Infrared cavity ringdown and integrated cavity output for trace species monitoring J. B. Paula, J. J. Scherera, and A. O’Keefea, L. Lapsonb, J. G. Andersonb, C. Gmachlc, F. Capassoc, and A.Y. Choc

aLos Gatos Research, 67 E. Evelyn Ave. Ste #3, Mountain View, CA 94041 bDept. of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138 cBell Laboratories, Lucent Technologies, 600 Mountain Ave., Murray Hill, NJ 07974

ABSTRACT

Although the ability of high finesse optical cavities to provide effective absorption path-lengths exceeding 10 km. has been known for quite some time, attempts to utilize this property for the purposes of high-resolution spectroscopy have often resulted in extremely complex experimental systems. Here, we demonstrate how off-axis optical paths through such cavities can be employed to produce relatively simple spectrometers capable of ultrasensitive absorption measurements. A proof-of-concept study using visible has achieved a normalized absorption sensitivity of 1.8*10-10 cm-1Hz-1/2. Additionally, quantum cascade lasers have been employed to extend this method into the mid- infrared region, where sensitivities of 1.2*10-9 cm-1Hz-1/2 have been obtained.

Keywords: absorption spectroscopy, quantum cascade , optical cavity, trace-gas detection

1. INTRODUCTION

In recent years, a highly sensitive absorption based method known as Cavity Ringdown Spectroscopy (CRDS) has been successfully implemented for infrared gas absorption analysis studies.1-7 In CRDS, the intensity decay rate of trapped in an optical cavity is used to obtain the associated total intra-cavity losses (per pass) as a function of the optical . When the cavity loss is dominated by cavity scatter and transmission, the resolved “loss” curve maps out the mirror reflectivity function. When a narrow band absorbing species is present, absolute atomic or molecular absorption intensities can be inferred by subtracting the baseline (non-resonant) loss of the cavity, which is determined while the laser is tuned off- with the transitions. The great utility of the CRLAS method lies as much in the extremely high sensitivity as in the simplicity of the technique. Absolute concentrations are easily inferred from the absorption data that CRDS provides. CRDS concentration detection limits for many species have been demonstrated to be in the part-per-billion to part-per-trillion range.8

Although this technique can provide a significant improvement in the sensitivity of infrared absorption analysis, pulsed IR light sources generally do not possess the high frequency resolution necessary for many IR applications, and often require too much power. In response to this, we have developed the Integrated Cavity Output Spectroscopy (ICOS) method,9,10 which allows narrowband continuous-wave (CW) lasers to be used in conjunction with optical cavities in a simple and effective manner. The absorption signal is obtained through the integration of the total signal transmitted through the optical cavity, in much the same fashion as in conventional absorption measurements. Single pass cavity losses are calculated from the measured cavity output, which is a function of the mirror reflectivity as well as scattering and absorption losses due to the presence of samples located between the . Previous attempts to use optical cavities as absorption cells often required complex hardware because they attempted to actively control the interaction between the narrowband light and the optical resonance modes of the cavity.11 The ICOS approach, in contrast, requires a relatively simple apparatus, because it is based on systematically disrupting the cavity , thereby recovering frequency averaged response of the cavity. It will be shown below that this property is extremely sensitive to the round trip intensity loss experienced by the intracavity light as it circulates through the cavity. Previously, we described a method for disrupting the resonances that involved modulating the cavity length and/or the laser frequency.9 Here, we describe a more effective method that we have recently developed, which is based on aligning the laser in an off-axis configuration with respect to the cavity. The principles of this technique are discussed below.

2. THE OFF-AXIS ICOS APPROACH

Off-axis paths through optical cavities are well understood,12,13 and in effect spatially separate the multiple reflections within the cavity until the “re-entrant” condition is fulfilled, which refers to the time at which the ray begins to retrace its original path through the cavity. The occurrence of this condition is dictated by the specific curvature and spacing of the mirrors forming the cavity. Any stable cavity geometry can produce stable off-axis paths through the cavity, where the stability condition (for a spherical 2-mirror cavity) is defined by the inequality

