The visible absorption spectrum of NO 3 measured by high-resolution Fourier transform spectroscopy

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. D3, 4077, doi: /2002jd002489, 2003 The visible absorption spectrum of NO 3 measured by high-resolution Fourier transform spectroscopy J. Orphal Laboratoire de Photophysique Moléculaire, CNRS, Orsay, France C. E. Fellows Laboratório de Espectroscopia e Laser, Instituto de Física, Universidade Federal Fluminense, Niterói, Brazil P.-M. Flaud Laboratoire de Photophysique Moléculaire, CNRS, Orsay, France Received 29 April 2002; revised 23 October 2002; accepted 20 November 2002; published 4 February [1] The visible absorption spectrum of the nitrate radical NO 3 has been measured using high-resolution Fourier transform spectroscopy. The spectrum was recorded at 294 K using a resolution of 0.6 cm 1 (corresponding to nm at 662 nm) and covers the cm 1 region ( nm). Compared to absorption spectra of NO 3 recorded previously, the new data show improvements concerning absolute wavelength calibration (uncertainty 0.02 cm 1 ), and spectral resolution. A new interpretation and model of the temperature dependence of the strong (0-0) band around 662 nm are proposed. The results are important for long-path tropospheric absorption measurements of NO 3 and optical remote sensing of the Earth s atmosphere from space. INDEX TERMS: 0317 Atmospheric Composition and Structure: Chemical kinetic and photochemical properties; 0340 Atmospheric Composition and Structure: Middle atmosphere composition and chemistry; 0365 Atmospheric Composition and Structure: Troposphere composition and chemistry; KEYWORDS: NO 3, nitrate, highresolution, spectroscopy, visible, atmosphere Citation: Orphal, J., C. E. Fellows, and P.-M. Flaud, The visible absorption spectrum of NO 3 measured by high-resolution Fourier transform spectroscopy, J. Geophys. Res., 108(D3), 4077, doi: /2002jd002489, Introduction [2] The nitrate radical NO 3 is amongst the most important atmospheric species [Wayne et al., 1991]. It participates in the cycles of stratospheric ozone destruction [World Meteorological Organization (WMO), 1998] and in tropospheric chemistry [Commission of the European Community (CEC), 2001]. Its concentrations vary with geographical location and altitude and present strong diurnal and seasonal variations [e.g., Platt et al., 1981; Solomon et al., 1989; Heintz et al., 1996]. In the present paper, new high-resolution laboratory measurements of the absorption spectrum of NO 3 at room temperature and a new interpretation of the temperature dependence of the strong (0-0) band around 662 nm are presented. [3] Atmospheric NO 3 has first been detected by Platt et al. [1980] and Noxon et al. [1980] using absorption spectroscopy of the strong electronic bands between nm. There is currently particular interest in monitoring tropospheric concentrations and vertical concentration profiles of NO 3 [Weaver et al., 1996; Aliwell and Jones, 1998; Allen et al., 1999; Fish et al., 1999; Renard et al., 2001; von Copyright 2003 by the American Geophysical Union /03/2002JD Friedeburg et al., 2002]. Recently the use of Cavity-Ring- Down spectroscopy has demonstrated high sensitivity for measuring very weak concentrations of NO 3 in situ [King et al., 2000; Ball et al., 2001]. [4] The absorption cross sections of NO 3 in the visible have been the subject of many laboratory studies [see Wayne et al., 1991; Yokelson et al., 1994; International Union of Pure and Applied Chemistry (IUPAC), 2001, and references therein]. The value currently recommended by international panels is a peak cross-section at 662 nm between ( ) cm 2 molecule 1 at room temperature [Wayne et al., 1991; DeMore et al., 1997; IUPAC, 2001]. A recent study has proposed that there could be an overestimation (as much as 17 percent) in the recommended peak cross-section value, derived from simultaneous DOAS and ESR measurements [Geyer et al., 1999]. These authors concluded that future experimental work on the NO 3 absorption cross sections would be desirable. However the most recent studies of absolute absorption cross sections at room temperature show very good agreement [Sander, 1986; Wayne et al., 1991; Yokelson et al., 1994; IUPAC, 2001, and references therein]. Therefore, it is possible that there are other sources for systematic errors than the value of the absolute absorption cross sections, including differences in the observing geometry [von Friedeburg et al., 2002]. ACH 1-1

2 ACH 1-2 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 [5] In addition, the temperature dependence of the absorption cross sections of NO 3 is well established [Sander, 1986; Yokelson et al., 1994; IUPAC, 2001] and needs to be accounted for in the analysis of atmospheric spectra. These variations, together with the fact that the atmospheric NO 3 absorption can be very small and is blended by strong solar Fraunhofer lines and atmospheric H 2 O absorption [see, e.g., Renard et al., 2001; von Friedeburg et al., 2002] necessitates very accurate reference data, in order to avoid systematic errors in the retrieved atmospheric concentrations. A recent sensitivity study on BrO (where similar absorption structures are used for atmospheric remote sensing) showed that, in addition to the absolute values of the absorption cross sections and their temperature dependence, the absolute wavelength calibration of the reference cross sections with respect to the wavelength calibration of strong absorbers is extremely important [Aliwell et al., 2002]. Wavelength calibration errors of a few tenth of a nm can lead to errors in the retrieved BrO column of up to 23 percent. A similar situation is expected for NO 3. [6] In the last years, new remote-sensing reference spectra (O 3,NO 2, OClO, BrO, SO 2,H 2 CO, IO,...) [see Orphal et al., 2002, and references therein] were recorded using high-resolution Fourier transform spectroscopy (FTS), with very accurate wavelength calibration and an instrumental line shape much narrower than that of state-of-the-art remote-sensing instruments. While in some cases these new spectra may be less accurate concerning absolute absorption-cross sections, they are an important element in establishing the absolute wavelength calibration of the reference absorption cross sections [e.g., Mohammed-Tahrin et al., 