Evaluation of a Photoacoustic Detector for Water Vapor Measurements

Institut für Chemie der Geosphäre (ICG I: Stratosphäre),. Forschungszentrum Jülich, P.O. Box 1913,. D-52425 Jülich, Germany. OTTMAR MOEHLER AND...
4 downloads 0 Views 74KB Size
Environ. Sci. Technol. 2001, 35, 4881-4885

Evaluation of a Photoacoustic Detector for Water Vapor Measurements under Simulated Tropospheric/Lower Stratospheric Conditions MIKLO Ä S S Z A K AÄ L L Department of Optics and Quantum Electronics, University of Szeged, Do´m te´r 9, H-6724 Szeged, Hungary Z O L T AÄ N B O Z O Ä KI Research Group on Laser Physics of the Hungarian Academy of Sciences, Do´m te´r 9, H-6724 Szeged, Hungary MARTINA KRAEMER AND NICOLE SPELTEN Institut fu ¨ r Chemie der Geospha¨re (ICG I: Stratospha¨re), Forschungszentrum Ju ¨ lich, P.O. Box 1913, D-52425 Ju ¨ lich, Germany OTTMAR MOEHLER AND ULRICH SCHURATH* Institut fu ¨ r Meteorologie und Klimaforschung, Forschungszentrum Karlsruhe, P.O. Box 3640, D-76021 Karlsruhe, Germany

Although water vapor is one of the most important and certainly the most variable minor constituent of the atmosphere, accurate measurements of p(H2O) with high time resolution are difficult, particularly in the cold upper troposphere/lower stratosphere. This work demonstrates that a diode laser-based photoacoustic (PA) water vapor detector is a viable alternative to current water vapor sensors for airborne measurements. The PA system was compared with a high-quality frost point hygrometer (FPH) and with a Lyman-R hygrometer in the pressure range of 1000100 hPa at frost point temperatures between 202 and 216 K. These conditions were simulated in a large environmental chamber for 14 h. Simultaneous measurements with the three instruments agreed within 6%. Nitric acid vapor interferes with the FPH measurements at low frost point temperatures but does not affect the other instruments. The sensitivity of the PA system is already sufficient for measurements in the upper troposphere, and straightforward improvements can extend its useful range above the tropopause. Rugged construction, extreme simplicity, small size, and potential for long-term automatic operation make the PA system potentially suitable for airborne measurements.

Introduction The paramount importance of water vapor as a minor constituent of the atmosphere (1), chemically as a source of * Corresponding author phone: +49-7247823943; fax: +497247824332; e-mail: [email protected]. 10.1021/es015564x CCC: $20.00 Published on Web 11/09/2001

 2001 American Chemical Society

OH radicals (2), physically by transporting latent heat from low to higher latitudes (3), and as a green house gas in the radiation budget of the atmosphere (4), is intimately related with its enormous variability as a function of latitude and altitude. Its volume mixing ratio, which can be as high as a few percent in the hot tropical boundary layer, drops sharply across the tropopause from around 100 ppmv in the upper troposphere to about 5 ppmv in the lowermost stratosphere. After a shallow minimum of about 4 ppmv, it increases again to about 6 ppmv at higher altitudes due to the oxidation of methane (5). Recently, commercial aircraft are being used on a routine basis to measure ozone, water vapor, and other parameters at flight altitude (6-8). This and many other applications require rugged, self-contained instruments capable of long-term unattended operation that measure p(H2O) over a wide dynamic range with constant interferencefree sensitivity and high time resolution. Although many types of humidity sensors exist, few if any meet all these requirements simultaneously. Two precision instruments are available for water vapor measurements in the upper troposphere/ lower stratosphere: the frost point hygrometer (FPH) and the Lyman-R hygrometer. The former measures the frost point temperature directly by optically detecting condensation/ evaporation of water vapor on a chilled mirror surface. Stateof-the-art FPH can measure frost points below -80 °C with an accuracy of (0.5 °C. This corresponds to an accuracy in p(H2O) of about 5% at -80 °C and of about 10% at -90 °C. However, FPH has further disadvantages. Above -35 °C, supercooled liquid water may condense on the chilled mirror instead of ice. Moreover, as it measures water vapor indirectly by detecting a phase transition, which is a kinetically controlled and therefore rather ambiguous process, it may be affected by the presence of co-condensable vapors. Yet, FPH offers the advantage of measuring temperature instead of partial pressure with high absolute accuracy and reproducibility and thus can be used under carefully controlled conditions to calibrate other humidity sensors. Lyman-R hygrometers are primarily research instruments (9, 10). An example is the FISH instrument (9), which uses a light source emitting the LR line of atomic hydrogen at 121.6 nm to photodissociate water vapor. The electronically excited OH* fragment emits molecular fluorescence around 310 nm, which is detected through an optical filter by a photomultiplier, or is quenched by collisions. The instrument is fast (time constant ∼1 s), measures directly water vapor volume mixing ratio, and has a high sensitivity (∼0.2 ppmv; 9). However, degradation of the MgF2 lamp window transmission requires frequent replacements of the LR radiation source and recalibrations of the system. Therefore, rather than performing absolute measurements over very long time periods, LR hygrometers are ideally suited for detecting rapid fluctuations of low water vapor mixing ratios. Detecting trace gases by optical absorption spectrometry with tunable diode lasers offers the advantage of fast, highly selective, andsdepending on the tuning and modulation schemesabsolute measurements of species concentrations (11, 12). Among spectroscopic methods, photoacoustic (PA) spectroscopy with room temperature tunable diode lasers is probably the simplest and most cost-effective (13). PA laser spectroscopy is based on the absorption of modulated laser light in a photoacoustic cell, which creates an acoustic wave at the modulation frequency when the laser is tuned to an absorption line of some trace gas. The amplitude of the acoustic wave, which is sampled with a sensitive microphone, is proportional to the concentration of the absorber. The technique is not only extremely sensitive but also has a very VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4881

