Anal. Chem. 1996, 68, 4194-4199
Dual-Inlet Chemical Amplifier for Atmospheric Peroxy Radical Measurements Christopher A. Cantrell,* Richard E. Shetter, and Jack G. Calvert
Atmospheric Chemistry Division, National Center for Atmospheric Research, Boulder, Colorado 80303
The state-of-the-art of peroxy radical measurements using the technique of chemical amplification has matured in recent years. The NCAR chemical amplifier has been improved over previous versions by employing a dual-inlet system that allows more rapid switching between measurement and background modes. Further improvement is realized by the use of two NO2 detectors. The dualinlet chemical amplifier (DICHAMP) employs two identical glass inlet reactor-NO2 detector combinations, one of which is operated in the background, while the other operates in the radical measurement mode. This instrument is less sensitive to fluctuations in ambient ozone than the single-channel version because of the continuous monitoring of the background and background plus radical signals. The single-inlet, dual-inlet with single-detector, and DICHAMP instruments are compared through theoretical calculations of the effect of noise at a given frequency and amplitude on retrieved radical levels. Laboratory experiments were conducted to support the theoretical results. Ambient radical concentrations were determined using these configurations to evaluate the performance under actual measurement conditions. Peroxy radicals (HO2 and RO2, where R is any organic group) play important roles in tropospheric photochemistry. These radicals oxidize NO to NO2 and thus impact the NO/NO2 ratio. If ambient NOx (NO + NO2) levels are sufficiently high (greater than about 50 pptv1), peroxy radicals can lead to in situ ozone formation. HO2 radicals also destroy ozone. Organic peroxy radicals are formed in the oxidation of carbon compounds. Reactions between peroxy radicals lead to hydrogen peroxide (HOOH) and organic hydroperoxides (ROOH). Peroxy radicals are formed when OH reacts with CO, SO2, and hydrocarbons, when ozone reacts with alkenes, and when the nitrate radical, NO3, reacts with aldehydes and other hydrocarbons. The relative strengths of these sources can vary greatly depending on the availability of ozone, hydrocarbons, NOx, and solar ultraviolet light. Chemical models2 and photostationary state calculations3 have yielded information on the dependence of peroxy radical concentrations on various environmental parameters. However, much more can be learned through direct measurements, preferably in conjunction with determination of the important controlling factors. (1) Liu, S. C.; Trainer, M.; Carroll, M. A.; Hu ¨ bler, G.; Montzka, D. D.; Norton, R. B.; Ridley, B. A.; Walega, J. G.; Atlas, E. L.; Heikes, B. G.; Huebert, B. J.; Warren, W. J. Geophys. Res. 1992, 97, 10463-71. (2) Trainer, M.; Hsie, E. Y.; McKeen, S. A.; Tallamraju, R.; Parrish, D. D.; Fehsenfeld, F. C.; Liu, S. C. J. Geophys. Res. 1987, 92, 11879-94. (3) Ridley, B. A.; Madronich, S.; Chatfield, R. B.; Walega, J. G.; Shetter, R. E.; Carroll, M. A.; Montzka, D. D. J. Geophys. Res. 1992, 97, 10375-88.
4194 Analytical Chemistry, Vol. 68, No. 23, December 1, 1996
The chemical amplifier has been used for several years to measure peroxy radical concentrations in urban,4 rural continental,5,6 marine boundary layer,7 and remote tropospheric environments.8 HO2 radicals are measured by conversion to NO2 in a chain mechanism involving NO and CO, which produces 50-200 NO2 molecules per HO2 in a few seconds of reaction time. The NO2 that is produced is measured by luminol chemiluminescence.9 NO2 that is produced from sources other than the radical chain reaction (the background signal) is measured when nitrogen (N2) is substituted for CO or when the CO addition point is moved downstream from the NO addition point. RO2 radicals are measured in large part because they can be converted to HO2 in the NO-rich environment of the instrument inlet. These conversions are not perfectly efficient because several reactions can remove radicals before the primary reaction that forms HO2, the reaction of the alkoxy radical (RO) with O2, can take place.9 Alkoxy radicals react with NO at rates that can compete with reaction with O2 for the conditions normally encountered in the instrument inlet (3 ppbv NO, 10% CO). Alkoxy radicals can also decompose or isomerize, complicating the evaluation of the measurement efficiency of particular radicals. Overall, for clean to moderately polluted atmospheres, theoretical estimates predict that the total peroxy levels measured should be greater than 90% of the actual levels, based on estimates of the types of peroxy radicals likely to be present.9,10 The chemical amplifier chemistry is shown schematically below.