2001; Orphal, 2002]. The goal is to have all reference spectra for atmospheric remote sensing on a common wavelength scale, accurate to 0.01 nm or better, in order to avoid spectral shifting of the individual reference spectra in the data analysis [Platt, 1999]. This is particularly important for the automatic processing of atmospheric spectra obtained from multispectral satellite sensors that is required due to the high data rates of these instruments. Recent satellite-borne instruments like GOME [Burrows et al., 1999], SCIAMACHY [Bovensmann et al., 1999], and SAGE-III [SAGE-III ATBD Team, 2002] aim to detect atmospheric NO 3 from space, including measurements of NO 3 vertical columns from Nadir-viewing geometry (GOME), determination of vertical concentration profiles using limb sounding (SCIAMACHY) or using lunar occultation (SAGE-III). [7] For these reasons we decided to record new absorption spectra of the NO 3 electronic bands between nm by FTS. In particular, the only previous set of spectra of NO 3 recorded with FTS [Cantrell et al., 1987] is not available in digital form as mentioned in the paper by Yokelson et al. [1994]. Although it has been known for over 40 years that the NO 3 absorption features do not show very high-resolution structure due to strong predissociation [Ramsay, 1963; Marinelli et al., 1982; Kim et al., 1992], the advantages of accurate wavelength calibration and of high spectral resolution provided by FTS remain. As alraeady said, these advantages can significantly reduce systematic errors in the remote-sensing data analysis [Mohammed- Tahrin et al., 2001; Aliwell et al., 2002]. Because the absolute value of the absorption cross sections of NO 3 are known to high accuracy, we did not attempt to improve them in the present study. However, we have reanalyzed the available data concerning the temperature dependence of the strong (0-0) band around 662 nm and we propose a new formula to obtain absolute absorption cross sections at different temperatures, based on a physical model that takes into account the changing population of the vibrational ground-state of NO 3. [8] The remainder of the paper is organized as follows: first, the experimental setup and procedure are described. Thereafter, the data reduction is presented and the results of the present study are discussed and compared with previously published work. Finally we discuss the temperature dependence of the strong (0-0) band around cm 1 (662 nm) together with some conclusions for atmospheric remote sensing. 2. Experiment [9] For recording the NO 3 spectra, the Bruker IFS-120 HR Fourier transform spectrometer at LPPM Orsay was used. The instrument was equipped with a Quartz beamsplitter and a Si diode detector. The wavelength calibration is accurate to 0.02 cm 1 based on simultaneous coverage of the O 2 A-band absorption lines around 762 nm that were used for absolute calibration with the ICLAS data of [O Brien et al., 2001] (see Figures 1 and 2). The spectral resolution was 0.6 cm 1 (unapodized) but some spectra were also recorded at higher resolutions (up to 0.06 cm 1 ) to validate that no saturation effects were present (linear range of the Beer-Lambert law). Except for increased noise, no differences in the spectral shape of the NO 3 absorption were observed at higher resolution, in agreement with previous studies [Ramsay, 1963; Marinelli et al., 1982]. Typically several blocks of 200 interometer scans each were recorded during 1 2 hours and averaged after inspection. As white light source, a Quartz-Tungsten-Halogen lamp (Osram, 50 W, 12 V) was used with a stabilized power supply. Therefore, the baseline stability of the setup was very good (better than 1%). As shown in Figure 1, there is no error in the phase correction of the interferograms that would lead to asymmetric line shapes. [10] The absorption cell was 300 cm long and 34 mm in diameter, made of Pyrex glass and equipped with CaF 2 windows. Relatively high concentrations of NO 3 radicals inside the cell were obtained by static mixtures of N 2 O 5 and O 3 using a method employed already by Jones and Wulf [1937] and by Graham and Johnston [1978]: Small quantities of NO 2 (Air Liquide, 98% stated purity) were introduced into the absorption cell and mixed with much higher amounts of O 3 diluted in O 2 (Air Liquide, 99.98% stated purity). This leads to constant concentrations of NO 3 assuming steady state conditions, as described in detail by Graham and Johnston [1978] and by Marinelli et al. [1982]. [11] The O 3 was produced with a commercial silentdischarge ozone generator (Sander Germany). Typical NO 3 concentrations were about molecule cm 3 in mixtures of up to molecule cm 3 of O 3, using initial NO 2 concentrations of molecule cm 3 or less. These numbers are in good agreement with calculations based on the steady state equations proposed by Graham and Johnston

3 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 ACH 1-3 Figure 1. The O 2 A-band recorded simultaneously together with the NO 2 spectra for wavelength calibration. The upper trace shows the raw (unapodized) data, the lower trace shows the data after apodization with a Norton-Beer (medium) function. The raw spectrum was shifted upward for clarity. [1978]. The reaction of NO 2 with NO 3 is fast so that steady state conditions were always achieved within a few seconds after closing the cell. The only problem encountered were initial H 2 O impurities on the wall of the Pyrex cell that efficiently destroyed the N 2 O 5, as already described by Marinelli et al. [1982]. After passivation of the cell, high stability of the NO 3 inside the closed reactor was observed, in good agreement with the studies by Graham and Johnston [1978] and by Marinelli et al. [1982]. The total pressure in the cell varied between hpa, measured using calibrated capacitive manometers (MKS Baratron). [12] The stability of the NO 3 samples inside of the absorption cell was monitored using a diode-array spectrometer (Ocean Optics PC2000) covering the nm region at a spectral resolution of 1.1 nm. Using this device, absorption spectra were displayed every 2 seconds during the high-resolution interferometer scans. Once chemical equilibrium was achieved, the NO 3 peak absorbance (typically between , i.e % transmittance) varied by less than 0.01 (corresponding to changes of less