Anal. Chem. 1980, 52, 1971-1974
I n conclusion, t h e deactivation results from this investigation accentuate t h e disconcerting fact that the relative degree of passivation for glass surfaces by present deactivation techniques is dependent on the types of analyzed compounds a n d the nature of the glass surface.
LITERATURE CITED Farwell, S. 0.; Gu&, S. J.; Bamesberger, W. L.; Schutte, T. M.; Adams, D. F. Anal. Chem. 1979, 5 1 , 609. Adams, D. F.; Farwell. S. 0.; Pack, M. R.; Bamesberger, W L. J. Air Pollut. Control Assoc. 1979, 2 9 , 380. Adams, D. F.; Farwell, S. 0.; Robinson, E ; Pack, M. R. Project Report 856-1; Electric Power Research Institute: Paio Alto, CA. Feb 1979. Gluck, S. J. M.S. Thesis, University of Idaho, Moscow, ID, 1979. Sandra, P.; Verzele, M. Chromatographia 1977, 10, 419. Sandra, P.; Verstappe, M.; Verzele, M. HRC CC J . High Resoiut. Chromatogr. Chromatogr. Commun. 1978. 1 , 28. Malec, E. J. J. Chromatogr. Sci. 1971, 9 , 1319. Aue, W. A.; Hastings, C. R.; Kapiia, S. J . Chromatogr. 1973, 9 1 , 25. Blomberg, L. J. Chromatogr. 1975, 115, 365. Grob. K.: Grob. G. J. Chromatoar. 1976, 125, 471. Franken, J. J.; DeNijs, R. c. M.; schulting, F. c. J . Chromatogr. 1977, 144, 253. DeNijs, R . C. M.; Franken, J. J.; Dooper, R. P. M.;Rijks, J. A. J . Chromatogr. 1970, 167, 231. Blomberg, L.; Wannman, T. J . Chromatogr. 1978, 148, 379. Daniewski, M. M.; Aue, W. A. J. Chromatogr. 1976, 147, 395. Grob, K., Jr.; Grob, C.; Grob, K. HRC CC J. High Resoiut. Chromatogr.
1971
Chromatogr. Commun. 1978, 1 , 149. (16) NovotnY, M.: Tesarik, K, Chromatogrwhia 1968s 1 , 332. (17) Welsch, Th.; Englewald, W.; Kiaucke, Ch. Chromatographia 1977, 10, 22. (18) Arkles, €3. CHEMTECH 1977, 7, 766. (19) Grob. K.; Gob, G.; &ob. K., Jr. HRC C C J . High Resoiut. Chromatogr. Chromatoar. Commun. 1979, 2 , 31. (20) Madani, Cy; Chambez, E. M.; Rigaud, M.; Durand, J.; Chebroux, P. J. Chromatogr. 1976, 126, 161. (21) Madani, C.: Chambez, E. M.; Rigaud. M.; Chebroux. P.: Breton, J. C. Chromatographia, 1977, 10, 466. (22) Franken, J. J.; Rutten, G. A. F. M.; Rijks, J. A. J . Chromatogr. 1976, 126, 117. (23) Novotny, M. Anal. Chem. 1978. 5 0 , 16A. (24) Rutten, G. A. F. M.; Luyten, J. A. J. Chromatogr. 1972, 74, 177. (25) Petrarch Systems Silicon Compound Catalog, Levittown, PA, 1979. (26) Schutte, T.M. M.S. Thesis, Washington State University, Pullman, WA, 1978. (27) Lee, M. L.; Vassilaros, D. L.; Phillips, L. ‘V.; Hercules, D. M.; Azumaya, H.; Jorgenson, J. W.; Maskarinec, M. P. Anal. Left. 1979, 12, 191.
RECEIVED for review January 30, 1980. Accepted J u n e 24, 1980. This work was by the Ele&ic ~~~~~~~h Institute under Contract No. 856-2. Reference to commercial products is for identification purposes only and does not constitute a n endorsement of these products by E P R I or the University of Idaho.
