High precision measurements of line mixing and collisional induced absorption in the O2 A-Band Erin M. Adkins, Melanie Ghysels, David A. Long, and Joseph T. Hodges Material Measurement Laboratory National Institute of Standards & Technology Gaithersburg, MD 20899, USA

Oxygen A- Band Spectroscopy for Remote Sensing Molecular oxygen (O2) has a well-known and uniform molar fraction within the Earth’s atmosphere. Consequently, the O2 A-band is commonly used in satellite and remote sensing measurements (GOSAT, OCO-2, TCCON) to determine the surface pressure-pathlength product for transmittance measurements that involve light propagation through the atmospheric column. For these missions, physics-based spectroscopic models and experimentally determined line-by-line parameters are used to predict the temperature- and pressure-dependence of the absorption cross- section as a function of wave number, pressure, temperature and water vapor concentration. In order to achieve the desired 0.25% CO2 concentration accuracy for the OCO-2 mission, high level O2 spectroscopy with advanced lineshape models is required. OCO-2 will not be measuring CO2 directly; but actually, the intensity of the sunlight reflected from the presence of CO2 in a column of air. The OCO-2 spectrometers will measure sunlight reflected off the Earth's surface. The sunlight rays entering the spectrometers will pass through the atmosphere twice - once as they travel from the Sun to the Earth, and then again as they bounce off from the Earth's surface to the OCO-2 instrument.  To make sure that we have an accurate derivation of Xco2, we also do a comparative absorption measurement of a second atmospheric gas, O2. The concentration of molecular oxygen O2 is constant, well known, and uniformly distributed throughout the atmosphere. Therefore, O2 is the best candidate for reference measurements. The O2 A-band wavelength channel, in the vicinity of 0.76 µm, will provide the required absorption spectra. The O2 A-band spectra indicate the presence of clouds and optically thick aerosols that preclude full column measurements of CO2. Observations from this band will be used to infer the total atmospheric pressure, as well as to measure of solar light path length as it passes through the atmosphere. NASA JPL OCO-2

O2 A- Band Spectroscopy Isolated Line Profiles Voigt Profile is characterized by: Doppler Broadening - ΓD Collisional Broadening – Γ0, Δ0 Advanced lineshapes can also account for higher order effects such as: Collisional induced velocity changes (VC) - νVC Hard collision models - NGP Soft collision models - GP Speed dependence of relaxation rates (SD) – aw (Γ2/ Γ0) as (Γ2/ Γ0) Correlation between velocity and rotations state changes due to collisions – η Line Mixing (LM) Non-isolated lines caused by exchange of quanta through collisions Collisional Induced Absorption (CIA) Intermolecular interactions lead to collisionally- induced dipole Dicke narrowing – narrows the doppler width at low pressures SD modifies the Lorentzian component from a symmetric velocity distribution Line mixing collisions that exchange quanta during absorption, leads to a reduction of absorption in the wings to the enhancement of the through at intermediate pressures and to the narrowing of the profile when the lines have merged. Transfer from weak absorption regions to stronge ones, induces a narrowing of spectral structures Collisional effects on molecular spectra: laboratory experiments and models, consequences for applications. Elsevier: 2008.

O2 A- Band Spectroscopy for Remote Sensing Studies of the impacts of O2 A-band VC effects, line mixing, and CIA on CO2 atmospheric retrievals found the following importance rankings with respect to measurement uncertainty: CIA LM VC effects – Dicke narrowing Recent multispectrum analysis of O2 A-band spectra collected with CRDS and FTS developed a model integrating advanced lineshapes, line mixing, and CIA to minimize retrieval uncertainty. In order to optimize the model additional studies focusing on CIA and LM are needed. Specifically: CRDS data in the R-branch Full-band spectra at low-temperatures Multispectrum analysis of the oxygen A-band. Journal of Quantitative Spectroscopy and Radiative Transfer 2016. On spectroscopic models of the O2 A‐band and their impact upon atmospheric retrievals. Journal of Geophysical Research: Atmospheres 2012, 117 (D12). Multispectrum analysis of the oxygen A-band. Journal of Quantitative Spectroscopy and Radiative Transfer 2016.

