Environ. Sci. Technol. 2000, 34, 3031-3040
A New, Portable, Real-Time Ozone Monitor DOUGLAS R. BLACK AND ROBERT A. HARLEY* Department of Civil and Environmental Engineering, University of California, Berkeley, California 94720-1710 SUSANNE V. HERING AND MARK R. STOLZENBURG Aerosol Dynamics Inc., 2329 Fourth Street, Berkeley, California 94710
New methods are needed to assess microenvironmental and personal exposures to ozone. A real-time ozone sensor using a piezoelectric quartz crystal has been developed and evaluated. The crystal is coated with polybutadiene which reacts irreversibly with ozone, resulting in a mass increase on the surface of the crystal which, in turn, alters its natural oscillation frequency. The rate of change in frequency is proportional to the concentration of ozone and is recorded by a datalogger with a time resolution of 10 min. Experiments were conducted to investigate sensor response, useful lifetime, and potential interferences. Monitor components such as a pump, power supply, and datalogger were arranged to fit within a compact enclosure. Five prototype ozone monitors were assembled and evaluated in the laboratory and in a field study of two offices and two residences in Southern California. Field measurements made with two quartz crystal-based ozone monitors exhibited precision ranging from 6 to 27% (13% on average) and bias ranging from -37 to 2% (-13% on average) when compared to a reference method (UV photometric detection) for real-time ozone measurement. The monitor has a sensitivity of 3 ppb ozone (limit of detection for S/N ) 3) and a useful lifetime of ∼20 h at an average ozone concentration of 50 ppb (1000 ppb-h). This prototype monitor is smaller, lighter, and more affordable than existing UV ozone analyzers and with further development may be suitable for real-time personal ozone monitoring.
1. Introduction Ozone is a common air pollutant that is known to have adverse effects on human health. There is extensive evidence of acute effects at concentrations within the typical range of ambient levels (1, 2). Limited data also suggest chronic effects (3, 4). Both the epidemiological evaluation of the adverse human heath effects associated with ozone exposure and the development of effective ozone control strategies require monitoring in all environments in which people are exposed to ozone. Ozone concentrations are routinely measured outdoors. However, in-transit and indoor exposures may dominate the total ozone exposure and, perhaps, total ozone dose. Time activity studies indicate that California residents spend an average of ∼90% of all time indoors or in-transit (5). Therefore, detailed knowledge of indoor and in-transit exposure levels is needed to assess acute ozone exposures accurately. Neither of these environments is well-represented 10.1021/es991453a CCC: $19.00 Published on Web 06/09/2000
2000 American Chemical Society
by outdoor measurements of ozone concentrations, and indoor-outdoor concentration ratios are highly variable. Indoor occupational exposures from copy machines and computer laser printers also may be significant (6, 7). With current ozone monitoring data it is difficult to estimate individual exposure levels. Indoor ozone concentrations are not measured routinely. Special studies, such as those of Avol et al. (8), Weschler et al. (9), Nazaroff and Cass (10), and Shair and Heitner (11), indicate that indoor levels range from 10 to 80% of outdoor levels, depending on the type and rate of ventilation, among other factors. Knowledge of indoor ozone concentrations is also needed to determine the indoor rates of chemical reactions (e.g., ozone-olefin reactions) that form other secondary pollutants (12, 13). Because ozone is formed photochemically in the atmosphere, there is a strong diurnal cycle in ambient ozone concentrations. Research has shown that an individual’s response to ozone is more strongly associated with the shape of the ozone concentration profile than the total exposure dose (14). Therefore, high temporal resolution is useful for both microenvironmental and personal ozone monitoring. To quantify the exposure of a subject population to ozone, small and unobtrusive, real-time, and inexpensive indoor and personal monitors are needed. Improved time resolution in tracking ozone concentrations in indoor