1. Long Term Stability in CW Cavity Ring-Down Experiments
Haifeng Huang and Kevin Lehmann
64th OSU Symposium on Molecular Spectroscopy
June 26, 2009

2. Cavity Ring-Down Spectroscopy
Laser
Detector
Cavity
Modulator
154.66±0.08µs, χ2 = 0.99
(µs)
(mV)
τ insensitive to laser power fluctuations
long effective absorption length >10 km
Cavity-Ringdown Spectroscopy, ACS and Oxford University Press, 1999

3. Experimental Setup Lens He-Ne laser DFB diode laser
Laser control board
AOM
AOM driver
Detector
Isolator
Computer
3PZTs
Flat mirror
Curved mirror
Mode matching optics
Cavity
Trigger signal

4. Trace Methane Detection
Methane R4 lines, 2ν3 band, fitted with HITRAN
Total pressure torr, methane concentration 33.0 ± 0.6 ppb

5. Trace Methane Detection

7. Drift of τ
Cavity was under vacuum. Ring-down data was recorded in 7 hours. The ring-down rate is 4.9 Hz for both panels. Δν is 23 GHz.

8. Sensitivity in CRDS
The signal averaging process is limited by the drift of CRDS system, caused by slow changes of experimental environment.
The minimum measurable difference (k – k0) determines the sensitivity.
Absorption coefficient:
Differential measurement between k and k0 can improve the sensitivity by the cancellation between the drift of k and that of k0.
Long term stability of CRDS setup determines the final sensitivity in real experiments.

9. Differential Measurement
Absorption coefficient:
CCB AD D FI PC
PZT
SM1
SM2 CV TS M
OPM
MML
AOM OS
Laser 2
Laser 1 C
D/A
Δν adjustable
Switching time 13 ms ν1 ν2

10. Drift of τ
Cavity was under vacuum. Ring-down data was recorded in 7 hours. The ring-down rate is 4.9 Hz for both panels. Δν is 23 GHz.

11. Allan Variance
Allan plot: log-log plot of Allan variance versus average size p
Normal algorithm, m = integer part of N/p
For white noise dominated signal, Allan plot slope is -1.
For linear drift dominated signal, Allan plot slope is +2.
Allan plot gives the optimal signal average size p.
The minimum of Allan variance gives final sensitivity in real experiments.
D. W. Allan, Preceedings of the IEEE 54, 221 (1966); P. Werle et al., Applied Physics B 57, 131 (1993)

12. Modified Algorithm
Modified algorithm, m = N – 2p + 1

13. Drift Cancellation Four hour data Pressure < 0.1 torr
Both λ on peak
Ring-down rate 6.9 Hz for each laser
For each k, optimal p ~500
For k1 – k2, optimal p is
Optimal integration time increased from 72 sec to 41 min.
Final sensitivity: 2.8 × cm-1

14. However…
Cavity pressure ~20 mtorr
Optical interference?

15. Optical Feedback Detector not tilted Short term Stdτ about 0.6 μs
With an isolator between output mirror and the detector, short term Stdτ ~0.07 μs

16. Effect of Ambient Pressure Change
Panel A: lab pressure change
Cavity under vacuum
Data averaged by every successive 20 decays
Seven hour data
The cell temperature ± 0.03 °C
No correlation between decay rate drift and the lab temperature change has been observed.
Mechanical deformation of the cavity by pressure change. Mirror reflectivity is not spatially uniform.

17. Other Noise Partial cancellation Possible reason: mechanical vibration
Sensitivity: 8.1 × cm-1 in 8.7 min

18. Noise in Baseline Δν = 23 GHz, pressure < 0.1 torr
Average size ~ 25, platform in one of the Allan variance
Ring-down rate 8.3 Hz for each channel

19. Noise period ~ 200 decays
Noise reason still unknown

20. Methane Detection Limit
Low thermal expansion: Invar plate
Mechanical vibration isolation
Stable baseline of decay transients
Alignment minimizing optical interference
Sensitivity: 5.6 × cm-1 in 15.4 min
Methane detection limit (3σ) at 1652nm:
0.3 ppbv at 20 torr
37 pptv at 760 torr

21. Conclusions
Low concentration (~0.2 ppb) methane in N2 has been measured in our lab.
Allan variance is used to characterize the drift of CRDS systems. Long term stability determines the final sensitivity in CRDS.
Noise factors include ambient pressure, optical interference, mechanical vibration, thermal stability and baseline noise. Further studies are needed.
With differential measurement, very high sensitivity in CRDS, e.g., 5.6×10-12 cm-1with 15.4 min averaging time, can be realized.

22. Acknowledgements Funding:
Princeton Institute for the Science & Technology of Materials (PRISM), PU
University of Virginia
Prof. Brooks Pate and group members
Other Lehmann group members
Charles Lam of machine shop

23. Thank you!

24. Methane Detection Limit
4.4 × cm-1
Stable time 32.8 min
N2 flow 20 sccm
Δν = 23 GHz
PZT mod 80 MHz
Pressure 1 atm

25. Single Shot Sensitivity Limit of CRDS
Noise density:
Detector noise limited CRDS:
Shot noise limited CRDS:
K. K. Lehmann and H. Huang, Frontiers of Molecular Spectroscopy, Chapter 18, Elsevier 2008

26. SOA as Light Modulator Semiconductor optical amplifier (SOA):
0th order
1st order to cavity
RF power 80 MHz
AOM crystal
Isolator
Output coupler
SOA
Fiber coupler
95% 5%
1512 nm laser diode
Optical fiber
Trigger signal
Current source
Advantages:
Highest extinction ratio (> 80 dB) when used as light modulator
Fast speed: ns or sub ns
Broadband gain media: ~70 nm
Optical fiber connected, no extra alignment needed when λ tuning
M. J. Connelly, Semiconductor Optical Amplifier, Kluwer, Boston, 2002.

27. Modified Setup Lens He-Ne laser DFB diode laser Laser control board
AOM
AOM driver
Detector
Isolator
Computer
3PZTs
Flat mirror
Curved mirror
Mode matching optics
Cavity
Trigger signal
Polarizer or
Pockel’s cell
λ/2 plate