MEASUREMENT OF THE TERRESTRIAL OZONE CONCENTRATION BY ABSORPTION UV SPECTROSCOPY

Transcription

1 Journal of Applied Spectroscopy, Vol. 72, No. 6, 2005 MEASUREMENT OF THE TERRESTRIAL OZONE CONCENTRATION BY ABSORPTION UV SPECTROSCOPY L. M. Bolot ko, A. N. Krasovskii, * A. M. Lyudchik, and V. I. Pokatashkin UDC /.08; The conditions allowing one to measure the concentration of terrestrial ozone by the open-path DOAS method using a zero line are considered. A technique based on measuring signals at two wavelengths, at least one of which belongs to the ozone absorption UV band, is described. The error of measurements has been estimated. A block-diagram of the optical "open-path" meter for measuring the terrestrial ozone concentration, TrIO-2, is presented. Keywords: absorption spectrum, terrestrial ozone, "open-path" meter, probing UV radiation. Introduction. Ozone (O 3 ) a small fraction of the atmosphere plays a key role in the photochemical and chemical processes developing in the troposphere and exerting a marked effect on its qualitative composition. By virtue of its high reactivity, this unstable molecular formation is a product and a participant of diverse branched photochemical and chemical reactions. The concentration of ozone in the terrestrial layer of the atmosphere changes substantially depending on the season, time of day, and weather conditions and is very sensitive to anthropogenic polluting emissions into the atmosphere. By exceeding the natural levels of concentration and being one of the main components of summer smog appearing in industrial regions, ozone can accumulate in the terrestrial atmosphere [1]. In toxicity it relates to the most hazardous substances (of the first class of hazard). The majority of ozonometric and Global Atmospheric Watch (GAW) stations that monitor the state of the terrestrial atmosphere are equipped with optical analyzers of terrestrial ozone concentration. Such analyzers have a substantial drawback associated with the necessity of taking samples of air for analysis, which leads to uncontrolled destruction of a portion of the ozone and increase in the error of measurement of its concentration. Moreover, to determine variations of natural (background) concentrations of terrestrial ozone of the order of 40 µg/m 3 (20 ppb) when using an optical path of 1m it is necessary to register changes in the intensity of UV probing radiation at a level of , which is a rather complex technical problem. At the present time, open-path systems based on differential optical absorption spectrometry (DOAS stations or DOAS open-path gas analyzers) are being used more and more widely for determining the concentration of small constituents of the terrestrial atmosphere [2]. The open-path method ensures a noncontact process of measuring the path-averaged concentration of the terrestrial ozone, when there is no sampling of the air mass, which substantially decreases the error of measurements of such an unstable substance as ozone. Measurement Procedure and Description of the Setup. At the National Scientific-Research Center of Ozone-Sphere Monitoring at Belarusian State University, an optical open-path meter of the concentration of terrestrial ozone, TrIO-1, has been developed, metrologically attested, and put into operation [3]; it operates on the principle of DOAS equipment. As a result of improvement of its optical circuit, a TrIO-2 optical open-path meter was created, which employs the previous technique of determining the concentration of terrestrial ozone. In contrast to the DOAS systems in which narrow absorption lines of the gas component of the atmosphere are analyzed, the most important characteristic feature of the meters named is the possibility of determining the concentration of the terrestrial ozone * To whom correspondence should be addressed. National Scientific-Research Center of Ozone-Sphere Monitoring, 7 Kurchatov Str., Minsk, , Belarus; nomrec@bsu.by. Translated from Zhurnal Prikladnoi Spektroskopii, Vol. 72, No. 6, pp , November December, Original article submitted May 19, /05/ Springer Science+Business Media, Inc. 911