environments is needed to further develop and validate exposure assessment models and to develop cost-effective strategies to reduce ozone exposure (15, 16). Standard methods for measuring ozone in ambient air include ultraviolet (UV) photometric detection and chemiluminescence (17). These monitors, while well-tested and reliable, are expensive, large, and noisy. As such, they are not ideal for microenvironmental sampling and are unsuitable for personal monitoring. As an alternative, diffusive and pumped ozone badges have been developed for microenvironmental and personal monitoring. A diffusive ozone badge has been developed by Koutrakis et al. (18). Sodium nitrite impregnated on a filter is oxidized to nitrate in the presence of ozone. The sampler measures 4 × 3.4 cm and is easily clipped onto the subject’s shirt. The detection limit of the badge, defined as three standard deviations of the blanks, is 200 ppb-h, which corresponds to 25 ppb over 8 h of sampling. The total capacity of the badge, defined as 5% of the conversion of the total nitrite ion, is 20 000 ppb-h. One disadvantage is that the effective sampling rate can vary by a factor of 2 depending on ambient air face velocity, as shown by laboratory wind tunnel tests (18). A timed exposure diffusion (TED) ozone sampler was developed for and used in a children’s health study conducted in Southern California (19). The TED sampler provides controlled airflow across two sodium nitrite-impregnated filters. The sampler uses a 7-day timer that can be set to sample for specific hours of the day on specific days of the week. The TED sampler responded to nitric acid and hydrogen peroxide in laboratory tests (19). In field evaluations the sampler measured ozone with a +6% bias and (12% precision on average compared to a continuous monitor (19). Because of the limitations of the diffusive badge design, in particular the influence of face velocity on measured ozone, Koutrakis and co-workers have developed an active sampler consisting of a pumped hollow-tube diffusion denuder. The hollow etched Pyrex tube is 10 cm long, has an inside diameter of 1.3 cm, is pumped at 65 mL/min, and is coated with the same sodium nitrite reagent used in the diffusive badge design described above. Under laboratory conditions the sampler demonstrated a limit of detection of 10 ppb-h and measured VOL. 34, NO. 14, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3031
ozone with a precision of 4.1-6.5% and a bias ranging from 0 to -6% when compared to UV photometric measurements (20). A limitation of all of the above badge methods is that they only yield a single integrated ozone measurement over averaging times that are long relative to time scales over which ozone concentrations change. While such measurements may be useful in calculating long-term average ozone exposures, they do not supply sufficient information regarding ozone peaks for estimating acute ozone exposures. Most badge samplers are sensitive to face velocity (i.e., measured concentrations can change with wind speed and how fast the subject is moving). Furthermore, the time and expense involved in extracting and analyzing the badge samples can be high. An alternate approach to ozone measurement involves use of a chemically coated piezoelectric crystal. When an electric field is applied across a piezoelectric crystal, it mechanically vibrates at a stable resonant frequency. Massbased sensing using piezoelectric crystals is possible because mass changes on the surface of the crystal alter the resonant frequency. For example, piezoelectric crystals have been used to measure airborne particulate matter mass concentrations (21, 22). Gaseous pollutant concentrations have been measured using chemically coated piezoelectric crystals, for a variety of compounds including ammonia, formaldehyde, hydrogen cyanide, hydrogen sulfide, and phosgene, among others (23). There is one previous study (24) reporting the use of polybutadiene-coated piezoelectric crystals for ozone measurement over short periods of time (∼15 min). The objective of the present research was to develop and evaluate a piezoelectric crystal-based sensor to measure real-time ozone concentrations over longer time periods, so as to provide useful information for microenvironmental and personal ozone exposure assessment.