than 2% in the NO 3 steady state concentrations) during 1 2 hours. [13] Note that previous studies used different sources for the NO 3 production, including techniques that avoid other absorbers in the spectral range of the NO 3 absorption, like flash photolysis of ClONO 2 [Sander, 1986] and the reaction between fluorine atoms and nitric acid [Yokelson et al., 1994]. However, the only strong absorbers observed in our setup were NO 3 and O 3, and as will be shown below it is possible to use the differential structures of O 3 in the nearinfrared (the red wing of the Chappuis band [Bogumil et al., 2001]) to scale the reference cross sections of O 3 for accurate spectral substraction. [14] The NO 2 concentrations were always very small once chemical equilibrium was reached, and N 2 O 5 has no strong absorption features in the spectral region nm [DeMore et al., 1997]. The only species observed independently of NO 3 was O 3, the NO 2 absorption being too small for accurate concentration determination (see below). Therefore, we estimated that the uncertainties in the determination of the concentrations of all relevant species (NO 3,NO 2,N 2 O 5,O 3 ) in our setup, together with the uncertainties of the rate coefficients for the chemical reactions involved [DeMore et al., 1997; IUPAC, 2001], would not lead to accurate absolute cross sections of NO 3 in our system. Titration with NO might have been possible; however, as stated above, the goal of the present study was not to improve the accuracy of absolute cross sections of NO 3 but the wavelength calibration of the NO 3 reference

4 ACH 1-4 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 Figure 2. Spectral calibration using absorption lines in the O 2 A-band together with reference line positions from O Brien et al. [2001]. The upper plot shows the linear relation between the observed line positions and the reference data, the lower plot shows the absolute differences. The root-mean-square deviation between the O 2 line positions in the calibrated spectrum and the reference line positions is 0.02 cm 1. data for atmospheric remote sensing. Although rapid-scan Fourier transform spectroscopy in the ultraviolet and visible is able to deal with time-dependent spectral sources (like flash-photolysis or modulated light sources) (O.-C. Fleischmann et al., Ultraviolet absorption cross sections of BrO measured by time-windowing Fourier transform spectroscopy, submitted to Journal of Photochemistry and Photophysics A Chemistry, 2002), we have not applied this technique in the present study. 3. Data Reduction, Results, and Discussion 3.1. Absorption Spectra [15] Reference spectra of the empty cell were recorded before and after each NO 3 spectrum. Typically, the stability of the I 0 baseline was better than 1%. Absorption spectra were obtained by application of the Beer-Lambert law. The raw absorption spectra (see upper trace of Figure 3) show the strong NO 3 electronic bands between cm 1 and in addition the Chappuis band of O 3 as an underlying continuum. One can also recognize the O 2 A- band around cm 1 (see inset) and the Wulf bands of O 3 in the near-infrared [Bogumil et al., 2001]. The sharp peak around cm 1 is straylight from the wavelengthstabilized He-Ne laser used for operation of the FTS. Some spectra were also recorded with pure O 3 /O 2 mixtures in the cell to validate that the structures observed in the nearinfrared were indeed only due to the Chappuis bands of ozone and the O 2 A-band. Note that the noise is increasing toward higher wavenumbers due to the decreasing total signal of the white light source. [16] To correct the raw spectra for the underlying O 3 absorption and for the reference laser straylight, the following procedure was applied. First, the NO 3 absorption and reference spectra were both corrected for the laser straylight by subtraction of a laser spectrum recorded without light source. Then, the O 3 optical density was determined using the absorption features of the Wulf bands in the nearinfrared (see inset in the upper trace of Figure 3) together with the O 3 reference spectra recorded with the SCIA- MACHY instrument before launch [Bogumil et al., 2001]. The latter data were chosen because of their high signal/ noise ratio and coverage of the entire O 3 spectrum up to 1000 nm, at sufficient resolution to resolve the differential

5 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 ACH 1-5 Figure 3. Correction of the raw spectra for the underlying O 3 absorption. The O 3 amounts were determined using the Wulf bands in the near-infrared (see inset) together with the reference data of Bogumil et al. [2001]. The upper plot shows the raw spectrum and the O 3 optical density used for the correction, the lower plot shows the spectrum after spectral substraction of the O 3. structures free of spectral artifacts (see the discussions of Orphal [2002]). The corrected FTS spectrum of our study is shown in the lower trace of Figure 3. Note that this procedure is completely independent of the absolute absorption cross-section of O 3. [17] The well-known NO 2 electronic bands can contribute to the absorbances at wavelengths above cm 1 [see Sander, 1986] and therefore the O 3 -corrected spectra were also corrected for NO 2 (see Figure 4) using scaled and convoluted optical densities from FTS work at high resolution [Voigt et al., 2002] that show very good wavelength accuracy as validated by comparison with other data [Orphal, 2002; Vandaele et al., 2002]. This correction changes the optical densities only significantly above cm 1 and can be neglected at lower wavenumbers, because the resulting optical density of NO 2 is less than 0.01 at wavenumbers below cm 1 and less than at wavenumbers below cm 1. [18] The overall uncertainty of this procedure is estimated to be smaller than 0.02 in the optical density at the top of the Chappuis band (610 nm), corresponding to a relative uncertainty that is less than 3% in the region of the 662 nm peak. In regions with weaker O 3 absorption the uncertainty is becoming even smaller. However, for regions where the NO 3 absorption cross sections are small (valleys between the strong peaks), the relative uncertainty introduced by the possible error in the O 3 correction becomes larger. This is clearly a drawback of the production method we employed in this study, compared to flash photolysis of chlorine nitrate or reaction of fluorine atoms with HNO 3. [19] In contrast, note that in the entire region cm 1, the noise accounts only for an optical density of about (peak to peak), which is a good result for a spectrum at spectral resolution (FWHM) of 0.6 cm 1 (i.e. about nm). However, at the high energy end (above cm 1 i.e. at wavelengths below 525 nm) the signal/ noise ratio in our spectra becomes rather small at high spectral resolution due to the decreasing intensity of the white-light source together with the smaller sensitivity of the diode detector at higher energies. Convolution of the NO 3 spectrum to the resolutions typically used for atmospheric remote sensing ( nm, i.e cm 1, FWHM of the instrumental line shape of the grating spectrometers) would improve the signal/noise ratio due

