Anal. Chem. 1893, 65, 209-214
208
Measurement of Ambient Ozone Using a Nitrite-Coated Filter Petros Koutrakis; Jack M. Wolfson, Arnold Bunyaviroch, Susan E. Froehlich, Koichiro Hirano,t and James D. Muliki Department of Environmental Health, Harvard School of Public Health, Boston, Massachusetts 02115
Standard ozone monltorlng technlques utlllze large, heavy, and expenslve Instruments that are not easlly adapted for personal or mlcroenvlronmental monltorlng. For largescale monltorlng projects, where spatlal varlatlons of a pollutant and human exposure assessments are of Interest, passlve sampllng devlces can provide the methodology to meet monltorlng and statlstlcal goals. Recently we developed a coatedfllter for ozone collectlonthat we usedIn a commerclally available passive sampllng devlce. Results from the ozone sampler valldatlon tests are presented. The paoslve ozone sampler used In fleld and laboratory experlments consists of a badge clip supportlng a barrel-shaped body whlch contalns two coated glass flber fllters. The principal component of the coatlngis nltrlte Ion, which Inthe presenceof ozone Is oxldlzed NOS- 0 2 . to nltrate Ion on the fllter medlum, N01- 0 3 After sample collectlon, the filters are extractedwlth ultrapure water and analyzed for nltrate Ion by Ion chromatography. The results from laboratory and fleld valldatlon tests lndlcate excellent agreementbetweenthe paslve method and standard ozone monltorlng technlques. We have determined that relative humldlty (ranglng from 10 to 80%) and temperature (ranglng from 0 to 40 "C)at typlcal amblent ozone levels (40-100 ppb) do not Influence sampler performance. Face veloclty and sampler orlentationwith respect to wlnd dlrectlon affected the sampler's collectlon rate of ozone. By using a protectlve cup which acts as both a wlnd screen and a raln cover, we were able to obtain a constant collectlon rate over a wlde range of wlnd speeds.
+
-
+
INTRODUCTION Ozone is an atmospheric oxidant formed through photochemicalreactions of volatile organic compounds and nitrogen oxides. Although a great deal of effort has been aimed at decreasing emissions of these ozone precursors, ambient concentrations of ozone have only decreased approximately 10% over the last decade.l Daily maximum 1 h ozone concentrations can range from 50 to 300ppb, often exceeding the National Ambient Air Quality Standard (NAAQS) of 120 ppb. In fact, more than half of the US. population reside in areas that are out of compliance with the current NAAQS for ozone.2 At these ambient concentrations, ozone exposure can cause respiratory health effects, including changes in lung capacity, flow resistance, and epithelial permeability.2 It was traditionally thought that indoor ozone concentrations would be negligible due to rapid depletion on indoor surfaces. However, recent studies suggest ozone is present indoors and it may represent a significant fraction of a person's + Current address: Air Pollution Division, Yokohama City Research Institute of Environmental Science, Yokohama, Japan. f Current address: US. Environmental Protection Agency, Atmospheric Research & Exposure Assessment Laboratory, Research Triangle Park, NC 27711. (1)U.S.Environmental Protection Agency "National Air Quality and Emissions Trends Report," EPA-450/4-89-001, 1989. (2) Lippmann, M.J. Air Waste Manage. Assoc. 1989,39,672-94. 0003-2700/93/0365-0209$04.00/0
total ozone e x p ~ s u r e .Indoor ~ ~ concentrations were found to be as high as 60-80% of the outdoor concentration (in an art gallery and in officebuildings) and to depend upon building geometry, interior surfaces, ventilation system, and indoor sources, such as electrostatic precipitators and copying machines. Thus, the investigation of human health effects associated with ozone exposure requires the knowledge of both outdoor and indoor concentrations as well as measurementa of personal exposures. While there is a great deal of information about outdoor ozone concentrations, very little is known about indoor concentrations and personal exposures. One reason for this is the lack of lightweight, inexpensive, and reliable ozone monitors suitable for personal or indoor monitoring. The indoor studies discussed above required bulky instruments which can be invasive on site as well as labor intensive. Although the techniques used by previous indoor studies provided accurate data, the effort necessary to implement such studies seriously limits the scope of investigation. Passive monitoring devices offer the flexibilityto undertake more ambitious sampling schemes, which can include both indoor and personal monitoring. Nonetheless, ozone is difficult to measure on a passive monitor due to its high reactivity and conditioning characteristics and the potential for interference. While there are many passive devices to monitor other pollutants (volatile organics,NOz), particularly at occupational concentrations, passive ozone monitoring has rarely been attempted until recently. Only a few passive ozone monitors employing conventional chemical analysis techniques have been developed recently. A passive ozone monitor developed by Monn and Hangartner' is based on ozone reacting with 1,2-di(4-pyridyl)ethyleneto produce an ozonide, eventually forming an aldehyde which is determined spectrophotometrically. Although their