wide dynamic range. Optical alignment of a PA detector is not critical, in contrast to tunable diode laser absorption systems, which require multi-reflection arrangements to achieve long optical path lengths (12, 14). This results in an exceptional robustness of the PA detector. Fast response, wide dynamic range, automatic operation, and compact and rugged construction, which have already been proven for a ground-based system (15), make PA detection potentially suitable for trace gas measurements on mobile platforms. In the work presented here, a PA water vapor detection system was calibrated against a FPH and applied for measurements under simulated upper tropospheric/lower stratospheric conditions. The system performance was compared with a FPH (General Eastern HYGRO dew point monitor, model 11311DR) and with the FISH LR hygrometer (9).

Experimental Section

FIGURE 1. Gas handling system used for calibrating the photoacoustic system. Arrows indicate direction of gas flow. MFC stands for mass flow controller.

Photoacoustic System. The light source of the system presented in this paper is a DFB diode laser (courtesy of Thomson CSF, France), with an emitted light power of ∼1 mW. It was temperature-tuned to the (1,0,1)51,5 r (0,0,0)41,4 rovibrational line of water vapor at 7339.846 cm-1. This line was chosen because of its intensity (line strength 1.7 × 10-20 cm2 cm-1 molecule-1 at 21 °C) and because other atmospheric trace gases do not interfere with its measurement (16). The PA cell used in this work was milled into a 250 × 120 × 30 mm-sized stainless steel block. It contains two resonators of 40 mm length and 4 mm diameter that are placed between acoustic filters. The acoustic resonance frequency of the cells is ∼4 kHz. A hearing aid microphone (Knowles, EK 3209) is attached to the middle of each resonator. The laser light goes through only one of the resonators. Therefore, by measuring the microphone signals differentially, outside electrical and acoustic noise as well as noise generated by the flowing gas is largely eliminated without reducing the PA signal. The PA detector has a total volume of only about 10 cm3, including resonators and acoustic filters, and can be operated in a continuous flow-through mode up to a volume flow rate of about 1000 cm3 min-1. This type of cell design was already successfully applied for trace gas detection (17). The laser beam enters and leaves the PA cell through antireflection-coated quartz windows and then passes through an identical PA cell, which is filled with humid air and sealed. This reference PA cell was used for tuning the laser to the maximum of the water vapor absorption line. A detailed description of the homemade electronics can be found in ref 15. It consists of a temperature and current controller for the diode laser and a microphone preamplifier. A DSP- (digital signal processor) based central unit, which is fully programmable through a personal computer, controls the electronics. Through its 14-bit DAC (digital to analogue converter) output, it generates the laser modulation waveform for PA signal generation, which is fed into the external modulation input of the diode laser current controller. The PA signal, amplified by the microphone preamplifier, is measured by the 16-bit ADC (analogue to digital converter) input of the DSP unit. The DSP unit was tested by replacing it with a state-of-the-art digital lock-in amplifier (SR 850 Stanford Research) and comparing the signal-to-noise ratios for both configurations. Calibration Unit. The gas handling system used for calibrating the PA sensor at different total pressures is shown in Figure 1. Synthetic air entered the system through two flow controllers. Different water vapor mixing ratios were adjusted by selecting flow rates between 0 and 100 cm3 min-1 with flow controller MFC1, which passed through a thermostated glass vessel containing liquid water. The humidified air was diluted with 5000 cm3 min-1 of dry synthetic air, which was set by flow controller MFC2. Gas in excess of the