Reduction-Distillation Method for Sulfate Determination Darryl
D. Siemer
Exxon Nuclear Idaho Company, Inc., Idaho Falls, Idaho 8340 1
T h e determination of total sulfur in calcined nuclear fuel reprocessing waste is a difficult problem. The majority of the material consists of highly fired mixed oxides and/or fluorides of zirconium, aluminum, and calcium, and it is consequently extremely refractory t o the often recommended Leco combustion sulfur determination technique. Other approaches requiring sample solubilization (e.g., ion chromatography) prior t o the actual determination are rendered difficult by the extreme difficulty of dissolving the material completely without both severe dilution and contamination problems. Solution samples encountered in the process streams usually contain large amounts of nitric acid. This report describes a n improvement of the reduction/distillation/spectrophotometric technique first described by St. Lorant (I),by Johnson a n d Nishita (Z),and later by Gustafsson (3,4 ) that is widely used in t h e nuclear industry (5,6). The procedure involves t h e reduction of oxidized forms of sulfur to hydrogen sulfide with a hot solution containing hydroiodic acid, sodium hypophosphite, and acetic acid. The hydrogen sulfide is sparged from the reaction mixture with nitrogen which is then bubbled through a dilute zinc acetate solution which traps the sulfide. T h e addition of acid, p-aminodimethylaniline, and ferric chloride t o the zinc acetate trapping solution quantitatively converts the sulfide to methylene blue which is then measured by absorption spectrophotometry at 667 nm. T h e most serious practical limitations of this approach for general application to nuclear industry samples are, first, the necessity of nitrate and water removal before analysis and, second, a complicated methodology requiring both long reduction times and partial disassembly of the apparatus between each determination. I n this version of the technique, the samples and standard are added serially to a single, relatively large batch of the boiling reduction solution. The apparatus is not cooled down 0003-2700/80/0352-1971$01.00/0
and disassembled between each sample run and the reduction solution is kept a t a boil a t all times and is continuously sparged with nitrogen gas. This reduces the total analysis time to a fraction of that necessary with the previous methods.
EXPERIMENTAL SECTION Apparatus. The reduction apparatus used (Figure 1)is quite similar to that used by Gustafsson ( 3 , 4 ) except that provision is made for adding sample to the reduction flask while it is connected to the rest of the gas train. A thermometer well is also added to permit monitoring of the solution temperature. One-half milliliter capacity sample spoons made of a 5-mm glass rod are sealed to the ends of “T’ head stoppers. The bottom side of the “T” is ground flat to prevent the spoon from rolling over and dumping the sample out during the weighing operation. Tygon tubing is used to connect the reduction flask to the gas wash bubbler and the bubbler to the 1-mL plastic pipets used as gas dispersers. Washers made of 1.2 cm diameter disks of 2 mm thick polyethylene plastic were first perforated with a paper punch and then placed over the ends of the 1-mL plastic pipets in order to promote better dispersion of gas bubbles in the zinc acetate trapping solution. Other apparatus used includes: heating mantle for the reduction flask; nitrogen tank, regulator, flowmeter, and needle valve to supply the sparging gas flow; ball mill to grind solid samples; 0.01 mg sensitivity analytical balance capable of weighing the approximately 20-g sample spoons; spectrophotometer with a 5 nm or less band-pass and a pair of matched 1-cm cuvettes; lo-,25,50-, 1W,and 2WpL micropipets with disposable plastic tips. Reagents. Reducing Solution. Place 20 mL of 56% HI, 40 mL of glacial acetic acid, and 5 g of NaH2P02.2H20into the reaction flask. Connect the flask to the reflux condenser and turn on the cold water. Heat the solution with the mantle at the boiling temperature (107-110 “C) and sparge continuously with nitrogen at 300-400 mL/min for about 0.5 h before using for analyses. Then add 30 mL of water to the gas washing bubbler and connect it in series with the reflux column. @ 1980 American Chemical Society
1972
ANALYTICAL CHEMISTRY, VOL. 52, NO. 12, OCTOBER 1980
P
w L w
5
0
.8
;/
.6r
v
w L
e=
.-
C
-e f
0
v
f
Flgure 1. Reduction flask,gas wash bubbler, and gas trapping cyclinder: a, nitrogen inlet; b, 20 cm long reflux condenser; c, 15 cm long gas wash bubbler; d, 250-mL flask; e, sample spoon; f , heating mantle; g, thermometer; h , pipet gas disperser; i, 50-mL glass-stoppered graduated cylinder of zinc acetate solution.
