Cavity-Enhanced, Frequency-Agile Rapid Scanning
Spectroscopy: Measurement Principles
J.T. Hodges, D.A. Long, G.W. Truong, K.O. Douglass,
S.E. Maxwell, R.D van Zee, D.F. Plusquellic
National Institute of Standards and Technology,
Gaithersburg, MD
joseph.hodges@nist.gov
250 spectra in 0.7 s
68th Ohio State University International
Symposium on Molecular Spectroscopy
June 17-21, 2013, Columbus OH
Cavity ring-down spectroscopy (CRDS)
recirculating field
Fabry-Pérot resonator
Single-mode excitationlow-loss
with locked
cavitymirror
dielectric
incident laser beam
detector
Attributes:
compact volume
insensitive to atmospheric absorption and laser intensity noise
long effective pathlength, leff = lcav(Finesse/)
potentially high spectral resolution & negligible instrumental broadening
readily modeled from first principles
spectra based on observation of time and frequency
A little history …
multi-mode cavity ring-down spectroscopy (CRDS)
signal with pulsed excitation
transform-limited pulse
transverse mode beats
Signals are dominated by transverse and longitudinal mode beating effects, resulting in
suboptimal statistics and severely compromised frequency resolution.
CRDS with continuous wave lasers
single-mode decay signals
Excitation bandwidth << free-spectral range (FSR)
Ring-down signal (V)
1.2
300 shot average:
 = 3.7097(53) s
empty-cavity
 = 6.3783(82) s
0.8
0.4
absorbing medium
Residuals (V)
0
0.006
0
-0.006
0.006
0
-0.006
0
20
40
time (s)
cavity mode spectrum
60
80
cw-CRDS scanning methods
technique
RD signal
amp; acq.
rate (Hz)
frequency
detuning
meas.
frequency
res.
other
dither cavity length,
step tune laser via
current, pzt or temp
low; 10 - 100
external
etalon, lmeter
laser
bandwith,
>> cavity
linewidth
dither laser frequency
through FSR at fixed cavity
length, step tune laser via
current, pzt or temp
low; 10 - 100
external
etalon, lmeter
cavity mode slow scan,
spacing,
no cav. pzt
>> cavity
req’d
linewidth
rapidly sweep laser
frequency via current tuning
low; ~5 kHz
mode
spacing
cavity line
width
RD signal
distortion
optical feedback lock of
laser to cavity, scan cavity to
drag laser frequency
high; ~5 kHz
pzt tuning of
cavity mirror
cavity line
width
can’t use
2-mirror
cavity,
non-linear
tuning
axis
std.
approach,
slow scan
Frequency-stabilized Cavity
Ring-Down Spectroscopy (FS-CRDS)
cw probe laser
optical resonator
pzt
decay signal
frequency-stabilized
reference laser
(a)
time
cavity stabilization servo
absorption spectrum
FSR = 200 MHz
stabilized comb of
resonant frequencies
(b)
frequency
Enables high-fidelity and high-sensitivity measurements of transition areas,
widths & shapes, positions and pressure shifts
High-spectral fidelity of FS-CRDS
Saturation dip spectroscopy
of blended H2O spectra
Line shape effects in O2
Voigt Profile
Galatry Profile
Systematic errors arise
from overly simplistic line shapes
The problem of slow frequency tuning
To record a spectrum in FS-CRDS you typically tune the laser frequency by
using a grating, pzt-actuated mirror or by temperature tuning
These approaches limit the spectrum acquisition rate to ~5 s/jump
optical frequency
Rapid Step Scanning of Laser Frequency
Frequency-agile, rapid scanning (FARS) spectroscopy
Method:
• Use waveguide electro-optic phase-modulator (PM) to generate tunable sidebands
• Drive PM with a rapidly-switchable microwave (MW) source
• Fix carrier and use ring-down cavity to filter out all but one selected side band
MW source
side-band spectrum
ring-down cavity
cw laser
gas analyte
phase modulator
Advantages:
• Overcomes slow mechanical and thermal scanning
• Links optical detuning axis link to RF and microwave standards
• Wide frequency tuning range (> 90 GHz = 3 cm-1)
Detector
FARS measurement principle
FSR
cavity
resonances
frequency
scanning
C+d
C+d+FSR
C+d+2FSR
How well does the cavity filter out sidebands?
