Introduction to CEAS techniques D. Romanini Laboratoire Interdisciplinaire de Physique Université Grenoble 1/CNRS
Outline : Interest of optical cavities in spectroscopy and related applications (through trace gas analysis ) General properties of optical cavities. How to measure the cavity losses, thus the absorption spectra? (ringdown, transmission ) How to inject light into a high finesse cavity using different light sources? => Different CEAS methods
Absorption spectroscopy Laser (λ) Sample cell Detector 1 ~ I out Detector 2 ~ I in I out = I in exp (- α L ) α L << 1 I / I = (I out I in ) / I in ~ - α L Higher sensitivity? Decrease noise on I and/or increase L (path).
The absorption coefficient carries all information on the sample α(σ) = Ν k Φ(σ) Detection of species at small concentration Detection of weak transitions Analysis of line profiles (collisions ) When N is known (at given P,T), absorption spectroscopy allows measuring k for different absorption lines Then, using this data allows quantitative trace detection (at given P,T) no calibration is in principle required! Φ(σ) Countless applications σ
Increasing the absorption path Multipass cell IN OUT 100 x path (10km!) volume / 100 IN Optical cavity OUT R = 1-T- Losses ~ 99,999% A. Kastler Appl. Opt. 1 (1962) 17 Atomes à l intérieur d un interféromètre Perot-Fabry
General properties of optical cavities.
Optical cavity = Resonator High sensitivity measurements most often exploit some kind of resonator or resonant filtering (lock in detection, quartz clock, atomic clocks ) A Fabry-Perot cavity is indeed an photon resonator It can accumulate an incoming field as a standing wave which has a persistence after the excitation ends. The buildup/decay time represents the memory time of the system during which the incoming signals can interfere. As any oscillator : It amplifies the effect of periodic signals at a resonant frequency It rejects perturbations out of resonance (it acts as a spectral filter)
Examples of resonators Mechanical watches : their oscillations measure time with a drift as small as 1 second/day ( ~10-5 ) Foucault s pendulum : it reveals earth rotation thanks to the large decay time of its oscillation Radio tuner : a tunable electronic resonant circuit filters out a communication channel from the oscillations of electrons in an antenna, excited by weak radio waves Quantum clocks : narrow transitions in cold and isolated atoms or molecules are exploited as ultra stable resonators (<10-14 ). Lasers : special resonators which include an active amplification which maintains the oscillation indefinitely. However a memory time applies given by decoherence from spontaneous emission
The optical resonator IN L~1m OUT Notes: 1) If R approaches 1 : modes get narrower 2) When intra-cavity losses are small : Transmission maxima approach 100% R = 1-T-Losses ~ 99,99% %T 100 50 2L=N λ Off resonance Finesse : F = FSR/δν ~ π/(1-r) FSR = c/2l δν 1/τ 2L=(N+1) λ 0 σ
Real life : Longitudinal & transverse modes When tuning a single-frequency laser, transmission by a cavity shows a periodic pattern of peaks : 1 out( ω) When looking at the profile of the cavity output beam : 0.5 0 ω TEM 00 TEM 01 TEM 02 Peak intensities will depend on alignment and focusing (mode matching...) thus, different transverse modes are excited at different frequencies.
How to measure the cavity losses, thus the absorption spectra? Ring-down, transmission, (phase shift)
Cavity based absorption spectroscopy Miroir R~99.999% laser photodiode Gas sample CEAS CRDS Cavity transmission τ 0 lo n g u e u r d 'o n d e τ(ν) temps Cavity Enhanced Absorption Spectroscopy Cavity Ring-Down Spectroscopy
CRDS & Cavity Enhanced CRDS : 1/τ = (A+T) c/l+ α(ν) c absolute measurement CEAS : I = I 2 α L T e (1 R e ) 0 α L 2 => I/I ~ (2F/π) α(ν) L value of τ is needed for F A, R, T = mirrors Absorption, Transmission, and Reflectivity L =cavity Length
Performance indicators in absorption spectroscopy 1 - Smallest detectable / noise-equivalent absorption (NEA*) : α min [cm -1 ] = r.m.s. noise in absorption spectrum The averaging time T is important: There is usually an optimum (stability time of the apparatus), then NEA increases 2 - Acquisition time: α min can be normalized by the square-root of T to obtain the NEA in 1 Hz bandwidth : α min [cm -1 Hz -1/2 ] = NEA * T 1/2 This is equal to the noise for 1s averaging if the system stability is longer 1s (usually so). 3 - Number of datapoints N in a spectrum. This provides a figure of merit : α min [cm -1 Hz -1/2 ] = NEA * T 1/2 / N 1/2 units are same as the previous indicator... Beware : this cannot be taken as an estimate of some smallest measurable absorption! NEA(1s) NEA(optimal) * Alternatively,the Minimum Detectable Absorption through the sample may be given : MDA = L x NEA
Different CEAS schemes : pulsed CRDS
Classic CRDS (pulsed laser) pulsed laser L gas (α) R+T+P=1 Detector Absorption par passage [ 10-6 ] At each pulse, fit exp(-γt) to the exp decay at cavity output to obtain photon lifetime τ = 1/γ 40 20 0 γ = c (T+P)/L + c α time λ [nm] 500 520 540 560 580 cavity output Loss per pass : ~ P+T ~ 10-5 Detectable limit : 1% => 10-7 /pass!! large spectral coverage!