2. Materials and Methods 2.1. Ozone Sensor. The ozone monitor developed in this study consisted of a polybutadiene-coated piezoelectric crystal. A jet of sample air was directed at each side of the crystal, and the polybutadiene coating on the crystal reacted irreversibly with ozone. This reaction caused a mass increase on the crystal which was sensed as a shift in the beat frequency between the coated crystal and an uncoated reference crystal. The rate of change of the beat frequency was proportional to ozone concentration in the sample air. Piezoelectric Crystals. The piezoelectric material used in the present study was alpha quartz, the most widely used material for piezoelectric crystal sensors (22). The frequency characteristics of quartz crystals depend on the cut of the quartz plates. Quartz plates cut with what is known as an AT cut achieve the best mass response (22). The electric field necessary to drive the quartz crystal was applied by means of electrodes attached to each side of the crystal. Each electrode was circular with a tab that connects to a wire contact/support on one side of the crystal. The amount and distribution of the electrode material determines the resonant frequency of the crystal. In the present study AT cut quartz crystals with electrode diameters of 14 mm were used. Crystals used in this study had a resonant frequency of 10 MHz (International Crystal Manufacturing Co., Oklahoma City, OK). The beat frequency was defined as the difference between the coated crystal frequency and the uncoated crystal reference frequency. The electronic components that drive the crystal frequencies and mix the beat frequency were configured on a circuit board supplied by California Measurements, Inc. (Sierra Madre, CA). The beat frequency was generated by subtracting the sensing crystal oscillator circuit 3032
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 14, 2000
frequency from the frequency produced by the reference crystal oscillator circuit. Polybutadiene. Polybutadiene is an unsaturated polymer with one double bond for every four atoms of carbon and has been shown to be a suitable crystal coating for sensing ozone (24). Ozone reacts with the double bond to form an unstable ozonide. The ozonide breaks down into a variety of species including aldehydes and alkyl radicals (25). It is not clear what products of the ozone-polybutadiene reaction remain attached to the crystal surface, but it has been shown that the mass of the coating increases upon reaction with ozone (24). Polybutadiene used in the present study consisted of 80% 1-4 addition and 20% 1-2 addition polymers and was supplied in the form of a viscous liquid with an average molecular weight of 5000 (Aldrich Chemical Co., Milwaukee, WI). Coating Procedure. Polybutadiene was applied to the crystal with a fine-bristled artist brush, while the crystal was installed in the electronics fixture, and the beat frequency shift was measured, in units of kHz, to indicate the applied mass. An amount of coating greater than desired was spread evenly. Excess coating was then removed by wiping the brush with a lint-free tissue between brush strokes that were made in an even pattern. This method of polybutadiene application proved to be reliable and repeatable. The optimal amount of coating was determined to be 8 kHz (4 kHz on each side of the sensing crystal). Coated crystals were covered with metal caps that snapped over the base of the crystal to prevent dust contamination during transport between the coating stand and the POM sensor housing. Coated crystals were cleaned after each use so that the crystal could be reused. Used polybutadiene was removed with a cotton swab and toluene. Sample Air Flow Rate and Configuration. Jets of sample air were directed at each side of the coated sensing crystal. After impacting the sensing crystal the sample air flowed across the reference crystal. Total sample air flow rates (both jets) of 0.13 and 0.25 LPM were examined. Conditioning. To ensure that the sensor was operating in its linear range the sensor was conditioned by exposing it to 300 ppb of ozone for 40 min. 2.2. Portable Ozone Monitor (POM). The POM consisted of the sensor, filters and flow control, pump, power supply, and data logger (Figure 1). The POM prototypes measured 25 cm × 35 cm × 13 cm and weighed 5 kg. Sensor. The base of the sensor consisted of an aluminum plate with four legs (Figure 2). The printed circuit board containing the oscillation and mixing electronics that produce the beat frequency output was mounted to the underside of the base plate. A platform mounted on top of the base plate held four gold sockets, two for each crystal. The round sockets were press fit through holes in the platform, and wire leads were soldered to the sockets. The leads from the sockets passed through an opening in the base plate and connected to the circuit board. The sensing and reference crystals were covered with a rectangular aluminum housing. Sample air was directed through jets in the side of the housing aimed at each side of the sensing crystal. After sample air impacted on the sensing crystal it passed over the reference crystal before exiting through a fitting mounted in the end of the housing. Sample Air Flow Control and Filtering. Before sample air reached the sensor housing it passed through a 25 mm diameter Teflon filter (Savillex, Minnetonka, MN) and a gas dryer which consisted of a Nafion tube inside a polypropylene tube (Perma Pure, Toms River, NJ). Captured water molecules were passed from inside to outside the Nafion material allowing the Nafion to continuously remove water vapor from sample air. The filter and dryer limited sensor interferences
FIGURE 1. Portable ozone monitor (POM) component layout.