This paper reports on the performance of the DICHAMP instrument compared to our single-inlet and dual-inlet instruments (4) Hu, J.; Stedman, D. H. Environ. Sci. Technol. 1995, 29, 1655-9. (5) Cantrell, C. A.; Lind, J. A.; Shetter, R. E.; Calvert, J. G.; Goldan, P. D.; Kuster, W.; Fehsenfeld, F. C.; Montzka, S. A.; Parrish, D. D.; Williams, E. J.; Buhr, M. P.; Westberg, H. H.; Allwine, G.; Martin, R. J. Geophys. Res. 1992, 97, 20671-86. (6) Cantrell, C. A.; Shetter, R. E.; Calvert, J. G.; Parrish, D. D.; Fehsenfeld, F. C.; Goldan, P. D.; Kuster, W.; Williams, E. J.; Westberg, H. H.; Allwine, G.; Martin, R. J. Geophys. Res. 1993, 98, 18355-66. (7) Weissenmayer, M. Peroxyradikalmessungen in der bodenahen Brenzschicht u ¨ ber dem Atlantik. Ph.D. Dissertation, University of Mainz, Germany, 1995. (8) Cantrell, C. A.; Shetter, R. E.; Gilpin, T. M.; Calvert, J. G. J. Geophys. Res. 1996, 101, 14643-52. (9) Cantrell, C. A.; Shetter, R. E.; Lind, J. A.; McDaniel, A. H.; Calvert, J. G.; Parrish, D. D.; Fehsenfeld, F. C.; Buhr, M. S.; Trainer, M.; Murphy, P. C. J. Geophys. Res. 1993, 98, 2897-909. (10) Madronich, S.; Calvert, J. G. J. Geophys. Res. 1990, 95, 5697-715. S0003-2700(96)00639-7 CCC: $12.00
© 1996 American Chemical Society
with one NO2 detector in terms of time response, stability, and sensitivity to changes in ambient ozone, using results from theoretical calculations, laboratory tests, and ambient measurements. EXPERIMENTAL DETAILS The single-inlet chemical amplifier has been discussed in detail in the literature.4,8,9,11 Since the technique involves switching between background and radical measurement modes, the actual background is interpolated between consecutive measured background signals and subtracted from the radical plus background signal. In the case of constant or only slowly varying ozone levels, the background determined by this method should be reasonably accurate. However, when the ozone level is rapidly changing, then significant noise can be introduced into the radical signal because the measurements of the background are only an approximation of the true background levels during the radical plus background measurement. Some improvement in the performance of the instrument can be accomplished by use of smaller inlet reactors along with shorter modulation periods.11 The inlet of Hastie et al.11 consists of about 4 m of 1/4-in.-o.d. Teflon tubing. The size of the glass inlets that we employ (about 1-L volume8), used to minimize radical wall contact, also precludes modulation cycles shorter than about 3 min. Improvement is also expected if two inlets are employed, with one in the background mode and the other in the radical plus background mode. The signal measured by the NO2 detector is selected through valves located close to the detector itself. Finally, if two NO2 detectors are employed to measure the signal from the two inlets described above, background fluctuations are constantly measured and can be removed from the measured radical plus background signal. These three instrumental configurations are compared in this study. Two modulation approaches reported in the literature are displayed graphically in Figure 1. A 10-min-long, asymmetric modulation cycle that we employed during the Mauna Loa photochemistry experiment 2 (MLOPEX-28) and other recent studies is displayed at the top of the figure. The upper part of the square wave represents periods of radical plus background measurement, and the lower part represents background measurement. Also shown is a 1-min-long, symmetric cycle used in our deployment of a dual-inlet, single-NO2 detector instrument