1. INTRODUCTION

A. Motivation

The CubeSat design standard was developed in 1999, initially for university-level space science and graduate student training purposes [1]. The basic concept was to develop a minimally sized functional satellite body in the form of a 10 cm cube, along with a standardized deployer capable of reliably launching multiple CubeSats from a larger parent spacecraft. Since its inception, the CubeSat standard has expanded to include larger spacecraft bodies defined in 10 cm cube “units” (e.g., 3U, 6U, 12U, 27U [2]) and matching deployers. The popularity of the standard for academic and commercial satellites has led to the development of a wide variety of commercially available communication, propulsion, attitude control, and other subsystems. Aside from enabling low-cost scientific and commercial endeavors, the CubeSat standard provides a means for space agencies to flight test new systems and instruments before their inclusion in larger-scale missions. Although CubeSat deployments have previously been limited to Earth orbit, the recent launch of Mars Cube One [3] as part of the InSight Mars probe [4] will serve as proof of concept for interplanetary travel with CubeSats. A successful flight will mark the next step towards interplanetary deployment of CubeSat scientific missions to a variety of bodies, both as orbital platforms and as transport for small atmospheric probes or landers.

CubeSats provide an ideal platform from which to build upon the success of the Tunable Laser Spectrometer (TLS) instrument included in the Sample Analysis at Mars (SAM) instrument suite [5] of the Mars Science Laboratory (MSL) rover, more commonly known as Curiosity. SAM and TLS have provided a wealth of valuable information on Martian atmospheric composition and variability [6–9], notably including long-term measurements of seasonal variations and temporary spikes in methane levels within the Gale crater [10]. Trace gas and isotope ratio measurements would be similarly valuable for other planetary and small-body targets, and instruments based on TLS have been included in several recent mission proposals [11–13]—examples of science goals include determination of oxygen isotope ratios to constrain Saturn’s location of formation [13] and trace measurements of chemically active gases controlling atmospheric chemistry below the Venus cloud deck [11,12]. Despite strong scientific motivation, the instruments included in MSL and the lost Mars Polar Lander remain the only examples of laser spectrometers included in planetary science missions to date. Furthermore, improvements in spectroscopic methodology and electronics design since the MSL era offer dramatic improvements in sensitivity, which in turn can open up possibilities for study of low-abundance targets or clumped isotope species. Using modernized, miniaturized TLS designs for CubeSat missions can provide opportunities for high-impact, low-cost planetary science while providing flight-tested designs for larger scale missions in the future.

The focus of this paper is the development of a cavity ring-down spectroscopy (CRDS) instrument intended as a laboratory proof of concept for future implementation in CubeSat-based planetary science missions. The water absorption band near 1.39 μm was selected due to the wide variety of terrestrial, planetary, and small-body targets for water isotope measurements and the relative ease of sourcing components in this spectral range; however, in order to keep the design adaptable to other targets, no components that lack functional equivalents throughout the 1–5 μm range were used. Performance was evaluated primarily in terms of noise-equivalent absorption coefficient comparisons with existing instrumentation; water vapor isotope ratio measurements were also performed for ${\mathrm{H}}_2^{16}\mathrm{O}$, ${\mathrm{H}}_2^{17}\mathrm{O}$, and ${\mathrm{H}}_2^{18}\mathrm{O}$.

Although the original scope of this development project included only a demonstration at 1.39 μm, the hardware was later retrofitted for operation at 3.27 μm using equipment available on hand. This did not yield a fully functional instrument within the time and budget constraints of the project, as inadequate optical isolation prevented reliable wavelength scanning [14]. However, single-wavelength measurements yielded noise-equivalent absorption coefficients within a factor of 4 of those achieved at 1.39 μm; this could easily be improved upon in future implementations. Data and observations from this effort are included in Appendix A; the difficulties with optical isolation are easily solvable with standard approaches, and work will continue if additional development time and funding become available in the future.