0 < (1 - d/R1)(1 - d/R2) < 1, (1) where d is the mirror spacing and R1 and R2 are the mirror curvatures. The multiple reflections appear on the mirrors as a series of spots in an elliptical pattern. The per-pass rotation (θ) around the ellipse is again determined purely by the geometry of the cavity, and is given by

cosθ = 1 - d/R, (2) assuming R=R1=R2. When 2mθ = 2πn, where m equals the number of optical round-trip passes and n is an arbitrary integer, the pattern becomes re-entrant. For certain cases, this occurs after only a few passes, however, for others the number of passes can approach infinity. In many respects, the properties of the cavity, including the free spectral range (FSR), become similar to one that is m times longer. For example, a 0.5 m cavity normally has a FSR of 300 MHz, however, when aligned in a 100 pass configuration (50 round-trips) has a FSR of only 6 MHz. The reentrant condition can easily be lengthened to over 1000 passes by using astigmatic mirrors, which results in a Lissajous spot pattern.13 For the cases of CRDS and ICOS, since the light is not extracted from the cavity by means of a hole in the mirror as in multi-pass absorption methods, a specific pattern (or even a known pattern) is not required. This fact removes much of the complications associated with using astigmatic mirror cavities.

Once a condition is achieved where the effective cavity FSR is significantly narrower than the laser bandwidth, the “fringe contrast” ratio approaches unity, implying that the energy coupled into the cavity ceases to be a function of the laser wavelength.14 In practice, the more important ratio is between the cavity FSR and the bandwidth of the absorption feature of interest, as once the effective FSR is narrower than the absorption feature the laser frequency can be dithered, or more simply, rapidly scanned linearly through the cavity modes to reduce the fringe contrast while retaining a sufficient number of data points to define the absorption feature.

With these conditions met, all wavelength and electric-field phase information can be neglected, leading to a simplified description of the intracavity optical intensity. In this case, a source term is added to the standard differential equation used to describe the ringdown event,15 resulting in the following rate equation describing the change in the intracavity power (traveling in each direction):

dI c =−−[]ICT21 I() R , (3) dt 2L Lp where IL is the incident laser power, Cp is a cavity coupling parameter, R and T are the mirror intensity reflection and transmission coefficients, L is the cavity length, and c is the . The factor of 2 in the loss term accounts for the fact that the light leaves through both mirrors, while only enters through one. Cp has a value between 0 and 1, and generally depends on the mode quality of the light source and the degree of mode matching between the cavity and the laser. For pulsed lasers, this value is often fairly low (~0.1), but it can approach unity for a well matched TEM00 CW laser. Assuming a stable (i.e. time invariant) light source, the solution to Eq. 3 for an initially empty cavity is

ICTLp  − t  I = 1− e τ  . (4) 21()− R  

When the laser is switched on, a “ring-up” occurs with the same time-constant as the ring-down, given by τ = L/(c(1-R)).

Steady-state is reached when I = ILCpT/(2(1-R)) in each direction, i.e. the amount transmitting through the rear mirror is ≈ ILCpT/2 (assuming R+T≈1). In other words, at steady-state half of the laser power coupling into the cavity leaves through each mirror, as required by energy conservation. In general, this result represents the broadband steady-state transmission of any high finesse etalon, as it also corresponds to the integral of the Airy function over one FSR. Once sufficient laser power is leaving the cavity, the laser can be quickly interrupted to observe the ringdown decay. As the intensity buildup occurs predictably and on a well defined timescale, this can be done with a passive device such as a mechanical chopper.

0 10

-1 10

4 G=10 -2 10 3 G=10

2 G=10 -3 10 Measured absorption

-4 10

-5 10 -8 -7 -6 -5 -4 -3 -2 -1 10 10 10 10 10 10 10 10 Sample absorption (ε)

Figure 1. Effective cavity as a function of the gain parameter (G) and intracavity absorption.

