Continuous Automated Measurement of Hexavalent Chromium in

Zhang Genfa, and Purnendu K. Dasgupta , Michael W. Martin and William F. ... Raffaela Sagl , Jiri Duchoslav , Roland Steinberger , Bernhard StrauÃ...
0 downloads 0 Views 160KB Size
Anal. Chem. 2001, 73, 2034-2040

Continuous Automated Measurement of Hexavalent Chromium in Airborne Particulate Matter Gautam Samanta, C. Bradley Boring, and Purnendu K. Dasgupta*

Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, Texas 79409-1061

An automated continuous instrument for the collection and measurement of aerosol Cr(VI) is described. The system alternately collects the sample on one of two glass fiber filters. After 15-min sample collection on one filter, the sampling switches over to the second filter. The freshly sampled filter is washed for 8.5 min, and the washings are preconcentrated on a minicolumn packed with anion exchange resin. The washed filter is dried with filtered hot air for the next 6.5 min so that it is ready for sampling at the end of the 15-min cycle. The preconcentrated Cr(VI) on the column is eluted with 0.1 M sodium perchlorate and then reacted with sym-1,5-diphenylcarbazide prior to absorbance detection with a light emitting diode-based dedicated flow-through absorbance detector. The detection limit (S/N ) 3) is 5 ng of Cr(VI)/m3, orders of magnitude lower than current regulatory requirements. The instrument operates unattended over long periods. In continuous round-the-clock operation, filter replacement frequency is every 24-72 h, depending on dust loading. The system is portable for facile field deployment and permits fully automated rapid determinations at a very low analysis cost per sample. Chromium is a naturally occurring element present in measurable quantities in air, water, soil, and biological materials. Of its two important oxidation states, Cr(III) is considered to be essential for metabolism of glucose, lipid, and protein in mammalian systems.1 In contrast, Cr(VI) is highly toxic and this is compounded by its high aqueous solubility.2-4 Diffusion through cell membranes can occur in the anionic form, resulting in oxidation of biological molecules in the interior of a cell.2,5 Hexavalent chromium is a human carcinogen. It has been specifically classified by the U.S. Environmental Protection Agency (EPA) as a group A inhalation carcinogen.3 Especially when occupational exposure is likely, inhalation is believed to be a major human (1) Guthrie, B. E. The Nutritional Role of Chromium. In Biological and Environmental Aspects of Chromium; Langa˚rd, S., Ed; Elsevier: Amsterdam, 1982; pp 117-148. (2) Cohen, M. D.; Kargacin, B.; Klein, C. B.; Costa, M. Crit. Rev. Toxicol. 1993, 23, 255-281. (3) Health Effects Assessment for Hexavalent Chromium; EPA 540/1-86-019; U.S. Environmental Protection Agency (EPA): Washington, DC, 1984. (4) Singh, J.; Pritchard, D. E.; Carlisle, D. L.; Mclean, J. A.; Montaser, A.; Orenstein, J. M.; Patierno, S. R. Toxicol. Appl. Pharmacol. 1999, 161, 240248. (5) Flora, S. D. Carcinogenesis 2000, 21, 533-541.