in one study (peroxy radical intercomparison exercise, PRICE 112) and similar to that used by Hastie et al.11 and others. This figure shows that more rapid modulation allows frequent measurement of the background. If the background is fluctuating, the changes can be better accounted for, minimizing the noise that might be imparted into the radical signal. In the case of the DICHAMP configuration, there is no modulation required, but the background and background plus radical signals can be averaged to any desired time interval for comparison with the other instrument configurations and modulation cycles. Figure 2 shows a power spectrum of 1 week of data collected once per minute (dotted line) during a ground-based measurement (MLOPEX-213) to illustrate the behavior of noise in typical ambient ozone data. The other curves shown in Figure 2 will be discussed later in this paper. (11) Hastie, D. R.; Weissenmayer, M.; Burrows, J. P.; Harris, G. W. Anal. Chem. 1991, 63, 2048-57. (12) Cantrell, C. A.; Shetter, R. E.; Calvert, J. G. Proceedings of International Union of Geodesy and Geophysics XXI General Assembly, Boulder, CO, July, 1995, B253. (13) Atlas, E. A.; Ridley, B. A. J. Geophys. Res. 1996, 101, 14531-41.
Figure 1. Graphical representation of two modulation cycles for the chemical amplifier. The upper part of each trace represents a period of radical plus background measurement, and the lower part represents background measurement. The upper trace (dashed line) is a 10-min asymmetric cycle that has been used in several ambient measurement campaigns (e.g., Cantrell et al., 1996, ref 8). In this cycle, 1 min of background data are averaged, the radical amplification is activated followed by a stabilization period of 1.5 min, and then six 1-min average radical plus background signals are collected. The radical amplification is deactivated, followed by another 1.5-min stabilization period. The lower trace (solid line) shows a 1-min symmetric modulation cycle that could be used for small inlets or larger inlets in a dual-inlet/single-detector configuration. Here, the background and the radical plus background signals were averaged for 24 s (0.4 min) with stabilization periods of 6 s (0.1 min) between.
Figure 2. Theoretical effect of sinusoidal noise with an amplitude 10% that of the background level on the retrieved radical level for three modulation cycles, as a function of the period of the sinusoidal noise. The solid line is for a symmetric 10-min modulation cycle, the medium dashed line is for a symmetric 1-min modulation cycle, and the short dashed line is for the asymmetric 10-min cycle shown in Figure 1. The dotted line shows the relative amplitude of fluctuations in an authentic ambient ozone measurement as a function of the noise period, also shown offset and multiplied by 10 for periods less than 20 min.
Features of the DICHAMP Instrument. The instrumental layout is shown schematically in Figure 3. The instrument is essentially identical to our earlier instrument,9 with the addition of a second glass inlet, associated reagent gas delivery apparatus, and a second luminol-based NO2 detector. The use of two inlets also requires modifications to the calibration gas delivery system so that both inlet/NO2 detector combinations can be readily Analytical Chemistry, Vol. 68, No. 23, December 1, 1996
4195
Figure 3. Schematic diagram of the DICHAMP instrument that is configured with two matched inlets, two matched and calibrated detectors, and series of solenoid valves that allow direction of the reagent and calibration gases to be controlled as described in the text.