With an absorbing medium between the mirrors, R is replaced by R’, given by

−αω RRe' =⋅ (), (5) where α(ω) represents the absorbance of the medium over the length of the cavity. Thus, the intracavity absorption can be determined by comparing the cavity decay times with and without the absorber present (thereby determining both R - and R’), as comparing Eq. (5) with the Beer-Lambert absorption formula for a single pass through the medium (I/Io = e α(ω) ) reveals that I/Io = R’/R. Eqs. (4) and (5) also show that absorption information is contained in the steady-state cavity output intensity, which is the basis for ICOS. From these equations, it is easy to show that the change in steady- state output due to the presence of an absorbing species is given by

∆I GA = , (6) I 1+ GA −α(ω) where A = 1 - e and G = R/(1-R). For weak absorption (GA << 1), the cavity provides a linear absorption signal gain (Fig. 1). Therefore, G will be referred to as the gain of the cavity. Physically, G equals the number of optical passes occurring within cavity decay time (G ≈ τ c/L), and is also simply related to the cavity finesse (G ≈ F/π). For example, a gain of 104 (1-R = 100 ppm) would result in a 1% change in cavity output power for a single-pass absorption (A) of 1 ppm. It is also clear from Fig. 1 that as the absorption becomes comparable to the intrinsic cavity loss, the gain “rolls- off” due to a saturation effect.

The above analysis shows that the cavity provides tremendous signal gains even if the resonances are completely suppressed. In fact, the signal gain is only reduced by a factor of two compared with the expected gain for the case of resonantly coupling the laser to a single cavity mode. This factor relates to the relative change between the peak and the area of an individual Lorentzian cavity mode for a given cavity loss figure. The transmitted power level, on the other hand, is reduced by a factor of ~T/2, or 2⋅104 for of the case of T=100 ppm mirrors. In terms of the ultimately achievable shot-noise limit, however, the difference between the two methods is only T /2 for a given laser power level, or a factor of 140 in this case.

3. RESULTS

3.1 Visible laser- proof of concept

Figure 2. Schematic diagram of the experimental apparatus used for visible and near-IR studies.

A schematic diagram of the experimental apparatus used in the initial proof-of-concept study is shown in Fig. 2. The optical cavity consisted of two identical 6 m. radius of curvature, 1 in. diameter mirrors, spaced 67 cm. apart. Other cavity configurations can be used, as long as the stable condition given by Eq. 1 is met. The use of long focal length optics compared with the cavity length was chosen following numerical simulations of a variety of cavity designs, including near confocal, and the commonly used spacing near the limit of stability of slightly less than 4f, where f equals the mirror focal length. While each of these possibilities could be used to create a dense cavity mode spectrum, the long focal length condition was found to be the least sensitive to changes in both alignment and mirror spacing. Additionally, the pattern of light emerging from such a cavity could be most easily focused to a small area. While this is not a big concern when large area photomultiplier tubes are available, it becomes important for infrared applications where small detector element sizes are preferred to reduce noise.

The mirrors in this case provided a per mirror intensity loss of 235-300 ppm (G≈4000) at 630 nm, depending upon their state of cleanliness. The ringdown cavity and a 2 mW external cavity diode laser (ECDL, New Focus 6300 series) were mode matched using a two lens telescope. A standard mechanical chopper placed at the focal point between the telescope lenses interrupted the beam with a 50% duty cycle at a 3 kHz repetition rate. This arrangement provided a shut-off time of the laser into the cavity of <0.5 µs. Light exiting the cavity was focused by a lens, filtered by a narrow band-pass filter, and detected by a photomultiplier tube (PMT). 3.1.1. Off-axis injection

(a) (b)

Figure 3. Images obtained by a video camera looking through the rear mirror of the cavity showing the multipass cavity alignment. a) Spherical mirrors. b) Astigmatic mirrors.

Fig. 3 shows images observed by a video camera through the rear cavity mirror. The pattern in Fig. 3a closely matches that expected from calculations, although individual spots are not observed because the resulting image is actually the sum of thousands of slightly displaced spots. With slightly astigmatic mirrors, the expected Lissajous pattern (Fig. 3b) is observed. Here, the effective reentrant condition is lengthened compared with the spherical case, resulting in an even closer cavity mode spacing. Even with the astigmatic cavity, however, the power fluctuations through the cavity were on the order of 50% of the peak power with the wavelength of the laser held constant. These fluctuations, which can be attributed to the extremely long coherence length of the ECDL (>2 km.), occurred on a millisecond timescale, compared with the microsecond events observed for on-axis alignment. Presumably, larger diameter mirrors could be used to increase the power stability in the cavity. Despite these fluctuations, at no time would the cavity output drop to zero, such that at any time the laser could be interrupted and a ringdown event observed.

