Anal. Chem. 1996, 68, 3059-3062
Miniaturized Ultraviolet Ozonesonde for Atmospheric Measurements John A. Bognar and John W. Birks*
Department of Chemistry and Biochemistry and Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado, Boulder, Colorado 80309-0216
The design and test results of a simple single-beam ultraviolet photometer for ozone measurements are presented. The instrument has several features that distinguish it from former UV absorption instruments, all of which contribute to a smaller size and much lower power consumption while retaining the advantages of the UV measurement technique. A novel airflow system and advanced electronics are among the most important changes from earlier reported systems. The instrument fits into a package 10 cm in diameter by 60 cm long and weighs under 1.5 kg. Hence, it is well-suited for lightweight airborne applications using kites and balloons, as well as other portable measurement needs. Independent measurements of ozone are made every 4 s, with a sensitivity of 0.3 ppbv ozone (limit of detection for S/N ) 3) and a precision of (2%. The vertical distribution of ozone in the atmosphere has been a subject of investigation for several decades. The first measurements were made with ground-based spectrophotometers, but only in the 1950s did the technology emerge for in situ measurements using balloon-borne ozonesondes.1,2 The electrochemical concentration cell (ECC) ozonesonde is an improved derivative of the early technology. It was developed by Komhyr in the 1960s, based on the measurement of current upon oxidation of I- to I3in half of a short-circuited concentration cell.3 This instrument has been, and remains, the smallest and least expensive of the ozone-measuring instruments in routine use. Unfortunately, those assets are offset by its susceptibility to interferences from species such as NO2 and SO2 in polluted air.4 Accuracy is further degraded by changes in the pumping speed with temperature, pressure, and battery voltage driving the air pump. Therefore, measurements made by an ECC sonde are subject to a significant amount of error, typically given as (10%.5 Ultraviolet photometers have also been extensively employed for ozone measurements. The technique is the preferred method listed by the Environmental Protection Agency for ambient ozone measurements, as it is highly immune to the effects of other trace gases in air samples.6 This is because the measurement is made (1) (2) (3) (4)
Brewer, A. W. New Sci. 1957, 2, 32. Mast, G. M.; Saunders, H. E. Inst. Soc. Am. Trans. 1962, 1, 325-328. Komhyr, W. D. Ann. Geophys. 1969, 25, 203-210. Finlayson-Pitts, B. J.; Pitts, J. N. Atmospheric Chemistry; John Wiley and Sons Inc.: New York, 1986. (5) Barnes, R. A.; Bandy, A. R.; Torres, A. L. J. Geophys. Res. 1985, 90, 78817887. (6) Bowman, L. D.; Horak, R. F. A Continuous Ultraviolet Absorption Ozone Photometer. In Air Quality Instrumentation; Scales, J. W., Ed.; Instrument Society of America: Pittsburgh, PA, 1972; Vol. 2.
S0003-2700(96)00418-0 CCC: $12.00
© 1996 American Chemical Society
with the primary (>98% of output) 253.7-nm line from a lowpressure mercury lamp, which lies at the maximum on the ozone absorption spectrum. At this wavelength, the absorbance of ozone in an air sample is at least 2 orders of magnitude greater than that of any likely interfering species.6 Furthermore, interfering gases are not removed by the ozone destruction filter to any significant extent, so that their contributions may be removed by making a reference measurement with ozone selectively removed.6 Thus, this technique is highly specific for ozone. However, ultraviolet absorption has not been the method of choice for routine balloon-based atmospheric vertical profiling work in which the instrument is sacrificed on each flight due to the substantial weight and cost of the instruments. Various ultraviolet instruments have been flown on balloons to obtain vertical ozone profiles, both during ascent and descent portions of the flights. The first flights were made in the early 1970s, using Dasibi Corp. Model 1003A ozone photometers.6,7 These instruments weighed up to 15 kg with batteries, which is far more than a typical ECC sonde package of 0.6 kg. Later instruments, such as that developed at the National Oceanic and Atmospheric Administration (NOAA) Aeronomy Laboratory by Proffitt and McLaughlin, offered much faster response times and better accuracy than the Dasibi instruments but still weighed 14 kg and cost much more than other typical balloon-borne instrument packages.8 We employ high-performance kites and tethered balloons to obtain simultaneous vertical profiles to altitudes up to 7.5 km of several chemical species, including ozone.9,10 Electrochemical sondes were initially used for making the ozone measurements, but a substantial number of large signal spikes were often present in the profiles.10 These spikes are due to the sloshing of the electrochemical sonde’s solutions in their cells as the instrument package is buffeted by winds. The spikes greatly reduced the precision of the electrochemical sonde’s profile measurements. This, coupled with the relatively long preparation time required by an ECC sonde of 1-2 h, its low accuracy as compared to an ultraviolet photometer, and difficulties encountered in shipping the required chemicals by airline, led us to develop an ultraviolet ozonesonde which could be flown on our kites, balloons, and light aircraft. (7) Hilsenrath, E.; Ashenfelter, T. E. NASA Tech. Note 1976, D-8281. (8) Proffitt, M. H.; McLaughlin, R. J. Rev. Sci. Instrum. 1983, 54, 17191728. (9) Balsley, B. B.; Birks, J. W.; Jensen, M. L.; Knapp, K. G.; Williams, J. B.; Tyrrell, G. W. Nature 1994, 369, 23. (10) Balsley, B. B.; Birks, J. W.; Jensen, M. L.; Knapp, K. G.; Williams, J. B.; Tyrrell, G. W. Environ. Sci. Technol. 1994, 28, 422A.