Stock Zinc Acetate. Dissolve 40 g of Zn(C2H30J2in 1 L of water. Store in a plastic wash bottle. A m i n o Reagent. Add 200 mL of 6 M HCl t o 1 g of NH,C,H,N(CHg)2*HCl. Ferric Chloride. Add 2 g of FeCl3-6Hz0to 100 mL of 6 M HCl. S u l f a t e Standard. Dissolve 2.958 g of dried (110 “C for 2 h) NaZSO4in distilled water. Dilute t o 1 L. This solution contains 100 pg of sulfate in a 50-pL aliquot. Trapping Solution. Add 10 mL of the stock zinc acetate solution to about 30 mL of water contained in a 50-mL graduated cylinder. Procedure. Solid samples are first ground t o a fine powder (less than 320 mesh). Then a quantity sufficient to contain a total inorganic sulfur level equivalent to from 10 to 100 fig of sulfate is weighed directly into the sample spoon. Liquid samples are pipeted into the sample spoon. A plastic pipet gas disperser is fmt connected to the tubing leading from the gas washing bubbler and then is inserted into a graduated cylinder containing the trapping solution. The spoon is rapidly inserted into the entrance port of the reaction flask and is turned to dump the powder into the liquid. After a suitable reduction period (2-3 min for liquid samples, 5-10 rnin for solids), the tubing to the pipet is disconnected and the cylinder and the pipet is set aside. To develop the methylene blue, pipet 2 mL of the amino reagent through the pipet gas diffuser into the solution. The gas-diffusing pipet should be held so that its lower end is just above the surface of the liquid during this step. The last drop should be blown out with a minimum of either displacement of air in the top of the cylinder over the solution or agitation of the solution itself. The differences in densities of the amino reagent and the trapping solution and the geometry of the cylinder provide adequate mixing of the two solutions at this point. Then 1 mL of the ferric chloride reagent is added and the cylinder is immediately stoppered and shaken for 15 s. Finally, the cylinder is filled to 50 mL with distilled water, restoppered, shaken to mix its contents, and set aside for at least 5 min before the methylene blue concentration is measured in the spectrophotometer at 667 nm.
R E S U L T S AND DISCUSSION Figure 2 shows the fraction of sulfur recovered vs. reduction time plots for two finely ground (320 mesh) solid samples and one solution sample. With each sample, four analyses were performed by using differing reaction times. The ordinate of each point is based on the relative absorbances of the final colored solution after the chosen reduction time to that seen with a 10-min reduction time. T h e first solid sample (solid line) is a n artificial mixture of barium sulfate ground up in a barium carbonate matrix. The second (dotted line) is a typical calcined nuclear waste. I t has the following approx-
.41
I
c 21
I
.
0
2
4
o
6
minutes
8
L
1
0
Figure 2. Sulfur recovery with respect to reaction time: solid line, solid samples (7-11 mg) of a mixture of 50 parts of BaCO, to 1 part BaS0,-points normalized to response/mg of sample; dotted Ilne, samples of a solid calcined nuclear waste (15-22 mg eachppoints normalized to responsehg of sample-0.51 % sulfate in each sample; dashed line, 50-pL aliquots of a 2000 mg/L liquid sulfate standard.
Table I. Effect of Matrix Constituents on the Determination of 100 rg of Sulfate concomitant ZrOC1, (in HF) Pb(NO,),
AlCl , CaCl, FeCl HglNO,h C$$N 0 J j;
u (NO,), HNO, HF HC1 H W ,
concn, M
effect
6 2 6 10 2 0.2 0.2
+1.8
-0.7 0.0 +0.9
1 8 29 12 6
+0.9 +1.2 -0.1 +1.8
%Q,b
-0.9
-1.2 +1.2
-0.2
The relative standard deviation of five determinations The “effect” of the matrix-free standard was 4.3%. figure is based on comparing the average of three analyses done both with and without the concomitant. a