2034 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

exposure pathway for Cr(VI).6 Exposure to airborne Cr(VI) is associated with various industrial operations such as metal plating, leather tanning, chromite ore processing, dye and pigment manufacturing, and cleaning of various metal parts prior to protective painting, especially in the aircraft and automotive industry.7,8 Significant exposure damages the skin, the mucous membranes, and the respiratory tract, affects reproduction, and increases frequencies of both sister chromatid exchange and chromosomal aberration.2-5,9 The large difference in toxicity of Cr(VI) compared with Cr(III) mandates that Cr(VI) be selectively measured in environmental samples of concern. Measurement of aerosol Cr(VI) in environmental and workplace samples is generally accomplished by discrete filter sampling followed by extraction and analysis back in the laboratory. Potential problems are caused by instability of the Cr(VI) during extraction or sample preparation and even during sampling due to reaction on the filter with cocollected material.10 The latter problem obviously increases with prolonged sampling. Hexavalent Cr is less stable than Cr(III) and converts to Cr(III) in the presence of oxidizable material (even high chloride concentrations, for example), especially at low pH.11 The typical half-life for Cr(VI) in ambient particulate form is 13 h.12 In recent years, a significant advance has been made by Wang et al.8 to develop a field extraction and analysis method that can process 24 sampled filters in 90 min. The thrust in the present work is not toward the analysis of sampling filters worn by individual workers but to (6) Langa˚rd, S. Absorption, Transport and Excretion of Chromium in Man and Animals. In Biological and Environmental Aspects of Chromium; Langa˚rd, S., Ed; Elsevier: Amsterdam, 1982; pp 149-169. Winer, A. M.; Cohen, Y. Development of Intermediate Transfer Factors for Hexavalent Chromium. In Proceedings of the 86th Annual Air & Waste Management Association Meeting and Exhibition; Air & Waste Management Association, Pittsburgh, PA, 1993; Paper 93-RA-116A. O3. (7) Paustenbach, D. J.; Rinehart, W. E.; Sheehan, P. J. Reg. Toxicol. Pharmacol. 1991, 13, 195-222. Paustenback, D. J.; Sheehan, P. J.; Lau, V.; Meyer, D. Toxicol. Ind. Health 1991, 7, 159-196. (8) Wang, J.; Ashley, K.; Marlow, D.; England, E. C.; Carlton, G. Anal. Chem. 1999, 71, 1027-1032. (9) Cohen, M. D.; Zelikoff, J. T.; Chen, L. C.; Schlesinger, R. B. Inhal. Toxicol. 1997, 9, 843-849. Valsecchi, R.; Cainelli, T. Contact Dermatitis 1984, 10, 252-253; Na, K. J.; Jeong, S. Y.; Lim, C. H. Arch. Toxicol. 1992, 66, 646651. (10) Gray, C. N.; Goldstone, A.; Dare, P. P.; Hewitt, P. J. Am. Ind. Hyg. Assoc. J. 1983, 44, 384-388. Dyg, S.; Anglov, T.; Cristensen, J. M. Anal. Chim. Acta 1994, 286, 273-282. Spini, G.; Profumo, A.; Riolo, C.; Beone, G. M.; Zecca, E. Toxicol. Environ. Chem. 1994, 41, 209-219. (11) Dasgupta P. K.; Petersen, K. Anal. Chem. 1990, 62, 395-402. (12) The Fate of Hexavalent Chromium in the Atmosphere; RTI/3798/00-01F; Research Triangle Institute (RTI): Research Triangle Park, NC, 1988. 10.1021/ac001337m CCC: $20.00

© 2001 American Chemical Society Published on Web 03/22/2001

provide a stand-alone “sentry” instrument located strategically in the workplace that can provide the aerosol Cr(VI) concentration without human intervention. EXPERIMENTAL SECTION Reagents and Standards. All chemicals were of reagent grade and used without further purification. Potassium dichromate (MCB), disodium hydrogen phosphate (Aldrich), sodium perchlorate (Fisher), sodium hydroxide (Baker), acetone (Merck), phosphoric acid (Mallinckrodt), sym-diphenylcarbazide (DPC, Aldrich), and ammonium sulfate (Sigma) were obtained as indicated. Nanopure deionized water (Barnstead) was used throughout. For preconcentration of Cr(VI), weak base type anion exchanger AG 4-X4 resin (100-200 mesh size, Bio-Rad, 75-150 µm) mixed with five times the weight of glass beads (250-300µm diameter, Ace Glass, Vineland, NJ) and packed into a minicolumn (3.5 mm diameter × 20 mm active length). Chromium(VI) aerosol of welding dust on glass fiber filters (Certified Standard BCR 545) was obtained from the European Institute for Reference Materials and Measurement.13 Aerosol Generation. A vibrating orifice aerosol generator (model 3450, TSI Inc., St. Paul, MN) was used to generate monodisperse aerosols containing Cr(VI). If a pure compound is nebulized from a droplet-based aerosol generator, the eventual dry particle size is proportional to the cube root of the feed solution concentration.14 To decouple concentration from particle size, we used 1 mM (NH4)2SO4 as the feed solution matrix. This solution was doped with K2CrO4 to the extent of 0-5000 µg of Cr/L, insignificant in mass concentration compared to that of the (NH4)2SO4 solution (132 mg/L). The aerosol generator was operated with a 20-µm-diameter orifice, 60-mL syringe capacity, syringe pump speed 5.0 × 10-4 cm/s (flow rate 0.165 cm3/min), drive frequency 51.27 kHz, primary airflow 1.5 L/min, and dilution air flow 30.0 L/min. Aerosol free “zero” air was generated by a pure air generator (model 737-14, Aadco, Clearwater, FL, capacity 100 L/min). The liquid droplets generated by the primary airflow were diluted and dried with the secondary airflow. The aerosol then passed through the Kr-85 neutralizer (model 3054 aerosol neutralizer, TSI Inc., St. Paul, MN) to allow the aerosol to attain equilibrium Boltzmann charge. The aerosol stream was then put through two sequential 20-L volume polyethylene chambers for drying to be completed. Using branching Y-splitters, the flow was divided. The presently developed particle analyzer and a comparison side filter (47-mmdiameter glass fiber filter, Whatman type GF/A, prewashed thoroughly with water and dried) were both used with sampling rates of 4 L/min each. (The side filter was always sampled for 4 h and extracted according to standard procedures and the extract analyzed by preconcentration-DPC colorimetry (vide infra).) The length and diameter of the tubing connecting the particle instrument and the side filter were maintained identical to equalize aerosol loss in the lines. A laser-based optical particle counter (model A2212-01-115-1, Met-One, Grant’s Pass, OR) was used to characterize the aerosol off another branch, and the excess flow was vented off this branch as well. The experimental arrangement is shown in Figure 1. (13) http://www.irmm.jrc.be/. (14) Hinds, W. C. Aerosol Technology; Wiley: New York, 1982; p 381.