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996 3059
Figure 1. Schematic diagram of the ultraviolet ozonesonde.
EXPERIMENTAL SECTION A schematic diagram of the ultraviolet ozonesonde is shown in Figure 1. The instrument is centered around a 30.5-cm-long PTFE absorption cell, which is 12.7 mm in outside diameter with a 6.5-mm bore. A relatively short (10 cm) piece of thin-walled 6.5-mm-diameter PTFE tubing serves as both the inlet and outlet of the instrument, and its high conductance allows the internal pressure of the cell to rapidly equilibrate with ambient atmospheric pressure. The other end of the cell is connected to the scrubber/ pump assembly. The pump is constructed of Delrin and is a simple reversible-syringe design. In operation, the pump piston is drawn backward, which pulls fresh ozone-bearing air into the absorption cell and fills the cylinder with ozone-scrubbed air. After the light intensity (I) passing through the sample is measured, the piston is driven in the reverse direction, expelling cylinder air back through the ozone scrubber (Mine Safety Appliances Co., Part 463532) and into the absorption cell. The light intensity (Io) passing through the ozone-free air is then measured. This entire measurement cycle takes ∼4 s to complete, yielding an independent measurement of ozone each time. The pump volume is 10 times that of the sum of the absorption cell and tubing. This ensures that the cell is completely flushed before each absorption measurement, even if only exponential dilution of the sample is occurring; however, the consistent 6.5mm diameter of the cell and tubing produces a nearly laminar and highly efficient purge. This system may be contrasted with previous double-beam instruments, where dual pumps and a switching valve were required to alternately fill two separate cells, one each with reference and sample air.8 This also required pressure measurements in both cells, as a pressure drop existed between the ozone destruction filter, located ahead of the cell, and the suction pump. 3060
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
Supporting sensors on the instrument may include a thermistor and a pressure transducer. The thermistor (Keystone Carbon, AL03006-11.7K-98-G1) is located in the gas flow path between the cell and scrubber to measure gas temperature in the cell. A pressure transducer (Motorola, MPX2101GS) may be located there as well. However, in aerodynamically static conditions around the inlet, such a transducer is not required. This is due to the cell being open directly to the atmosphere, and atmospheric pressure is reported separately in our system via a radiosonde package. The instrument’s optical arrangement is a single-beam absorption photometer, with an isolated reference beam path. The lamp (Sankyo Denki, type G4T5 low-pressure mercury lamp) irradiates the cell through the silica window (Melles Griot, 02 WLQ 105), which forms one end of the cell. It is not necessary to block out shorter, ozone-forming mercury lines (e.g., 188 nm) as the lamp is of such low power that no measurable quantity of ozone would be formed from it during the time a sample is exposed. This is in contrast to the much higher power lamps used in other instruments, which require a Vycor shield to eliminate ozoneproducing spectral lines.6 The interference filter (Melles Griot, 03 FIM 016: center wavelength 253.7 nm, fwhm 10 nm) forms the other end of the cell and serves to isolate the 253.7-nm mercury line and to shield the photodiodes (Hamamatsu, type S1226-18BQ) from ambient light. To correct for lamp intensity fluctuations during a measurement, a second light path is provided to the reference photodiode through a 30.5-cm length of fiber optic (Oriel Corp., type 77513). This transmits light directly from the lamp through the filter to the reference photodiode, so that fluctuations in ambient air do not affect the reference measurements. The electronics consist of three primary sections: a microcontroller, the lamp power supply, and the photodiode amplifiers and analog-to-digital converters. The microcontroller (Microchip, PIC16C57) carries out several functions. It controls the pump, makes temperature and pressure measurements, processes the photodiode signals, and communicates in either analog or digital format with the radiosonde package (Vaisala, RS80-15) it is flown with. The lamp is powered by a compact inverter circuit which was included with the lamp (Ultraviolet Products Inc., Model (UVG4). This inverter in turn is powered by a regulated 6-V supply to ensure a constant output over time after the lamp is warmed up. The lamp runs cool, so that only a short warm-up time of less than 15 min is required. The photometer circuit is shown in Figure 2. The photodiode currents, which are ∼2 nA, are amplified in high gain (109) currentto-voltage amplifiers (U7 and U8; Harris Semiconductor, CA3140E). The resulting voltages enter an analog voltage divider (U10, Analog Devices, AD734AN), which provides the ratio of the two signals to yield a normalized cell signal corrected for any lamp fluctuations. The voltage follower (U9, Analog Devices, OP177GP) and associated resistor (R10) are included for proper operation of the divider.11 The signal is then taken into an analog-to-digital converter (Analog Devices, AD7701AN), where it is digitized and sent to the microcontroller. The microcontroller averages absorption measurements for 1 s in order to further reduce any pressureor temperature-dependent noise in the system. The averaged, digitized signals may be directly output in real time for ozone (11) AD734 Data Sheet; Analog Devices: Norwood, MA.