Stable power output from the cavity could be achieved by slightly modulating the laser wavelength using a 1 kHz sine wave applied to the laser PZT to produce a 150 MHz frequency modulation. Aside from sharp insensity spikes occurring near the turning points of the sine wave, where the laser end mirror is momentarily stationary, stable power transmission can be observed. In fact, scanning the laser at a rate exceeding 1 cm-1/s eliminated all traces of cavity resonances, even with a detection bandwidth exceeding 100 kHz, which implies an effective cavity mode spacing on the order of 3 MHz. This estimate assumes that scanning the laser through several free spectral ranges within the ringdown time is sufficient to suppress resonant energy buildup in the cavity.

3.1.2. CRDS

With the chopper wheel spinning, the buildup-ringdown events occur at a high repetition rate, in this case chosen to be 3 kHz (Fig. 4). An average of 20 individual cavity decay waveforms is shown in Fig. 5, which exhibits a smooth exponential function. Scanning the laser frequency produces an absorption spectrum such as the one shown in Fig. 6a, in which P(8) line of the oxygen γ band has been detected directly in the laboratory atmosphere with a 67 cm open-path cavity. Each point in this scan represents the average of 50 individual decay waveforms. This spectrum corresponds to 7 seconds of actual data collection (21,000 shots at 3 kHz). The RMS fluctuations in these data are 0.8 ppm, which corresponds to an uncertainty in the time-constant determination of 3*10-3, and an RMS fractional absorption sensitivity of 1.5*10-9 cm-1Hz-1/2. While this sensitivity is consistent with the limitations of an 8-bit digitizer, unfortunately, one with a higher bit-depth was not available. As expected, this system was found to be largely insensitive to vibrations and slight changes in alignment. 0.20

0.15

0.10

0.05

0.00 0 200 400 600 800 µs

Figure 4. The buildup-ringdown cycle measured with the chopper activated.

1.0 0 -1 0.8 -2 -3

0.6 ln (signal) -4 -5 0.4 0 5 10 15 20 25

signal (arb. units) signal (arb. µs 0.2

0.0 0 5 10 15 20 25 µs

Figure 5. Measured ringdown decay signal using the present multipass cavity ringdown method. Inset: The natural logarithm of the signal is linear, indicating a single exponential decay.

20 20

15 15

10 10

5 5

0 0

15841.0 15841.5 15842.0 15842.5 15843.0 15841.0 15841.5 15842.0 15842.5 15843.0

(a) (b)

Figure 6. a) The P(8) line of the molecular oxygen γ band as measured with off-axis CRDS. Also shown is the HITRAN prediction based on an atmospheric O2 abundance of 21% b) The same line measured with off-axis ICOS. 3.1.3. ICOS

To record an ICOS spectrum, the laser was scanned over a 2 cm-1 frequency interval at a repetition rate of 10 Hz with the chopper wheel deactivated. The noise level during a single sweep corresponded to fractional absorption of ~1 ppm, such that the 21.5 ppm absorption feature could be easily observed in real time with a 0.1 s update interval. Averaging for 10 sec. (100 sweeps) produced the data shown in Fig. 6b. The rms S/N demonstrated here is >200, which corresponds to a per-pass fractional absorption sensitivity of 1*10-7. The best absorption equivalent noise levels observed were as small as 6.5*10-8, or 1.8*10-10 cm-1Hz-1/2, considering the 3 kHz detection bandwidth and the number of averaged cycles. This result is within a factor of 50 of the shot-noise limit considering our best estimate of the received optical power.