results were promising, the evaluation presented by their research does not fully validate their sampler. There is also the possibility of interference by UV light in the method, which would limit its usefulness as a personal monitor. Grosjean and Hisham8developed a colorimetric passive ozone monitor using indigo carmine as the colorant. However, it displays a positive interference from NO2 (approximately E%), which is another important atmospheric oxidant often present in high concentrations. Kanno and Yanagisawagdeveloped a passive ozoneloxidant monitor which is based on ozone reacting with potassium iodide to liberate iodine, which is determined via constant-currentcoulometry. The drawback to this method is that it is a total oxidant monitor. It is not ozone-specificand displays a positive interference from NO2. (3)Weschler, C. J.; Naik, D. V.; Shields, H. C. J. Air Waste Manage. Assoc. 1989,39,1568-76. ( 4 ) Davies, T. D.; Ramer, B.; Kaapyzok, G.; Delany, A. C. J.Air Waste Manage. Assoc. 1984,31, 135-37. (5)Allen, R. J.; Wadden, R. A. Enuiron. Res. 1982,27,136-49. (6) Allen, R. J.; Wadden, R. A.; Ross, E. D. Am. Ind. Hyg. Assoc. J . 1978,39,466-71. (7)Monn, C.; Hangartner, M. J. Air Waste Manage. Assoc. 1990,40, 357-58. (8) Grosjean, D.;Hisham, M. J. Air Waste Manage. Assoc. 1992,42, 169-73. (9)Kanno, S.;Yanagisawa, Y. Enuiron. Sci. Technol. 1992,26,744-49. 0 1993 Amerlcan Chemical Society
210
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3. FEBRUARY 1. 1993
The passive ozone sampling method presented here allows for the specific collection and measurement of ozone. Prior to the passive sampler validation tests, the composition of the coating solution was tested using active samplers, with both diffusion denuders and coated filters. Promising results prompted us to apply the method to a passive design. Laboratorytests conducted by Fehe'osuggested that nitrite collected by alkaline-coated denuders was oxidized to nitrate when ozone was present in the air sample. Further, Sickles and Hodson" noted oxidation of nitrite, presumably by ozone, on sodium carhonate-coated filters after exposure to ambient air. The nitrite-based oxidation reaction, confounding one sampling methodology by producingnitrate, became the focus in developing an ozone-specificsampling medium. The solution used to coat the collecting filters includes sodium and potaasium salts of nitrite and carbonate, respectively, in a solution of glycerol, methanol, and water. In the presence of ozone, the nitrite ions are oxidized to nitrate ions:
NO;
+ 0,-NO, + 0,
Results from collection efficiency tests showed ozone reacts optimally with nitrite when the nitrite and carhenate come from salts of differentmetals. This may he explained hy the fact that the mixed sodiumipotassium crystals formed on the coated glass fiher filter collection medium are more hygroscopic. Increasing the number of water molecules a t the surface of the crystals enhances the oxidation reaction of nitrite by ozone. For this reason, the hygroscopic compound glycerol is also added to the solution. Therefore, we speculate that the reaction between ozone and the collecting medium occurs through homogeneous aqueous reactions which take place inside microscopic droplets. Solution components were chosen to ensure that oxidation would be ozone-specificand not he caused by other gaseous pollutants. Rate equations for aqueous nitrogen chemistry12 indicate that the above reaction is pH-dependent, with a rate constant that increases with pH. Thus, potassium carbonate is used to keep the collecting medium alkaline. Since the oxidation of nitrite by hydrogen peroxide is fast only a t low pH, the coating is insensitive to the presence of this other important atmospheric oxidant. Prior to the validation testa, in establishing the potential for the coating chemistry, UV light exposure tests were done withozone-exposed fitem. There werenodifferencesbetween these filters and those not exposed to UV light.'" We believe NO2 does not affect ozone measurements using this passive device, as preliminary results using active samplers showed no interferences. During development of the coating solution a t Harvard School of Public Health (HSPH), 100 ppb NO2 in zero air was forced through a filter pack prepared with the nitrite-based coating solution and through a filter pack prepared with a sodium carbonate and glycerol coating solution. The slight reductions in NO2 downstream from each filter pack were quantitatively identical, indicative of no reaction of NO?with nitrite on the filter medium.13 Later, in separate evaluation tests of the active sampler conducted by the U.S.EPA, no difference was seen hetween the active sampler and a continuous analyzer when sampling 30 ppb ozone in the presence of 76 pph NO2 in a testchamber?4 (10) Febo,A,; De Santis, F.; Perrino, C. In Proceedings of PhysicoChemical Behaviour of Atmospheric Pollutants; 14th European S p posium, Strew, Italy, 1986; Angeletti, G.,Restelli, G., Eds.; D. Reidel: Dordrecht, 1986; pp 121-125. (11) Sickles,J. E.;Hodson, L. L. At-. Enuiron. 1989.23, 2321. (12) Seinfeld, J. H. Atmospheric Chemistry ond Physics of Air Pollution; John Wley & Sons: New York, 1986. (13) Koutrakis, P.;Wolfson, d. M.; Slatm, J. Unpublished results. (14) Koutrakis, P.;Mulik, J. D. Unpublished results.
