor characteristics based on flow patterns in impactors. Of great importance is the average flow velocity, the diameter of the jet, D, and the distance between the jet and the graphite furnace, L (impaction surface). The practical inside diameter, D, of the jet could not exceed 1.50 mm (2.00 mm 0.d.) as the injection port of the graphite tube would need to be greatly enlarged which would effect the lifetime of the tube. The minimum size of the inside jet diameter considered was 0.5 mm, as smaller than this may become easily blocked with larger particles. The maximum flow rate of the impactor system was 15 L min-l with lower flow rates possible. The closest distance the jet could be accurately positioned to the graphite furnace was 0.5 mm with larger distances up to a maximum of 8.0 mm. At 5.0 mm the jet was just positioned outside the injection port of the graphite tube. In Table I,
theoretical calculations of various jet diameters, distance of the jet from the impaction surface and flow rates from the Stoke’s number, minimum size of particle deposited, and Reynolds number are shown. At high flow rates, all three jets give a submicrometer particle size deposited and high Reynolds numbers (>2000) indicating turbulent flow. As the flow rate is decreased the particle size deposited is increased and the flow Reynolds number decreases. At flow rates of 1 L m i d the 1.0 and 1.5 mm jet diameters give Reynolds < 2000 indicating laminar flow. The smallest distance of the jet from the surface gives the smallest theoretical particle size collected.
u,
EXPERIMENTAL SECTION Apparatus. The impactor system attachment to an Instrumentation Laboratory 655 Temperature Controlled Furnace (TCF) consists of two parts and is shown schematically in Figure 3. An aluminum rod has an initial inside diameter of 20.0 mm and tapers to an inside diameter of 1.0 mm over a length of 110.0 mm. The gradual decrease in diameter is to prevent any build up of aerosol particles at sharp edges. A tantulum jet of inside diameter 1.0mm and outside diameter 2.0 mm was machined after consideringvarious jet diameters and pressed into the aluminum rod with a 4.0 mm length protruding as shown. A 40.0 mm piece of the aluminum rod was threaded and when matched with the aluminum plate allowed movement of the jet in and out of the graphite tube to study the effect of distance from the impaction
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
surface (graphite tube wall). The second part consists of a 10.0 mm thick aluminum plate which replaces the plate on the 655 TCF as supplied by Instrumentation Laboratory. Particular attention was taken in exactly positioning the plate such that the aluminum rod/tantulum jet precisely fitted the graphite furnace injection port and by means of the threaded rod on the 2 axis (in and out) could be varied. A quartz viewing window allowed visual alignment of the jet and two cylinder outlets of inside diameter 5.0 mm and 15.0 mm in height were pressed into the plate. These two outlets were connected via a tee piece to a flowmeter (Chemtron,0-15 L mi&, Medical Products, St. Louis, MO) and then to a Duo Seal Vacuum pump (Sargent-Welch, Skokie, IL) with vacuum tubing used in these connections. This system allowed a maximum flow rate of 15 L m i d with lower flow rates possible. All contact surfaces were sealed with a rubber seal to prevent leakage and was periodically checked throughout this work. The 655 TCF with impactor system was inserted in an Instrumentation Laboratory 457 atomic absorption/atomic emission spectrometer. When the laboratory atmosphere was sampled, the graphite furnace temperature and pressure were that of ambient air. It is well known that if a temperature gradient exists in a gas, then an aerosol particle in that gas will experience a force in the direction of decreasing temperature (thermophoresis). Peak height (absorbances)were measured in a Linear Instruments Model 500 variable-range, strip chart recorder. Several experiments were performed with peak area measurements and the results were comparable with peak height measurements for lead. Throughout the course of this work, peak height measurements were used. Eppendorf micropipets with disposable plastic tips were used for sample injection of standards. A stock solution of 262 fig mL-’ was made by dissolving 0.262 g of pure lead in the minimum amount of nitric acid and diluting to 1L with deionized water with standard solutions diluted daily as required. The collection surface of the graphite tube was covered by introducing a sucrose solution by means of the Instrumentation Laboratory Fastac I1 aerosol sampling autmampler. The graphite tube was heated to 120 O C and addition of the sucrose solution corresponded to a 10-rL volume addition which had the aqueous removed on contact. Electron micrographs were obtained by use of a Philips Scanning Electron Microscope (SEM 501B) with attached Sternheil Camera with Polaroid back and using a Polaroid 107C film. The sample was prepared with a Kinney KSE-2 Vacuum evaporator.