FSR
d
Lowest order of a spurious
sideband close to a cavity mode is
1- N where,
N = Round(R=FSR/ d)
carrier
selected sideband
Cavity filtering (fixed TEM)
In general for unwanted sideband orders,
local detuning/cavity linewidth = e*finesse/N
where e = R – N (non-integer remainder)
In the absence of dispersion, this level of
discrimination does not change as the modulation
frequency is stepped in increments of the FSR
If there is a spurious overlap, one can readily
change carrier detuning to avoid this situation
Sideband filtering for our spectrometer
R = 203.076 (MHz)/13 (MHz) = 15.621 so that N = 16 and
epsilon = 0.3788, meaning that the m = -15 sideband would be
the first one to come near a resonance of the cavity.
For our finesse of 20,000, the local detuning would be about
485 times the cavity line width, showing that we have
nearly perfect frequency discrimination (assuming perfect mode
matching into TEM00).
We have never observed any evidence of spurious
coupling into other sidebands.
Independent methods for characterizing frequency axis
of PDH-locked FARS-CRDS setup
1. Measure frequency, f, of probe laser with optical frequency comb and
count change in mode order, q.
Gives absolute frequencies and cavity free spectral range (FSR).
2. Measure FSR from differences in microwave frequencies
corresponding to transmission resonance peaks.
3. Measure FSR with dual sideband method of Devoe & Brewer.
Methods 2 & 3 give agreement in FSR at 2 Hz level
and yield dispersion
Absolute frequencies are ~5 kHz and are limited by
10 kHz stability of I2-stabilized HeNe reference laser
Dual-sideband FSR measurement scheme
q-1
q
D
D-d
w0 –(w1+w2)
w0 -w1
w0 –(w1-w2)
Two sets of sidebands:
w1 at FSR= s
w2 for PDH lock
D = w0 – qs
d = w1 – s
Devoe & Brewer, PRA 30, 2827 (1984).
q+1
w0 -w2
w0
w0 +w2
D+d
w0 +(w1-w2)
w0 +w1
w0 +w1+w2
Demodulation of heterodyne beat at w1 - w2 gives
dispersion signal g(d) centered about d=0, where
d = w1 – s.
Accuracy of FARS-CRDS frequency axis
cavity dispersion
gDD  40 fs2
Measuring losses in terms of cavity line width
Due to the quality of our
frequency axis we can record
the shape and width of
individual cavity resonances
The width of the resonances
provides an equivalent
measure of the absorption in
the frequency domain,
α = Δω1/2/c
~130 Hz relative laser linewidth
Uncertainty of the fitted
resonance frequency ~1 Hz
Uncertainty of the fitted width
of the resonances ~0.04%
Effect of beam extinction ratio on
ring-down time measurement statistics
Extinction ratio = 10 log( Id/Il)
Id = “decay” intensity
Il = “leakage” intensity
cavity decay signal = Idexp(-t/)
Id
Il
0
t
cw leakage signal = Il
Ideal case (infinite extinction ratio):
Il = 0,  exponential decay
Actual case:
leakage intensity interferes with decay signal
to yield noisier and/or non-exponential decay
Measured FARS-CRDS decay signals
Noise in residuals is insensitive to
extinction ratio (phase-locked case)
Systematic deviations become important
for extinction ratios < 50 dB
Effect of extinction ratio on measurement precision
This work
Huang & Lehmann, Appl. Phys. B 94,
355 (2009)
With DFB laser leakage intensity introduces
excess noise in ring-down signal
phase locked case, small amount of excess noise
s/ = 8x10-5
FARS-CRDS has been demonstrated with:
waveguide electro-optic phase modulator
1) distributed feedback diode laser (DFB)
2) single-mode fiber laser
3) external cavity diode laser (ECDL)
with high-bandwidth Pound-Drever-Hall lock