e ....
w i r e Nor,, screen C ~ O t O dm t s r
\oitf"'i:"
End=*!,
Flgm 1. Exploded diagram of the Qawa passive sampler.
In both of these cases, the amount of NO2 contacting the activelysampledcoatedfilterwasordersofmagnitudegreater than that expected for the filter in a passive device. Validation tests in a controlled environment are crucial to understanding passive sampler response and the limitations of its use. Brown15 and PalmeslB provided the primary guidelines for our passive sampler test design. Ozone diffusion and reaction on the filter can he affected hy various sampling conditions which must be simulated in the laboratory. As an oxidant, ozone reacts with surfaces near the filters. Also, passive samplers can he highly affected by variation in wind face velocities. Some passive samplers are inappropriate for the lower exposures encountered in an ambient, rather than an occupational, environment. These issuesmust be consideredwhendesigningthevalidationtests and interpreting their results. The development and use of this ozone sampling monitor providesuswiththepotentialto (1)assesspersonalexposures to ozone, (2) monitor a variety of environments (confmed workspaces, remote sites) which may be inaccessible to standardmonitoringtechniques,and (3) measureozoneusing simple and relatively inexpensive techniques.
THEORY The sampling mechanism in passive devices is diffusion of the gaseous pollutant through some diffusion zone to the collectionmedium. The thcoreticalcolledionrateforpassive devices, defined by DAIL, is given by Fick's First Law of Diffusion: DA J A = -C L where J = mass flux of m n e [pg/cm2sl, D = diffusion coefficient [cm2/sl,A = cross-sectional area of diffusion zone [cm21,L = length of diffusion zone [em], and C = ambient ozone concentration [pg/cm31. For the Ogawa passive sampler, the diffusion zone is defmed as the volume of the holes drilled into both diffusion endcaps. Thus, the theoretical collection rate is 24.5 cm3/min. This value was a point of reference for comparison with the collection rates observed in the validation tests. EXPERIMENTAL SECTION All validation tests were performed at HSPH unlesa otherwise noted. Allsampleswereanal~datHSPHregardlesaof~xposure location. Each test run consisted of 3-6 colocated passive samplers. Passive Sampling Device. The passive sampler (Ogawa & Co., USA, Inc., Pompano Beach, FL) consists of a cylindrical polymerbody(2-cmdiameterX 3cm)andaplasticpin-dipholder (4X 3 cm), shown in Figure 1. There are two cavities on the ends of the cylinder, each of which hold one coated filter between two stainless steel screens. Because the core of the body is solid, each cavity is isolated from the other. The diffusion harrier endcaps hold the screens and filters in place by friction fit. Prior (15)Bmm, R. H.; Harvey, R. P.; Purnell, C. J.; Saundsrs, K. J. Am.
Znd. Hve. Aaane. ~
~~~~1~
.I
1984.45. fi7-75.