calibrated. Valve C can direct either zero air or zero air containing NO2 from the temperature-controlled permeation tube to the inlets or to the radical calibration system. Valve D selects to which inlet the NO2/air mixture is diverted. In the radical calibration system, which is based on the thermal decomposition of peroxyacetyl nitrate (PAN), valve E selects whether the PAN output from a temperature-controlled diffusion tube is added to the NO2/air mixture or vented. Thus, valve D allows NO2 calibration of both inlets, and valve E allows testing for artifact signals possibly generated from the heating of the NO2/air mixture (we have found that a heater temperature less than about 120 °C reduces the artifact signals to negligible levels). In order for the use of two NO2 detectors to yield improvement in the signal-to-noise ratio of actual ambient peroxy radical signals, they must be carefully calibrated. Since only a slight error in the relative calibration of either detector will yield shifts (positive or negative) in the ambient radical signal, it is important that multiplepoint calibrations are performed regularly (to account for changes in calibration levels and changes in nonlinear response of the NO2 detector). In addition, it is prudent to compare the response of the two NO2 detectors at one calibration point on an even more frequent schedule (every few hours). On occasion, one should also check various components of each channel of the system by switching the inlet/reagent gas/sample line combination sampled by each NO2 detector. The latter is accomplished by simultaneous switching of the two valves labeled A in Figure 3. In the dualinlet/single-detector combination, these valves are switched in order to accomplish the modulation. The concentration of NO reduces the sensitivity of the NO2 detectors in an approximate inverse relationship (i.e., doubling NO approximately halves the sensitivity to NO2). In addition, the concentration of NO affects the sensitivity of the inlet chemistry to ambient peroxy radicals.9 4196
Analytical Chemistry, Vol. 68, No. 23, December 1, 1996
The reagent gas mixtures can be directed to one or the other inlet through the simultaneous switching of the two valves labeled B in Figure 3. These valves control the modulation in our singleinlet/single-detector instrument. The response time to switching valves A is quite short (99% response in 6 s), but the response time to switching valves B is much longer due to the relatively large volume in the glass inlets (99% response in 60 s). Therefore, switching of valves A can be performed frequently without loss of too much ambient data, whereas the switching of valves B necessarily involves loss of about 1 min of ambient data. Other factors that impact the sensitivity of the NO2 detectors are the composition of the luminol solution, the flow rates of the luminol solution and the sampled NO2 mixture,14 the cleanliness of the filter paper on which the NO2/luminol reaction takes place, and the presence of insoluble impurities in the luminol solution. The luminol solution has been optimized for conditions of the chemical amplifier.9 The instrument sensitivity is approximately inversely proportional to the liquid flow rate and relatively insensitive to the air/NO2 flow rate above 2 L/min (although there is an indirect effect on the sensitivity due to the dilution of the reagent NO gas). To minimize variations of these factors between the two NO2 detectors, both use the same luminol solution from the same reservoir, and both are pumped by the same peristaltic pump. Upon exiting the solution reservoir, the luminol solution is filtered with an inline, HPLC-like liquid filter. The filter papers that are viewed by the photomultiplier tubes in each detector are cleaned regularly. Flow rates of reagent gases and of the main instrument flows are controlled by mass flow controllers that are calibrated on a regular basis. It is also important that the times between air entry into the inlets and detection in the NO2 detectors (14) Wendel, G. J.; Stedman, D. H.; Cantrell, C. A.; Damrauer, L. Anal. Chem. 1983, 55, 937-40.