3.2. MID-INFRARED () RESULTS

The experimental apparatus used to obtain mid-IR results is shown in Fig. 1. Up to 50 mW of optical radiation near λ=8 µm is produced by the single-mode (distributed-feedback) quantum cascade (QC) laser when supplied with a current of up to 1 A. The laser is mounted to the cold-head of a LN2-cooled dewar, which maintains the temperature of the copper substrate of the laser near 80 K during CW operation. The current is supplied by a low-noise controller (Wavelength Electronics MPL-2500), which in turn is controlled by a standard laboratory function generator. These components typically supply a 10 Hz sawtooth current waveform to the QC laser that ranges from 0.25 A - 1.0 A, thereby scanning the laser frequency over ~2.5 cm-1. The lower end of this range is chosen to bring the laser below the in order to sample the zero-light baseline with each current sweep. The laser emission is collimated by L1 (a 1.1 in. dia. ZnSe aspheric f/1 AR-coated lens), and directed into a simple two-lens telescope formed by L2 and L3 (f/1.38 and f/1 ZnSe aspheres, respectively) to mode-match the light into the optical cell. A chopper wheel placed at the focal point of L2 can be used to interrupt the laser with a switching time of ~2 µs for cavity ringdown measurements. Following the telescope, a pair of mirrors are used to direct the laser beam into the optical cavity, which is pre-aligned using a He-Ne laser. The two mirrors forming cavity consist of multilayer dielectric coatings applied to 2 in. dia, 6 m. radius-of-curvature ZnSe substrates. These mirrors also seal the ends of a gas flow-cell against the atmosphere. The optical power leaving the cell is collected by L4 (2 in. dia. f/1 ZnSe meniscus lens), which focuses the cavity output onto a LN2-cooled, 2 mm dia HgCdTe photovoltaic detector with a built-in pre-amplifier module. The resulting electrical signal is filtered by a 0.1 Hz - 30 kHz bandpass filter, recorded by a digital storage oscilloscope, and processed by a standard PC. The alignment of the cavity optics could be confirmed by aligning the laser on-axis with respect to the cavity, which would result in large power transmission spikes as the laser randomly drifts in-and-out of resonance with the cavity modes. With the current sweep enabled and the laser aligned off-axis by ~0.6 in, the resonances could be effectively suppressed.

The first experiment involved filling the cell with 30 torr of a methane/air mixture (1.04 ± 0.01 ppm mixing ratio, Matheson Gas Products). The cavity length was carefully measured to be 82.3 cm. Fig. 8a shows the raw QCL-ICOS signal collected in 10 sec. at a 10 Hz repetition rate (100 averages). The overall sawtooth shape is the result of the current ramp applied to the QC laser. The large intensity drop near point #1700 reveals a mode-hop; the location of which is quite reproducible and therefore does not present a problem when interpreting the subsequent data. Following this, several absorption features are evident, the first of which is due to water. Two CH4 absorption features can be observed near the end of the current ramp. This region, indicated by a box, is shown expanded and scaled in frequency in Fig. 8b along with a least-squares fit of the data to the following algebraic variant of Eqn. 6

P()ω I ()ω = , (7) out 1+⋅GA()ω where P(ω) is a simple quadratic polynomial representing the baseline cavity output in the limit of zero absorption, G is the cavity gain parameter given by R/(1-R), where R is the mirror reflectivity, and A(ω) is the frequency dependent single-pass absorption. In terms of an integrated cross-section (S) such as those found in the HITRAN database

AScL()ωω=−1 exp( −⋅Γ () ⋅⋅ ), (8) where Γ(ω) is an area-normalized lineshape function, in this case given by a Voigt profile, c is the concentration of the absorbing species, and L is the cavity length. According to HITRAN, the small peak is weaker than the main peak by a factor of 70. This ratio was fixed at this value during the fitting procedure. The mirror reflectivity was measured immediately following the data collection by setting the laser current to 1.0 A, thus placing the laser wavelength sufficiently far from any molecular absorption, and employing the chopper wheel. The resulting ringdown decay waveforms determined the baseline per-pass cavity loss to be 148 ppm (G=6.76·103).

Current Function controller generator QCL Chopper L1 wheel

MCT MML2 L3

L4 ICOS/Flow cell

Digital oscilloscope Computer

Figure 7. Schematic diagram of the QCL-ICOS experimental apparatus. QCL = LN2 cooled distributed-feedback quantum-cascade laser, M = cavity mirror, L1-4 = Zinc-selenide lenses, MCT = LN2 cooled mercury-cadmium-telluride photovoltaic detector.