Where are the cavity modes? Output spectrum (σ) = T cav (σ) x Input spectrum (σ) ν laser >>1/ t laser Cavity modes Intensity 1/ t C = c / 2L 1/ t laser typical ns laser spectrum -1200-900 -600-300 0 300 600 900 1200 Frequency /MHz Cavity output
CW-CRDS (swept-crds, or Cavity Leak-Out Spectroscopy CALOS)
Laser spectrum (CW, single-frequency) CRDS with a CW laser (CW-CRDS) frequency Cavity modes Cavity modulation by piezo translator I out Laser Phase noise Threshold T h r e s h ofor ld laser interruption Ring-down events time I out la s e r O N L a s e r O F F -50 0 50 100 time Romanini et al. Chem Phys Lett264 (1997) 316
Passage through resonance : Monochromatic laser Ringing effects as a function of tuning speed Modeling cavity transmission η : tuning speed (adimensional)
Passage through resonance : Tuning speed effect... κ=10 Injection Efficiency I out /I in 0.3 0.2 0.1 0.0 0.1 0.0 0.1 0.0-100 -50 0 50 100-5.0-2.5 0.0 2.5 5.0 single event averaged profile <200> time [τ rd ] η=10-1 η=1-20 -10 0 10 20 η=10 1 Single-mode injection using faster and faster frequency sweeps Injection Efficiency I out / I in κ : ratio of laser to cavity mode linewidth η : normalized tuning speed 0.03 0.02 0.01 0.00 0.0002 0.0001 0.0000 κ=10-1 0 1 2 3 4 single event averaged profile <200> monochromatic field η=10 2 η=10 4-1 0 1 2 3 4 time [τ rd ] Morville et al. Appl Opt 41 (2002) 6980
Fast CRDS 2 ms! He and Orr Appl. Phys. B 79 (2004) 941 Debecker et al. Optics Express 13 (2005) 2906
The fibered CW CRDS spectrometer at Grenoble (1200~1800 nm) 6 nm/diode 70 diodes Diode laser Lambda meter ν=f(t,i) laser ON Laser OFF Threshold Optical Isolator Tap coupler Fast switch (acousto-optic) -50 0 50 100 Optical cavity photodiode
The N 2 challenge N 2 predictions : Li & Le Roy, JCP 126 (2007) 224301
Spectra averaging over 5 days
S(4) (3-0) of N S~ 1.5 10-31 cm/mol
Frequency stabilized CW-CRDS High precision spectroscopy of lineshapes The cavity length is referenced to a stable HeNe laser, so each ringdown event is precisely defined in frequency. This also reduces drifts and allows longer averaging with white noise dependence => talk on Tuesday Lisak, Hodges, and Ciurylo PRA 73 (2006) 1 13
Towards an ultimate spectrometer Atmosphere: very accurate line profile study Environment: accurate isotopic ratio Fundamental physics: k B constant Sustainable handbooks: precision spectroscopy Burkart, Romanini, Kassi Opt.Lett. 38 (2013) 2062
ICOS (swept CEAS) & off-axis ICOS The first is basic and has basic performance too. The second has excellent performance but it is not well explained in the literature (commercial developments!)
