Experimental observation of the ν1+3ν3 combination bands of 16O14N18O and 18O14N18O in the near infrared spectral region

The first observation of the ν1+3ν3 combination band of the nitrogen dioxide isotopologue 16O14N18O is presented. The band was measured using Fourier-Transform Incoherent Broad-Band Cavity Enhanced Absorption Spectroscopy (FT-IBBCEAS) in the region between 5870 cm−1 and 5940 cm−1. To confirm the assignment, the band was simulated using a standard asymmetric top Watson Hamiltonian using extrapolated rotational and centrifugal distortion constants. Furthermore, the first experimental observation of the ν1+3ν3 band of the 18O14N18O isotopologue is also reported. The positions of ro-vibrational lines of the ν1+3ν3 band of the naturally most abundant isotopologue 16O14N16O were used for wavenumber calibration of line positions.


Introduction
Nitrogen dioxide (NO 2 ) has been extensively studied spectroscopically owing to its importance as atmospheric trace gas constituent and as driver of a large number of atmospheric gas phase reactions [1][2][3].The detection of NO 2 in the atmosphere is commonly based on its strong absorption spectrum in the blue region of the visible spectrum between ~400 and 450 nm [4][5][6].In addition, the mid-infrared absorption bands of NO 2 at 3.4 and 6.2 μm have also been used [7,8].Apart from its atmospheric importance NO 2 is also interesting from a molecular spectroscopy point of view and therefore has been extensively studied by high-resolution laboratory spectroscopy in the infrared region .Based on the substantial amount of experimental data on NO 2 , Lukashevskaya et al. [30] established an extensive list of ro-vibrational line positions and intensities of N 16 O 2 from the far-to the near-IR (0.006-7916 cm − 1 ).NO 2 lines are also included in the HITRAN [36] and GEISA spectroscopic databases [37,38].In the near-infrared spectral region the strongest band of NO 2 is the ν 1 +3ν 3 combination band, which was first identified in 1958 by Arakawa and Nielsen [9].Olman and Hause also observed this band again in 1968 [10] and Blank et al. carried out a first detailed rotational analysis of this band in 1970 [11].More recently, Miljanic et al. [24] presented a very detailed experimental and theoretical study of the ν 1 +3ν 3 band of N 16 O 2 ; in Ref. [24] 1147 ro-vibrational transitions were identified with rotational quantum numbers N and K a of up to 47 and 8, respectively.Several years later, Naumenko et al. [39] published also a study of this band and reported experimental line positions and intensities of 3154 transitions with quantum numbers N and K a of up to 59 and 13, respectively.
The near infrared region is spectroscopically particularly interesting.Around only 10000 cm − 1 , the two lowest electronic states of NO 2 cross with a conical intersection [21,40].This conical intersection leads to unusual effects including a breakdown of standard spectroscopic theory [21], so that the line-by-line analysis of the rovibronic bands of the A-X transition in the near-infrared (although fully resolved at Doppler-limited resolution [21]) has not been successful until today.A proper understanding of the vibrational structure of the electronic ground state of NO 2 is therefore very important.For this, experimental observations of isotopically substituted species of NO 2 at high energies are most suitable.However, despite the extensive spectroscopic literature on the most abundant isotopologue N 16 O 2 , spectroscopic investigation of other isotopologues, especially those with singly and doubly substituted oxygen ( 16 O 14 N 18 O and 18 O 14 N 18 O) are sparse [41].In 1976, Hardwick and Brand [42] predicted the band centers of 25 bands of 18 O 14 N 18 O between 722 cm − 1 and 5854 cm − 1 .In 2006, Volkers et al. measured the high-resolution spectra of N 16 O 2 , 16 O 14 N 18 O, 18 O 14 N 18 O in the spectral region between 11800 and 14380 cm − 1 [40].More recently Marinina et al. [43] reported the Fourier transform IR spectra in the 1540-1640 cm − 1 spectral region were the ν 3 band of 16 O 14 N 18 O is located.
In this work, we report the first experimental observation of the ν 1 +3ν 3 bands of the isotopologues 16 O 14 N 18 O and 18 O 14 N 18 O in the near infrared region between 5840 cm − 1 and 6000 cm − 1 using Fourier Transform-Incoherent Broad-Band Cavity-Enhanced Absorption Spectroscopy (FT-IBBCEAS) [44,45].In comparison to the well-known ν 1 +3ν 3 band center of 14 N 16 O 2 at 5984.7 cm − 1 [24,39], the corresponding band centers of 16 O 14 N 18 O and 14 N 18 O 2 were observed at 5922 cm − 1 and 5854 cm − 1 , respectively.