These parameters, along with the results of the fitting routine, determined the methane concentration in the cell to be 1.02 ± .03 ppmv. This agrees well with the concentration data from the supplier of the gas mixture quoted above. The fitted parameters are summarized in Table 1, which show good agreement overall with expectations based on room temperature and the measured cell pressure. The error in the concentration measurement represents the statistical error generated by the fitting routine, which is largely due to the covariance between the fitted parameters. As expected, fixing the lineshape parameters to the expected values lowers the statistical error by an order of magnitude, which implies a detection limit of ~3 ppbv in 10 sec. This is consistent with the observed signal-to-noise ratio, which corresponds to a fractional per-pass absorption equivalent noise level of 1·10-6, or a normalized noise figure of 2·10-9 cm- 1Hz-1/2. Table 1. Lineshape parameters for the fitted CH4 absorption. Parameter Expected Measured Gaussian linewidth (HWHM) 65.3 MHz 66.3 MHz Lorentzian linewidth (HWHM) 68.1 MHz 85.7 MHz Concentration 1.04 ± .01 ppmv 1.02 ± .03 ppmv

Repeated measurements, however, reveal a systematic under determination of the concentration by ~3%. While there are several possible explanations for this, we believe it to be a manifestation of the ~20 µs cavity time-constant limiting the signal acquisition speed of the system. The slight distortions from the expected Voigt profile observed in the measured lineshapes are consistent with this explanation. While we have developed a model that accounts for this effect and accurately reproduces the measured lineshapes, further study will be required in order to confirm this as the origin of this effect. We have also identified a noise source in the laser current supply electronics that was causing a slight frequency jitter in the laser emission from sweep-to-sweep. This could account for the overly broad Lorentzian linewidth returned by the fitting routine, and may also have contributed to the lower than expected concentration measurement.

0.8 0.4 50 0.0 -0.4

40 50 30 40 20 signal (arb. units) signal (arb. 30 signal (arb. units) signal (arb. 10 CH4 H2O 20 0 1000 2000 3000 4000 0.00 -0.05 -0.10 -0.15 -0.20 -0.25 -0.30 data point # rel. wavenumber (a) (b)

Figure 8. (a) QCL-ICOS signal on the detector with 30 torr of a calibrated 1.04 ppmv CH4/air balance mixture in the detection cell and a 10 sec averaging time. (b) The boxed region above shown expanded and scaled in frequency along with the results of the fitting routine.

Next, a continuous flow of room air at a cell pressure of 10 torr was scanned at 15 Hz for 2 min. The cavity length in this case was 68.5 cm. Figure 9 shows the resulting total cavity loss spectrum, in which the methane -1 absorption line measured above is again clearly observed, along with a N2O transition at 1263.83 cm , and several other -1 13 features. The more interesting of these include a pair of HDO lines near 1263.7 cm , and a pair of CH4 lines marked in the figure by asterisks. The recovered concentrations were 1.78 ppmv and 297 ppbv for CH4 and N2O, respectively, which are reasonable values for room air. However, the faster tuning rate in this case led to more pronounced lineshape distortions, which gives us reason to believe the true concentrations were actually 5-10% higher. The peak-to-peak noise equivalent absorption (NEA) level in this spectrum is ~0.2 ppm, as seen in Fig. 9 (inset).

While the sensitivity demonstrated here is comparable to the best levels reported using a QC laser to date, we believe there exists considerable room for improvement. In the above work, the limiting noise source was detector noise. One of the main reasons for this appears to be the rather poor spatial mode quality of the particular QC laser used in this study, which led to inefficient coupling of the laser light into the cavity. In response to this, we have redesigned the optics used to focus the light leaving the cavity onto the detector in order to replace the 2 mm detector with a 1 mm element, which correspondingly reduces the detector noise by a factor of 2. This, combined with improved amplification and filtering schemes, have increased the detected laser power to a point were the limiting noise source has become fluctuations in the cavity output due to residual cavity mode resonances. While this was not a problem in the visible region, even with only 1 in. diameter cavity mirrors, the fact that the characteristic spot size on each mirror scales linearly with wavelength makes it extremely difficult to keep at least some of the radiation from coupling on-axis in the mid-IR. Potential solutions to this problem include using even larger cavity optics, spatially filtering the laser, and modulating the cavity length. Hopefully, implementing one or all of these solutions will bring the IR results on par with those observed in the visible region.