2 using the Hartley smooth diffuse absorption band. This is attained by taking a constant account of the state of the source of probing radiation at the "zero" open-path of the meter where the ozone concentration is known to be smaller than the error of its determination on the "operational" path. We find the relationships between the optical signals S 0λ and S lλ that traversed the "zero" and "operational" paths: S 0λ = I λ g 0 R λ χ λ, (1) S lλ I λ g l exp ( τ λ ) R λ χ λ, (2) where the subscripts 0 and l correspond to the signals that traversed the "zero" path not containing ozone and the "operational" path of length l; I λ is the spectral intensity of radiation from a source at the wavelength λ; R λ is the reflection coefficient of the retrodirective mirror of the "zero" and "operational" paths; χ λ is the spectral sensitivity of the registration system; τ λ is the optical thickness of the "operational" path; g 0 and g l are the geometric factors that characterize attenuation of radiation in traversing the "zero" and "operational" paths without allowance for absorption and scattering and that are independent of the probing-radiation wavelength. The logarithm of the ratio of signals (1) and (2) is Here the optical thickness of the "operational" path τ λ is defined by the sum ln (S lλ S 0λ ) = ln (g l g 0 ) τ λ. (3) τ λ = σ λ nl + β λ l, (4) where σ λ is the section for absorption of the probing radiation of wavelength λ by ozone; β λ is the volumetric coefficient of radiation attenuation due to molecular scattering at the same wavelength; n is the concentration of ozone. The difference of the logarithms of the ratio between the signals measured at two wavelengths is With allowance for (4) we obtain D = ln (S lλ1 S 0λ1 ) ln (S lλ2 S 0λ2 ) = τ λ2 τ λ1 = τ. (5) n = 1 l σ (D l β), (6) where σ = σ λ2 σ λ1, β = β λ2 β λ1. The final expression (6) that determines the concentration of ozone involves the measured, calculated, and tabulated parameters; it does not contain the spectral sensitivity of the registration path χ λ, reflection factors of the mirrors R λ, and the geometric factors g 0 and g l of the "zero" and "operational paths." To obtain an averaged concentration of terrestrial ozone over the probing path it is necessary to find the ratio of the signals at two wavelengths (5), one of which, as a minimum, must be within the absorption width of ozone. The measurements are relative and are performed for two opticals paths: "zero" and "operational" ones. The ozone absorption cross section at operational wavelengths is selected from tables [4]. The volumetric coefficient of molecular scattering β λ is determined from the semi-empirical expression for the optical thickness of the Rayleigh scattering of a vertical column of the atmosphere [5]: β λ = γ λ dp dz = γ λ p H, (7) where p is the pressure in atmospheres; H C 8 km is the height of a homogeneous atmosphere according to the barometric formula 912

3 Fig. 1. Results of measurements of the optical thickness of the atmosphere over the path: 1) optical thickness of absorption by ozone over the path (concentration of ozone 46 ppb); 2) spectrum of scattered solar radiation; 3) optical thickness of the atmosphere measured over the path of length 125 m. γ λ = λ λ λ. (8) In (8), the wavelength λ is given in micrometers. Ozone gives a number of absorption bands in the IR, visible, and UV regions of the spectrum. It is convenient to perform quantitative measurements in the Hartley band ( nm, Fig. 1, curve 1), which is characterized by a high absorption cross section in the maximum (σ λ = 253 nm C cm 2 ) [6]. The distinct advantage of the absorption UV spectrophotometry of ozone is the relatively small number of other natural and anthropogenic-origin gas components of the atmosphere, the absorption bands of which are overlapped with the absorption band of ozone. The most typical hindrance in measuring the concentration of terrestrial ozone can be sulfur dioxide, the absorption spectrum of which overlaps the entire longwave portion of the absorption spectrum of ozone. However, the maximum absorption cross section of sulfur dioxide in this spectral range and its possible concentration even at high gas pollution of urban industrial regions is two orders of magnitude lower than the same values for ozone. The concentrations of other gas components of the terrestrial atmosphere that are capable of influencing the value of the measured ozone concentration with their insignificant absorption cross sections are still lower. Consequently, in the Hartley absorption band there are practically no gaseous components of the atmosphere that substantially influence the error of measurement of the concentration of terrestrial ozone. The exception is oxygen, the concentration of which is eight orders of magnitude higher than the background concentration of terrestrial ozone and which has a number of absorption bands in the UV region of the spectrum [7]. Thus, the limit of convergence of the Herzberg band [8, 9] of oxygen is at nm; there is a weak continuum below. As the temperature increases (>0 o C), expansion of the continuum to the longwave side of up to 260 nm is observed. The absorption cross section of oxygen for the continuum is determined at extremely low gas pressures. At room temperature, at a wavelength of nm, σ D cm 2. At an insignificant absorption cross section of oxygen in the spectral region considered, its concentration and the temperature absorption spectrum extended to the longwave side do not permit one to carry out measurements of ozone concentration over the open path, on the shortwave slope, and in the region of the maximum of its absorption cross section (λ = nm). Oxygen virtually does not influence the error of measurement of the ozone concentration for λ > 260 nm, i.e., on the longwave wing of the absorption band of ozone. Farther to the longwave side, approximately from 300 nm, scattered solar UV radiation appears in the terrestrial atmosphere. Its intensity cannot be taken into account in the probing-radiation spectrum, since it depends randomly on the time needed for carrying out measurements, cloudiness, transparency of the atmosphere, and a number of other meteorological parameters. Thus, the spectral interval that can be used for path measurements of the concentration of terrestrial ozone is limited by two factors: absorption of the probing radiation by atmospheric oxygen and the presence of scattered solar radiation (Fig. 1, curve 2). According to the measurement procedure, in the nm range a pair (or several pairs) 913