Measurement method, components, and parameters
The general experimental setup has been published before [46][47][48] and thus merely the key experimental aspects are outlined here.The light source was a 2 W supercontinuum source (Fianium SC 450-2) operating at a repetition rate of 80 MHz.The broadband light (500-1800 nm) was passed through a long pass filter (Thorlabs FEL 1250-1) with a cut-off wavelength of 1250 nm (8000 cm − 1 ).The light was spatially filtered and collimated before entering the optical cavity of length d ~644 cm.The optical cavity consisted of two dielectric plano-concave mirrors (Layertec GmbH, Germany) with a reflectivity of ~0.999 between 5750 cm − 1 and 8000 cm − 1 .The optical cavity was attached to a vacuum chamber consisting of long stainless-steel pipes, which was evacuated by a turbo pump (Leybold Turbovac) to a pressure of approximately 10 − 5 mbar before injecting any gas samples.The experiment was carried out with a static gas and hence the cavity mirrors were not purged.The light exiting the cavity was coupled into a multimode fiber with an IR optimized achromatic doublet.The other end of the multimode fiber was connected to the entrance port (aperture diameter 0.5 mm) of a Fourier transform spectrometer (FTS; Bruker Vertex 80).The light transmitted by the cavity was measured without (evacuated cavity ~10 − 5 mbar), I 0 (λ), and with the sample, I(λ), in the cavity.From the ratio of the transmission intensities, the reflectivity of the mirrors R(λ), and the sample path length per pass, d, inside the optical cavity, the absorption coefficient of the sample α(λ) was evaluated using [49]: The instrumental line shape and spectral resolution of 0.08 cm − 1 was obtained from the measurement of a CO 2 spectrum at 10.7 mbar employing Norton-Beer weak apodization.The integration time used for measuring the spectrum was 120 min.For this integration time a signal-to-noise-ratio of >198 was achieved, which was evaluated on the basis of the 16 O 14 N 18 O absorption at ~5934 cm − 1 .

Materials and sample preparation
CO 2 (purity >99.90 %) was purchased from Irish Oxygen and used without further purification.H 2 18 O and NO 2 were purchased from Taiyo Nippon Sanso Corporation (purity >98 %) and Sigma Aldrich (purity >99.994 %), respectively.The H 2 18 O samples were degassed by several "freeze-pump-thaw" cycles before injection into the cavity.The cavity chamber was first evacuated to a pressure of ~10 − 5 mbar at room temperature.It was then primed with H 2 18 O vapor at a pressure of ~6.8 mbar.A waiting time of 1 h was provided for H 2 18 O to thermally equilibrate and passivate the chamber walls.After 1 h the pressure dropped to ~6.2 mbar.Subsequently NO 2 was injected (partial pressure ~3.5 mbar) into the chamber.The mixture was left to equilibrate for approximately 60 min to enable the 18 O exchange and formation of the NO 2 isotopologues 16 O 14 N 18 O and 18 O 14 N 18 O at room temperature through the formation and decomposition of 18 O-substituted HONO and HNO 3 .The overall pressure of the mixture at the start of the measurement reduced to 8.2 mbar.