400

120 CH4 H2O 119 Figure 9. QCL- 350 ICOS absorption 118 spectrum of 117 300 flowing room air 116 at 10 torr.

115 Prominent 250 features are N2O 1263.3 1263.4 labeled.

loss per pass (ppm) Asterisks mark 200 the locations of HDO 13 two CH4 150 absorption lines. * * 100 1263.2 1263.4 1263.6 1263.8 1264.0 1264.2 1264.4 wavenumber 4. CONCLU SIONS

Off-axis paths through high finesse optical cavities have been employed to produce high resolution, ultrasensitive absorption measurements in the visible and infrared spectral regions. Normalized absorption sensitivities of 1.8*10-10 cm-1Hz-1/2 and 1.2*10-9 cm-1Hz-1/2, respectively, have been achieved. Additionally, this method is relatively simple compared with previous attempts to use optical cavities for such purposes, and is robust and easy to use.

5. REFERENCES

1. A. O'Keefe and D.A.G. Deacon, “Cavity ring-down optical spectrometer for absorption measurements using pulsed laser sources,” Rev. Sci. Instrum. 59 (12), 2544-51 (1988). 2. A. O'Keefe, J.J. Scherer, A.L. Cooksey et al., Chem. Phys. Lett. 172, 214 (1990). 3. J. J. Scherer, D. Voelkel, D. J. Rakestraw et al., “Infrared cavity ringdown laser absorption spectroscopy (IR- CRLAS),” Chemical Physics Letters 245 (2-3), 273-80 (1995). 4. J. B. Paul, C. P. Collier, R. J. Saykally et al., “Direct measurement of water cluster concentrations by infrared cavity ringdown laser absorption spectroscopy,” Journal of Physical Chemistry A 101 (29), 5211-14 (1997). 5. J. B. Paul, R. A. Provencal, and R. J. Saykally, “Characterization of the (D2O)2 hydrogen-bond-acceptor antisymmetric stretch by IR cavity ringdown laser absorption spectroscopy,” Journal of Physical Chemistry A 102 (19), 3279-83 (1998). 6. J. J. Scherer and D. J. Rakestraw, “Infrared cavity ringdown spectroscopy in low-pressure flames,” CLEO '96. Summaries of Papers Presented at the Conference on Lasers and Electro-Optics. Vol.9. 1996 Technical Digest Series. Conference Edition , 52 (1996). 7. J. J. Scherer, K. W. Aniolek, N. P. Cernansky et al., “Determination of methyl radical concentrations in a methane/air flame by infrared cavity ringdown laser absorption spectroscopy,” Journal of Chemical Physics 107 (16), 6196-203 (22 Oct. 1997). 8. G. Berden, R. Peeters, and G. Meijer, “Cavity ring-down spectroscopy: Experimental schemes and applications,” Int. Rev. Phys. Chem. 19 (4), 565-607 (2000). 9. A. O'Keefe, J.J. Scherer, and J.B. Paul, “CW integrated cavity output spectroscopy,” Chem. Phys. Lett. 307 (5-6), 343-9 (1999). 10. A. O'Keefe, “Integrated cavity output analysis of ultra-weak absorption,” Chemical Physics Letters 293 (5-6), 331-6 (1998). 11. J. Ye, L.-S. Ma, and J. L. Hall, “Ultrasensitive detections in atomic and molecular physics: demonstration in molecular overtone spectroscopy,” J. Opt. Soc. Am. B 15 (1), 6-15 (1998). 12. D. R. Herriott, H. Kogelnik, and R. Kompfner, “Off-Axis Paths in Spherical Mirror Interferometers,” Appl. Opt. 3 (4), 523-26 (1964). 13. D. R. Herriott and H. J. Schulte, “Folded Optical Delay Lines,” Appl. Opt. 4 (8), 883-89 (1965). 14. K. K. Lehmann and D. Romanini, “The superposition principle and cavity ring-down spectroscopy,” J. Chem. Phys. 105 (23), 10263-77 (1996). 15. J. J. Scherer, J. B. Paul, A. O'Keefe et al., “Cavity Ringdown Laser Absorption Spectroscopy - history, development, and application to pulsed molecular beams,” Chem. Rev. 97 (1), 25-51 (1997).