amount sampled by flow controller MFC3 upstream of the measuring section was vented through a rotameter. A pressure controller (MKS 250) in combination with a pressure sensor and two flow controllers (MFC3 and MFC6) were used for setting and maintaining constant pressures in the measuring section of the gas handling system. The PA cell and the FPH were incorporated into the pressure-controlled part of the gas handling system. Thus, the PA detector could be directly calibrated against the FPH. As the PA system and the hygrometer were operated at different flow rates of ∼220 and 500 cm3 min-1, respectively, two additional flow controllers (MFC4 and MFC5) were required. Sample gas flow through the measuring section of the gas handling system was maintained by a vacuum pump. Simulation Chamber for Performance Tests under Extreme Atmospheric Conditions. The performance of the calibrated PA sensor was tested under upper tropospheric/ lower stratospheric conditions that were established in the large environmental chamber AIDA, which is short for Aerosol Interactions and Dynamics in the Atmosphere (18-20). This thick-walled aluminum chamber, which has a volume of 84.3 m3, was specially designed to simulate the full range of atmospheric pressures, temperatures, and humidities from ground level up to the middle stratosphere. The measurements reported here covered pressures between 1000 and 100 hPa, which corresponds to 16 km altitude in the real atmosphere, and wall temperatures between 216 and 202 K. The chamber walls were covered with an ice film, i.e., frost point temperatures equaled wall temperatures under static conditions. For simultaneous water vapor measurements with the PA system, with the FPH, and with the FISH hygrometer, chamber air was sampled through stainless steel tubing, which protruded 50 cm into the AIDA chamber. The sampling line was electrically heated to room temperature to minimize adsorption effects. Two side-by-side performance tests of the three water vapor detection systems were carried out, while upper tropospheric/lower stratospheric conditions were simulated in the AIDA chamber. The experiments were part of a comprehensive study of heterogeneous atmospheric processes, which will not be discussed in detail here. In the first test, rapid changes of chamber pressure and thus water vapor content were imposed by pumping from ambient pressure to 100 hPa, while the wall temperature dropped slowly from 216 to 214 K. In a second test at a constant wall temperature of 202 K, nitric acid was generated in situ by the reaction of NO2 with an excess of ozone in the presence of sulfuric acid aerosol. This leads to the formation of N2O5 in the gas phase, which is the anhydride of nitric acid and hydrolyses on the aerosol, forming supercooled H2O/H2SO4/HNO3 aerosol. These “ternary solution droplets” are instrumental in polar

4882

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, 2001

FIGURE 2. Calibration curves of the PA water vapor detector at different total pressures. The frost point hygrometer was used as a reference instrument.

FIGURE 3. Response of the dew point hygrometer (thick line) and of the PA system (thin line) to stepwise changes of the water vapor mixing ratio at 180 hPa total pressure. Volume mixing ratios (ppmv) indicated in the graph.

TABLE 1. Equations of Calibration Lines and Detection Limits of the PA System at Different Total Pressuresa

The detection limits listed in Table 1 make the diode laserbased PA system suitable for fast water vapor measurements throughout the troposphere, while the sensitivity is not yet sufficient to measure water vapor above the tropopause. However, significant improvements are still possible. Because of the linear dependence of the PA signal on the laser power, a 5-fold increase in laser power would improve the sensitivity of the PA system at sea level pressure to an equivalent frost point of -90 °C, which is comparable with the most sensitive frost point hygrometers. Table 1 shows that at 180 hPa total pressure the minimum concentration detectable by the PA system was somewhat lowered when the homemade DSP unit described in ref 15 was replaced by the lock-in amplifier. However this difference can be explained entirely by the different time constants of the two instruments (10 s for the lock-in and 3 s for the DSP unit). These comparable detection limits justify the use of our homemade electronics. From the calibration measurements above it can be concluded that in situations where the pressure of the measured gas is not constant (e.g., on an aircraft), the PA system presented here must be combined with a pressure sensor to correct the PA signal for its pressure dependence. From the data listed in Table 1, the following generalized equation for the pressure dependence of the PA system could be deduced:

