1. Temperature Dependence of Near-Infrared CO2 Line Shapes Measured by Cavity Ring-Down Spectroscopy
Mélanie Ghysels, Adam J. Fleisher, Qingnan Liu and Joseph T. Hodges
Chemical Sciences Division
National Institute of Standards and Technology
Gaithersburg, Maryland USA 20899
International Symposium on Molecular Spectroscopy, 72nd meeting June 19-23, 2017, Champaign-Urbana, IL

2. Application
Determination of sources and sinks of atmospheric CO2 from satellite-based and terrestrial platforms (e.g. GOSAT, OCO-2, TCCON) requires high-precision spectroscopic data and predictive models
Credits:NASA/OCO-2

3. The Orbiting Carbon Observatory (OCO-2) Mission
Grating spectrometer, 3 spectral regions
CO2 in the 4850 cm-1 (2.06 mm) band
CO2 in the 6220 cm-1 (1.6 mm) band
O2 in the cm-1 (0.76 mm) band
High-accuracy requirements for XCO2
Goal ~ 0.25% (1 ppm out of 400)
Goal for uncertainty of modeled radiance ~ 0.1 %
OCO-2 approach
multispectrum fitting of laboratory spectra
derive line shape parameters consistent with
assumed line shape
generate look-up tables with same model
Image: NASA / Orbital
For the OCO-2 mission, it was apparent early on that the spectroscopic databases available at that time would not be sufficient to meet the scientific goals of the mission

4. Previous observations of 1.6 & 2 mm band CO2 line parameters
Studies from [Hartmann et al., 2009; Thompson et al., 2012] show that inconsistencies remain in the retrieved total column of [CO2] using the 1.6 µm and 2 µm regions. The 1.6 µm region has been extensively studied at room temperature [Malathy Devi et al., 2007a, Malathy Devi et al., 2007b, Predoi-Cross et al., 2007a, Predoi-Cross et al., 2007b, Long et al., 2015] However, there are only two FTS-based studies at low temperatures, [Predoi-Cross et al., 2009; Devi et al., 2016] providing gair, dair, n, ( K)], neither of which reported the temperature dependence of non-Voigt parameters.

5. Previous line shape results at room temperature
CO2: R16e (30012) - (00001)
Long et al. JCP, (2011)
[1]
Dicke narrowing only
[2]
[3]
Ref line profile
[1] Long et al. JQSRT, 161, 35 (2015): CqSDNGP
[2] Predoi-Cross et al., J. Mol. Spec 246, 98 (2007): SDVP
[3] Devi et al., J. Mol. Spec. 242, 90 (2007): SDVP
Considering only SDVP causes
an overestimation of pressure broadening
Two narrowing mechanisms

6. Low-temperature, cavity ring-down spectroscopy
Only few low-temperature cavities have been reported in the literature (temperature stability ≈ ± 1K):
Phase shift CRD [Lewis et al., 2009; Perez-Delgado et al., 2006]: LN2 cryostat system
Exponential decay CRDS [Kassi et al., 2009]: LN2 cryostat system (80K)
Here we present a new variable temperature ( K) CRDS system with
a stability better than 1 mK over several hours, with axial differences < 50 mK.

7. Cavity mechanical design (220-290 K)
Monolithic 80 cm-Invar cell (thermal exp. ~ 1x10-6/K)
No piezoelectric actuator, frequency stabilized by temperature stabilization
Invar Optical cavity
Purge gas inlet
Sample gas inlet
Temperature-regulated enclosure
3D view of the inside of the cavity
20ppm mirror
AR coated window

8. Thermo-mechanical configuration
Refrigerants:
R %
R %
R508b 17.3 %
3-component mixture
enables single-stage
condenser over entire
temperature range

9. Optical Layout

10. Mode-by-mode spectral scanning
Frequency detuning dq
FSR
1. lock to local mode
2. acquire ring-down data
3. unlock laser
4. tune to next mode

11. Temperature dependence of the cavity frequency axis
Cavity of length L
FSR = c/2L
Frequency=q*FSR
Mode spacing (FSR) q
q +1
q-1
Laser frequency
Time
Cavity temperature

12. Dependence of cavity mode spacing on T & P
q = mode order
nF = mode spacing
nq = resonance frequency
nr = refractive index
B = coefficient of thermal expansion

13. Servo for measuring the absolute frequency axis stability
Method: measure the transmission of an I2-stabilized HeNe laser (+- 10 kHz)
through the ring-down cavity. Implement closed-loop feedback to an electro-optic phase modulator
(EOM) to maintain resonance with the ring-down cavity