6 ACH 1-6 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 Figure 4. Correction of the spectra for the underlying NO 2 absorption. The NO 2 amounts were determined using the optical density at wavenumbers above cm 1 together with the reference data of Voigt et al. [2002]. The upper plot shows the O 3 -corrected spectrum and the NO 2 optical density used for the correction, the lower plot shows the spectrum after spectral substraction of the NO 2. to the averaging properties of this mathematical process. In addition, the spectral region used for atmospheric remote sensing of NO 3 is generally limited to wavelengths above 600 nm (wavenumbers below cm 1 ) where the signal/noise ratio in our spectrum is higher than 100 (see below) Comparison to Previous Studies [20] The peak cross-section of the strongest band around cm 1 (662 nm) is dependent on spectral resolution and wavelength calibration [see, e.g., Burrows et al., 1985]. Therefore, in order to obtain absolute absorption cross sections we decided to calibrate the NO 3 absorbance spectrum of the present study using integrated absorption cross sections. The reason for this choice is that integrated absorption cross sections are much less sensitive to differences in spectral resolution than using peak cross sections [Orphal, 2002]. By recommendation of the IUPAC, the reference peak cross-section at 662 nm is (2.1 ± 0.2) cm 2 molecule 1 at room temperature [IUPAC, 2001] which is actually the value recommended by Wayne et al. [1991]. Note, however, that another international panel [DeMore et al., 1997] recommends a value of (2.0 ± 0.2) cm 2 molecule 1 and that, if the more recent value of Yokelson et al. [1994] is used instead of the value of Ravishankara and Mauldin [1986] in the averaging procedure, a value of (2.2 ± 0.1) cm 2 molecule 1 is obtained. If new and more accurate measurements of the peak absorption cross sections of NO 3 become available in the future, they can be used to rescale the data of Sander [1986], Yokelson et al. [1994], and of this work, if necessary. [21] In this study we used the following procedure for obtaining absolute NO 3 cross sections. First, we scaled the room temperature absorption cross-section of Sander [1986] by multiplication with a factor of to obtain the peak value recommended by IUPAC. Second, we calculated the integrated absorption cross-section between and cm 1 ( nm) using this scaled data. The value obtained in this way, cm molecule 1 is smaller than that obtained using the unscaled cross sections of Yokelson et al. [1994], cm molecule 1, corresponding to a difference of 3%. It is interesting to note that the peak cross-section of Yokelson et al. [1994] at 662 nm is actually 6% higher than the value recommended by IUPAC, namely (2.23 ± 0.22) cm 2 molecule 1. The reason why we used the cross-section of Sander [1986]

7 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 ACH 1-7 Figure 5. Comparison of the NO 3 cross sections at room temperature (shifted upward for clarity) with the data of Yokelson et al. [1994]. Note that the spectral resolution of the latter study is about 0.5 nm (interpolated to a 0.2 nm grid), while the spectral resolution of the FTS data is 0.6 cm 1 (about nm at 662 nm). The lower trace shows the relative difference between both sets of data. The feature around cm 1 is a residual due to the calibration laser in our setup that was not completely removed by spectral substraction but becomes much smaller during convolution of the cross sections. for calculation of the integrated absorption cross sections and not that of Yokelson et al. [1994] is that the data of Sander [1986] are explicitly recommended by IUPAC. The difference of only 3% in the integrated absorption cross sections shows that both studies agree very well within their experimental uncertainties. Finally, we normalized our integrated optical density to unity and multiplied the data with the integrated absorption cross sections obtained from the scaled cross sections of Sander [1986]. [22] To compare the absolute absorption cross-sections NO 3 of this work with previous studies, they have to be convoluted with the appropriate instrumental line shapes which are often not given in the literature for spectra recorded with other techniques than FTS. In addition the spectra need to resampled to the wavelength grid at which the other data are given. In addition, wavelength calibration errors can contribute to the differences. Accurate comparison of different laboratory spectra is therefore a difficult issue [Orphal, 2002]. However, in the case of NO 3, even without convolution and resampling, there is good agreement between the results of the present study and the room temperature cross sections by Yokelson et al. [1994] as shown in Figures 5 and 6. This is mainly due to the widths of the spectral features of NO 3 induced by predissociation broadening [Marinelli et al., 1982]. For these plots, the wavenumber scale of the data of Yokelson et al. [1994] was shifted toward the red by 6.27 cm 1 (0.275 nm) and the cross sections of Yokelson et al. [1994] were reduced by 3% for better agreement. Note that the stated wavelength accuracy in the data of Yokelson et al. [1994] was ±0.1 nm, in agreement with the value determined above. The remaining differences are less than 5% (except for some residual structures of the reference laser around cm 1,see Figures 5 and 6) and are larger only in regions with small cross sections. This latter problem might be due to small baseline differences together with the necessity of O 3 correction in the present work. The differences become larger (more than 10%) in the region below cm 1 but this is a region where our setup actually had a very good/signal ratio due to the strong output of the QTH lamp, so further work is required to clarify this issue. [23] The only previous broadband high-resolution study of the NO 3 absorption spectrum in the visible is the work by Marinelli et al. [1982], who used a tunable dye laser