(16) Palmes, E. D.; Lindenboom, R. H. Anal. Chem. 1979,51,24C+l.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3. FEBRUARY 1, lQQ3 211
to exposure, the assembled sampler is sealed in a plastic ziplock bag and placed in a polystyrene bottle. Filter and SamplerPreparation. Glass fiber filters (14mm, grade 30, Schleicher & Schuell) were successively soaked in concentrated solutions of nitric acid (2 h), chromic and sulfuric acid (1h), hydrochloric acid (0.5 h), and 10 N sodium hydroxide (1 h). Between soakings, they received several rinses with ultrapure water (Millipore Milli-Q). The fiters were placed on a porcelain funnel under vacuum for a final methanol rinse, followed by room air until dry. Each filter was coated with 100 pL of coating solution and dried in a glovebox purged with air filtered through potassium permanganate-coated alumina,activated charcoal,and silicagel. Passive sampler parts were cleaned with ultrapure water and dried. The samplers were assembled with coated filters inside the glovebox. Laboratory Chamber Tests. A rectangular Plexiglas chamber (94 X 41 X 53 cm, L X W X H)was used for the majority of these testa. A small fan controlled by a voltage transformer ensured good mixing within the chamber. Chamber access was via a sliding Plexiglas panel with ports drilled for tubing and a relativehumidity/temperature probe (VaisalaModelHMP 113Y). Tubing from the ozone generator was positioned immediately downwind of the fan. Samplers were clipped to strings hanging from the chamber ceiling 13-20 cm apart and at least 15cm from the chamber walls. For a few testa, a smaller glass aquarium (48 X 25 X 30 cm, L X W x H)was used in the same manner as the Plexiglas chamber. Before each test, chamber interiors conditioned and equilibrated to the target ozone concentration. Typically,this required 3-4 h. Placing samplers in the exposure chambers caused the interior ozone concentration to drop, but it would usually reequilibrate within 3 h. A subset of chamber testa at high ozone concentrations were performed by Los Amigos Research and Education Institute (Downey, CA). Their walk-in (93-m3)stainless steel chamber operated at 10air changes/h to ensure mixing and equilibration. Face Velocity Effects. A wind tunnel was constructed from two concentric Plexiglas cylinders, the smaller (21-cm diameter x 1.52-m length) centered within the larger (29-cm diameter X 1.85-m length). The test section inside the smaller cylinder was upwind of a fan which circulated the air in a closed loop, down through the test sectionwith return via the annular space between the cylinders. Individual probe porta were drilled through both cylinders for all tubing, the relative humidity/temperature probe, and an anemometer (Kurz Digital Air Velocity Meter Model 1440). Maximum wind velocity for the wind tunnel was nominally 300 cm/s. A removable plate upwind of the test section provided access to slide a sampling rack into the internal cylinder. In orientation wind velocity testa, a group of four samplers were clipped to the rack to form a cross-sectionalplane (top, bottom, left, and right). They were located 1m downwind from the entrance edge of the internal cylinder. In some testa, the samplers were protected with an inverted 16-02. polypropylenejar (8-cmdiameter at the mouth, suspended sampler barrel centered 4 cm abovethe mouth). Inside is a Teflon tubing-encased metal cross rod. In wind velocity tests using protective cups, three cups were located downwind at distances of 0.52, 0.83, and 1.13 m,respectively. Pretest conditioningand reequilibrationbehavior after sampler placement were similar to the laboratory chamber testa. Equilibration was always done with the fan running. The fan was off during the brief period of sampler placement. Humidity Effects. The wind tunnel was used for the relative humidity testa with the samplers clipped into protective cups. Water vapor was added to the air in the return annulus. A syringe pump forced ultrapure water through fine tubing onto paper toweling taped to the internal cylinder. The toweling wicked the water and functioned as an evaporation surface for excellent relative humidity control. The wind tunnel was given 12 h to equilibrate to relative humidity level before a test. Temperature Effects. A cylindrical polyethylene container (20-cm diameter, 19-cm height) was chosen for the temperature experimenta due to the size limitations of the incubator (Tenney
EngineeringProportioNull1300 Series). Alow-speed fan ensured good mixing and limited turbulence within the small interior space. Tubing from the ozone generator was positioned immediately downwind of the fan. This tubing was coiled within the incubator to thermally equilibrate the ozonated air before it entered the container. The samplers, suspended from a metal wire framework,were nominally equidistant from each other and the container walls. Pretest conditioningand reequilibrationbehavior after sampler placement were similarto the laboratory chamber testa. Because of the preheat coil, there was little temperature fluctuation at the test's beginning. Field Exposure Testing. Field testing was performed at the U.S. EPA Environmental Monitoring and Assessment Program (EMAP) test site, located at Prince Edward, VA. T w o passive Ozone samplers were housed under each of three A-frameshelters on top of a 10-m "tip" tower. Instrumentation. A Thermo Environmental Instruments Model 49PS Ultraviolet Photometric Ozone Calibrator served as a calibrating unit and