Reflectivity calibration
To retrieve absolute absorption coefficients with FT-IBBCEAS, the broadband mirror reflectivity must be established [45].The mirror reflectivity, R(λ), was calibrated by filling a well evacuated (~10 − 5 mbar) cavity with a known amount of CO 2 (~10.7 mbar) [44,46].Based on Eq. ( 1), the reflectivity was determined using the measured CO 2 absorption coefficients and the literature absorption cross-section of CO 2 from the HITRAN database [36].R(λ) was found to be almost constant in the wavenumber region from 5840 to 6000 cm − 1 with a maximum value of 0.9982 falling off by ca.0.0005 at the edges of the region.The uncertainty in the reflectivity calibration (based on errors in cross-sections and pressure measurement of CO 2 ) leads to an uncertainty in (1-R) of ~10 %.This uncertainty is the dominant systematic error contribution to the measured NO 2 absorption coefficients.Other uncertainties occur from the pressure measurements (~5 %) and from the average intensity fluctuation of the supercontinuum light source (~4 %) [28].The S. Chandran et al. total Gaussian absolute uncertainty contributing to the measured absorption coefficients was estimated to be ~12 %.

Results and discussion
Fig. 1 shows the experimental FT-IBBCEA spectrum of the absorption coefficient of NO 2 at 8.2 mbar (total absolute pressure) in the region of the ν 1 +3ν 3 band between 5840 cm − 1 and 6000 cm − 1 .The spectrum clearly exhibits three spectrally overlapping bands that are attributed to the isotopologues 18 O 14 N 18 O, 16 O 14 N 18 O, and 14 N 16 O 2 .The spectrum will be further discussed after a short description of aspects concerning the wavenumber calibration; the raw data are also available as supplementary material.

Line position accuracy
Wavenumber calibration is a crucial aspect in high resolution spectroscopy.As a first step, to minimize the error in line position, the wavenumbers of 20 strong ro-vibrational absorption lines of the 3ν 1 band of CO 2 within the reflectivity calibration spectrum between 6040.7 and 6106.7 cm − 1 were compared with the corresponding line positions reported in the HITRAN data base [36].The band center  of the 3ν 1 band of CO 2 is about 150 cm − 1 above the band center of the new 16 O 14 N 18 O band reported in this work.The observed average absolute discrepancy between the measured and the HITRAN line positions was 0.019 ± 0.003 cm − 1 .This discrepancy is about 4 times smaller than the instrumental resolution of 0.08 cm − 1 , which is indeed satisfying given the signal-to-noise-ratio and rather high line density in the experimental spectrum.
Based on the wavenumber calibration of the Fourier transform spectrum using the CO 2 spectrum, the wavenumber scale accuracy was independently verified by comparing 20 strong ro-vibrational absorption features of the ν 1 +3ν 3 band of N 16 O 2 reported in Ref. [37] with the experimental lines measured in the present study.Fig. 2 shows the selected lines in the ν 1 +3ν 3 band of N 16 O 2 reported by Naumenko et al. 2019 [39] (cut off intensity >1.8 × 10 − 24 cm/molecule) together with the FT-IBBCEAS spectrum in the region between 5955 cm − 1 and 6000 cm − 1 (see section (C) in Fig. 1).Note that, the cavity ring-down spectrum in Ref. [39] has a five times higher resolution (0.015 cm − 1 ) than our FT-IBBCEA spectrum (0.08 cm − 1 ).Consequently, in regions with very high line density, ro-vibrational features in the CRD spectrum generally overlap with the ro-vibrational absorption features of the FT-IBBCEA spectrum.The 20 lines were carefully selected in wavenumber regions where the spectrum is less congested.A typical example illustrating the match of the center wavelength of our data with those from Ref. [39] (stick spectrum representing line intensity) is shown in the inset of Fig. 2 (red rectangle magnified).Table 1 contains the positions from this study (ν exp ) and from Naumenko et al. (ν N ) [39], the position differences (Δν) and the rotational assignments (N, Ka, Kc) in the lower and upper state of the 20 selected ro-vibrational lines.The average absolute discrepancy between the measured and the literature line positions of N 16 O 2 was 0.018 ± 0.009 cm − 1 .The ν 1 +3ν 3 band of N 16 O 2 , used for wavenumber scale calibration, is adjacent (~60 cm − 1 ) to the ν 1 +3ν 3 band of the 16 O 14 N 18 O isotopologue.Therefore, the wavenumber calibration is expected to also hold in the region of the corresponding 16 O 14 N 18 O band assuming a calibration uncertainty of the new ν 1 +3ν 3 lines of 16 O 14 N 18 O to be about 0.027 cm − 1 (=max error for N 16 O 2 from the literature).Note that this uncertainty is again small in comparison with the spectral resolution of 0.08 cm − 1 , and in very good agreement with the value determined from the spectral calibration using CO 2 .