8 ACH 1-8 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 Figure 6. The strongest part of the NO 3 band cross sections at room temperature, comparing the data of the present study (shifted upward for clarity) with the data of Yokelson et al. [1994]. The conditions are the same as in Figure 5. There is clearly an asymmetry in the band contour of the (0-0) band around cm 1, in agreement with previous work [Marinelli et al., 1982]. Note that there are small vibrational/rotational features in the region around cm 1 (628 nm) that were much less pronounced in the spectra from previous studies and hidden by noise in the data by Marinelli et al. [1982]. The feature around cm 1 is a residual due to the calibration laser in our setup that was not completely removed by spectral substraction but becomes much smaller during convolution of the cross sections. (0.05 nm resolution FWHM) for measuring absorption spectra of NO 3 between cm 1 ( nm). In comparison to the present work (0.06 cm 1 FWHM), the data by Marinelli et al. [1982] show more noise (see Figures 5 and 6 of this paper compared to Figure 4 of the paper by Marinelli et al. [1982]) at lower resolution (about 1.2 cm 1 FWHM) and have a smaller wavelength coverage. Note that due to the higher signal/noise ratio in the data of the present study, some vibrational/rotational structure in the second band around cm 1 (628 nm) that is hidden by noise in the spectrum of Marinelli et al. [1982] is now clearly observed. We estimate the peak signal/noise ratio in our new data to about 400 (at a spectral resolution of 0.6 cm 1 i.e. about nm) while from the paper by Marinelli et al. [1982] one can estimate their peak signal/ noise ratio to about 50, using the 0-0 band around cm 1. In addition, the absolute wavelength calibration of the data by Marinelli et al. [1982] is only accurate to about 0.08 nm (about 2 cm 1 ) which is much less than the wavelength accuracy demonstrated in the present study (0.02 cm 1 i.e. less than 1 pm). [24] Marinelli et al. [1982] have proposed that the 0-0 band between can be modeled by two superimposed Lorentzian functions with linewidths (FWHM) of 96 and 60 cm 1, respectively. There are, however, several small peaks in this band (see Figure 5), in particular at the top of the band, which are not reproduced by this model. In addition, the asymmetry of the 0-0 band is significantly stronger than predicted by the Lorentzian model (see Figure 5 of the paper by Marinelli et al. [1982]), in particular around the peak and on the high energy tail of the 0-0 band. [25] The width of the (0-0) band of NO 3 around 662 nm is about 80 cm 1 but the structure is slightly asymmetric. Clearly, a spectral resolution (FWHM) of at least 10 cm 1 (i.e. better than 0.35 nm) is required to mimimize convolu-

9 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 ACH 1-9 tion effects leading to smaller differential structure. Note that the peak cross-section at 662 nm of this study ( cm 2 molecule 1 ) is slightly higher than the value recommended by IUPAC, as already observed by comparing the integrated cross sections and peak value of Yokelson et al. [1994] with the IUPAC recommendations, and might change only slightly with spectral resolution. However, for atmospheric remote-sensing applications, the detailed impact of wavelength shifts and spectral convolution should be evaluated by several independent groups, similar to the study recently published for BrO [Aliwell et al., 2002]. [26] The cross-section spectrum obtained in the present study is available in digital form (ASCII format) upon request to any of the authors. In addition, the data have been submitted to the journal as supplementary material to this paper. 1 1 Supporting data are available via Web browser or via Anonymous FTP from ftp://kosmos.agu.org, directory append (Username = anonymous, Password = guest ); subdirectories in the ftp site are arranged by paper number. Information on searching and submitting electronic supplements is found at Temperature Dependence of the Cross Sections [27] For some time there has been discussion about the temperature dependence of the cross-sections NO 3 [Sander, 1986; Ravishankara and Mauldin, 1986; Cantrell et al., 1987; Yokelson et al., 1994]. It is well established that there is a temperature dependence of the band shapes and absolute cross sections in the entire nm region. However, an interesting issue is that the (0-0) band between nm ( nm) shows an overall increase in cross-section but no change of shape when going to lower temperatures (see Figure 3 in the paper by Ravishankara and Mauldin [1986]). The measured increase toward lower temperatures ranges from an 18% change between 298 K and 230 K [Sander, 1986] to 42% between 298 K and 220 K[Ravishankara and Mauldin, 1986]. A more recent study obtained an increase of 36% when going from 298 K to 200 K[Yokelson et al., 1994]. [28] Up to now there is no clear understanding for the reasons of this behaviour. Ravishankara and Mauldin [1986], who were the first to show that the shape of the (0-0) band is independent of temperature, proposed that changes in the ground-state rotational population together with electronic transition probabilities that are strongly rotational quantum number dependent are responsible for the overall increase of the absolute absorption cross sections toward lower temperatures, while predissociation in the upper electronic state (in agreement with the paper by Marinelli et al. [1982]) leads to a band contour that is temperature independent. [29] There is, however, experimental and theoretical evidence [Kim et al., 1992; Hirota et al., 1997] for a low-lying vibration at 368 cm 1 in NO 3 : the in-plane bending vibration n 6 that is two-fold degenerate within the D 3h symmetry group. Hirota et al. [1997] have clearly demonstrated by rotational