4 Fig. 2. Optical-electronic circuit of the meter. of operational wave lengths is selected. The metrological attestation of the TrIO-1 meter was performed for the pair λ 1 = 268 nm and λ 2 = 297 nm. In a TrIO-2 meter in this very spectral interval several operational waves were selected that form several pairs for calculating the concentration of ozone. The final result of measuring the concentration of ozone over the path represents an average value. In both meters, to bind the concentration of terrestrial ozone there is an extended (more that 100 m) "operational" path over which a change in the optical density of the medium due to the UV-radiation absorption by the gas analyzed is determined relative to the "zero" path over which the gas analyzed (ozone) is absent. The path measurements make it possible to find minimum concentrations of small components of the atmosphere by recording relative changes of the order of 10 3 in the probing-radiation intensity, which does not impose high requirements on the complex of equipment used. Figure 2 presents an optical circuit of the TrIO-2 path meter of the concentration of terrestrial ozone. A halogen lamp of the KGM type with a filament temperature of 3000 o C is used as a source of probing radiation 1. Formation of a probing weakly diverging light beam is made by a spherical reflector 2. The "operational" path is formed by retrodirective mirror 3. To obtain the "zero" path, mirror 4 is introduced, which screens mirror 3. The mirrors of the "operational" and "zero" paths have the same reflection coefficient. To simplify the process of alignment and long preservation of the parameters of the "operational" path, retrodirective mirror 3 can be replaced by a quartz angular reflector, and its reflective characteristics must correspond to those of mirror 4 of the "zero" path or be taken into account in the process of processing of the optical signals recorded. The application of the retrodirective mirror 3 curtails twice the "operational" path, preserving the total length of the optical path (l = 2L), and allows one to place a source and receiver of probing radiation in one functional unit. The receiving recording unit consists of a focusing quartz lens 5, monochromator 6, FE U with a current-voltage converter, and an analog-to-digital (ADC) converter 7. To reduce the level of the dark current and decrease the influence of scattered light, a solar-blind photomultiplier tube is used. To exclude the direct incidence of light from a source of probing radiation onto the monochromator inlet, a protecting screen 8 is installed at the center of the reflector hole. The level of the scattered light falling on the focusing lens is taken into account in the process of measurement. Moreover, it is decreased to a minimum by applying diaphragms 9 and increasing the distance between the focusing lens and reflector. Monitoring of the meter and processing of the results are done with the aid of an intermediate microprocessor 10 and a personal computer 11. Estimation of the Measurement Error. The absolute error of measurement of the ozone concentration is calculated by the formula obtained from Eq. (6) by differentiation: δn = δ l σ D l σ [ε l + ε σ ] β σ [ε β ε σ ], (9) where ε l is the relative error of measurement of the operational path length; ε σ is the relative error of determining the difference of absorption cross sections of ozone at the two operational wavelengths used; ε β is the relative error of determining the difference between the coefficients of Rayleigh scattering, and δd is the absolute error of measurement of the value of D. Assuming that l C 100 m and σ C cm 2, for λ 2 near a wavelength of 260 nm and for λ 1 near 290 nm, as well as β/ σ C cm 3 for a pair of wavelengths in the nm range and summing the contributions to the net error, we, according to [4, 5], find 914