total pressure, P (hPa)

calibration curve, S (µV) ) aγ (ppmv) + b

detection limit (ppmv), 3σ/a

975 800 600 400 180

0.09125γ + 0.9754 0.08028γ + 0.7887 0.06241γ + 0.6183 0.04371γ + 0.5817 0.02246γ + 0.2591

2.4 2.7 3.6 5.1 9.6 5.5 (lock-in)

a P, S, γ, and σ denote total pressure, PA signal, water vapor mixing ratio, and background noise of the photoacoustic system, respectively. All measurements were done with the homemade electronics. Additionally, at 180 hPa a lock-in amplifier was used for comparison purposes.

stratospheric ozone loss (21, 22). In both experiments, adiabatic volume cooling occurred while the pressure was reduced by pumping, thus giving rise to ice supersaturation.

Results and Discussion Calibrations at Different Total Pressures. Water vapor measuring systems for atmospheric research in the upper troposphere/lower stratosphere must operate over a wide range of pressures, preferably to less than 200 hPa. To examine the performance of the PA system in this respect, calibrations were carried out at constant total pressures of 975, 800, 600, 400 and 180 hPa. The dew point temperatures indicated by the reference FPH were converted to partial pressures p(H2O) using the integrated Clausius-Clapeyron equation:

log p (Pa) )

A +B T (K)

(1)

with A ) -2663.5 and B ) 12.537 (23). Partial pressures were converted to volume mixing ratios γ by dividing p(H2O) through the total pressure. Measured calibration curves at different total pressures are shown in Figure 2. The equations of the fitted calibration curves are listed in Table 1. Detection limits, defined as three times the background noise level (σ) divided by the slope of the corresponding calibration line (a) at the given total pressure, are also listed. For the PA system, the background noise, mostly of acoustical origin, was typically in the 100 nV range. Compared with the system reported previously (15), the detection limit at normal pressure is lowered by about 1 order of magnitude. This is the consequence of using a diode laser operated at a stronger water vapor absorption line.

γ (ppmv) )

(

)

7413 + 3.58 S (µV) - 11.025 P (hPa)

(2)

where γ is water vapor volume mixing ratio, P is the total pressure, and S is the PA signal. This equation was used to eliminate the pressure dependence from the raw data measured with the PA system during the performance tests. Response Time. During the calibration at a given total pressure, it was also possible to compare the response time of the PA system with that of the FPH. By varying the flow rate through the humidifier (Figure 1), prompt and relatively large stepwise changes of the water vapor volume mixing ratio could be achieved. These measurements mimic situations when a research aircraft passes through air masses with different water contents. Figure 3 shows the response of the PA detector (thin line) and the FPH (thick line) to sudden changes of the water vapor mixing ratio at a fixed total pressure of 180 hPa. In some cases, large amplitude oscillations of the FPH signal occurred before the true mixing ratio was indicated after about 5 min. The PA system did not show such artifacts. The response times of both systems, defined as the time needed for the measured signal to increase from 10% to 90% of the VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4883