an ozone generator for the laboratory experiments. The calibrator was periodicallycalibrated against a Dasibi Model 1003 PC UV photometer configured as a U.S. EPA "Primary Standard".17 Two continuous ozone monitors were used in the laboratory testa, a Monitor Labs Ozone Analyzer Model 8410 using a chemiluminescent method and a Thermo Environmental Instruments Model 49 Ultraviolet Photometric Ozone Analyzer. FEP Teflon tubing was used throughout all exposure systems. Inline PTFE Teflon filters were placed on the sample input to the ozone monitors. Before and after each test, multipoint calibrations were performed with the filters in place. The output flowrate from the 49PS ozone generator varied depending on the conditioning demand of an exposure system, the target ozone concentration, and the need to maintain a low positive pressure within an exposure system. Strip-chart recorders recorded ozone concentration, relative humidity, and temperature. Face velocity and relative humidity testa were run with data acquisition software (ACQUIRE and PROCESS, 1986-1989versions,CommonwealthEnvironmental, Inc., Boston, MA) supported on an IBM-clonepersonal computer. For other testa, the true mean ozone concentration was calculated manually from the chart recorder trace. The mean concentration included the ozone level recovery time at the beginning of each test, which was short relative to the total exposure time. At the EMAP field site, a Thermo Environmental Instruments Model 49-103Ultraviolet Photometric Ozone Analyzer was used for continuous measurement. Precision checks were performed every few days. Multipoint calibrations using a "Transfer Standard"" Model 49-103 were performed quarterly. Final 1-h integrated ozone concentrations were provided by the U.S. EPA. Sampling and Analysis. Except during exposure testa, all samplers were kept in ziplock bags inside bottles. Exposed samplers and blanks were disassembled in the glovebox. The sampler's two coated filters were transferred to an 8-mL polyethylene bottle containing 5 mL of ultrapure water and sonicated for 15min. These sample extracts were stored at 5 OC (for typically less than 1 week) and later analyzed for nitrate concentration by ion chromatography (Dionex Model 2000i equipped with a conductivity detector). At least three blank samplers were prepared with each group of test samplers. Blanks were stored at room temperature conditions during the test. For each test, the mean nitrate concentration of the blank samplers was subtracted from the exposed sample value to obtain the net nitrate concentration.
RESULTS AND DISCUSSION There were two phases of the laboratory validation testa. The first phase was to test the effect on the collection rate of varying exposure times and ozone concentrations. This was done by a test series in the rectangular exposure chambers. (17) 'Transfer Standards for Calibration of Air Monitoring Analyzers for Ozone," Technical AssistanceDocument EPA-600/4-79-0&6, September 1979.
212
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3,FEBRUARY 1, 1993
Effectively, the passive monitor was calibrated by defining its collection rate under specific test conditions. Laboratory chamber tests were designed to meet the combinations of exposure times and concentrations the sampler would most likely encounter during initial ambient use. It was expected that the reference collectionrate from the chamber tests would be a constant value, and that with further investigation, it would be applicable to sampling conditions found in the field. The second phase of laboratory testing identified the potential effects on collection rate behavior from such parameters as face velocity, relative humidity, and temperature. These tests were conducted at ozone levels ranging from 40 to 100 ppb, again in the range of typical ambient concentrations. Also, a similar amount of nitrite ion was converted in each test, nominally equivalent to 2000 ppbsh (50 ppb ozone at 40 h). Field exposure tests began concurrently with the laboratory chamber tests. Coated filters prepared from the same batch were tested in both places, and their collection rates were compared. Precision within each set of tests was good, but the collection rates for outdoor field testa were markedly higher than the rates for laboratory chamber tests. Since passive samplers are integrated monitors, it was necessary to observe sampler performance over a dynamic range of concentration x exposure time. Combining the laboratory chamber and field exposure data filled a range from 400 ppb.h to 12 OOO ppbh. Data Analysis. The sampler was exposed to an average ozone concentration determined by the following equation:
where: Cp~ss= average ozone concentration, passive monitor [ppb], M = net NO3- concentration [pg/mL], V = extraction volume = 5 [mL], MW, = molecular weight [pg/pmoll, S = collection rate [cm3/minl~[m3/106 cm31, K = 0.0409 bmol/ ppb-m3], and T = sampling time [min] and where the numerator is the mass of ozone collected and the denominator is the “volume”of the sampled air, with unit conversionfactors (the value of K assumes 25 “C and 1atm). For each sample, all quantities were known except the sampler collection rate, S. The experimental collection rate, SE,is determined empirically using the following equation:
70-
7 0
7 6050 50-
40-
::V’, 1 30-
0 0
10
20
30
,
I
,
40
50
60
70
UV-Continuous Monitor [ppb 031
Figure 2. Laboratoryexposure chamber tests: Comparlson of passlve
samplers to a standard continuous monitor.