Figure 1. Aerosol generation and general test arrangement. Aerosols are generated from a vibrating orifice aerosol generator (VOAG) with pure air and liquid feeds (not shown). The aerosol passes through a Kr-85 charge neutralizer, through two sequential 20-L chambers (to allow complete drying) and then through branching Y-splitters that direct the flows to (a) the test instrument, (b) a side filter for comparison via manual extraction and analysis, (c) an optical particle counter, and (d) a vent.

Aerosol Collection/Extraction. Switching or directing flows in streams bearing aerosols is difficult to accomplish without aerosol loss. Commercially available multiway valves invariably use a construction where the aerosol can deposit by impaction on a bend. To avoid this, we constructed a large-bore (7.5-mm) rotary valve driven by a rotary solenoid (TRW/Ledex, Vandalia, OH, 30° movement). In this valve (see Figure 2a), the stator consists of three adjacent ports A-C, where A and C are connected in common to ambient air via a high-efficiency particle filter cartridge. Port B is connected to the aerosol source to be sampled. The rotor consists of two ports 1 and 2 that respectively connect to sampling filters 1 and 2. In one position of the rotor (“V1 off”), stator ports A and B are respectively connected to rotor ports 1 and 2. In the other position (“V1 on”), stator ports B and C are respectively connected to rotor ports 1 and 2. Thus, this valve allows switching the aerosol stream from one filter to another without aerosol loss. The two filters used for air sampling in the system are housed in custom Plexiglas filter holders, each of which contains push-fit air inlets (and outlets) for the 8.25-mm-o.d. perfluoroalkoxy (PFA) Teflon tubing (Zeus Industrial Products) that connect to V1. In addition, the holders also house two oppositely located 10-32 threaded ports at the top for introducing the wash liquid uniformly on the top of the filter. A single similarly threaded aspiration port is situated on the bottom of the holder and located at right angle to the inlet ports. The two inlet ports are connected to a tee via identical 5-cm lengths of 0.25-mm-i.d. tubing, which provide the necessary resistance to distribute the liquid flow evenly. Filters used in each holder are Whatman GF/B glass fiber filters (glass fiber filters are most commonly used to sample for aerosol Cr(VI)) backed up by Whatman grade 5 paper filters (the latter help to Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2035

a

b

Figure 2. (a) Aerosol collection and extraction arrangement. Airflow diagram schematically shown. Sample air is sampled in a mass flow controlled manner (MFC1, MFC2) on one filter, chosen by straight passage selector valve V1. At the end of the sampling cycle, V1 switches to begin sampling on other filter while the first filter is put under air lock (V2 and V3 or V4 off). Filter is washed in this fashion. After washing filter is dried in situ by hot filtered air (V2 and V3 or V4 on). (b) Liquid flow schematic for instrument. Five channels of flow are managed by peristaltic pump P. After sampling, filter is washed by NaOH (V5 on) and the extract pumped through V7/V8 through fiber trap filter (FTF) and membranebased cation exchanger/suppressor (regenerated by a flow of H2SO4) to preconcentrator (PC1/PC2). Water is used (V8 off) to wash PC1/PC2 after loading. When the concentrator is eluted by switching V9, NaClO4 elutes the analyte that reacts with DPC reagent and the product is detected by a LED-based detector.