FIGURE 2. Portable ozone monitor (POM) sensor housing. that could have been caused by particle impaction and water vapor adsorption. In the POM the dryer was configured in a counter-current flow scheme in which sample air first passed through the Nafion tube before entering the sensor housing. After exiting the housing the sample air was then routed through the larger outside tube. As the dry air flowed through the larger tube it collected the water molecules that passed from the inside to the outside of the Nafion tube. After the sample air passed through the particle filter and the dryer, it was split equally between two stainless steel inlets on the sides of the sensor housing. The rate at which air flowed through the inlets was controlled by a critical orifice (stainless steel) installed between the housing exit and the dryer return inlet. In the present study a flow rate of 0.13 LPM was used unless otherwise noted. The flow rate could be varied by changing the critical orifice. To minimize ozone loss upstream of the sensing crystal all “wetted” materials were either Nafion, Teflon, or stainless steel. Pump. For microenvironmental monitoring a quiet pump was desired. The POM used a diaphragm pump (KNF Neuberger, Trenton, NJ) that could pull a vacuum of at least 15 in. of Hg at flows up to 0.8 LPM. Power Supply. The sensor circuit board (5 VDC, 50 mA) and the pump (12 VDC, 250 mA) were powered with a switching power supply (Condor DC Power Supplies, Oxnard, CA). The power supply was isolated from the metal enclosure with rubber washers that were placed between the standoffs and the enclosure. Datalogger. The frequency output of the ozone sensor was logged by a programmable datalogger (Campbell Scientific, Logan, UT; model no. CR500). A dedicated 12 VDC
rechargeable battery (Powersonic, San Diego, CA) was used to supply power to the datalogger. The datalogger was configured to measure sensor beat frequency every 5 s with a resolution of 0.2 Hz. Each minute the average of the beat frequency measurements was stored to a register along with the date and time. Before the frequency data was converted to ozone concentration measurements the frequency data was smoothed by calculating its 10 min running average. 2.3. Laboratory Testing. Response to Ozone. The POM was tested to examine its response to varying levels of ozone. Teflon tubing was used to connect the output of an ozone generator (Thermo Environmental Instruments (TECO), Franklin, MA; model 146) to the inlets of two to five POMs configured in parallel. Ozone levels were monitored simultaneously with a UV photometric ozone analyzer (Dasibi, Glendale, CA; model 1008-AH). Interference Testing. Interference testing was conducted in the laboratory with the following compounds: water vapor, nitric oxide (NO), nitrogen dioxide (NO2), nitric acid (HNO3), and toluene. Mixtures of the potential interference test gases and ozone were delivered directly from the gas generation systems (described below) to the POMs via Teflon tubing. Water vapor was generated by pumping pollutant-free air through a ground glass bubbler submerged in deionized water. Sample air containing various levels of water vapor and ozone was generated by combining ozonated air from the ozone generator with humidified air from the bubbling flask in a glass plenum. Relative humidity (RH) and temperature were measured directly upstream of the POMs with a temperature/RH probe (Vaisala, Woburn, MA; model 50Y). The POMs were exposed to varying levels of ozone and RH levels of 20 and 80%. Also, RH levels were rapidly changed from 20 to 80% and 80 to 20%. Known concentrations of NO were produced by diluting NO in nitrogen from a calibration gas cylinder with pollutantfree air. Dilution was controlled by a gas calibration system (TECO Model 146). Pollutant-free air was provided using a zero-air supply (Dasibi, Glendale, CA; model 5011-B) equipped with an activated carbon canister, a Purafil II chemisorbent canister, and a heated palladium/alumina catalyst. Ozone generated by a UV lamp in the gas calibration system was mixed with NO from a calibration gas cylinder to produce NO2. NO and NO2 concentrations were measured using a chemiluminescence analyzer (TECO Model 42). An HNO3 concentration of 40 ppb was generated using a permeation tube (Vici-Metronics, Santa Clara, CA) placed in a 35 °C oven (TECO Model 146). Similarly, a toluene concentration of 50 ppb was generated using a permeation VOL. 34, NO. 14, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3033