B. Design Criteria

The overall goal of the project was to produce an instrument compatible with CubeSat design principles—low cost, adaptable design, and minimum risk to the parent spacecraft—while maintaining the high performance and science returns of the MSL spectrometer. Producing a general-purpose framework that can be adapted to different wavelength regions and science goals with minimal changes is particularly attractive, since the significant time and funding required to develop flight qualification and heritage would not need to be repeated for subsequent missions. For most applications, the spectrometer would be enclosed in a small descent probe that would be delivered to the planetary body by the CubeSat-based bus. The following points guided the design of the instrument and serve as metrics by which to evaluate the end product:

These guidelines have a number of consequences with regard to the design of a CRDS. The most obvious is the limited maximum cavity length, which in turn impacts instrument sensitivity and the cavity’s free spectral range (FSR). Because CRDS requires that the laser be on resonance with the cavity, we must either accept the cavity’s FSR as our minimum spectral resolution or include some means of altering the cavity length in order to adjust the FSR—this typically means including piezoelectric actuators to move one or both of the mirrors. A 15 cm cavity has an FSR of approximately 1 GHz, making the added weight and complexity of piezoelectric actuators an unfortunate necessity if we wish to observe Doppler-limited lines within our size constraints. Aside from FSR considerations, it is desirable to use the longest cavity that meets instrument size criteria. In addition to directly increasing the effective path length of the spectrometer, having a longer ring-down time constant improves the quality of data acquired in the face of limited detection bandwidth and laser switching speed. It should be noted that reducing the volume of the optical cell also reduces the size of the gas handling systems required to handle evacuation of the cell between measurements and the intake or production of sample gas; the specifics vary from mission to mission, but reducing the required sample volume can have a dramatic impact on the overall size of the instrument payload. Although this involves a tradeoff with regard to cavity length, it is desirable to reduce the diameter of the cell as much as possible. A diameter of 5 mm should be readily achievable for most on-axis ring-down cavities—for a perfectly aligned cavity, no measurable diffraction losses should occur within the specified wavelength range using the cavity geometry described below. At long wavelengths (i.e., 5 μm), slightly tighter mirror parallelism tolerances may be required to prevent diffraction losses; a tolerance of 0.05 deg or less should be sufficient in this case.

The choice of laser switching methods is also impacted by our design criteria. When a ring-down event is triggered, the light incident on the cavity must be attenuated strongly (by at least $-70\mathrm{dB}$ [15]) and quickly relative to the ring-down time constant. The most common choice in laboratory instruments is an acousto-optic modulator (AOM). Although AOMs have been used in spaceflight applications in the past [16], they consume significant power and generate waste heat, both of which should be avoided where possible. Where available, semiconductor optical amplifiers (SOAs) provide an excellent alternative [17]; however, commercial availability of SOAs is limited outside of narrow wavelength ranges. Various triggering schemes involving electro-optic modulators have also been used [18,19], but these suffer from similar limitations. An SOA trigger is likely the best choice when available, but the only option viable throughout the entire 1–5 μm range is to drop the laser injection current below threshold. This necessitates a waiting period between ring-down events while the laser re-stabilizes, which in turn limits the maximum collection rate; moreover, distortions to the beginning of the ring-down data are likely due to limited laser driver modulation bandwidth. These limit performance relative to AOM- or SOA-based instruments, but performance can still readily surpass that of non-cavity approaches.

Finally, maintaining adaptability across various wavelength regions and mission designs necessitates a flexible approach to instrument control and data acquisition. A parallel effort by the TLS team at the Jet Propulsion Laboratory (JPL) to produce a miniaturized, field programmable gate array (FPGA)-based control and data acquisition electronics suite provides an opportunity to achieve this while following our goal of minimizing the use of new custom hardware designs. This miniaturized electronics suite—built around the Xilinx Zynq system-on-a-chip—fulfills all the needs of the ring-down system described while easily fitting within CubeSat architecture. Because electronics development was still underway when the ring-down project began, a commercial PC/FPGA and high-speed transceiver system was selected that matches the capabilities of the Zynq-based electronics suite. This allowed the development of custom software and FPGA configurations discussed below.

2. EXPERIMENTAL

A. Optical Design

The monolithic aluminum optical cell, based in part on earlier work by Tang, Yang, and Lehmann [20], is shown in Fig. 1. The ring-down mirrors are separated by 12 cm, of which the central 8 cm are a tube with an outer diameter of 2 cm and a wall thickness of 2 mm. This relatively thin aluminum wall is easily stretched by two piezoelectric transducers (Thorlabs PAS005) mounted parallel to the optical axis and held in place by fine-thread adjustment screws. Over the full actuation range of 0–75 V, the cell can be stretched in excess of 2.45 μm; normal operation of the instrument requires a voltage range of approximately 22 V. The overall dimensions of the prototype cell are $19\times 5.4\times 3.6\mathrm{cm}$.