are precisely matched. If the volumes of the inlet and connecting tubing or total flow rates are even slightly different, then noise and possibly bias will be introduced into the radical signals due to the phase shift between the two detector signals. Instrumental Tests. The performance of the chemical amplifier in the DICHAMP configuration has been compared to the single-inlet and dual-inlet/single-NO2 detector configurations in theoretical calculations, in laboratory tests, and through ambient measurements. For the theoretical calculations, a sinusoidal value was applied to a square wave modulated signal such that the amplitude of the sine wave noise was 10% that of the square wave amplitude. Values were calculated every 5 simulated seconds. The two modulation types shown in Figure 1, as well as a 10-min symmetric modulation, were employed over about one simulated hour for sine wave periods of 0.4-200 min. The laboratory tests were accomplished by varying the flow of air passing over an unfiltered mercury pen-ray lamp, which photolyzes oxygen to produce ozone. A sinusoidal shape to the ozone concentration versus time profile was generated by applying a sine wave voltage to the mass flow controller control signal. The flow over the ozone source varied between 20 and 80 cm3(STP)/min and was then diluted with 10 L(STP)/min of zero air. The diluted mixture was delivered to both instrument inlets through the NO2 calibration port. The sine wave generator varied the frequencies for the experiments from 100 µHz to 0.1 Hz (signal cycle periods of 0.167-167 min). The ozone concentration was 40-50 ppbv, and the amplitude of the sinusoidal modulation was usually about 5 ppbv. An offset could be applied to one of the detector signals to simulate a radical signal, through addition of NO2 from the permeation tube system. For most experiments, this was about 2-3 ppbv, which resulted in a peroxy radical signal equivalent to 10-15 pptv. Notice that the peak-to-peak modulation on the ozone background was comparable to the amplitude of the radical signal. Experiments were performed in which the frequency and amplitude of the ozone variations and the amount of simulated radical signal were systematically varied to examine the behavior of the chemical amplifier in the DICHAMP configuration as compared to the other configurations. A sample of these laboratory data is shown in Figure 4. Here, the background ozone is modulated at 0.005 Hz (period of 3 min, 20 s), with a simulated radical level of about 15 pptv. The radical concentration was calculated using the three instrument configuration/modulation cycle combinations (single-inlet/single-detector 10-min asymmetric cycle; dual-inlet/single-detector 1-min symmetric cycle; and DICHAMP 1-min averages). The impact of the phase relationship between the sine wave noise and the modulation cycle is demonstrated in the positive bias of the results calculated with the 10-min modulation, because the cycle is an integer multiple of the noise period. For the ambient measurements, about 5 days of 0.5-min data were collected from the DICHAMP instrument. Radical signals were calculated using appropriate averaging and interpolation procedures to simulate the three arrangements. RESULTS AND DISCUSSION In the following, results of theoretical calculations are compared with those of laboratory studies of the effects of background noise on retrieved radical levels using various modulation schemes. Also, actual ambient data are presented to verify the performance of the DICHAMP instrument compared with other instrument configurations.
Figure 4. Sample data from laboratory studies of the effect of sinusoidal background noise on the retrieved radical level. Shown are data with noise at 0.005 Hz (3 min, 20 s period). The solid line is the radical plus background signal measured by detector 1, the dotted line is the background signal measured by detector 2, the filled circles are the calculated radical levels using the DICHAMP configuration, the open circles are the radical levels for the dual-inlet/single-detector instrument with the symmetric one-minute modulation cycle, and the open squares are for the single-inlet/single-detector configuration with the asymmetric 10-min modulation cycle.