tube (Vici-Metronics) placed in a 35 °C oven. Concentrations of HNO3 and toluene were estimated using the manufacturer’s stated output for each permeation tube at 35 °C. 2.4. Field Testing. Offices and Residences. The POM was evaluated in the field between Oct 27 to Nov 3, 1998 in Southern California. Sampling was conducted on 1 day each at two offices located in Riverside, CA and on 2 days each at two residences in Pasadena, CA. Indoor measurements were made at each location on each day with two POMs and a UV photometric ozone monitor. A temperature and humidity probe (Vaisala, Woburn, MA; model 50Y) was placed near the monitors, and T/RH data were recorded using one of the dataloggers. Measurements of outdoor ozone concentrations were made with one POM and a UV photometric ozone monitor at the residential locations only. Both monitors were placed indoors with Teflon sample lines running out a window. The sample line inlet was positioned about 30 cm from the outside wall. A temperature and humidity probe was positioned near the sample line inlet and was recorded by the POM datalogger. Indoor and outdoor measurements were made simultaneously on each day of sampling at each residence. All sampling was conducted from 10 A.M. to 5 P.M. each day at each location. Although both residences had gas stoves with pilot lights, the stoves were not operated during field sampling. Office 1 was a cubicle (2.5 m × 3.5 m) located within a large room (32 m × 42 m × 5 m) at the University of California Riverside Center for Environmental Research and Technology. The monitors were placed on a cart with their sample inlets facing the center of the cubicle at a height of 1 m. Office 2 was a small office (2.7 m × 4.3 m × 3.7 m) located on the second floor of Pierce Hall on the UC Riverside campus. Monitors were placed on the floor along the back wall with their sample air inlets facing away from the wall. Residence 1 in Pasadena was a two story 65 m2 apartment located in the back of a larger building. All monitors were placed in the living room of the apartment on the first floor. The monitors measuring indoor air were placed on chairs in the center of the room, and the monitors sampling outdoor air were placed on the floor with sample lines running out a nearby window. Residence 2 was a single story 75 m2 duplex house located in Pasadena near the foothills of the San Gabriel Mountains. The indoor monitors were placed on the dining room table with the sample air inlets facing the center of the living/ dining room, and the monitors measuring outdoor air were placed on the floor with sample lines running out a nearby window. Strong airflow was generated in the living/dining room when windows and doors in this room were opened. The windows and doors were opened and closed periodically to cause changes in indoor ozone levels. All sampling was conducted using three POMs with 8 kHz of coating and sample air flow rates of 0.13 and 0.25 LPM. The ozone generator, UV photometric ozone monitor, and pollutant-free air source (ambient air drawn through canisters of Dri-rite and activated carbon) were configured at a laboratory near the sampling sites to condition the coated crystals and generate calibration points. POMs were calibrated at the beginning and at the end of each sampling day. Calibrations were made with ozone levels of 30, 50, and 100 ppb except those for the offices which did not include the 100 ppb calibration point. Photocopy Room. Ozone measurements were made from 9 A.M. to 5 P.M. in the 7th floor photocopy room of Davis Hall on the University of California Berkeley campus with two POMs and a UV photometric ozone monitor. Sample lines for each monitor were positioned at typical breathing height (1.5 m) and about 0.3-0.6 m behind where the photocopier operator would stand. Operators were asked to record the time of day and the number of copies made on 3034
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 14, 2000
a log sheet attached to the copier. The room had a volume of 40 m3, was mechanically ventilated, and contained one Kodak Imagesource 85 photocopier. The copier was about 1 year old, copied at a rate of 85 pages per min, and was not fitted with an exhaust filter. The copying mechanism consisted of a three wire corona primary charger, dual component dry toner, magnetic roller developer, and a heated roller fuser. 2.5. Quality Control/Quality Assurance. The ozone generator (TECO model 146) used in this research was calibrated by the Bay Area Air Quality Management District (BAAQMD) in Richmond, CA. The ozone output was calibrated using an ozone analyzer that had been calibrated with an EPA-approved ozone transfer standard. Two UV photometric ozone analyzers (Dasibi 1008-AH) used in this study were calibrated at UC Berkeley using the calibrated multi-gas calibrator and five ozone settings. Calibrations were repeated every 3 months. The ozone analyzers were also calibrated on separate occasions at the California Air Resources Board (CARB) in Sacramento, CA and at the BAAQMD in Vallejo, CA. Both calibrations were performed with an EPA-approved ozone transfer standard. Calibration of the multi-gas calibrator was repeated after the ozone analyzers were calibrated at CARB and BAAQMD.