retain glass fibers washed from the GF filter). A multichannel peristaltic pump (Rainin Dynamax) was used for all liquid pumping at a fixed speed of 2.5 rpm. The aerosol collector/extractor operates in four steps. These steps are of 8.5-, 6.5-, 8.5-, and 6.5-min duration, respectively. Sections a and b of Figure 2 show the system air and liquid flow schematic, respectively. During state 1, the air sample is aspirated through filter 1 (V1, V3 on; V2, V4 off) while filter 2, which has just completed sampling in the previous step, is washed with NaOH. The air flow rate on filter 1 is controlled by mass flow controller 1 (MFC 1, AFC 2600D, 5 standard liters per min (SLPM) full scale, Aalborg Instruments, Orangeburg, NJ) to 4.0 SLPM. On the liquid side (V5-V8 all off), 10 mM NaOH is pumped at 0.9 mL/min (2.29-mm-i.d. pump tube; all pump tubes were Pharmed) through V5 and V6 over filter 2. (Note that NaOH is necessary for the extraction/solubilization of lead chromate. For many experiments involving soluble Cr(VI), water was used instead of NaOH.) Filter 2 remains under an airlock because V2 and V4 are off. This allows the pump to aspirate the filter 2 wash through V7 and V8 at 1.5 mL/min (3.20-mm-i.d. tubing) and pump it through a 5-µm pore size fiber trap filter FTF (Acrodisc CR, 25 mm, Pall-Gelman, serves as a final filter for stray glass fibers) into an externally resin-packed filament-filled helical Nafion membrane continuous cation exchanger with 10 mM H2SO4 flowing as regenerant on the outside. Construction details of such devices are in the literature;15 in the present case, a 50-cm, 0.82036 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

mm-i.d. membrane tube, filled with a 0.6-mm nylon monofilament was used. Commercially available “suppressors” (e.g., from Dionex), used for ion chromatography can be substituted. This device neutralizes the NaOH in the extract without loss of chromate. The neutralized effluent is pumped through anion exchange preconcentration column PC2, which constitutes one of the loops in an eight-port dual-loop rotary injector V9 (“on” position). V9 has just switched from off at the beginning of this stage so that the contents of PC1 are being eluted. In state 2 (total time 6.5 min), drying of the filter 2 starts after washing is complete (V1-V4 on). The sampling on filter 1 continues. The in-line siliconized heater (25 W, Watlow, St. Louis, MO) is hooked up electrically in parallel with V2 and turns on whenever V2 is on. Simultaneously, MFC2 is turned on completely open so that maximum flow of the drying air is available to dry filter 2. Note that when this step begins, although the analyte has already been all transferred to the preconcentration column, there is wash liquid still held up in the filter. The trap bottle between the filter and the mass flow controller protects the latter from the intrusion of water. The traps are emptied as a maintenance routine when the filters are replaced (every 24-72 h depending on dust loading). On the liquid side (V5, V8 on; V6, V7 off), the NaOH aspirated from the container is returned to the container through V5 while PC2 is washed with water (purified in-line by aspiration (15) Gupta, S.; Dasgupta, P. K. J. Chromatogr. Sci. 1988, 26, 34-38.

through minicolumn MB, 4 × 90 mm, filled with 20-50 mesh nuclear grade mixed-bed ion exchange resin) aspirated through V8. The water wash also helps remove any air bubbles pumped into the preconcentrator column. In addition, the optical flow cell used in this work is particularly immune to bubble trapping.17 States 3 (V1, V2, V4, V5, V8, V9 off, the rest on) and 4 (V1, V6, V7 off, the rest on) are functionally mirror images of states 1 and 2, with respect to the filters. The third state begins with analyte-loaded and water-washed PC2 going into injection mode and sampling beginning on now dry filter 2. The four-step cycle repeats indefinitely until the instrument is shut down. Valves V2-V4 are 1/2-NPT large orifice two-way normally closed solenoid valves (Skinner A10, Parker/Hannifin), V5 and V6 are miniature PTFE solenoid valves (161T031), V7 and V8 are pinch-type solenoid valves (161P091, both foregoing from Neptune Research, West Caldwell, NJ), all are 12 VDC. V9 is an eight-port two-position rotary valve with its own electrical actuator (C223188EH, VICI, Houston, TX). Although there are a large number of valves, there are only four states. Two relay outputs programmed by an inexpensive timer (ChronTrol, San Diego, CA) are configured to provide TTL outputs. These suffice to provide complete control for the system via a demultiplexer, a hex-inverter, and a number of logic-level MOSFET switches with the valves being isolated by diodes. Details are available from the authors on request. The entire instrument is contained in a tower style PC chassis. Analytical System. A reflective flow cell coupled by optical fibers to a light emitting diode (LED)-based miniature absorbance detector was used throughout the study. The LED was GaP type (HBG 5566X, Stanley Electric) with a nominal center emission wavelength of 555 nm. The LEDs were cut flat and polished on the emitter face. The details of the cell and the detector electronics have been previously described;16,17 it is also commercially available (Global FIA, Gig Harbor, WA). The detector output was acquired at 1 Hz by a Keithley Metrabyte DAS-1601 data acquisition board housed in a personal computer with software written in-house. The flow injection analytical configuration is also shown in Figure 2b. The carrier/eluent solution is 0.1 M NaClO4 (adjusted to pH 11.0 with NaOH) pumped at a rate of 270 µL/min (1.30mm-i.d. pump tubing). The DPC reagent was pumped at 40 µL/ min (0.38-mm-i.d. tubing) and merged with the carrier stream and followed by a mixing coil (0.75 × 1000 mm i.d., 1 m long, Super Serpentine type SS1,18 Global FIA) that provided a reaction time of 2.5 min prior to detection. The DPC reagent was made contained 7.5 mL of 1 w/v% DPC in acetone, 7.5 mL of concentrated H3PO4, and 35 mL of 0.1 M Na2HPO4 (pH 8.0). RESULTS AND DISCUSSION Solution-Phase Cr(VI) Measurement System. The chromogenic reaction of diphenylcarbazide with Cr(VI) has been known for almost 100 years.19 The principle of the method is the Cr(VI)-induced oxidation of the reagent to diphenylcarbadiazone,