 

Fig. 1. Images of the monolithic cell for the prototype CubeSat ring-down spectrometer. The top image (a) is a cutaway CAD drawing of the cell assembly. The most notable features are the hard aluminum seats for the ring-down mirrors, which include narrow gas flow channels, 0.1 mm clearance around the mirror diameter, and parallelism tolerances of 0.1°; and the monolithic aluminum cell body consisting of an aluminum tube with an inner diameter of 1.8 cm and an outer diameter of 2 cm for the central portion. Optics are held in place with retaining rings; the face seals of the vacuum windows use Viton O-rings. The aluminum cell body is stretched by two piezoelectric actuators as needed during the operation of the spectrometer. The lower image (b) is a photo of the assembled system.

Download Full Size | PDF

Because the cell is mounted at the center of the tube, the mirrors experience symmetrical displacement relative to this central point. This enables the use of the adjacent-resonance etalon cancellation technique described in Ref. [21]. Although a similar approach of asymmetrically modulating etalon lengths on one side of the cavity via a piezo-mounted folding mirror [22] is also effective and could be implemented on both sides of the cavity if desired, the symmetric approach used here offers some advantages. First, the symmetric approach requires no additional hardware if the cavity length is already being modulated for spectral resolution reasons—this is significant in the case of compact flight instruments. The more general advantage is that the symmetric approach is self-calibrating. The cavity resonance condition ensures that ring-downs are triggered only when the cavity mirrors are separated by increments of $\lambda /2$, which in turn means that each mirror is displaced by $\lambda /4$ from all other optical surfaces (aside from the vacuum windows in the current design—various means of dealing with the vacuum windows are discussed in Ref. [21]). This abrogates the need for careful calibration or closed-loop operation needed when using an extra mirror for etalon cancellation and eliminates a potential mode of instrument failure. The primary disadvantage of the symmetric approach is that it constrains the design of the optical cell and mounting hardware; it works well in this case of a small monolithic cell, but may be difficult to implement in larger cavities.

The front mirror is flat with a specified 99.97% reflectivity (Newport 10CM00SR.60F), and the back mirror is concave with a 1 m radius of curvature and 99.99% specified reflectivity (CRD Optics 901-0010-1400); both mirrors are 2.54 cm in diameter. The measured average reflectivity of the cavity mirrors is 99.981%, in good agreement with specifications, yielding a time constant of approximately 2.145 μs. In the original design, both mirrors were mounted on bare aluminum seats with 2.55 cm diameter and a surface parallelism tolerance of 0.1°. It should be noted that these machining tolerances were insufficient when tested with two concave mirrors; using one flat mirror negates the need for a tight concentricity/runout tolerance. Small channels machined at the edges of the mirror seats allow gas flow, preventing any pressure differential across the mirrors. The mirrors were held in place with retaining rings. Later, a 5 mm diameter aperture plate (Newport 70276) was placed between the front mirror and its seat to demonstrate that the instrument could be redesigned with a much smaller mirror diameter. All data presented in this paper were collected with this aperture in place; aside from a very slight increase in mirror separation, no changes in cavity properties were observed.

Additional optics are kept to a minimum. Two antireflection (AR)-coated N-BK7 plates act as vacuum windows for the cell. Laser light from a 10 mW fiber-coupled diode laser centered at 1.392 μm (Eblana Photonics EP1392-5-DM-B01-FA-C) is passed through an optical circulator (Thorlabs CIR-1310-50-APC) to provide back-reflection isolation (in addition to the laser’s integrated $-40\mathrm{dB}$ isolator) and coupled to an adjustable fiber collimator (Thorlabs CFC-2X-C) through an AR-coated FC/PC-terminated fiber. This fiber collimator is mounted in a five-axis mount, allowing simple control of the cavity coupling alignment. No additional mode matching optics are used. The cavity output is focused by an off-axis parabolic mirror onto an avalanche photodiode detector (Thorlabs APD110C). The light reflected back from the cavity and isolated by the optical circulator is used during alignment procedures, but not during data collection. Frequency calibration is presently reliant on absorption line positions of the sample gas; in a complete flight instrument, a small reference gas cell similar to those used in the MSL spectrometer would be included.