3. Results and Discussion 3.1. Ozone Sensor. Piezoelectric Crystals. While the resonant frequency of the quartz crystals proved to be exceptionally stable over time, it varied slightly when temperature changed. A type AT crystal typically has a temperature coefficient of about 0.05 ppm per degree Celsius over the range of outdoor temperatures (22). In other words, a temperature change of 2 °C will cause a 1 Hz frequency shift with a 10 MHz crystal. Therefore in the present study, both a chemically coated sensing crystal and an uncoated reference crystal were used. Both crystals were exposed to the sample air flow and should have responded similarly to changes in air temperature, while only the coated sensing crystal responded to ozone. Sensing and reference crystals were matched to prevent sensor drift as temperature varied. Crystal pairs that had AT cut angle values (provided by the manufacturer) within 0.1 degrees were considered to be temperature matched (26). Coating Procedure. The amplitude of the vibration of the piezoelectric crystal attenuates rapidly outside the perimeter of the electrode-covered portion of the crystal (27). Thus, only the electrode was coated with polybutadiene. The amount of polybutadiene applied to the surface of the sensing crystal was measured as a shift in the sensor’s beat frequency in units of kHz. There is an electrode on each side of the quartz crystals. Thus, when both sides of the crystal were exposed to sample air, coating both electrodes of the sensing crystal with polybutadiene was found to double the response to ozone compared to coating only one electrode. The total amount of coating applied on both electrodes is reported here as a single increase in beat frequency. The mass of coating was divided equally between the two sides of the crystal. Coating Amount. The reaction of ozone with polybutadiene is irreversible; therefore, the coating and in turn the ozone sensor were found to have a limited useful lifetime. Useful lifetime of the sensor was found to be proportional to the amount of polybutadiene coating applied. Experiments also showed that the sensor response became less stable at higher coating amounts. This is most likely due to poor coupling between the outer layer of the coating and the surface of the crystal’s electrode. The maximum amount of coating that produced stable sensor response was 14 kHz. Sensor lifetime varied with coating amount, but sensor response to a given ozone concentration did not. Laboratory and field experiments were conducted with coating amounts
FIGURE 3. POM measurement when exposed to laboratory generated ozone over 24 h. ranging from 8 to 14 kHz. The mass of coating and the thickness of the coating layer on the surface of the crystal electrode were calculated using equations derived by Sauerbrey (28). Coating quantities of 8-14 kHz correspond to total masses of 25-44 µg and thicknesses of 190-330 nm on each side of the crystal. Experiments showed that coatings of 8-14 kHz correspond to sensor lifetimes of 600-1000 ppb-h. Sample Air Flow Configuration. Experiments were conducted with both parallel and impactor flow configurations. In the parallel scheme, sample air flowed parallel to the surfaces of the coated sensing crystal. With impactor flow jets of sample air were directed at each face of the sensing crystal. Impactor flow was chosen in the final sensor design, because sensor response was more repeatable. Sensor response with parallel flow appeared to be sensitive to crystal position which can vary due to slight bending of crystal supports that can occur during coating and cleaning. Sample Air Flow Rate. At a constant concentration of ozone, the number of ozone molecules that come in contact with the coated surface of the crystal is proportional to the sample air flow rate. Therefore, the sensor’s limit of detection (LOD) is related to sample air flow rate. Sensor LOD is often defined to be three times the noise level of the sensor response. As described earlier, the rate of change of the sensor’s beat frequency is proportional to the presented ozone concentration. The sensor