the enol form of which then reacts with the Cr(III) formed to yield a red-purple product (λmax 540 nm). Some metals such as Fe, Cu, Mo, and V and oxidants such as chlorine can interfere.18,20 Nevertheless, overall the diphenylcarbazide-based colorimetric system is not only highly selective for Cr(VI) over other metals and over Cr(III) but it readily allows low-nanogram to subnanogram detection limits to be attained.9,18,21 The selectivity is further enhanced because any cationic species washed from the filter sample is removed by the membrane-based proton exchange device; also only anionic species can be concentrated on the anion exchanger.9,18,22 The redox instability of Cr(VI) is strongly pH dependent. Washing a filter with NaOH and then drying it essentially results in a base-impregnated filter, and this provides acid neutralization capacity for any acidic aerosol and stabilizes the Cr(VI) contained in such an aerosol. This is important because often the inhalation of Cr(VI) is associated with the use of chromic acid. Chromate is very strongly retained by most anion exchange resins. This may make preconcentration facile, but makes it difficult to elute it as a sharp peak from the sorbent. In an earlier paper, Amberlite IRA-68 was chosen as a suitable stationary phase.18 Further work during this study established that a similar weakly basic but more easily available resin AG-4X4, especially when diluted by an inert carrier, is particularly well suited. Elution is accomplished with a perchlorate-containing eluent that is adjusted to a distinctly alkaline pHsperchlorate is a strongly pushing anion and an alkaline pH further accelerates elution from a weak base type exchanger. Experiments were carried out with variations in both the exchanger capacity (by diluting the ion exchange resin with an inert glass bead carrier) and the eluent perchlorate concentration. Peak heights and widths obtained with a 0.1-1.0 M NaClO4 pH 11 eluent are shown in Figure 3. Normally, as the width of a peak decreases, the height increases. However, in this case, while the width decreases continuously with increasing perchlorate concentration, the peak height reaches a maximum value with 0.1 M NaClO4. These data suggest that excessive perchlorate concentration in the eluent results in a high enough perchlorate loading on the stationary phase that it interferes with quantitative uptake of the chromate during loading. Note that the dissolved solids content of the membrane-based cation exchanger is very low, and this matrix remains in the preconcentration column at the time it is switched to the inject mode. The refractive index mismatch between this liquid and the eluent increases with large perchlorate concentrations. The perchlorate concentration was therefore chosen at the desirably low 0.1 M level and matched with a low enough column capacity and dimensions to permit a reasonable elution time and peak width. The linearity of the system was studied by loading 0.95 mL of standard solutions containing 0, 1.0, 2.5, 5.0, 10.0, 25.0, 50.0, and 100 µg of Cr(VI) L-1 on to the preconcentration column with a minimum of triplicate measurements. The RSD value was as low as 0.23% and increased to 8.1% at the lowest test concentration. The calibration was linear (r2 ) 0.9986, improving to 0.9999 if the

(16) Jambunathan, S.; Dasgupta, P. K.; Wolcott, D. K.; Marshall, G. D.; Olson, D. C. Talanta 1999, 50, 481-490. (17) Jambunathan, S.; Dasgupta, P. K. J. Soc. Leather Technol. Chem. 2000, 84, 63-73. (18) Waiz, S.; Cedillo, B. M.; Jambunathan, S.; Hohnholt, S. G.; Dasgupta, P. K.; Wolcott, D. K. Anal. Chim. Acta 2001, 428, 163-171. (19) Moulin, A. Bull. Chem. Soc. Paris 1904, 31, 295.