Optical-Feedback CEAS (& CRDS) NICE OHMS => Talks on Wednesday and today
Cavity injection by Optical Feedback (OF) V-cavity : OF at resonance φ OF κ = P OF /P in Diode Laser Direct reflection OF 1.5 Cavity output Power input Fast & optimized TEM 00 modes injection Mode-by-mode freq. calibrated spectra Ringdown calibration of absorption scale Photodet. signals (V) 1.0 0.5 => talk on Wednesday 0.0 0 10 20 30 40 50 time (ms) Morville et al. APB 80 (2005) 1027 Ringdown
Noise-Immune Cavity-Enhanced Optical Heterodyne Molecular Spectroscopy F = 100,000 Laser locked to 1mHz rel. linewidth! L = 47cm α min (1s)= 10-14 /cm => next talk Ye, Ma, Hall JOSAB 15 (1998) 6
BB-CEAS Increasing interest for Broad Band coverage for several applications. More spectral resolution => less light per spectral element! Spectral quality and stability of the BB source is primordial: Dye lasers and lamps are not very good, contrary to LEDs and ML lasers. ML lasers have high beam quality (mode matching ), allow large spectral coverage by nonlinear optics (harmonic or parametric generation, fibers ), and produce a comb of frequencies!
Wavelength Multiplexed Spectroscopy Transmission broad-band modelocked LASER sample gas CCD array laser frequency
CEAS with a lamp Fiedler et al. CPL 382 (2003) 447
CEAS with a LED Ball, Langridge and Jones CPL 398 (2004) 68
Comb CEAS general comments and first demonstration (back in 2002) See other talks on new developments
Spectrum of a mode-locked laser... a coherent ensemble of single-frequency lasers Time domain : Periodic train of pulses δt t 1/δt Mode spacing 1/ t Fourier T. => Spectrum of mode-locked laser Typically a ML spectrum has ~10 5 modes
Spectrum of a mode-locked laser A closer look http://www.mpq.mpg.de/~haensch/comb/research/combs.html
Cavity transmission spectrum Comb-CEAS basic idea Modelocked laser spectrum Output Spectrum = T cav x Input Spectrum The spectrum transmitted by the cavity depends on the overlap of the two combs of modes : by changing the cavity length, we observe comb beatings, with different periods When combs are in tune, the laser spectrum is efficiently transmitted without beatings => Magic point
CEAS with a Mode-Locked laser : first demo Millennia pump laser, 5W Ti:Sa 100fs, 80Mhz optical isolator L 1 Photodiode M 2 PinHole M 1 L 2 PZT Spectrograph Translation stage M 1,2 : high reflectors around 800 nm CCD Gherman and Romanini, Opt. Expr. 10 (2002) 1033
Beating between mode combs 3.0 1.5 0.0 4 2 0 1000 µm 500µm Transmitted spectra for different displacements of cavity length from the «magic point» ( smaller peaks are from transverse cavity modes ) 20 10 0 150 µm 2000 1000 0 848 849 850 851 852 853 854 855 Wavelength [nm] 00 µm Gherman and Romanini, Opt. Expr. 10 (2002) 1033
Cavity transmission as a function of cavity length 0.15 0.10 150 µm 0.50 0.25 0.00 40 µm distance from «magic point» 1.0 0.5 0.0 10 «magic point» 20 µm NOTE: When looking at these signals on a fast timescale, one can recover the pulse train again... 5 0 00 µm 0 1 2 3 4 Time [ms] ( Cavity length tuning by a PZT ) Gherman and Romanini, Opt. Expr. 10 (2002) 1033
Why we see more than one comb resonances? 1 f=0 frequency RF THz Mid IR Near IR Laser emission 2 3 1 3 2 fine cavity tuning by piezo These secondary resonances have different intensity : Dispersion + f0
ML-CEAS : A first demonstration laser spectrum Spectrum transmitted by cavity filled with air Spectrum transmitted by cavity filled with HCCH 857 858 859 860 861 862 nm Moderate cavity finesse F~360 (effective absorption length F x L /π 120 m) Cavity modulated by >1FSR @ ~1KHz Acquisition time ~ 10 ms (averaging a few passages through resonance) Gherman and Romanini, Opt. Expr. 10 (2002) 1033
ML-CEAS : Conclusions Robust, simple, spectrally multiplexed technique Real-time, high sensitivity (10-10 /cm/hz 1/2 per spectral element) Quantitative (calibration of cavity finesse by ringdown) Access to the UV (efficient frequency upconversion) Cavity modulation is robust and avoids cavity dispersion effects More schemes of comb-ceas have been developed since. see other talks in this conference!
Acknowledgements. All my collaborators and more III-V labs