imate gross composition: 20% Zr, 32% Ca, 8% Al, 1.1%P, 23% F, with the balance being oxygen and trace elements including sulfur as sulfate. The liquid sample (dashed line) was a 50-pL aliquot of a 2000 ppm sulfate (as sodium sulfate in water) standard. The reaction rate of the solution aliquot is higher than those of the solids. The curves in Figure 2 approach an asypmtotic limit exponentially with reaction half-times on the order of 40 s for the solids and 20 s for the liquid. In practice this means that for 99% conversion (7 half-times) a 5-min reaction time for these solids is sufficient and for liquid samples 2.5 min is adequate. These times compare very favorably with the 10- ( 4 ) to 35-min reduction times (5) recommended by previous authors. T h e fact that this apparatus need not be cooled, disassembled, reassembled, and reheated for each determination saves a t least another 5 min per sample or standard run. The performance of the method on complex solution samples was investigated by analyzing 50-pL aliquots (100 Fg of sulfate) of sodium sulfate solutions both with and without added matrix constituents. This was done by pipeting t h e standard into the sample spoon and then adding 5O-kL aliquots of t h e matrix concomitant solution prior to inserting the spoon into the reduction vessel. A 3-min reduction time was used in each case. Table I gives the concentrations of the concomitants in the matrix solutions added and t h e results of the experiments. None of the added materials either singly
ANALYTICAL CHEMISTRY, VOL. 52, NO. 12, OCTOBER 1980
Table 111. Comparison of Ion Chromatographic Results with Those Obtained by the Sulfate Reduction Technique
Table 11. Study of Nitrate Effecta amt of 2000 ppm run no. 1 2
3 4 5
so,
amt of 8 M HNO,,
standard, pL
pL
0 0 50 50 50
0 100
0 100 0
%
absorbance (1-cm cuvette ) 0.014 0.013 0.723 0.728 0.718
Add 0.6 mL of 8 M HNO, to Flask-Wait 3 minb 6 50 0 0.692 7 50 0 0.720 Add 0.6 mL of 8 M HNO, to Flask-Wait 1 0 minb 50 0 0.725 50 100 0.721 50 400 0.650 50 0 0.715
8 9 10 11
Add 0.6 mL of 8 M HNO, to Flask-Wait 12 50 0 13
50
100
1973
1 0 minb
0.721 0.719
a 3 min reduction reaction time used in all cases. is the time allowed before the next sample is run.
This
or in combination caused a significant signal purturbation a t the 95% confidence level. A series of 10 experiments was performed in which 100 Fg of sulfate standard was alternately run alone or with the addition of 15-20 mg of dry CuCl2.2H20powder added to the sample spoon prior to its being inserted into the reduction flask. There was no apparent (greater than 2 % ) reduction in sulfide recovery caused by the added copper. Apparently the extremely high acidity of the hot hydroiodic acid solution prevents the formation of the extremely insoluble copper(1) or copper(I1) sulfides. The added copper leaves no apparent precipitate in the flask and is probably present as an anionic copper(1)-iodo complex. Of especial significance is the fact that no signal loss was noted in the presence of nitric acid. Previous workers have noted significant nitrate interferences ( 3 , 4 , 7 ) and have recommended procedures for its removal before attempting the sulfur determination. For example, Sinclair et al. reported that when using a conventional batch process sulfate reduction system, 310 pg of nitrate caused a 25% reduction in the apparent recovery of sulfate (7). Table I1 shows the results of a series of experiments run to estimate how much nitrate can be tolerated with the system described in this paper. T h e data in Table I1 indicate that the addition of 50 mg of nitrate (0.1 mL of 8 M HNOJ along with the sulfate has no appreciable effect on the sulfur determination. However, the coaddition of 200 mg of nitrate does appreciably reduce the reduction efficiency. The addition of 300 mg of nitrate affects sulfate determinations attempted as soon as 3 min after the nitrate is added. However, if the reduction solution is allowed t o boil for 10 min before the next sample is run, it.s ability to quantitatively and rapidly reduce sulfate is restored. Eventually, however, repeated additions of nitric acid (or any oxidizing agent) will cause the permanent oxidation of and subsequent loss of iodide as iodine. This can be prevented by periodically adding more sodium hypophosphite (about 2.5 g for every 8 mmol of nitrate added) to the reduction flask. A series of 10 sulfate standards were run to test the tolerance of the system for water. In each case, l mL of water was added to the reduction flask immediately before inserting the sample spoon containing a 50-pL aliquot (100 pg of sulfate) of standard sulfate solution. There was no significant (greater than 2 % ) difference in the results obtained with the extra water added as compared with the standard aliquot alone.