Odeg x
Cup
theoretically predicted rate (24.5 cm3/min), which may be a result of the influence of starvation effects. Figure 2 compares the HSPH laboratory chamber tests with the standard continuous monitor concentrations. The coefficient of variation (CV) for these tests is *9.8%. Figure 2 is shown with error bounds of *20%, where 15 out of 16 tests predicted ozone levels within 20% of the actual concentrations. For the higher concentration Loa Amigos laboratory tests, the three sampler groups predicted ozone levels within 7% of the actual concentrations. Face Velocity Effects. The first set of tests was performed with the diffusion endcaps of the passive samplers MV(MWoJMW,o,) SE = (3) oriented Oo to the wind direction (the wind was blown over CTRUET the surface of the endcaps, with the plane of the flat plastic where: SE= experimental collectionrate [cm3/min]and C ~ U E holder perpendicular to wind direction). In this orientation, = true mean ozone concentration [pg/cm31. The “true mean the mean collectionrates increased exponentially (21-46 cm3/ ozone Concentration” was determined by averaging the data min) with velocity (18-270 cm/s), as shown in Figure 3. The from the standard ozone continuous monitor. Actual colthree highest velocity tests displayed an increase in the CV lection rates were determined for all exposed samples using to greater than 10%. eq 3. A mean collection rate was calculated for each group The mathematical relationship between wind speed and of colocated samplers in a single test. By averaging the collection rate at this orientation was not determined because colocated mean collection rates, a reference collection rate, applying this information to actual field use of the sampler SR,was obtained. The average ozone concentration for each would be impractical and hopelessly complicated. However, sampler was calculated from eq 2 using the reference collection to understand the essence of these results, consider that the rate. boundary layer over the endcaps will be part of the ozone diffusion path, L. Therefore, with increasing wind speeds, L Laboratory ChamberTests. The chamber test exposures decreases, and from eq 1, the collection rate increases. ranged from 17 to 62 ppb ozone for 24-167 h. Higher A test series at the opposite orientation, 90° (wind blowing concentration tests (approximately 270 ppb ozone) with directly into one diffusion endcap), was run to understand exposure times between 11 and 48 h were performed a t the this sampler geometry’s face velocity effects a t two extremes Los Amigos laboratory. The reference collection rate from of orientation. For 90° orientation, the low collection rates the colocated means, 18.1 1.9 cm3/min,was used to calculate (20-28 cm3/min), although exhibiting high CV values at the the ozone concentrations for both sets of laboratory chamber higher velocities, were similar to collection rates at low wind experiments. Note that this rate is slightly lower than the
*
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, 1993
-8
80-
n n
70-
4
60-
-&
E, v)
1
50
e
-b
50-
40-
30-
m
9
'I
213
(I)
.-8
3020'O1
/
o Y v 0
20-
v)
10
I
I
,
20 30 40 50 60 70 80 UV-Continuous Monitor [ppb 031
90
II
I a 10-
100
Flgure 4. Wind tunnel velocity experiments: Comparlson of passive samplers to a standard continuous monitor for samplers which were clipped into protectivecups (1) at velocities greater than 25 cm/s, and (2) between 10% and 80% relative humidii.
speed Oo orientation tests. Separate analysis of the front and back filters revealed only a slight difference (nominally 2 cc/min) in upwind and downwind collection rates. These results indicate that convective transport of air into the upwind diffusion barrier was minimal, if it indeed occurred. A possible explanation for the collection rate relative stability shown in the 90° tests is that a stagnation point exists at the upwind end of the sampler, a blunt body in the windstream. The velocity at a stagnation point is zero. Further, it is possible that the turbulent eddies in the wake directly adjacent to the downwind endcaps do not change dramatically within the range of wind speeds tested. Once the face velocity effects for the unprotected samplers were determined, a protective cup was used to attempt to reduce these effects. The results from this test series, also shown in Figure 3, demonstrate the protective cup's effectiveness in attenuating the effects of high wind velocities. The testa, performed at 50 % nominal relative humidity from 40 to 90 ppb ozone, had a reference collection rate of 21.1 f 1.9 cm3/min for the wind speeds ranging from 25 to 295 cm/s. The data suggest a minimum wind speed of 25 cm/s to avoid starvation effects. The cup geometry makes wind direction relatively independentof sampler placement within it. Ozone concentrations for the protected samplers are compared with the standard ozone monitor in Figure 4. All tests predicted ozone levels within 20% of the actual concentrations. Unfortunately, using the polypropylene cups for outdoor passive monitoring was unsuccessful. In an ozone monitoring study in State College, PA, during summer 1991, monitors clipped into the cups demonstrated a high and variable positive interference. Its cause is undetermined, although because the cup is translucent, overheating during sun exposure is suspected. Side-by-side monitoring was done with a second passive ozone monitor protected under opaque white PVC pipe caps which provided successful State College results.ls Current outdoor monitoring conducted by HSPH uses PVC covers with geometry similar to the wind tunneltested polypropylene cup. Results from these studies may indicate whether the protective cup collection rate can be applied to field use. Humidity Effects. Relative humidity tests were performed at a fixed wind speed (170 cm/s). Two teats were performedat 10-20%, twoat7+80%,andoneat48% relative humidity. The collection rates were unaffected by relative humidity and averaged to 20.9 f 0.4 cm3/min. (18)Liu, L.-J. S.;Koutrakis, P.; Suh, H. H.; Mulik, J. D.; Burton, R. F. Use of personal measurements for ozone exposure assessment - A pilot study. Submitted to J . Air Waste Manage. Assoc.