Theoretical Calculations. The retrieved radical levels from the theoretical calculations were scaled to that expected with no noise added. These relative retrieved radical levels are unity for some low and high added noise frequencies but depart from unity at intermediate frequencies. The value at these intermediate frequencies depends on the frequency and symmetry of the modulation cycle used. Also, the degree to which background noise is transferred to noise in the radical level depends on the phase relationship between the noise and the instrument modulation cycle. The results of calculations of the variation of the frequency, phase, and modulation cycle on retrieved radical signal are shown in Figure 2. The lines represent the envelope of results for relative radical levels greater than 1, but there is a symmetry in this envelope about unity. The long dashed line shows results for the 1-min modulation cycle, the medium dashed line is for the asymmetric 10-min modulation cycle, and, for comparison, the solid line is for a symmetric 10-min modulation cycle (5 min of background and 5 min of background plus radical signal). Notice that, for the symmetric modulation cycles, the maximum effect of background noise occurs at a period about 1.3 times the length of the modulation cycle, and a 10% noise amplitude (peak to peak) leads to noise in the retrieved radical level of up to (13%. The asymmetric modulation cycle is affected by a much wider range of background noise frequencies. This is because the background is measured for only 1 min, so higher frequency fluctuations have more impact than in the symmetric 10-min modulation. The asymmetric 10-min modulation cycle would be acceptable for extremely stable ozone concentrations, but it is more susceptible to a wider range of fluctuations that the symmetric modulation cycles. The maximum amplitude of the background-induced noise is also somewhat higher in the asymmetric modulation cycle. By plotting the relative amplitude of fluctuations in ambient ozone on this same figure, one can see that the noise induced from the Analytical Chemistry, Vol. 68, No. 23, December 1, 1996
4197
Figure 6. Summary of standard deviations for the single-inlet/singledetector instrument with asymmetric 10-min modulation and the dualinlet/single-detector configuration with symmetric 1-min modulation cycles ratioed to the standard deviations for the DICHAMP calculations, for the same experiments. The curves are least-squares fits to the ratios for the 1-min cycles (solid line) and the 10-min cycles (dashed line).
Figure 5. Relative radical levels retrieved for the three modulation configurations in laboratory tests as a function of the sinusoidal noise period. The upper panel represents data for the single-inlet/singledetector instrument with asymmetric 10-min modulation cycle, the middle panel shows data for the dual-inlet/single-detector configuration with symmetric 1-min modulation cycle, and the bottom panel shows 1-min data for the DICHAMP configuration.
1-min modulation cycle overlaps the larger fluctuations less than that from the 10-min modulation cycle. Laboratory Studies. The results of the laboratory measurements are shown in Figure 5 for the three instrument configuration/modulation cycles considered. Plotted are the relative radical levels retrieved as a function of the period of the background sine wave noise (like the calculations in Figure 2). Multiple points are shown at each noise period for different phase relationships between the noise the modulation cycle. The residual frequencyindependent noise comes from other sources. These laboratory measurements are in general agreement with the calculations in that the maximum in the retrieved radical levels is at longer periods as the modulation period is increased for the singledetector configurations. For the DICHAMP arrangement, the noise is lower than that for the single-detector arrangements and is relatively insensitive to the ozone modulation frequency. For the single-detector configurations, the shorter the modulation period, the less variations in ambient ozone affect the retrieved radical signals. The noise in the radical concentration introduced by fluctuations in ozone is random and can be reduced through averaging. The noise level is still lower in, for example, a 30-min average of data from 1-min modulations than in a 30-min average of data from 10-min modulations. The ratios between the standard deviations of 1 h of data of the two single-detector modulation cycles and the standard deviations of 1 h of data for the DICHAMP cycle as a function of the noise period are shown in Figure 6. The curves are least-squares fits to the data using 4-7-parameter equations determined using the TableCurve 2D fitting program (Jandel Scientific). The noise in the single-detector configurations 4198 Analytical Chemistry, Vol. 68, No. 23, December 1, 1996