Figure 2. Photometer signal processing circuit.
calculations in the ground computer or else may be run through an on-board math processor to give an analog output, which may be telemetered to ground in addition to or in place of the digital signal. The overall instrument fits into a cylinder 10 cm in diameter by 60 cm long. The weight, including lithium batteries for 4 h of continuous operation, is 1.5 kg. The overall power consumption is 8 W, budgeted as follows: pump, 4 W; lamp, 2 W; and electronics, 2 W. The cost of materials for construction of the instrument is U.S. $470, based on the purchase of single components. Future modifications to the design to further reduce size, weight, and cost are addressed briefly in the Conclusion. RESULTS AND DISCUSSION The recovery of data is based on the Beer-Lambert absorption law:
I ) Ioe-σnl The intensities I and Io are those measured with and without ozone present in the cell, respectively. Each is normalized to correct for lamp intensity fluctuations by dividing the cell photodiode signal by the reference photodiode signal. The absorption cross section σ for ozone at 253.7 nm is 1.147 × 10-17 cm2 molecule-1. This cross section is dependent to a very small extent on pressure and temperature, but corrections for those factors may be neglected for most tropospheric and stratospheric measurements.12 The number density of ozone is represented by n, in molecules per cubic centimeter, and the cell length of 30 cm is
represented as l. Substitution of known values into this equation yields n, which may be converted to a mixing ratio by using the pressure and temperature of the sample. A laboratory comparison was made between this instrument and a commercial Thermo Electron Corp. Model 49 ozone photometer. Both instruments simultaneously sampled air with varying concentrations of ozone, from zero to 350 parts per billion by volume (ppbv), using a homemade photolytic ozone source. A linear relationship was found between the two systems, with a correlation coefficient of 0.985. The very small deviations observed are suspected to be due to a coupling of fluctuations in the ozone source and different sampling intervals for the two instruments. The ultraviolet ozonesonde has a limit of detection of 0.3 ppbv ozone (S/N ) 3) and precision of (2%, based on these tests. Flight tests were also carried out at Boulder, CO, using a light aircraft (Piper Apache) as the sampling platform. The flight goals were to study the instrument’s behavior through pressure and temperature changes and to obtain a comparison against an ECC sonde. In this series of tests, the ultraviolet ozonesonde was not compared with the commercial ultraviolet instrument, as it was not possible to power the Model 49 on the aircraft. Figure 3 is a plot of ozone concentrations versus flight time for the two instruments on a flight which began at Boulder (1630 m above (12) DeMore, W. B., Ed. Chemical Kinetics and Photochemical Data for Use in Stratospheric Modeling; JPL Publication 94-26; National Aeronautics and Space Administration; Government Printing Office: Washington, DC, 1994.
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
3061
Figure 3. Ozone measurements versus flight time for the UV and ECC ozonesondes. Flight data February 15, 1996; local time 14: 00-15:20.
sea level), reached a maximum altitude of 5200 m over the Rocky Mountains, and ended by passing over Denver and returning to Boulder. Excellent agreement is found between the two ozone instruments, especially considering that both instruments were independently calibrated. The deviation in the first 15 min reflects a known phenomenon we have observed with ECC sondes, in that an ECC sonde often reads low for up to 30 min after being removed from a zeroing filter. The deviation in the center of the plot is a result of different aerodynamic effects at the inlets of the two sondes during a tight sustained turn, with the UV instrument more significantly affected due to its wider bore sample tube. This is an effect that should be negligible in our intended applications but has been easily corrected for by adding a pressure sensor in the cell, as described above.
ultraviolet absorption technique, while approaching the standards set by electrochemical sondes for weight and cost. This instrument will replace electrochemical sondes for our tropospheric ozone measurements using kites, balloons, and light aircraft and may be suitable for other surface-based applications, such as at remote stations, where power consumption and low maintenance are critical. The instrument is inexpensive enough that it could be manufactured for release sondes. Another important application could be tropospheric measurements on-board commercial airliners. Alterations to the design presently being investigated include the use of compact, high-speed fans to replace the syringe pump assembly, which is the single largest and heaviest (0.6 kg) component of the present system. The mercury lamp will also be replaced by a more compact model (TEC/WEST (USA), Inc., Model GTL3). This will yield an instrument with a substantially smaller overall diameter, reduced weight and power consumption, and lower cost. ACKNOWLEDGMENT This work was supported by the National Oceanic and Atmospheric Administration Global Change Program (Grant NA56GP0238) and the Environmental Protection Agency (Grant R82-1252-010). We thank Bill Ingino for the construction of the syringe pump and absorption cell and Kurt Radzay for piloting the aircraft.
Received for review April 25, 1996. Accepted June 27, 1996.X AC9604189
CONCLUSIONS A new, compact design for an ultraviolet ozone photometer has been developed and tested. It retains the advantages of the
3062
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
X
Abstract published in Advance ACS Abstracts, August 1, 1996.