0
10
20 30 40 50 UV-Continuous Monitor [ppb 031
60
Flgure 5. EMAP field tests: comparison of passlve samplers to a standard continuous monitor.
Temperature Effects. The passive sampler exhibited no change in collection rate due to temperature variation from 0 to 40 "C. Test exposures were for approximately 36 h with an average 41-53 ppb ozone and approximately 1025% relative humidity. The collection rates averaged to 19.5 f 0.85 cm3/minwhich agrees, within experimental error, with the mean colocated collectionrates reported for the laboratory chamber testa at room temperature. Copollutant Interferences. Other environmental pollutants may adversely affect this ozone sampling method. For example, nitric acid gas is collected simultaneously on the alkaline filters during sampling. However, under typical ambient conditions this positive interference probably represents less than 5% of the nitrate formed during the nitrite/ ozone reaction.l9 The possible reaction of ozone with organic aerosols collected on the coated filters may result in an underestimate of the ozone concentration. However, due to the amount of nitrite on the coated filters relative to the probable amount of organic aerosols collected, this interference is also expected to be insignificant. Interference from particles is expected to be negligible because they have diffusion coefficientsorders of magnitude less than gases. Interference from sulfur dioxide is not expected because this gas is not a strong oxidant. Blank correcting the samples accounts for any negligible interference effecta from oxygen, a weak oxidant, as well as from nitrate formed due to filter aging. Peroxyacetyl nitrate (PAN), as a strong oxidant, could oxidize nitrite to nitrate. Since ambient concentrations of PAN are typically 10-20 times smaller than ozone concentrations, significant interference in most locations is not expected.20 Ozone's diffusion coefficient is about 1.6 times that of PAN, further minimizing the PAN'S interference. Field Exposure Testing. Each sampler group was exposed for 7 days. Weekly data were collected from November 13, 1990, to October 15, 1991. The reference collection rate, 29.0 f 2.7 cm3/min was significantly higher than that calculated for the laboratory chamber testa. Average ozone concentrations calculated with the reference collection rate are compared with the continuous ozone monitor results in Figures 5 and 6. Data from the February 12 sampling week were not averaged into the reference collection rate because rain (19) Koutrakis, P.; Mueller, P. In Proceedings of the 82nd Annual Meeting of the Air and Waste Management Association; 1989, Paper NO.89-71.4. (20)Finlayson-Pitta,B. J.; Pitts, J. N. Atmospheric Chemistry; John Wiley & Sons: New York, 1986.
214
ANALYTICAL CHEMISTRY, VOL. 65, NO. 3, FEBRUARY 1, 1993
01, I , I , , I r 11/13/90 1/8/91 12111 215 I
I
I
I
I
I
I
I
I
,
I
I
I
I
I
I
I
I
,
, I
I
4/30
3/5 4/2
,
,
I
I !
6/25 5/28
'
I
7/23
'
I
I
"
8/20
'
3
'"1
9/17
Sampling Week Figure 6. EMAP field tests: comparison of passlve samplers to a standard continuous monitor for 11 months of sampling.
contamination was suspected. Also, a criterion of 90% data capture from the continuous monitor over the sampling period was met by only 38 of the 44 sampling weeks. Figure 5 illustrates close agreement of the passive sampler response with that of the standard continuousozone monitor. The CV value is f8.7 % The results show excellent linearity of the samplingtechnique and suggest that after establishing a collection rate for the sampler under given exposure conditions, reasonable accuracy can be expected from the analytical technique. Predicted ozone levels were within 20% of the actual concentrations for the 40 out of 43 sampling weeks where the continuous ozone monitor was funtioning for at least 50% of the time. Limit of Detection (LOD) and Sampler Capacity. The LOD was defined as 3 times the standard deviation of the blanks in a batch of filters. Using the most conservative estimate, i.e. the lowest reference collection rate of 18.1 cm3/ min, the LOD is approximately 201 ppb-h, or 17 ppb at 12 h and 8 ppb at 24 h. The sample capacity was defined as the conversion of 5 % of the total nitrite ion on the coated filters to nitrate ion, using the higher reference collection rate, 29.0 cm3/min. The capacity was approximately 19 865 ppbeh, or 1655at 12 hand 828ppb at 24 h. Note that the 5 % conversion is a conservative estimate. Storage Stability. Reanalysis of some EMAP field samples show a maximum change of f0.5 ppm nitrate concentration at 10 weeks after filter extraction, which corresponds to f 4 ppb ozone. These results suggest that the samples are reasonably stable in solution.