FS-CRDS O2 A Band System 2% O2 – 133 kPa 20% O2 – 133 kPa

FS-CRDS: System Schematic Finesse: 15,000 Effective pathlength: 3.5 km (0.74m cavity) Temperature: 70 mK stability Gradients < 0.1 K Pressure: 0.01% relative standard deviation across 5 kPa to 133 kPa range Acquisition speed: ~315 pts/hour FSRP=0: 198.9788(4) MHz HeNe Stability: 1MHz αmin = 4.6 x 10-12 cm-1 1. lock to local mode 2. acquire ring-down data 3. unlock laser 4. tune to next mode

FS-CRDS: Scan Lock to laser to cavity mode using high precision wavemeter Acquire ring-down data Unlock laser and calibrate wavemeter versus HeNe Check if next mode has losses above threshold based on HITRAN simulation Over threshold – tune laser grating to next mode under threshold Under threshold – tune laser with piezo to next mode If the next point is outside of PZT tuning range, reset PZT and tune the grating 5) When tuning grating Tune grating to next resonance Search for a laser current that provides a mode hop free tuning range across the PZT scan Frequency FSR Threshold Resonant Si pressure gauge calibrated against NIST secondary pressure standar, maximum relative standard deviation less than 0.01% across the 5kPa - 133 kPa range

FS-CRDS O2 A Band System 2% O2 – 133 kPa 20% O2 – 133 kPa

FS-CRDS O2 A Band System 2% O2 – 133 kPa 2% O2 – 133 kPa

FS-CRDS: Lineshape Measurements For system validation: 2% O2 balance N2 P21P21 and P21Q20 0.4% O2 balance N2 P11P11 and P11Q12 – 20% O2 P5P5 and P5Q6 – 20% O2 R5R5 and R5Q4 – 20% O2 R11R11 and R11Q10 – 20% O2 Pressures: 1.33, 3.33, 6.67, 13.3, 39.2, 66.7, 133 kPa

FS-CRDS: Modeling and Fitting Spectrum P11P11 and P11Q10 – 39.9 kPa 0.4% O2 SNR 12345 SNR 10896 1 𝑐𝜏(Δ 𝑣 𝑞 ) = 𝛼 0 + 𝛼 𝑒𝑡 Δ 𝑣 𝑞 + 𝛼 𝑎𝑏 (Δ 𝑣 𝑞 ) Δ 𝑣 𝑞 = 𝑣 𝐹 𝑇,𝑃 Δq VP: QF 975 GP: QF 5558 NGP: QF 4556 SDVP: QF 4344 Frequency Detuning (MHz)

FS-CRDS: Γair O2 A-band line parameters to support Atmospheric Remote Sensing. Journal of Quantitative Spectroscopy and Radiative Transfer 2010. Multispectrum analysis of the oxygen A-band. Journal of Quantitative Spectroscopy and Radiative Transfer 2016.

FS-CRDS: νvc with GP P Branch R Branch 𝜈 𝑉𝐶 = 𝜈 𝑉𝐶 ∞ −κ𝛾 𝐽 ′ 𝜈 𝑉𝐶 = 𝜈 𝑉𝐶 ∞ −κ𝛾 𝐽 ′ O2 A-band line parameters to support Atmospheric Remote Sensing. Journal of Quantitative Spectroscopy and Radiative Transfer 2010.

FS-CRDS: aw with SDVP 𝑎 𝑤 = 1−𝑛 2 3 ⋅ 𝑚 𝑝 𝑚 𝑎 1+ 𝑚 𝑝 𝑚 𝑎 = 0.087 𝑎 𝑤 = 1−𝑛 2 3 ⋅ 𝑚 𝑝 𝑚 𝑎 1+ 𝑚 𝑝 𝑚 𝑎 = 0.087 Multispectrum analysis of the oxygen A-band. Journal of Quantitative Spectroscopy and Radiative Transfer 2016. Quadratic speed dependence of collisional broadening and shifting for atmospheric applications. Journal of Quantitative Spectroscopy and Radiative Transfer, 2015.

FS-CRDS: SDVP with linemixing Multispectrum analysis of the oxygen A-band. Journal of Quantitative Spectroscopy and Radiative Transfer 2016.

Conclusions/Next Steps A FS-CRDS system was developed that can scan the entire A-band skipping over the high density line cores with the focus of studying CIA and LM effects in the O2 A-band Initial tests show good agreement with literature values for νvc , aw , Y in the GP, SDVP, and SDVP-lm lineshapes respectively. This system is built on the same table as a low-temperature cavity ring-down cell, which will be used to study the temperature dependence of lineshape parametrers, linemixing, and CIA in the O2 A-Band.

FS-CRDS O2 A Band System 2% O2 – 133 kPa 20% O2 – 133 kPa

FS-CRDS: System Characteristics αmin = 4.6 x 10-12 cm-1 Resonant Si pressure gauge calibrated against NIST secondary pressure standar, maximum relative standard deviation less than 0.01% across the 5kPa - 133 kPa range