(20) Clesceri, L. S.; Greenberg, A. E.; Eaton, A. D. Standard Methods for the Examination of Water and Wastewater; American Public Health Association, Washington, DC, 1998; pp 3-66. (21) Wang, J.; Ashley, K.; Kennedy, E. R.; Neumeister, C. Analyst 1997, 122, 1307-1312. (22) Vogel, A. I. A Textboof of Qualitative Inorganic Analysis, 4th ed.; Longmans Green: London, 1954; p 210.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2037

Figure 3. Analyte response peak height and width as a function of eluent NaClO4 concentration.

Figure 4. Lack of influence of Cr(VI) concentration in feed on particle size distribution or number counts.

50.0 µg/L data were excluded) with a slope of 0.4 mAU/ng of Cr(VI). The limit of detection (LOD) on a S/N ) 3 basis (where N is the standard deviation of the blank) was 0.22 ng. Others22 have previously quoted an LOD of 0.11 ng. However, in that measurement, the chromophore was preformed and simply injected into a methanol water carrier. Therefore, it was not subject to either chemical noise from the reaction system or dispersion due to elution from a preconcentration column. Overall, the LODs in the two studies are therefore quite comparable. Aerosol Cr(VI) Determination. (1) Aerosol Characterization. In a vibrating orifice generator, the primary droplet volume is computed simply by the liquid flow rate divided by the vibration frequency. This produces a primary droplet volume of 53.6 pL under the experimental conditions. A droplet of this size containing 1 mM (NH4)2SO4 represents a solute mass of 7.1 pg. Below a relative humidity of 80%, (NH4)2SO4 exists in the anhydrous mascagnite form with a density of 1.77. The volume of the dry particle is thus computed to be 4 fL with a spherical equivalent diameter of 1.96 µm. Figure 4 shows the particle counts obtained with the optical particle counter. (Note that counts for particles of >3 µm in size were too small to be displayed.) These results are in excellent agreement with the theoretical predictions and further show that the Cr content had no influence on the particle size. (2) Recovery of Sampled Cr(VI). To evaluate recovery of sampled Cr(VI), 2.5-100 ng of Cr(VI) were spiked as solution standards on the glass fiber filters contained in the instrument and the instrument was put through its normal operating cycle while sampling zero air. The recovery (n ) 3 at each concentration level) averaged 101.7 ( 2.9% (97.6 ( 2.4, 101.2 ( 3.9, 99.6 ( 5.3, 103.0 ( 1.5, 105.4 ( 2.7, and 103.6 ( 1.4%, respectively, at 2.5, 5, 10, 25, 50, and 100 ng of Cr(VI) loaded on the filter). Similarly, PbCrO4 was spiked on the filter from a suspension of known concentration that was continuously stirred. Ultrasonic extraction with NH3/(NH4)2SO4 has been recommended for the extraction of lead chromate. We found that dilute sodium hydroxide is perfectly acceptable for this purpose because it

solubilizes lead, forming sodium plumbite.22 This was more acceptable in terms of chromate recovery than any other approaches we have tried (various acids) or ammonium sulfatecontaining media recommended in the literature9,22 because these extractants obligatorily lead to much higher concentrations of other anions that cannot be removed with an ion exchanger without removing chromate. If they are allowed to remain in the effluent, they compete with the chromate for available sites and even for 10 mM methanesulfonic acid (methanesulfonate has poor affinity for the anion exchange sites), the recovery of chromate is decreased. It is possible that in low-pH media the loss of Cr(VI) is additionally or alternatively affected by redox processes, the exact cause of low recoveries was not studied. Two different amounts of PbCrO4 (85.3 and 170.6 ng) were spiked on the filter to check the recovery Cr(VI) from this common insoluble form of Cr(VI). The recovery was quantitative within experimental error, 100.9 ( 6.0 (n ) 6) and 98.2 ( 5.9% (n ) 4), respectively. The relatively high standard deviation is reflective of the difficulty in keeping the dense lead chromate as a homogeneous suspension, rather that any intrinsic problem with the method. Calibration with Cr(VI)-Containing Aerosols. Detection Limits. As previously mentioned, Cr(VI) aerosols were generated by doping a known amount Cr(VI) into 1 mM (NH4)2SO4 feed solutions that were nebulized. The doped Cr(VI) concentrations were 0, 50, 100, 250, 500, 1000, and 5000 µg/L and resulted in a final aerosol stream concentration of 0-8 µg of Cr(VI)/m3. The linear correlation coefficient between the feed Cr(VI) concentration and the generated aerosol Cr(VI) concentration (as measured on the basis of the instrument signals) was excellent, 0.9996. To evaluate the system LOD for aerosol Cr(VI), the blank value and the blank noise was first evaluated using aerosols generated from the (NH4)2SO4 feed solution to which no Cr(VI) was (at least deliberately) added. The resulting signals, which include the refractive index effects, had a magnitude of 1.48 ( 0.05 mAU (n ) 12). The analytical signal from low levels of Cr(VI) in the aerosol phase was determined by generating an aerosol with a feed