so,
% so,
samplen
ICb>C
reductiond
F 64 F 65 4944-113 2483
0.52 1.60 5.70 3.21
0.53 1.60 5.65 3.19
Samples contained 5-7% Al, 1!3-35% Ca 8-15% Zr, Dionex 0.2-7.2% Na, 0.8-1.19% B, and 6-13% F. Model 1 0 ion chromatograph, 500 mm standard anion separator colunin with 0.0024 F Na,CO, and 0.003 F 1 : 5 HCliwater solvent; metal cations NaHCO, eluent. removed prior t o injection into IC, some CaF, and/or AlF, precipitate left after sample solubilization step. Five minute reduction reaction time used. However, as is the case with nitrate, repeated additions of water will eventually dilute the reagents to the point that lower hydrogen sulfide recoveries are obtained. T o counter this, acetic anhydride in the ratio of five volumes to each volume of sample solvent water introduced can be added to the reduction flask to convert the water to acetic acid. I t is best, however, to avoid the unnecessary addition of water to the flask by concentrating very dilute aqueous samples by evaporation prior to analysis. The presumed reasons for this method’s superior insensitivity to both nitrate and water interferences are, first, t h a t the sample is instantaneously heated to the reaction temperature and, second, that a far greater ratio of reductant (hydroicdic acid) to sample is used. The recoveries of sulfate in laboratory prepared artificial solid matrices consisting largely of barium carbonate, calcium carbonate, and lead dioxide were both rapid and quantitative. This agrees with Gustafsson’s observations ( 4 ) . Some alternate values (obtained by Teflon bomb dissolution with HC1 followed by ion chromatography) for the sulfate contained in “real” solid calcined nuclear waste samples are compared with those gotten by this method in Table 111. The sample solutions injected into the ion chromatograph were first passed through a short, hydrogen-form, strong acid cation exchange column (Bio Rad 100-200 mesh AG50W-Xl6, 5 cm long) in order t o remove cations which would have been precipitated by the ion chromatographic eluent. The agreement between values is good and sulfate reduction procedure takes less than one-fifth as long to perform because the sample solution and cleanup steps are not required. Sulfite is also quantitatively reduced and converted to hydrogen sulfide. However, organically bound sulfur in the form of sulfamic acid, thiourea, and sulfosalicyclic acid was not quantitatively recovered with a 5-min reduction period. For this reason, samples containing considerable organic matter should be oxidatively decomposed prior to analysis as recommended by Gustafsson ( 3 , 4 1 . The prior removal of heavy metals from the zinc acetate trapping reagent was not necessary with the reagent lots available for this project. If analytical results are found to be either inexplicably low (less than 0.67 absorbance unit for 100 pg of sulfate) or variable, it may be necessary to use Gustafsson’s ( 3 ) purification procedure to remove the heavy metals. On the other hand if very high results are obtained, it may prove necessary to remove traces of hydrogen sulfide from the nitrogen gas used to sparge the reduction solution. The most likely sources of negative errors are, first, gas leaks in the system and, second, loss of hydrogen sulfide during the addition of the amino and ferric chloride reagents. All glass joints must be firmly seated and checked repeatedly. However, if this is done i t is not necessary to use silicone grease on the joints as recommended by Chstafsson ( 4 ) . A drop of
1974
Anal. Chem. 1980, 52, 1974-1977
acetic acid on the joints prior to assembly will suffice to seal them and its use will prevent “creeping” of the grease throughout the entire system. Higher and more reproducible results were obtained in this laboratory by using the tall form cylinders (instead of flasks) for collecting the hydrogen sulfide and the method of reagent addition specified in the Procedure section of this paper rather than with Gustafsson’s recommended methodology ( 3 , 4 ) . It is very important that the trapping solution, once it has been acidified with the amino reagent, not be violently mixed or even allowed to stand for long before the addition of the ferric chloride reagent. The overall time for the analysis of a single solution sample is on the order of 10 min including the delay for color development. Solid samples require 2 or 3 min more. The
relative standard deviations obtained on finely ground solid calcined nuclear waste containing from 0.1 t o 10% sulfur as sulfate are typically on the order of 3-5%.
LITERATURE CITED St. Lorant, I. Hoppe-Seyler’s 2. Physiol. Cbem. 1929, 85, 245. Johnson, C. M.; Nishita, H. Anal. Cbem. 1952, 24,736-742. Gustafsson, L. Talanta 1960, 4 , 227-235. Gustafsson, L. Talanta 1960, 4 , 236-243. Annu. Book ASTM Stand. 1979, Part 45, C698, 312-314. Thorn, L. E.; Bryan, R. G.; Waterbury, G. R. Report LA-5884; Los Alamos Scientific Laboratory: Los Alamos. NM, June 1975. (7) Sinclair, A.; Hall, R. D.; Burns, D. T.; Hayes, W. P. Talanta 1971, 76. 972-976.
