1. Introduction

Gas-driven shock tubes have a long history of providing well-controlled high-temperature conditions for the study of combustion chemistry [1]. Here we present the first use of mid-infrared (MIR) cavity-enhanced absorption spectroscopy (CEAS) to improve detection sensitivity for in situ measurements of time-resolved species concentrations in a shock tube. Briefly, such shock tubes have a high-pressure section separated by a diaphragm from a low-pressure gas mixture of precursors to the chemistry being studied. The diaphragm bursts and a shock wave travels through a gas mixture, thereby heating, compressing and accelerating the gas; the shock wave then reflects off the end wall of the tube, stagnating the gas as it heats and compresses it a second time. The resulting uniform high-temperature gas mixture has been nearly instantaneously heated to a temperature and pressure that are precisely known from measured shock velocity, providing an ideal test bed to study combustion chemistry. The post-reflected-shock test-time for high-temperature chemical kinetics experiments is limited by arrival of other reflected waves, and this test time can be short, often ranging from 0.5 to 3ms, although recent advances have shown that steady gas conditions can be maintained for up to 100ms [2]. Monitors of the changing gas composition thus often require time resolution of ten microseconds (10μs) or better. Species-selective, continuous wave, in situ laser absorption is an ideal diagnostic for shock tube chemistry experiments [3]. The sensitivity of laser absorption measurements (in the absence of interference by other species) is the product of path length and the absorption strength (per unit length) of the transition used; thus, detection sensitivity can be increased by selecting a stronger transition and/or increasing the path length. Here we report the demonstration of MIR CEAS to increase the effective path length and thereby to improve species-selective in situ absorption measurements in a shock tube.

CEAS using a high-finesse optical cavity formed with highly reflective mirrors has long been used to increase the effective path length for absorption measurements in a gas cell [4–9]. However, the residence time of light in high-finesse cavities can limit the time response (e.g., a 15cm cavity with reflectivity R = 0.99999 mirrors has a time resolution ~100μs). Faster time response requires a lower finesse cavity. Recently, we demonstrated in situ shock tube measurements using near-infrared (NIR) CEAS with an effective path length gain of 83 using a low-finesse (R~0.988) cavity, where the time resolution was not determined by the cavity [10]. In that experiment, the cavity mirrors were installed as windows on a gas-driven shock tube (diameter 15cm), and used for in situ time-resolved absorption measurements of C₂H₂ using a telecommunications laser to access transitions in the NIR combination bands. Here we extend this work to the MIR using a DFB quantum-cascade laser (QCL) to access the R(13) transition of the CO fundamental vibrational band.

2. CEAS for in situ measurements

The cavity was formed by a pair of concave dielectric-coated mirrors (CaF₂, radius of curvature −1m, diameter 0.025m, and reflectivity 98.9% at 4.559μm, backside-polished and antireflection-coated, manufactured by Rocky Mountain Inc, USA). As shown in Fig. 1 of [10], the cavity mirrors were also the vacuum/pressure windows on the 0.15m diameter shock tube (resulting in a cavity length of 0.15m). The light from a DFB QCL near 4.56µm (Alpes Lasers) was transmitted through the cavity and lens collected and focused onto a TE-cooled photovoltaic HgCdTe detector (Vigo System, 0.002m diameter) whose output was digitally sampled at 10MHz. The laser bandwidth is estimated to be 0.001cm⁻¹ produced by the noise on the injection current.

 

Fig. 1 Left panel: Transmitted laser intensity for a single cycle of the scan without absorption (blue) and laser wavelength (red); Right panel: Transmitted laser intensity for a single scan cycle with 10ppm CO absorption (T = 1499K, P = 1.51atm.).

Download Full Size | PDF

Off-axis injection of light into the cavity has been shown to be insensitive to changes in optical alignment, temperature, reflective index of the gas flow and mechanical vibrations [11]. The alignment procedure followed [11] although instead of filling the entire cavity, here the beam was aligned to trace a narrow ellipse (short dimension aligned with the shock tube axis), because, as discussed in [10], the time resolution is limited by the transit time of the shock wave through the light in the cavity. The width of the light pattern exiting the cavity was measured to be less than 0.005 m by translating a beam block across the exit mirror. For the ~1000m/s shock speeds used in this experiment, this corresponds to ~5μs measurement time resolution.