B. Instrument Control

Instrument control and data acquisition tasks are handled via a PC with integrated FPGA and high-speed 16-bit analog transceiver components (National Instruments PXIe-7972R and NI 5783). A block diagram of the instrument control system is shown in Fig. 2.

 

Fig. 2. Block diagram of the prototype CubeSat ring-down spectrometer. Light from a fiber-coupled diode laser is collimated via fiber collimator (FC) and passes through a vacuum window (W) before injection into a two-mirror (M) ring-down cavity. An optical circulator (OC) isolates the laser from reflections. Light exiting the cavity is focused by an off-axis parabola (OAP) onto an avalanche photodiode detector (APD) and analyzed by a combined CPU/FPGA system, which also controls the laser and piezos. Figure reproduced from [21].

Download Full Size | PDF

Both the laser injection current and the piezo voltages are controlled by independent direct digital synthesis waveform generators implemented in the FPGA, which pass analog signals via the transceiver to the laser (ILX Lightwave LDC-3724C) and piezo (Thorlabs MDT693B) controllers. Because the transceiver outputs have a range of $\pm 1\mathrm{V}$, a voltage preamplifier (Stanford Research Systems SR560) is used to enable full-scale operation of the piezos. Although a variety of modulation approaches were tested, the best results were obtained when the laser current was held stable (aside from ring-down triggering) and the piezos were positioned according to the adjacent resonance etalon cancellation technique described in Ref. [21]. In this technique, piezos stretch the cavity to be on resonance with the laser, along with a very small $(<0.1\lambda )$ sinusoidal dither. Once the desired number of ring-down events is collected (50 points in this work), the cavity is stretched by an additional $0.5\lambda$, and a second set of ring-downs is collected (also 50 points). Because both mirrors move by $0.25\lambda$ relative to most other optical surfaces in the system, the majority of fringing effects are 180° out of phase and cancel when the two data sets are averaged. The piezos then return to their previous positions, with a small offset scaled to the wavelength step size, and the process is repeated for the next scan point.

Ring-down events are detected, triggered, and recorded by the FPGA system. The output of the photodetector is coupled to the analog input of the transceiver. When this signal crosses a user-specified threshold, the laser injection current is dropped below lasing threshold via the external modulation input of the laser controller. After a user-specified period of time has elapsed, data recording begins. The recorded ring-down transient is passed to software for least-squares fitting. The laser is then brought back to its previous injection current, with a slight delay for stabilization before another ring-down event can be triggered. The lowest noise-equivalent absorption was obtained with a trigger threshold of 1 V, a recording delay time of 400 ns, and a stabilization delay of 2 ms. A single datum point, composed of 100 ring-down events, could be recorded in roughly 0.25 s.

The dithered-cavity approach yielded a ring-down event rate of approximately 400 Hz, limited primarily by the laser stabilization delay. Higher repetition rates, in excess of 3,000 Hz, were achieved by dithering the laser rather than the cavity. However, this approach required that the laser be shifted slightly off resonance to trigger ring-down events (rather than being dropped completely below lasing threshold) in order to decrease re-stabilization time and take advantage of the higher ring-down rate. This proved to degrade the quality of recorded ring-downs, and as such the optimized dithered-cavity approach had a noise-equivalent absorption coefficient three times lower despite the different ring-down collection rates.

Data recording, instrument monitoring, and user interface are handled by a software layer that interfaces with the FPGA hardware. The offsets, modulation amplitudes, and modulation frequencies of the piezo voltage and laser current waveform generators are controlled by the software during scanning. The software requests and receives ring-down events from the FPGA, then performs least-squares fitting on the decays received. To enable high-speed data processing, three parallel fitting channels are included—this approach is more than sufficient for the highest ring-down collection rates observed ($>3,000\mathrm{Hz}$) and could easily be expanded with more parallel channels if needed. Outliers can be rejected on the basis of poor fitting residuals or non-physical values for time constant or amplitude. Because all instrument inputs and outputs pass through the PC/FPGA system, a wide array of metadata can be recorded for each ring-down point if desired—trigger time, piezo and laser modulation phases, fitting residuals, amplitude, etc. These are helpful as diagnostics, but can be omitted if data storage or transmission capabilities are limited.