response has units of Hz min-1. In the case of the ozone sensor presented in this study the noise level was measured as the average of the sensor response to pollutant-free air and has units of Hz min-1. Multiplying the noise level by a factor of 3 provides the LOD in units of Hz min-1. The LOD is converted to units of ppb by multiplying by a calibration factor that has units of ppb min Hz-1. The sensor calibration factor was determined by plotting ozone concentration vs sensor response for various ozone levels presented to the sensor. Experiments conducted with different sample air flow rates showed that a rate of 0.13 LPM corresponded to a sensor calibration factor of 58 ppb min Hz-1 and a sensor noise level of 0.015 Hz min-1 which results in a LOD of 3 ppb. Conditioning. Experiments showed that the POM sensor was less sensitive to initial exposure to ozone and that the initial conditioning time needed for sensor response to a constant ozone concentration to stabilize was proportional to the amount of coating. Coating amounts of 8, 10, and 14 kHz required preconditioning with an ozone concentration of 300 ppb for 25, 40, and 75 min, respectively. To decrease
the time needed for conditioning during field sampling a coating amount of 8 kHz was used. Temperature Operating Range. The ozone sensor has been tested in the laboratory and in the field where temperatures ranged from 16 to 28 °C. Sensor performance outside of this temperature range has not been evaluated. Future work will be directed toward evaluating the POM at temperatures up to 35 °C. 3.2. Laboratory Testing. Sensor Response to Ozone. In the laboratory the POM accurately measured ozone concentrations from 30 to 320 ppb (Figure 3). Sensor conditioning with ∼300 ppb of ozone is shown at the left (between 1400 and 1500 h) in Figure 3; the POM measurement increased from 10 to 320 ppb, while the actual ozone level was constant at 320 ppb. Following the conditioning period the ozone sensor responded rapidly to changes in ozone levels. POM and UV monitor measurements were logged every minute and presented as 10 min running averages. When step changes were made to the ozone test concentration the POM reached the test level within 1 min. The sensor responded linearly until the polybutadiene coating became saturated. Coating saturation was apparent after 11:30 A.M. the next day when the sensor measurements steadily decreased even though the actual ozone levels were constant (Figure 3). Several laboratory experiments were conducted in which POMs were exposed to varying levels of ozone. Coating amounts varied from 8 to 14 kHz. A calibration factor was obtained by plotting the actual ozone concentrations vs the sensor response in Hz min-1 (Figure 4). The sensor response was linear from 30 to 300 ppb for both flow rates of 0.13 and 0.25 LPM. The dependence of sensor response on sample air flow rate indicates that the uptake of ozone by the polybutadiene coating is mass transfer limited and not reaction rate limited. Interference Testing. NO2 was chosen for interference testing because its level in outdoor air is related to ozone levels, and because NO2 concentrations are typically higher in indoor environments containing gas appliances (29). Measurements of NO2 made in museums located in Southern California indicated peak levels of ∼100 ppb (30), while measurements made in homes showed peak levels up to 1000 ppb when gas appliances were operated (29). The POM exhibited 5% and 2% interference in the presence of 200 and 800 ppb of NO2, respectively (Figure 5). This was determined by dividing the difference between the measured and actual ozone concentrations by the NO2 VOL. 34, NO. 14, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3035
FIGURE 4. POM calibration curve with sample air flow rates of 0.13 and 0.25 LPM and coating amounts of 8-14 kHz.
FIGURE 5. POM response when exposed to varying levels of ozone, nitric oxide, and nitrogen dioxide. concentration. In other words, sensor response to NO2 was 0.02-0.05 times that for ozone. Extreme indoor NO2 levels (750-1000 ppb) would cause interferences of