To further suppress cavity coupling noise, the laser wavelength was rapidly scanned by varying the injection current using a 50kHz sinusoidal waveform. The low-injection-current part of the scan was below the laser threshold, which allowed detection of any thermal emission from the hot gas on every laser scan. In this work, the thermal emission at the highest temperatures was only a few per cent of the detector signal. The light transmitted by the cavity without absorption from a single scan is shown in the left panel of Fig. 1. (The detector signal was verified to be linear in light intensity.) The cavity transmission was ~1% and the laser power >30mW; thus detector noise was not significant even with the 10MHz detector bandwidth. The MIR cavity transmission was unchanged by the passage of the shock waves, and although there is some schlieren-driven beam steering when the shock wave (or reflected shock wave) passes, the transmission recovered within a few µs (consistent with the transit time of the shock wave across the light-filled volume of the cavity). Figure 1 also illustrates the phase delay between the scan of the laser wavelength compared to the laser intensity. Note the increase in cavity mode noise at the peak of the sinusoidal scan when the laser wavelength scan rate is near zero, which illustrates the suppression of the cavity mode noise by the laser wavelength scan. The laser wavelength scan range was ± 0.32cm⁻¹, although this corresponds to an average scan rate of 0.064cm⁻¹/μs;, the laser scan rate was 0.1cm⁻¹/μs in the region of the sinusoidal modulation where CO absorption was measured. In the near-linear portions of the sinusoidal scan where the CO absorption is measured, Fig. 1 shows no evidence of cavity coupling noise even with the small (0.025m diameter) mirrors. Previous work using high-finesse cavities, low wavelength-scan rates, and a narrower laser bandwidth, recommended using at least 0.1m diameter mirrors for MIR CEAS [12,13]. Here we show that fast wavelength scanning, coupled with a low-finesse cavity and a relatively broad bandwidth laser, has a significant advantage in the reduction of cavity mode noise (even with small mirrors and a partially filled cavity).

When the cavity coupling noise is fully suppressed and the light transmission is uniform with wavelength, the expression of the total transmitted light intensity reduces to the simple form [14]:

3. Absorption measurements of CO

The QCL probed the R(13) CO transition in the fundamental ro-vibrational band, and the spectroscopic parameters of this transition are listed in Table 1. The CEAS in situ measurement scheme was tested by shock-heating non-reactive gas mixtures of CO (2-100ppm), H₂ (1%), and argon, which provides test gas at precisely known gas composition, temperature and pressure. Hydrogen was added to the test gas for fast vibrational energy transfer in CO to insure that the CO vibrational distribution was thermal.

Table 1. Spectroscopic parameters at a reference temperature of 296K for R(13) CO transition in the fundamental vibrational band. Line center, line strength and lower-state energy are from HITRAN 2012 [15]; the collision broadening parameters 2γ and n, and line strength confirmation are from [16].

View Table

A single scan of shock-heated CO (10ppm at 1499K and 1.51atm.) shown in the right panel of Fig. 1 data was collected in the stagnated high-temperature and pressure region behind the reflected shock wave. Note the phase delay of the wavelength tuning of the laser causes the CO absorption to be asymmetric compared to the laser intensity modulation; up-scan and down-scan data must be separately evaluated. The gain factor G must be determined for quantitative CEAS. Known amounts of CO in a bath gas were used to determine G by calibration; alternatively G was calculated from the mirror specification. Both methods returned G = 91. No variation in G was observed over the entire measurement campaign (two weeks and more than 30 shock events).

The absorbance from the data in the right panel of Fig. 1 is shown in the left panel of Fig. 2. Each such absorbance curve was fit to a Voigt lineshape and the peak absorbance used to determine mole fraction. The right panel of Fig. 2, shows the CO measured over the first 1ms of the post-reflected shock test time. Note that the expected CO mole fraction was obtained with a standard deviation of only 0.11ppm. This corresponds to a minimum detectable absorbance α_CEAS of 2.4 × 10⁻⁵ with a 10μs measurement time; note the three standard deviation value would be three times this value. To check the repeatability and dynamic range of the sensor, measurements were conducted at different shock conditions for temperatures from 1100 to 2100K, pressures from 1.2 to 1.6 atm, and CO mole fractions from 2ppm to 100ppm. The measured and prepared CO mole fractions are shown in Fig. 3.

 

Fig. 2 , The absorbance from 10 ppm CO @ 1499K obtained from the data in Fig. 1; CO mole fraction for the first 1ms after the reflected shock (10ppm CO, 1% H₂, balance argon, 1499K, P = 1.51atm).