.
The face velocity and orientation experimentssuggest that, barring the influence from other unknown interferences, the dramatic difference between laboratory chamber and field collection rates is primarily due to face velocity effects. The influence of face velocity on collection rates has been observed in other passive monitor validation studies. In fact, the Oo orientation results display a trend similar to the Palmes tube 45O wind velocity tests.*l After attenuation from the protective cup, the wind velocity results are similar to Lautenberger's,zzdisplaying starvation effects at low velocities until some threshold velocity, after which the collection rate is stable. Since the field samples were minimally protected, they were exposed directly to varying wind velocities and directions typically encountered outdoors. Conversely, lower wind speeds encountered in the laboratory chamber tests were employed to reduce concentration gradients and starvation effects. Future outdoor field work is necessary to assess the effectiveness of the protective cup, which attenuated face velocity effects in wind tunnel tests. Because of starvation effects at low face velocities, there are risks in using the sampler indoors, where still air w i l l frequently be encountered. It is possible that the reference collection rate derived from the chamber testa could be applied to indoor monitoring, but further experimental work simulating indoor microenvironments is necessary. Alternatively, the passive monitor could be used indoors if placed in the windstream of a small fan. Then, a collection rate could be estimated from the Oo orientation tests in Figure 3, based on a measured face velocity. Because of concern that direct moisture may affect performance, future studies should include collection rate stability in fog or mist conditions. With respect to copollutants, tests to investigate the interference of PAN should be undertaken.
ACKNOWLEDGMENT
The difference between laboratory chamber test and field reference collection rates, 18.1f 1.9cm3/min(20observations) and 29.0 f 2.7 cm3/min (37 observations), respectively, prevented applying the chamber test collection rate to the monitor during field use, as originally intended. By using separate reference collection rates, however, excellent agreement was demonstrated between the sampler response and a standard ozone analyzer. The EMAP results indicate these samplers are well suited for outdoor applications of 1day or longer.
We would like to thank those people who have contributed and continue to work on this project: Carl Lamborg (HSPH), Eileen Abt (HSPH), Martin Barthel (HSPH), Stella Papasawa (HSPH), Benjamin Rosenthal (HSPH), Keith Kronmiller and Dennis Williams (ManTech Environmental Services, formerly NSI Technology Services), and Karen Anderson (Loa Amigos Research and Education Institute), and James Slater (University of Steubenville) for his participation in the original tests. Thanks to Yukio Yanagisawa for his wind tunnel design. Also, thanks to Jerry Varns and Eugene Brooks (U.S. EPA) for input concerning the EMAP field testing. In addition we would like to thank Hiroshi Ogawa and Donald Schaeffer of Ogawa & Co., USA, Inc., Pompano Beach, FL, who supported us throughout this project. We would also like to acknowledge MarthaRichmond (Health Effects Institute) and Robert Burton (U.S. EPA) for their collaborative effects in bringing this project together. Finally, funding was provided by the Health Effects Institute (Contract 019326-01) for the laboratory validation testing and by the US. EPA under a cooperativeagreement for field testing. A patent application has been fiied for the fiiter coating method tested in this project (Serial No.07/662,164; filed: February 28,1991). Although the information in this document has been funded wholly or in part by the United States Environmental Protection Agency under Cooperative Agreement CR 816740-01 to the Harvard School of Public Health, it does not necessarily reflect the views of the Agency, and no official endorsement should be inferred.
(21)Palmes, E. D.; Gunnison, A. F.; DiMattio, J.; Tomczyk, C. Am. Ind. Hyg. Assoc. J. 1976, 37, 570-77. (22) Lautenberger, W. J.; Kring, E. V.; Morello, J. A. Am. Ind. Hyg. Assoc. J. 1976, 37, 737-43.
RECEIVED for review April 6, 1992. Revised manuscript received September 16, 1992. Accepted November 2, 1992.
CONCLUSIONS