14. Temperature regulation performance
Pond Engineering, CO
Temperature measured by two NIST-calibrated PRT’s
Long-term cavity temperature record
4 mK
~ 4hrs
20 hrs
Thermal Stability
Short-term (2 hrs):
± 5 mK, as small as ± 1 mK after 3 days of stabilization.
Target acquisition time: 10 min
± 0.2 mK stability
Set point
Temperature gradient across the cavity: 5-50 mK

15. Temperature stability
Measured via EOM-sideband
tracking at l = 633 nm
Dfprobe  70 kHz over
scan time of 30-min

16. Allan deviation and averaging statistics
Decay time relative standard deviations: 0.02%-0.04%
Typical room temperature CRDS :
cm-1 [Hodges and Lisak, 2006; Havey et al., 2009; Mikhailenko et al., 2011; Song et al., 2010, Mondelain et al., 2010, Long et al., 2012 and others]
FARS-CRDS: cm-1 [Long et al., ; Long et al., 2014]
[Lin et al., 2015]: cm-1
Recently: long-term (4-day) averaging of spectra: cm-1 [Kassi et al., 2013]

17. Hartmann-Tran (HTP) line shape parameters
GD Doppler width
G0 Lorentzian width
G2 speed-dependent relaxation rate
D0 line shift
D2 speed-dependent line shift
nvc frequency of velocity-changing (vc) collisions
correlation between vc & dephasing collisions
For multispectrum fits acquired over a range of
temperature and pressure, need appropriate constraints
n: temperature exponent for broadening
The ratio aw = G2/G0 is predicted to be independent of T
for n  0.7, and CO2 in air, aw  0.08

18. Lines Measured:
12C16O2 (30013 – 00001) band
(cm-1)
P10e
R16e
R34e
Pressure range: (40 – 450 Torr) 4 to 6 each T
Temperatures: (220, 240, 260, 290 K)
Multispectrum fits in pressure at each temperature

19. Multispectrum fit of G0 g = FWHM/(2P) SDVP T=240 K
( ) R16e line of 12CO2 at cm-1
g = FWHM/(2P)
SDVP
ur(g) = %
T=240 K

20. Fits of (30013-00001) R16e line of 12CO2 at 6240.104478 cm-1
peak SNR 15,000:1
ur (slope) = 2.7%
At each T, spectra at the following pressures
were acquired: (5, 9, 13, 21, 43, 59) kPa
Multi-spectrum fits were carried out at each T

21. qSDNGP fits (HTP with no correlation)
R16e transition
aw theoretical =
[16] Lin et al., JQSRT 161, 11 (2015)
[56] Long et al., JQSRT 161, 35 (2015)

22. qSDVP fits (HTP with no correlation and no Dicke narrowing)
R16e transition ( ) band
[53] Predoi Cross et al., JCP 87, 517 (2009)
[54] Devi et al., JQSRT 177, 117 (2016)

23. P10 transition (qSDNGP fit)
g/g0 = (T0/T)n
g0 = (42) cm-1/atm
n = (26)
nHT = 0.71
Pressure broadening
expected value
speed dependent
narrowing parameter
Dicke narrowing
A*T/D(T)
(K)
HT12 values
g0,HT = (cm-1/atm)

24. Diffusion-based collisional (Dicke) narrowing frequency

25. Proposed constraints on collisional narrowing and speed dependent
relaxation rate terms for 2-D multispectrum (pressure and temperature) fitting
of Hartmann-Tran Profile
Equivalent to quadratic speed dependent Nelkin-Ghatak profile (qSDNGP)
float b term
float temperature exponent, n and g0
Effective frequency of
velocity-changing collisions
Speed dependent relaxation rate

26. Summary
New variable-T CRDS system with ~ 0.5 mK (30-min time scale) stability, exhibiting
an Allan deviation of ~ 7x10-12 cm-1 and spectral SNRs (>10,000:1) that enables precise determination of temperature dependence of line shape parameters.
Frequency stability performance is comparable to actively-stabilized FS-CRDS systems
For the three transitions of the 1.6 mm CO band investigated:
Both speed dependent narrowing and Dicke narrowing need to be considered
The ratio of speed dependent narrowing to pressure broadening (aw= G2/G0) is
nearly independent of temperature and consistent in magnitude with the measured
temperature exponent and theoretical predictions

27. Thank you for your attention !
This work was supported by the NIST Greenhouse Gas Measurement and Climate Research Program and
the NASA/JPL Orbiting Carbon Observatory Science Team.