5 δn δ + n [ε l + ε σ ] [ε β ε σ ], (10) where an approximate equality n C D/(l σ) is used for concentrations at the background level. The first term in (10) represents the error associated with random errors of registration of four optical signals δd. The second and third terms determine the contribution of the persisting (not excluded) systematic errors of determination of the path length l, the difference between the cross sections of ozone absorption σ, and the difference between the coefficients of molecular scattering. The error of setting monochromator wavelengths as well as the error associated with the displacement of the zero level of the registration system and other errors that influence the determination of the concentration of terrestrial ozone and having a random character can be taken into account in the final result as errors of determination of the probing-radiation intensity and be evaluated by direct testing of the measurement system. The absolute error of determination of the "operational" path length l using an RL-50 laser tape measure is ±1 cm and practically does not influence the error of determination of ozone concentration. The error of experimentally measured absolute values of the cross sections of absorption of UV radiation by ozone is estimated to be 4% [8], although the error of measuring the relative spectral dependence of the absorption cross section is considerably lower (of the order of 1%). Allowance for the temperature dependence σ within climatic scattering of temperatures does not bring out the final error ε σ outside the threshold adopted. According to [5], the relative error of determination of the coefficient of molecular scattering β does not exceed 3%. Consequently, the contribution of a systematic error to the error of determination of concentration appears to be less than cm 3 (less than 1 ppb) in measuring concentrations close to the background ones (of the order of 20 ppb) and can be considered insignificant. Thus, the main error of measurement of concentration is determined by the errors of registration of intensity and by probingradiation instability. To estimate the value of δd we shall avail ourselves of the generally accepted procedure [10], which, subject to Eq. (5), yields 2 δd C ε i 2ε, (11) i=1 4 where ε i is the relative error of measurement of each of the four signals and ε is the maximum value of this error. The construction of the meter ensures the closeness of all four signals in absolute value within one order of magnitude. A small increase in the relative error of measurements of a signal over the operational path is due to the optical-thickness fluctuations caused by turbulence and convection. The relative error of measurement of precisely these signals is taken as the upper limit of the registration error. Since the signals registered differ little in intensity, there is no need to take into account the possible nonlinearity of the measuring path. Thus, a good estimate of the absolute error of measurement of the concentration of terrestrial ozone that differs little from the phase one is δn C ε. (12) Whence it follows that the absolute error of measuring the concentration of ozone does not exceed 5 ppb under normal conditions (δn cm 3 ) if the maximum possible relative error of registration equals approximately 0.5%. Conclusions. According to the program developed and the procedure adopted, metrological attestation of the measurement means, "TrIO-1 Optical Open-Path Meter of the Concentration of Terrestrial Ozone," has been carried out. According to the protocol, TrIO-1, intended for determining the concentration of ozone in the air atmosphere under natural conditions (without taking samples), has the following basic technical characteristics: range of the measured concentrations of ozone ppb; limits of the basic absolute error ±1.45 ppb. It is intended to carry out metrological attestation of a TrIO-2 meter. In the process of investigations, the relationships between the recorded spectra of the "zero" and "operational" paths for different weather conditions corresponding to different contents of ozone in the terrestrial atmosphere have been established. The spectra of absorption of UV radiation in the nm range have been analyzed. For a path of length 125 m background concentration of terrestrial ozone of ( cm 1 = 20 ppb), the relative intensity of 915

6 the spectrum of the "operational" path in the maximum of absorption decreases by C5%. The difference of the logarithms of the intensities of the spectra of the "zero" and "operational" paths (3) must reproduce (accurate to the shift caused by the difference of geometrical factors) the spectral behavior of absorption by ozone over the path distorted by the errors of spectra registration. Such a reproduction is possible in the absence or insignificant influence, on the spectral intensity, of the path signal of other absorbers or sources of parasitic illuminations. The spectra of the "operational" path have an appreciable noise component caused by the turbulence-caused instability of the density of the atmosphere, convection flows, etc. In fact, the optical thickness measured over the path practically coincides with the optical thickness of absorption by ozone (Fig. 1, curve 3) only in the range of wavelengths D nm. During motion to the shortwave and longwave sides from the indicated spectral interval, deviations from the shape of the contour of the absorption band of ozone, which is probably caused by the absorption of oxygen and by the contribution of solar radiation scattered in the atmosphere to the signal measured are observed. REFERENCES 1. Atmosphere: Handbook [in Russian], Gidrometeoizdat, Leningrad (1991). 2. U. Platt, D. Perner, and H. W. Patz, J. Geophys. Res., 84, (1979). 3. L. M. Bolot ko, V. I. Pokatashkin, A. N. Krasovskii, and V. L. Tavgin, Opt. Zh., 71, (2004). 4. Technical Assistance Document for the Calibration of Ambient Ozone Monitors, EPA-600/ (1970). 5. C. Frohlich and G. E. Shaw, Appl. Opt., 19, (1980). 6. V. V. Lunin, M. P. Popovich, and S. N. Tkachenko, Physical Chemistry of Ozone [in Russian], Izd. MGU, Moscow (1998). 7. H. Okabe, Photochemistry of Small Molecules [Russian translation], Mir, Moscow (1981). 8. V. Hasson and R. W. Nickolls, J. Phys. B: Atom. Mol. Phys., 4, (1971). 9. M. Ogawa, J. Chem. Phys., 54, (1971). 10. G. P. Gushchin, Meteorol. Gidrolog., No. 4, (1984). 916