analysis that NO 3 has D 3h symmetry in its ground electronic state. Using these observations we have calculated the population of the ground vibrational state as a function of temperature (Figure 7), taking into account the two lowest vibrations at 368 cm 1 (together with its twofold degeneracy) and at 762 cm 1, and we obtained a population increase of 34% between 298 K and 200 K, and of 24% between 298 K and 230 K. This is extremely close to the experimental values of the cross-section increase (in particular in excellent agreement with the most recent study of Yokelson et al. [1994]) and proves that the reason for the temperature dependence of the (0-0) band is indeed a changing population of the ground vibrational state. This is also additional evidence for the D 3h symmetry of NO 3 in its electronic ground state, as already demonstrated by the rotational analysis of several vibrational bands [see Hirota et al., 1997, and references therein]. [30] The calculation is simply based on the following formula sðtþ=sð298kþ ¼ 1 e 1096:4=T 2 e 529:5=T 1 e 1096:4=298:0 2 e 529:5=298:0 where the values of and correspond to the vibrational energies of 368 cm 1 and 762 cm 1 divided by the Boltzmann constant ( cm 1 K 1 ) and T is the absolute temperature in degrees K. The agreement with the experimental data [Sander, 1986; Ravishankara and Mauldin, 1986; Yokelson et al., 1994] is indeed excellent, see Figure 7. The error analysis shows that an uncertainty in the vibrational energy of the n 6 state of 20 cm 1 changes the ratio of the vibrational ground state population only by less than 3% between 298 K and 200 K. This model (that is based on simple physical reasons) is probably more accurate than the experimental data, that have relative errors in absolute cross sections of at least 4% at each temperature (which is, considering the fact that NO 3 is a reactive and unstable radical, a very high accuracy). The only assumption made here is that the absorption due to electronic transitions arising from excited vibrational states can be neglected when looking at the (0-0) band around 662 nm. We believe that this simple model is useful for modeling the cross sections of the (0-0) band at temperatures below room temperature. [31] The consequences are rather important for atmospheric remote sensing. First, concerning the band around 662 nm, one can use the room temperature cross sections scaled to the corresponding values (obtained from the ground-state population differences) for all temperatures below room temperatures, at instrumental resolutions larger than 0.4 nm, without systematic errors due to the spectroscopic reference data. Second, if one wants to obtain atmospheric NO 3 concentrations from spectra recorded in other geometries than tropospheric long-path absorption experiments or space- and balloon-borne limb scanning of the atmosphere, then particular care must be taken to avoid the correlations between the temperature dependence of the cross sections, the air mass factors, and the diurnal variation of the NO 3 concentrations [Weaver et al., 1996]. For these applications it would be better to use large spectral regions (including at least the peak around 623 nm that shows a different temperature dependence than the 662 nm band [Yokelson et al., 1994]) for the retrieval. However for applications like limb scanning of the atmosphere or optical measurements from space in solar or lunar occultation, the scaling of the room temperature cross sections of the (0-0) band using the model described above will provide very

10 ACH 1-10 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 Figure 7. Calculation of the change in the population of the vibrational ground state as function of temperature (see text for details), compared to the experimentally observed variations in the absorption cross-section of the (0-0) band around 662 nm with temperature. The agreement with all recent studies is good, in particular with the values of Yokelson et al. [1994]. Note that no single parameter was fitted to obtain this agreement. The dashed line shows a calculation based on the empirical model proposed by IUPAC [2001] assuming a linear variation of the cross sections with temperature. accurate reference spectra in the region cm 1 ( nm). 4. Conclusion [32] New absorption spectra of the nitrate radical NO 3 have been recorded at room temperature using Fourier transform spectroscopy. The accurate wavelength calibration of the new data and the narrow instrumental line shape of the FTS are important advantages for atmospheric remote sensing. The spectra are in good agreement with previous studies but show higher spectral resolution (0.6 cm 1 ), higher signal-to-noise ratio (up to 400 which is rather high taking into consideration the high resolution used in this study) in the spectral region below cm 1 (above 500 nm), and much higher absolute wavelength accuracy (0.02 cm 1 ). The temperature dependence of the absorption cross sections of the strongest (0-0) band around 662 nm is explained by an increasing population of the vibrational ground state due to depopulation of the degenerate in-plane bending vibration at about 368 cm 1 and the vibration at 762 cm 1. Future experimental work would be helpful to improve the absolute values of the absorption cross sections, to validate their temperature dependence at very high resolution, and to better understand the nature of the vibronic interactions in the electronic ground state of the nitrate radical. [33] Acknowledgments. This work was supported by the CNRS and the CNPq in frame of the scientific exchange program between France and Brazil (contract 10214, 2001). The authors wish to thank M. Vervloet and Q. Kou for helpful support and discussions. We also thank J. B. Burkholder and S. P. Sander for sending us original cross sections in digital form, and C. E. Miller for pointing out an error in our initial understanding of the D 3h symmetry group. We acknowledge the comments and suggestions of three unknown referees who helped to improve the paper. References Aliwell, S. R., and R. L. Jones, Measurements of tropospheric NO 3 at midlatitudes, J. Geophys. Res., 103, , Aliwell, S. R., et al., Analysis for BrO in zenith-sky spectra: An intercomparison exercise for analysis improvement, J. Geophys. Res., 107(D14), doi: /2001jd000329, 2002.