FIGURE 4. Water vapor mixing ratios in air sampled from the AIDA chamber, measured with the PA detector (thin line), the FPH (up triangles), and the FISH instrument (open squares). Note that for clarity only every 15th measurement point of the FPH and every 20th measurement point of the FISH instrument are shown. Also shown is the pressure variation in the AIDA chamber (thick line). final signal after a stepwise change of p(H2O), were in the order of 100 s. Measurements at different total pressures gave practically the same response times. The response time of the PA system has three components: the first arises from the time needed to flush the PA cell, which is in the order of a few seconds, depending on the sample flow rate. The second comes from the averaging time of the electronics (10 s in the present case). The largest contribution comes from adsorption/desorption phenomena of water vapor on the walls of the gas handling system and of the PA cell. Therefore reducing the effect of adsorption/ desorption by increasing the gas flow rate and/or by heating the PA cell (24) can be expected to result in shorter response times of the system. The PA system can thus be used as a water vapor monitor when its mixing ratio undergoes rapid changes under upper tropospheric/lower stratospheric conditions. It has basically the same response time as the FPH but operates more reliably in case of sudden concentration changes. Effect of Pressure Variations. The performance of the PA system was tested under simulated atmospheric conditions by sampling air directly from the AIDA chamber, as described above. Throughout the tests, the PA system was run side by side with the FPH and the FISH instrument. The results of the first experiment are shown in Figure 4. After the chamber walls had been coated with a thin ice film under slightly warmer conditions, the temperature was slowly reduced to 216 K until the experiment started at t ) 0 h. At an initially constant pressure of 1000 hPa, the wall temperature continued to decrease slowly, as indicated by the drop of the water vapor mixing ratio in equilibrium with the ice-covered chamber wall. Pumping started at t ) 4 h, reducing the pressure from 1000 to 180 hPa. As a consequence, ice evaporated from the chamber walls to maintain the saturation pressure above the ice film, thus causing the water vapor mixing ratio in the chamber to increase from 11 to about 60 ppmv. After the pressure had remained constant at 180 hPa for about 3.5 h, the pump was started again, and the pressure was reduced to a final value of 100 hPa. This time the water vapor mixing ratio started to increase again but could not reach the expected final value of 108 ppmv corresponding to ice saturation, partly because the ice film had completely evaporated and partly because dry air was added 34 min before the experiment terminated after 14 h. As shown in Figure 4, the water vapor mixing ratios measured with the PA system, the FPH, and the FISH instrument are continuously in excellent agreement. Indeed, for any two of these instruments the relative differences between their readings, averaged over 14 h, was as low as 4884

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, 2001

FIGURE 5. Water vapor mixing ratios in air sampled from the AIDA chamber measured by the PA detector (thin line), the FPH (thick line), and the FISH instrument (open squares). Note that for clarity only every 20th measurement point of the FISH are shown. Also shown is the pressure variation within the AIDA chamber. Vertical dashed lines indicate injections of 0.01 Pa of NO2 into the chamber. 6%. This proves that all variations of the water vapor mixing ratio that were observed during and after the pumping phases were in fact real and not an artifact of either instrument. The variations reveal deviations from thermodynamic equilibrium, which are caused by the transport-limited evaporation rate from the ice-covered walls into the chamber volume. Effect of Nitric Acid Vapor. In the following performance test, the selectivity of the PA system could be demonstrated, as can be seen on Figure 5. Before the experiment, 0.1 Pa of ozone had been added to the chamber air at a total pressure of 180 hPa. The temperature of the ice-coated wall was 202 K, constant throughout the experiment. Ice saturation at this temperature corresponds to a water vapor mixing ratio of ca. 12.5 ppmv at 180 hPa total pressure, in agreement with the measurements. As can also be seen in Figure 5, the situation changed abruptly after the first injection of 0.01 Pa of NO2, which gave rise to a continuous increase of the FPH signal, while the signals of the PA sensor and the FISH instrument remained unaffected. The mismatch is due to the heterogeneous nucleation of nitric acid trihydrate (NAT) on the chilled mirror of the FPH, which occurs above the frost point of ice (25). The evolution of the FPH signal is due to the fact that nitric acid continues to be formed at a nearly constant rate since the reaction of NO2 with ozone is very slow at 202 K. This gives rise to a continuous increase of the NAT point while p(H2O) remains unchanged, as indicated by the PA detector and the FISH instrument. Also shown in Figure 5 is the impact of pumping on the water vapor volume mixing ratio. The evolution of the PA signal can be interpreted analogous to the measurements shown in Figure 4, except for a transitory signal during the adiabatic cooling phase: this is due to condensation of water vapor on the sulfuric acid aerosol during the adiabatic cooling phase, which re-evaporates in the heated sampling line. A detailed interpretation of these processes would be beyond the scope of this paper. However, it can be clearly seen that a second addition of 0.01 Pa of NO2 causes the FPH signal to increase further, while the PA detector and the FISH instrument are not affected. This shows that the presence of nitric acid (and of other condensable vapors) may significantly interfere with optical frost point measurements. The problem does not occur when water vapor is measured by the PA effect, demonstrating an important advantage of the PA system.

Acknowledgments A visiting scientist grant by Forschungszentrum Karlsruhe for M.S. as well as partial funding of the work by the Hungarian

National Research and Development Programme under Contract 0361/2001 are gratefully acknowledged.