approaches that of DICHAMP at high and low frequencies. Symmetric modulation cycles have a narrower distribution compared to the asymmetric one shown, in agreement with the theoretical results shown in Figure 2. The standard deviation of the DICHAMP radical levels is typically less than 0.5 pptv. This implies (barring significant other noise sources) that our chemical amplifier in the DICHAMP configuration can have a detection limit of 1 pptv or less. Additional laboratory experiments were performed to see if background noise impacted the linearity of the retrieved radical signals. At a noise frequency of 0.005 Hz (3-min, 20-s period), the simulated radical level was systematically varied from zero to about 62 pptv, and the levels retrieved by three modulation cycles were compared. The results are shown in Figure 7. All three modulation configurations show linearity with slopes that agree quite well, on average. The uncertainty in the slopes, however, is much larger for the single-detector configurations, with the 10min asymmetric cycle having the largest uncertainty. Ambient Measurements. The DICHAMP instrument was set up for ambient measurements at the National Center for Atmospheric Research (NCAR) near the outskirts of Boulder, CO. The site was not ideally suited for ambient chemical measurements because it is located near a loading dock, and thus it is frequently impacted by fresh emissions from delivery trucks and other vehicles. In any case, these data serve to demonstrate the improvement of the performance of the instrument for actual ambient measurement conditions. Ten-minute average peroxy radical concentrations for a few days in early May, 1996, are shown in Figure 8, for DICHAMP, the dual-inlet/single-detector instrument operating with a 1-min modulation cycle, and the singleinlet/single-detector instrument operating with the asymmetric 10-min modulation cycle. Changes to the data collection software had not yet been implemented to automate frequent instrument NO2 calibrations, so they were performed manually three times throughout this period. This explains why there are some periods of slight negative radical concentration. Visually, it is clear that the noise level of the measurements improves significantly from the single-inlet/single-detector configuration to the DICHAMP
Figure 7. Retrieved peroxy radical level versus amount of NO2 added to inlet 1. Each data point represent a 1-h average, with the standard deviation shown as error bars. The squares are for the single-inlet/single-detector configuration with asymmetric 10-min modulation cycle, the triangles are for the dual-inlet/single-detector configuration with symmetric 1-min modulation cycle, and the circles are for the DICHAMP calculations. The lines represent least-squares fits to each of the data sets. All the slopes are the same within the error retrieved from the fit, and the uncertainties of the slopes are smallest for the DICHAMP and largest for the asymmetric 10-min cycle.
configuration. To quantify this improvement, averages and standard deviations were calculated for three periods of 1.5-2 h duration. The standard deviations for the dual-inlet/singledetector results are 3-9 times smaller than those for the singleinlet configuration. The DICHAMP arrangement results in an additional 20-30% improvement for these conditions. Typical standard deviations for DICHAMP data that were relatively stable (e.g., 400-600 h on 4 May) are less that 0.5 pptv. For comparison, the instrument background and standard deviation of 10-min average background levels are also plotted in Figure 8, showing the correlation between background standard deviations and the noise in the single-inlet/single-detector data. CONCLUSIONS A variant on the single-channel, single-detector chemical amplifier has been developed and characterized. The new instrument is much less sensitive to fluctuations in ambient ozone than the previous instrument. The responses of chemical amplifiers in the single-channel and dual-channel configurations have been investigated as a function of the frequency and amplitude of background ozone fluctuations, both theoretically and through
Figure 8. Ten-minute average ambient peroxy radical levels that were collected near the NCAR building in early May, 1996. In the upper graph, the instrument background divided by 10 (dashed line) and standard deviation of 10-min average background levels (solid line) are plotted versus time for 4-9 May, 1996. In the lower graph, radical concentrations were calculated for the 10-min modulation cycle with the single-inlet/single-detector instrument (inverted triangles) offset vertically by 40 pptv, the 1-min modulation cycle with a dualinlet/single-detector configuration (open circles) offset vertically by 20 pptv, and DICHAMP (filled circles). NO2 calibrations were performed on the mornings of 3, 6, and 9 May and interpolated for the time periods in between.
laboratory studies. The single-channel instruments are particularly sensitive to ozone fluctuations at a frequency about 1.3 times the modulation frequency of the detector. The DICHAMP configuration removes the dependence on the frequency of the background noise or the length of the modulation cycle, since both the background and radical plus background signals are continuously monitored. The information gained in the theoretical calculations and the laboratory tests is confirmed in the ambient measurements. ACKNOWLEDGMENT Thanks to Drs. Eric Apel and Elliot Atlas for helpful comments in the preparation of this manuscript. NCAR is sponsored by the National Science Foundation. Received for review June 27, 1996. Accepted September 9, 1996.X AC960639E X
Abstract published in Advance ACS Abstracts, October 15, 1996.
Analytical Chemistry, Vol. 68, No. 23, December 1, 1996
4199