3.2.
The ν 1 +3ν 3 band of 18 O 14 N 18 O Section (A) of Fig. 1 shows the ν 1 +3ν 3 band of 18 O 14 N 18 O in the region between 5840.0 cm − 1 and 5868.0 cm − 1 .In 1976, using an anharmonic force field calculated from experimental spectra, Hardwick and Brand [42] predicted the center of the ν 1 +3ν 3 band of 18 O 14 N 18 O (see Tab. 3 in Ref. [42]) to be at 5853 cm − 1 , which is indeed in excellent agreement with our observation (see the spectrum in Fig. 1) at ~5854 cm − 1 .Our measurement is hence an experimental corroboration of a prediction made almost 50 years ago.The results in Ref. [42] were based on an anharmonic potential calculation from experimental data of the N 16 O 2 isotopologue.

Table 1
Comparison of 20 ro-vibrational line positions of the (1 0 3) ← (0 0 0) band of N 16 O 2 measured using FT-IBBCEAS (column 1) and CRDS from Ref. [39] (column 2).The differences between the experimental line positions in this work and the line positions from the literature are shown in column 3. Rotational assignments are given in columns 4 and 5.
FT-IBBCEAS this work CRDS Naumenko et al. 2019 [39] Difference Rotational quantum numbers [39] ν spectrum of this band (lower panel, blue trace).The simulation was made using an A-reduced Watson-type Hamiltonian, using the values of Bird et al. [50] for the principal rotational constants, and of Miljanic et al. [24] for the centrifugal distortion constants, of the ground state.Based on the vibrational dependence of these constants from Miljanic et al. [24] the rotational constants of the excited (1 0 3) state (see Table 2), were calculated.The intensities were calculated using the rigid-rotor approximation based on the symmetric top wavefunctions obtained from the line position calculation.Although the simulation is not based on a line-by-line fit of the observed spectrum, the overall agreement is satisfying, since it clearly confirms the assignment to the ν 1 +3ν 3 band of 16 O 14 N 18 O.By comparing the observed spectrum with the simulation, the band center was identified to be located at 5922.3 cm − 1 .A list of line positions is also given in Table S1 in the supplementary material.A line-by-line analysis of this combination band was not possible since the spectral resolution of our data is limited and the electron spin splitting constants for 16 O 14 N 18 O are not known for the lower state.

Conclusions
In this publication, the first experimental observation of the ν 1 +3ν 3 combination bands of 16 O 14 N 18 O (5922 cm − 1 ) and 18 O 14 N 18 O (5854 cm − 1 ) in the near infrared is reported using FT-IBBCEAS.The band center for the doubly substituted species confirms the prediction of Hardwick and Brand [42] made nearly 50 years ago.The study demonstrates again the potential of FT-IBBCEAS for studying weak absorption bands in the near-infrared at high spectral resolution and at large spectral bandwith.

Fig. 2 .
Fig. 2. Comparison of 20 line positions in the ν 1 +3ν 3 band of N 16 O 2 between 5955 cm − 1 and 6000 cm − 1 from section (C) of Fig. 1 (black trace) with the corresponding ro-vibrational features of the same band reported in Naumenko et al. 2019 [39] (blue trace, cut off intensity >1.8 × 10 − 24 cm/molecule).The inset shows a magnified view of a line at 5971.92 cm − 1 (red rectangle) illustrating a typical match of position.

Fig. 3 .
Fig. 3.The upper panel shows the ν 1 +3ν 3 band of 16 O 14 N 18 O (also see section (B) of Fig. 1) measured using FT-IBBCEAS.The lower panel (blue trace) shows the simulated spectrum of the same band.