Literature Cited (1) Randel, D. L.; Vonder Haar, T. H.; Ringerud, M. A.; Stephens, G. L.; Greenwald, T. H.; Combs, C. L. Bull. Am. Meteorol. Soc. 1996, 77, 1233-1246. (2) Ehhalt, D. H. Phys. Chem. Chem. Phys. 1999, 1, 5401-5408. (3) Peixoto, J. P.; Oort, A. H. Physics of Climate; Springer: New York, 1992. (4) Inamdar, A. K.; Ramanathan, V. J. Geophys. Res. 1998, 103, 32177-32194. (5) Zo¨ger, M.; Engel, A.; McKenna, D. S.; Schiller, C.; Schmidt, U.; Woyke, T. J. Geophys. Res. 1999, 104, 1817-1825. (6) Suhre, K.; Cammas, J.-P.; Ne´delec, P.; Rosset, R.; Marenco, A.; Smit, H. G. J. Nature 1997, 388, 661-663. (7) Brenninkmeijer, C. A. M.; et al. J. Atmos. Oceanic Technol. 1999, 16, 1373-1383. (8) Zahn, A.; Brenninkmeijer, C. A. M.; Crutzen, P. J.; Fischer, H.; Heintzenberg, J. Air Pollution Research Report 74; EUR 19428; European Commission: Brussels, 2000; pp 33-36. (9) Zo¨ger, M.; Afchine, A.; Eicke, N.; Gerhards, M.-T.; Klein, E.; McKenna, D. S.; Mo¨rschel, U.; Schmidt, U.; Tan, V.; Tuitjer, F.; Woyke, T.; Schiller, C. J. Geophys. Res. 1999, 104, 1807-1816. (10) Hintsa, E. J.; Weinstock, E. M.;. Anderson, J. G.; May, R. D.; Hurst, D. F. J. Geophys. Res. 1999,104, 8183-8189. (11) Fehe´r, M.; Martin, P. A. Spectrochim. Acta 1995, A51, 15791599. (12) Hovde, D. C.; Hodges, J. T.; Scace, G. E.; Silver, J. A. Appl. Opt. 2001, 40, 829-839. (13) Miklo´s, A.; Hess, P. Anal. Chem. 2000, 72, 30A-37A.

(14) Kessler, W. J.; Allen, M. G.; Davis, S. J.; Mulhall, P. A.; Polex, J. A. Proc. SPIE 1999, 3537, 139-149. (15) Bozo´ki, Z.; Sneider, J.; Gingl, Z.; Moha´csi, AÄ .; Szaka´ll, M.; Bor, Zs.; Szabo´, G. Meas. Sci. Technol. 1999, 10, 999-1003. (16) Rothman, L. S. HITRAN2000 Database, 2000, http://cfa-www.harvard.edu/HITRAN. (17) Liang, G.-C.; Liu, H.-H.; Kung, A. H.; Mohacsi, A.; Miklos, A.; Hess, P. J. Phys. Chem. A 2000, 104, 10179-10183. (18) Kamm, S.; Mo¨hler, O.; Naumann, K.-H.; Saathoff, H.; Schurath, U. Atmos. Environ. 1999, 33, 4651-4661. (19) Saathoff, H.; Naumann, K.-H.; Riemer, N.; Kamm, S.; Mo¨hler, O.; Schurath, U.; Vogel, H.; Vogel, B. Geophys. Res. Lett. 2001, 10, 1957-1960. (20) Zink, P.; Knopf, D. A.; Schreiner, J.; Mauersberger, K.; Mo¨hler, O.; Saathoff, H.; Seifert, M.; Tiede, R.; Schurath, U. Geophys. Res. Lett. (in press). (21) Carslaw, K. S.; Peter, Th.; Clegg, S. L. Rev. Geophys. 1997, 35, 125-154. (22) Fahey, D. W.; et al. Science 2001, 291, 1026-1031. (23) Marti, J.; Mauersberger, K. Geophys. Res. Lett. 1993, 20, 363366. (24) Beck, H. A.; Bozo´ki, Z.; Niessner, R. Anal. Chem. 2000, 72, 21712176. (25) Hanson, D.; Mauersberger, K. Geophys. Res. Lett. 1988, 15, 855858.

Received for review June 15, 2001. Revised manuscript received September 27, 2001. Accepted October 17, 2001. ES015564X

VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4885