(1) (2) (3) (4) (5) (6)
RECEIVED for review February 19, 1980. Accepted July 14, 1980.
Sampling of Tetraalkyllead Compounds in Air for Determination by Gas Chromatography-Atomic Absorption Spectrometry Waiter R. A. De Jonghe, Dipankar Chakraborti,’ and Fred C. Adams’ Department of Chemistry, University of Antwerp (U.I.A.), E-26 70 WiIrQk,Belgium
Cantuti and Cartoni first described a procedure for the direct collection of tetraethyllead from polluted air and its chromatographic determination at ppm levels by using electron capture detection ( I ) . Their technique was later modified and improved in order to apply it to the analysis of city air (2). In this way detection limits of about 0.05 and 0.5 pg/m3 were reported for the determination of tetraethyllead and tetramethyllead. Since both methods use electron capture detection, they lack the specificity required for environmental samples, as these contain a variety of other compounds with high electron affinity. By use of cooled gas chromatographic column packing material for the collection, the tetraalkyllead compounds (TML = tetramethyllead, TMEL = trimethylethyllead, DMDEL = dimethyldiethyllead, MTEL = methyltriethyllead, and T E L = tetraethyllead) have also been determined with gas chromatography/atomic absorption spectrometry (GC/ AAS). This method was first reported by Chau et al. ( 3 ) . However, the low analytical sensitivity combined with a very slow air sampling rate (13Ck150 mL/min) made this procedure unsuitable for the analysis of ambient air ( 4 ) . In later modifications, the sensitivity was greatly improved ( 5 , 6) but the determination of tetraalkyllead compounds in street air still required collection times of 16 h or more. Other reports also mentioned the possibilities of the GC/AAS technique for the analysis of tetraalkyllead compounds in the air, without entering into details about the method used for sampling (7) or analysis (8, 9). In the latter work only samples of polluted atmospheres with organic lead concentrations of 0.1 pg/m3 or higher were adequate in order to separate the different alkyllead components. Reamer e t al. (10) developed a method using a gas chromatograph/microwave plasma detector, by which the trapped tetraalkyllead compounds (TAL) were not directly eluted into the gas chromatograph but removed from the adsorbent by a freeze-drying technique. The limit of detection for an individual organolead species was quoted as about 0.5 ng/m3 P r e s e n t address: D e p a r t m e n t of C h e m i s t r y , J a d a v p u r U n i v e r India.
sity, Calcutta-32,
0003-2700/80/0352-1974$01 .OO/O
for a 2-h sample. Despite the high sensitivity, this method is not practical for pollution-control measurements in view of the lengthy period (more than 1 2 h) required for the analysis of a single sample. With the exception of the work of Laveskog ( I I ) , no method published so far is able to monitor alkyllead species in ambient air within a reasonably short time. Laveskog’s method, however, involved gas chromatographic/mass spectrometric instrumentation, which is not readily available t o most workers. Atomic absorption spectrometry, being sensitive, relatively free of interferences, and widely available in trace metal laboratories, appears to be more suitable as the detection system (12),provided it is used in conjunction with a sampling technique other than sample collection on tubes filled with a GC adsorbent. This report describes the development and utilization of a suitable sampling system for the analysis by GC/AAS. Sampling periods of 1 h or less proved to be sufficient, even for the determination of alkyllead species in relatively nonpolluted air. The major difficulty in collecting the compounds from air samples on GC column packing material is the condensation of moisture in the trap. Ice condensation on the column material leads t o clogging of the pores and a sharp decrease in the air flow rate. The volume of air that can be sampled, is therefore limited. Several authors pass the air before it enters the adsorption tube through a predeposition trap, t o condense the excess of atmospheric water. By use of sampling apparatus with an empty U-tube a t -15 “C, maximal reported volumes are about 7 5 L (6),as water condensation on the adsorption tube is not completely prevented. An empty impinger held a t -78 “C is still not capable of extending the sampling volumes above 200 L (9, 10). To design a more efficient water condensation trap, we investigated the use of a large U-tube filled with glass beads. In this way the predeposition of water would be much improved as a result of a better cooling efficiency of the air. From preliminary experiments it appeared that this predeposition trap was really effective, only if temperatures of about -80 “C or lower are used. However, a t these low temperatures, a substantial fraction of the tetraalkylleads is retained also, 1980 American Chemical Society