11 ORPHAL ET AL.: VISIBLE ABSORPTION SPECTRUM OF NO 3 ACH 1-11 Allen, B. J., N. Carslaw, H. Coe, A. Burgess, and J. M. C. Plane, Observations of the nitrate radical in the marine boundary layer, J. Atmos. Chem., 33, , Ball, S. M., I. M. Povey, E. G. Norton, and R. L. Jones, Broadband cavityringdown spectroscopy of the NO 3 radical, Chem. Phys. Lett., 342, , Bogumil, K., J. Orphal, J.-M. Flaud, and J. P. Burrows, Vibrational progressions in the visible and near-ultraviolet absorption spectrum of ozone, Chem. Phys. Lett., 349, , Bovensmann, H., J. P. Burrows, M. Buchwitz, J. Frerick, S. Noel, V. Rozanov, K. V. Chance, and A. P. Goede, SCIAMACHY-Mission objectives and measurement modes, J. Atmos. Sci., 56, , Burrows, J. P., G. S. Tyndall, and G. K. Moortgat, Absorption spectrum of NO 3 and kinetics of the reactions of NO 3 with NO 2, Cl, and several stable atmospheric species at 298 K, J. Phys. Chem., 89, , Burrows, J. P., M. Weber, M. Buchwitz, V. Rozanov, A. Ladstaetter-Weissenmayer, A. Richter, and M. Eisinger, The Global Ozone Monitoring Experiment (GOME): Mission concept and first scientific results, J. Atmos. Sci., 56, , Cantrell, C. A., J. A. Davidson, R. E. Shetter, B. A. Anderson, and J. G. Calvert, The temperature invariance of the NO 3 absorption cross section in the 662-nm region, J. Phys. Chem., 91, , Commission of the European Community (CEC), European tropospheric research, edited by G. Angeletti and Øystein Hov, Air Pollut. Res. Rep. 75, Brussels, DeMore, W. B., S. P. Sander, D. M. Golden, R. F. Hampson, M. J. Kurylo, C. J. Howard, A. R. Ravishankara, C. E. Kolb, and M. J. Molina, Chemical kinetics and photochemical data for use in stratospheric modeling, evaluation number 12, JPL Publ., 97-4, Fish, D. J., D. E. Shallcross, and R. L. Jones, The vertical distribution of NO 3 in the atmospheric boundary layer, Atmos. Environ., 33, , Geyer, A., B. Alicke, D. Mihelcic, J. Stutz, and U. Platt, Comparison of tropospheric NO 3 radical measurements by differential optical absorption spectroscopy and matrix isolation electron spin resonance, J. Geophys. Res., 104, 26,097 26,106, Graham, R. A., and H. S. Johnston, The photochemistry of NO 3 and the kinetics of the N 2 O 5 -O 3 system, J. Phys. Chem., 82, , Heintz, F., U. Platt, H. Flentje, and R. Dubois, Long-term observations of nitrate radicals at the Tor station, Kap Arkona (Rügen), J. Geophys. Res., 101, 22,891 22,910, Hirota,E.,T.Ishiwata,K.Kawaguchi,M.Fujitake,N.Ohashi,and I. Tanaka, Near-infrared band of the nitrate radical NO 3 observed by diode laser spectroscopy, J. Chem. Phys., 107, , International Union of Pure and Applied Chemistry (IUPAC), Subcommittee on Gas Kinetic Data Evaluation, evaluated kinetic data, Data Sheet PNOx5, Basel, Switzerland, updated 16 July (Available at www. iupac-kinetic.ch.cam.ac.uk.) Jones, E. J., and O. R. Wulf, The absorption coefficient of nitrogen pentoxide in the ultraviolet and the visible spectrum of NO 3, J. Chem. Phys., 5, , Kim, B., P. L. Hunter, and H. S. Johnston, NO 3 radical studied by laserinduced fluorescence, J. Chem. Phys., 96, , King, M. D., E. M. Dick, and W. R. Simpson, A new method for the atmospheric detection of the nitrate radical (NO 3 ), Atmos. Environ., 34, , Marinelli, W. J., D. M. Swanson, and H. S. Johnston, Absorption cross sections and line shape for the NO 3 (0-0) band, J. Chem. Phys., 76, , Mohammed-Tahrin, N., A. M. South, D. A. Newnham, and R. L. Jones, An accurate wavelength calibration for the ozone absorption cross-section in the near-uv spectral region, and its effect on the retrieval of BrO from measurements of zenith-scattered sunlight, J. Geophys. Res., 106, , Noxon, J. F., R. B. Norton, and E. Marovich, NO 3 in the troposphere, Geophys. Res. Lett., 7, , O Brien, L. C., H. Cao, and J. J. O