Download Full Size | PDF

 

Fig. 3 CO measured by CEAS behind the reflected shock versus known CO in mixtures with 1% H₂ and the balance argon with gas pressures ranging from 1.2 to 1.6atm.

Download Full Size | PDF

To demonstrate the utility of in situ MIR CEAS for shock tube kinetics studies, a chemically reacting experiment was performed; 10ppm of acetone, (CH₃)₂CO, in argon was shock heated to 1380K (at 1.5atm), and the time-resolved appearance of CO was monitored by CEAS as illustrated in Fig. 4. At this low acetone mole fraction, there is no significant secondary chemistry in the first few ms. Thus, one CO molecule is expected for each acetone molecule that dissociates. At 1380K, complete thermal decomposition of acetone is expected. The CO production is fit with the thermal decomposition rate constant of 3.3 × 10³s⁻¹ [17], which is within the uncertainty of the published value of 2.6 ± 0.7 × 10³s⁻¹ [18] that was performed with 250 times more acetone (and hence greater influence of secondary reactions and temperature change from the endothermic acetone decomposition).

 

Fig. 4 CO formation (red) during the thermal decomposition of shock heated acetone with exponential fit (green) to determine the acetone decomposition rate constant [17]; also shown (dashed) is fit using the rate constant from [18].

Download Full Size | PDF

The current MIR CEAS time resolution was limited to 10µs by the 10MHz analog-to-digital card and the need for at least 100 points to fit the absorption baseline and lineshape, but faster hardware is readily available. Ultimately however, the measurement time resolution will be limited by the time the shock wave takes to cross the width of the beam pattern in the cavity, which for typical shock speeds is currently 5-10µs. For example, an incident shock wave at 1000 m/sec takes 5μs to cross the laser filling of the CEAS cavity, and for argon bath gas, the reflected shock speed would be roughly 550 m/s and thus take 9μs to cross the CEAS cavity.

4. Comparison with previous NIR CEAS in a shock tube

In situ shock tube NIR CEAS measurements of C₂H₂ were performed recently using a combination band transition near 1.537μm [10] with a minimal detectable absorbance (MDA) of 1.6 × 10⁻⁴ and 20μs measurement time. Notably, the MIR MDA of 2.4 × 10⁻⁵ with a 10μs measurement time demonstrated here was significantly better than that demonstrated by us in the NIR, for nearly identical cavities (R~0.989 MIR and 0.988 NIR and ROC −1m for both). The laser scan rate in the MIR of 0.1 cm⁻¹/μs was slightly faster than the NIR 0.07cm⁻¹/μs. However, we attribute the improved performance to the fast scan coupled with the laser bandwidth, which for the MIR was 0.001 cm⁻¹ and hence at least two orders-of-magnitude broader than for the NIR. The MIR bandwidth was dominated by injection-current noise, and noise driven laser bandwidth has been shown to reduce cavity-coupling noise [19]. Using the same MIR laser, the MDA determined for single-pass absorption of CO in the shock tube, using CaF₂ windows instead of the CEAS mirrors, was 2 × 10⁻³. This is nearly the same as single-pass MDA for the CEAS experiment (CEAS MDA × G = 2.4 × 10⁻⁵ × 91 = 2.2 × 10⁻³). Note the comparison of NIR and MIR MDA used values for one standard deviation. Thus, in the rapid scan region of the sinusoidal modulation the cavity adds very little to the noise of single-pass absorption.

5. Conclusions

A fast MIR CEAS technique, based on a rapidly scanned quantum-cascade laser, was developed and demonstrated for CO in a shock tube. The measurement scheme was validated for CO mole fraction ranging from 2 to 100ppm, temperatures from 1000 to 2100K, and pressures from 1.2 to 1.6atm. The absorption gain of 91 with the CEAS cavity was equal to the value determined from the mirror specification and remained constant through the entire measurement campaign without any need to clean the mirror. The CO detection sensitivity at 1500K was measured to be 0.1ppm. The sensor was demonstrated to quantitatively detect CO products of acetone decomposition at high temperature illustrating the potential use of CEAS in combustion chemistry experiments. This new capability will enable studies at very low reactant concentrations, thereby reducing the influences of secondary chemistry [17] or of temperature perturbations associated with chemical change. The current experiments revealed very low noise levels suggesting the potential for further improvement in the detection limit by using mirrors of larger reflectivity.