2038 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

Figure 5. System output at a level of 1.5 µg/m3 Cr(VI) being sampled.

solution containing 50 µg/L Cr(VI) in 1 mM (NH4)2SO4. As measured by the side filter on two successive days, the aerosol Cr(VI) concentration was 0.078 ( 0.005 µg/m3. Simultaneously, the analytical signal produced by the instrument was recorded. On the basis of 16 measurements on each of 2 days, the S/N ) 3 LOD was calculated to be 5.3 ng/m3. In contrast, NIOSH method 7600 achieves an LOD of 500 ng/m3 after 2-6 h of filter sampling, extraction, and spectrophotometry after DPC-based color development.23 Sheehan et al.24 reported an LOD of 0.1 ng/m3 but this method requires sampling at 15 L/min for 24 h (a 360 times greater sample volume) followed by manual extraction and ion chromatography with postcolumn reaction detection with DPC. It should be noted that quantitative stability of aerosol concentrations is not as good as solution standards. Standard deviations of 5-10% on short-term samples are common. Figure 5 shows the instrument output from a continuous sampling experiment at a concentration of 2.0 µg of Cr(VI)/m3. Data over a period of 4+ h are shown. The total peak height uncertainty is 2.05%; note that alternate peaks represent alternate sampling channels with their own uncertainties. The system provided excellent linearity of response over a wide range of aerosol Cr(VI) concentration. For the entire test range (0.08-24 µg/m3, 5 different concentrations, 6-13 measurements each, not including blank), the peak height from the instrument was linearly correlated with the aerosol-phase concentration as determined by the side filter (n ) 2-3) with an r2 value 0.9999 and a slope of 25.2 mAU µg-1 m-3. The RSD values at each of these concentrations ranged from 1.35 to 4.85%, much of which can be attributed to the short-term fluctuations in the aerosol generation system. If the signals from the present instrument are interpreted in terms of the liquid-phase calibration conducted with it, we can (23) Manual of Analytical Methods, 3rd ed.; NIOSH Pub. No. 84-100, Method 7600; National Institute for Occupational Safety and Health (NIOSH): Cincinnati, OH, 1984. (24) Sheehan, P.; Ricks, R.; Ripple, S.; Paustenbach, D. Am. Ind. Hyg. Assoc. J. 1992, 53, 57-68.

Figure 6. Sample carryover determined for a high level of Cr(VI) loading.

directly compute the aerosol Cr(VI) concentrations being sampled. At two different concentrations, these data were compared with concentrations determined by the side filter (0.928 and 13.5 µg/ m3). The calculated concentration from the instrument output was 98.4 ( 1.6% of that given by the parallel side filter, essentially the same within experimental error. Sample Carryover and Breakthrough. Sample carryover in an aerosol collection and analysis system occurs from a number of sources: aerosol deposition in the internal lines and valves and resuspension, incomplete washing of the filter or other components, etc. It is therefore of importance to determine this quantitatively. This experiment was conducted with a generated aerosol concentration of 10-15 µg/m3 of Cr(VI), well above the regulatory limit of 1 µg/m3. Immediately following the sample collection, the aerosol inlet was disconnected and a high-efficiency filter was put in its place to filter out any particles from the sampled air. The instrument was allowed to go through several cycles until the signal obtained from the initial collection filter in successive cycles was indistinguishable from the blank. In all cases, no signals over blank were measurable by the fourth wash. The pooled results for six separate experiments are shown in Figure 6 with the original signal normalized to unity. As is true in most systems, the carryover decreases exponentially. The overall carryover is too small to be of concern. To confirm the quantitative removal of Cr(VI) by the instrumental collection system, a 47-mm glass fiber backup filter was put in immediately ahead of the sampling mass flow controller. Several experiments were carried out with aerosol Cr(VI) concentrations of 10-20 µg/m3. After 2 h of sampling, this filter was taken off for manual extraction and analysis. In no case was any detectable Cr(VI) found on the backup filter. Effect of Relative Humidity. Without additional humidification, the sample RH was 21%. Because of the difficulties in fully humidifying the large flow rate of the dilution air, we were able to attain a maximum RH of 68% for the humidity dependence experiments. Humidification promotes moisture uptake by the Cr(VI)/(NH4)2SO4 aerosol, and the resulting growth in particle size Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2039