Brien, Molecular constants for the v = 0, b 1 g + excited state of O 2 : Improved values derived from measurements of the oxygen A-band using intracavity laser spectroscopy, J. Mol. Spectrosc., 207, , Orphal, J., A critical review of the absorption cross-sections of O 3 and NO 2 in the nm region, ESA Tech. Note MO-TN-ESA-GO-0302, Eur. Space Res. and Technol. Cent., Noordwijk, Netherlands, Orphal, J., et al., Laboratory spectroscopy in support of UV-visible remotesensing of the atmosphere, in Recent Research Developments in Physical Chemistry, vol. 6, pp , Transworld Res. Network, Trivandrum, India, Platt, U., Modern methods of the measurement of atmospheric trace gases, Phys. Chem. Chem. Phys., 1, , Platt, U., D. Perner, G. W. Harris, A. M. Winer, and J. M. Pitts, Detection of NO 3 in the polluted troposphere by differential optical absorption spectroscopy, Geophys. Res. Lett., 7, 89 92, Platt, U., D. Perner, J. Schroeder, C. Kessler, and A. Toenissen, The diurnal variation of NO 3, J. Geophys. Res., 86, 11,965 11,970, Ramsay, D. A., Optical spectra of gaseous free radicals, in Proceedings of 10th Colloquium on Spectroscopy International, pp , Spartan, Washington, D. C., Ravishankara, A. R., and R. L. Mauldin III, Temperature dependence of the NO 3 cross section in the 662-nm region, J. Geophys. Res., 91, , Renard, J.-B., F. G. Taupin, E. D. Riviére, M. Pirre, N. Huret, G. Berthet, C. Robert, and M. Chartier, Measurements and simulation of stratospheric NO 3 at mid and high latitudes in the Northern Hemisphere, J. Geophys. Res., 106, 32,387 32,399, SAGE-III ATBD Team, SAGE III Algorithm Theoretical Basis Document (ATBD)-Solar and lunar algorithm, version 2.1, LaRC Doc , NASA Langley Res. Cent., Hampton, Va., Sander, S. P., Temperature dependence of the NO 3 absorption spectrum, J. Phys. Chem., 90, , Solomon, S., H. L. Miller, J. P. Smith, R. W. Sanders, G. H. Mount, and A. L. Schmeltekopf, Atmospheric NO 3, 1, Measurement technique and the annual cycle at 40 N, J. Geophys. Res., 94, 16,423 16,427, Vandaele, A.-C., C. Hermans, S. Fally, M. Carleer, R. Colin, M.-F. Merienne, and A. Jenouvrier, High-resolution Fourier transform measurement of the NO 2 visible and near-infrared absorption cross section: Temperature and pressure effects, J. Geophys. Res., 107(D18), 4348, doi: / 2001JD000971, Voigt, S., J. Orphal, K. Bogumil, and J. P. Burrows, The temperature and pressure dependence of the absorption cross-sections of NO 2 in the nm region measured by Fourier-transform spectroscopy, J. Photochem. Photophys. A Chem., 149, 1 7, von Friedeburg, C., T. Wagner, A. Geyer, N. Kaiser, B. Vogel, H. Vogel, and U. Platt, Derivation of tropospheric NO 3 profiles using off-axis differential optical absorption spectroscopy measurements during sunrise and comparison with simulations, J. Geophys. Res., 107(D13), doi: / 2001JD000481, Wayne, R. P., et al., The nitrate radical: Physics, chemistry, and the environment, Atmos. Environ., Part A, 25, 1 206, Weaver, A., S. Solomon, R. W. Sanders, K. Arpag, and H. L. Miller Jr., Atmospheric NO 3 5. Off-axis measurements at sunrise: Estimates of tropospheric NO 3 at 40 N, J. Geophys. Res., 101, 18,605 18,612, World Meteorological Organization (WMO), Scientific assessment of ozone depletion: 1998, Rep. 44, Global Ozone Res. and Monit. Proj., Geneva, Yokelson, R. J., J. B. Burkholder, R. W. Fox, R. K. Talukdar, and A. R. Ravishankara, Temperature dependence of the NO 3 absorption spectrum, J. Phys. Chem., 98, 13,144 13,150, C. E. Fellows, Laboratório de Espectroscopia e Laser, Instituto de Física, Universidade Federal Fluminense, Niterói, Brazil. P.-M. Flaud and J. Orphal, Laboratoire de Photophysique Moléculaire, CNRS UPR 3361, Bât. 350, Centre d Orsay, Orsay Cedex, France. (Johannes.Orphal@ppm.u-psud.fr)