RESULTS AND DISCUSSION To evaluate the distance of the jet from the impaction (graphite) surface, an aerosol generated by a CRA 90 graphite rod atomizer using 20-pL aliquots of 1% copper nitrate plus 20 ng mL-l of lead was sampled at different distances from the surface. The sample flow rate was fixed at 8 L min-l and the amount of lead measured in terms of peak height absorbance was compared to the original content of the solution injected into the CRA 90. This can be expressed as minimum efficiency, Emin where Emin=
Mcolleded
Minjected
(4)
where M m ~ isdthe amount of lead measured after collection in the graphite tube and M h is the ~ amount of lead injected into the CRA 90. The minimum efficiency, Emin, includes all the sample losses in both aercaol generation and transport and is only used to express the relative differences in jet distance. An optimum lead absorbance signal was obtained when the jet was positioned outside the graphite furance injection port. Moving the jet away from the port reduced the signal. Moving the jet inside the injection port showed a similar lead absorbance signal as positioning on the outside of the port but had the inconvenience of having to remove the jet because it blocked the light from the hollow cathode lamp and prevented measurement. It must be recognized that these results only apply to particle size ranges generated by a CRA 90.
1985
In the theoretical discussion the calculations are only the conditions for the particles to reach the wall of the graphite tube. For particle collection, it is necessary that the collision with the wall of the tube should be effective, that is, would lead to their sticking to the tube. As the energy of adhesion of spherical particles to a plane surface is roughly proportional to their diameter, while their kinetic energy is proportional to the third power of their diameter and to the square of the velocity, the “critical” particle velocity, above which the particles rebound from the wall, should be roughly inversely proportional to their size. This may account for a poor collection efficiency at the large particle size. Dzubay et al. (16) have reviewed particle bounce errors in cascade impactors and concluded that the impaction surface has a significant effect on apparent size distribution, and they compared a number of different surface types. Preliminary results showed that an uncoated graphite tube gave a better Eminvalue than a pyrocoated graphite tube and may be atrributed to the more open structure of the uncoated graphite tube. It may be possible to increase E m i n with different atomization surfaces or coat the collection surface with a “sticky” material, which will be easily removed at the ashing stage of the analysis. This was achieved by introducing a solution of sucrose to the graphite furnace using the autosampler. Using samples introduced by the CRA 90 gave a small improvement in Eminover not using a sticky material. In order to investigate the particle size range collected by the impaction system, electron micrographs of the graphite collection surface after collection of aerosol generated by the CRA and laboratory generated dust were obtained. After collection of the aerosol or dust the graphite tube was removed and cut to expose the collection surface. Sample preparation involved evaporative coating using gold. Large agglomerates of particles in the range of twenty to several hundred micrometers were found. Further experiments involving mounting a 0.8-pm Millipore filter opposite the sampling hole and at the ends of the graphite tube were investigated. Experiments with an aerosol generated by the CRA had particle sizes on the order of 1pm whereas laboratory dust had particle sizes in the range of 1to 50 pm. In one trial a 90-pm particle size was found after a vigorous sweep of the flow near the collection system. In both cases the filters a t the ends of the graphite tube has particle sizes on the order of 0.1 pm, suggesting that the impaction system was capable of collecting particles in the range of 1to 90pm with good efficiency. While this method of sample collection is not identical with the use of the graphite tube, it does provide some idea of the particles which enter the tube. The porosity of the graphite tube and filter paper will not be identical, but both have an open structure which would assist the collection of particles. Further experimentsusing an aerosol with known particle sizes and a more efficient sample preparation of the collection surface using a sputtering coater are currently under investigation to fully characterize the particle size collection of the impactor system. To calculate the concentration of lead in the laboratory atmosphere, a calibration curve is used equivalent to injected standard solutions. This assumes that all particles are collected and that the atomization process is identical for a sample introduced as an ion in solution or a dry particle. A major problem in solid analysis in ETAAS has been obtaining a suitable standard. The concentration of lead is calculated as follows: (5) where Cpb is the equivalent concentration of lead in the laboratory atmosphere for the standard in ng m-3, Mpb is the mass
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
1986
ONSET OF ACTIVITY
50-
p:
: -
”;