causes more line losses at higher humidities. So it is not possible to simply compare the instrument output at a given liquid feed concentration while varying the dilution air RH. However, it is possible to compare the aerosol concentrations generated by the instrument with that measured by the side filter. The following values (µg/m3, n g 3 each) were obtained with two different aerosol generation feed compositions: 21% RH: 0.458 ( 0.027 (instrument), 0.457 ( 0.015 (side filter); 1.835 ( 0.069 (I), 1.934 ( 0.151 (SF), 68% RH: 0.220 ( 0.038 (I), 0.226 ( 0.008 (SF); 1.045 ( 0.070 (I), 1.061 ( 0.105 (SF). We conclude that there is no statistically measurable effect of RH. Validation with Welding Fume Standard. Currently, the only relevant certified standard available is a welding dust/fume, sampled on a 25-mm glass fiber filter, from the European Institute for Reference Materials and Measurement. Unfortunately, the filter contains over 100 µg of Cr(VI), nearly 2 orders of magnitude more than the highest loading on our instrumental filter measured during these experiments. Therefore, we have used this standard in the following fashion. Each filter was cut into quadrants. The quadrant was placed on one of the sampling filters in the instrument, and the air inlet instrument was protected with a particle filter. The instrument was allowed to go through its normal sampling and wash cycle, but instead of the ion exchange device effluent being loaded onto the preconcentration column, the effluent was collected in a 100-mL volumetric flask and subsequently made up to the mark. A 250-µL aliquot from this solution is diluted to 5 mL in another volumetric flask. This volumetric flask’s content was aspirated through the tube that normally aspirates the filter wash. At least three replicates were run. The results of the four quadrants from each filter were added up. Filter B14-2 contained a certified amount of 2.841 mg of dust, containing 40.2 g of Cr/kg of dust; thus the total amount of Cr(VI) on the filter was 114.2 µg. The amount found was 103.8 ( 1.5 µg, resulting in a 90.9 ( 1.3% recovery. Filter B14-3 contained 2.844 mg of dust with the same certified Cr(VI) content, thus resulting in a total

Cr(VI) loading on the filter of 114.3 µg. The amount found was 105.1 ( 1.9 µg, resulting in a recovery of 92.0 ( 1.6%. Considering that the first wash produces ∼95% of the total recovery (see Figure 6) and the more involved steps necessary to analyze these highly loaded filters, these results are in good agreement. In summary, an affordable, simple, and readily fieldable instrument was developed for the determination of Cr(VI) in airborne particulate matter. The instrument offers good time resolution with minimal carryover and operates in a fully automated manner. Obviously, the instrument is capable of being used for other, more general purposes, e.g., coupled to ion/liquid chromatographs for more complete analysis of the collected aerosol. With appropriate filters, filter housing, and minor changes in plumbing, it should be readily possible to incorporate stationary (rather than percolating) extraction methods, possibly even ultrasonic and thermally aided digestion methods. Relative to previously developed vapor condensation aerosol collection approaches25 there are advantages in that no steam is used and hightemperature hydrolytic reactions are avoided. These aspects will be discussed in a future paper.

(25) Simon, P. K.; Dasgupta, P. K. Anal. Chem 1995, 67, 71-78.

AC001337M

2040

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

ACKNOWLEDGMENT This research was supported in part by the U.S. Department of Defense, through the Strategic Environmental Research and Development Program (SERDP) under a cooperative agreement with the USAF, Institute for Environment, Safety, and Occupational Health, Brooks AFB, Texas. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies of endorsements, either expressed or implied, of the 311 HSW/IERA or the U.S. Government. We also acknowledge the help, assistance, and moral support provided by our colleague George P. Cobb. Received for review November 15, 2000. Accepted February 14, 2001.