C. Gas Handling

As this instrument was intended primarily as a proof of concept to demonstrate the signal-to-noise performance of a miniature CRDS instrument, control over sample gas temperature and isotopic fractionation is limited. To perform a water vapor isotope analysis scan, the cell is first evacuated via a turbomolecular drag pump to $<0.1\mathrm{mbar}$. After closing off the pump line, a needle valve connected to a vial of laboratory air and liquid water is opened until the desired cell pressure is reached. A 0.5 μm particulate filter is used to prevent dust or water droplet contamination of the mirrors during evacuation or sample intake. Although accurate knowledge of sample temperature and prevention of isotopic fractionation during ingest or measurement are clearly critical to the function of a complete isotope analyzer, they are outside the scope of this work.

D. Performance Evaluation

To evaluate the performance of the instrument, both long-term empty-cell measurements and measurements of water vapor isotope ratios at realistic integration times for an atmospheric descent probe were performed. No environmental or vibrational controls were applied.

Empty-cell measurements were taken after evacuating the cell via turbomolecular drag pump and continuing to pump down the cell for 24 h prior to the measurement; pressure was off scale at $<0.1\mathrm{mbar}$. While continuing to apply vacuum, the laser was moved away from any water absorption features and ring-downs were collected and processed for a further 24 h.

Water isotope ratio measurements were taken after evacuating the cell and re-filling to 1 mbar cell pressure with humid laboratory air as described above. This yielded a measured water vapor partial pressure of 45 μbar. Measurements were taken over the range of $7183.2–7185.2{\mathrm{cm}}^{-1}$, with a total scan time of 320 s. The region of interest, $7183.5–7184.0{\mathrm{cm}}^{-1}$, was observed in under 50 s. From these data, the absorption peaks corresponding to ${\mathrm{H}}_2^{16}\mathrm{O}$, ${\mathrm{H}}_2^{17}\mathrm{O}$, and ${\mathrm{H}}_2^{18}\mathrm{O}$ were fit to Gaussian profiles. The amplitudes of these Gaussian profiles were compared to HITRAN spectra [23] simulated at 296 K to determine isotope ratios. The standard deviations of the Gaussian amplitudes were used to approximate the precision limit for isotope ratio determination imposed by spectral/fitting noise.

3. RESULTS AND DISCUSSION

A. Empty-Cell Performance

Time series and Allan deviation [24] data for long-term stability scans are shown in Fig. 3. At an integration time of 1 s, the noise-equivalent absorption coefficient is $3.72\times {10}^{-9}{\mathrm{cm}}^{-1}{\mathrm{Hz}}^{-1/2}$; at the minimum of the Allan deviation plot, the value is $1.08\times {10}^{-9}{\mathrm{cm}}^{-1}$. Given an effective cavity length of 645 m, the noise equivalent absorbance values are $2.40\times {10}^{-4}{\mathrm{Hz}}^{-1/2}$ and $6.96\times {10}^{-5}$, respectively. Although the etalon cancellation technique employed significantly reduces long-term drifts coupled into the system by etalon effects (laser frequency drift, thermal expansion along the optical axis, low frequency motions, etc.), optical feedback from the vacuum windows allows these drifts to dominate the noise spectrum beyond 20 s. This can easily be addressed in future designs, as discussed in Ref. [21].

 

Fig. 3. 24-h stability data. The CubeSat ring-down spectrometer was allowed to collect data on an evacuated cell at a wavelength removed from any water absorption lines for just over 24 h to characterize long-term noise performance and stability of operation; the instrument operated without user intervention for the entirety of this period. The upper graph (a) shows the Allan deviation of the measured absorption coefficient, while the lower graph (b) shows a time series of the recorded time constants. Over long integration periods, environmental fluctuations cause the Allan deviation to increase; this can also be seen in the slower time constant drift outside of normal working hours (i.e., hours 4–20 in the time series data).

Download Full Size | PDF

B. Water Vapor Isotope Ratio Measurements

Data from water vapor isotope ratio measurements are presented in Fig. 4 and Table 1, with the absorption coefficients and standard deviations presented in Table 1 corresponding to the amplitudes of the Gaussian fits. The full scan range covers $7183.1–7185.0{\mathrm{cm}}^{-1}$ and includes a number of absorption lines representing ${\mathrm{H}}_2^{16}\mathrm{O}$, ${\mathrm{H}}_2^{17}\mathrm{O}$, and ${\mathrm{H}}_2^{18}\mathrm{O}$. The oxygen isotope peaks between 7183.6 and $7183.8{\mathrm{cm}}^{-1}$ were selected for isotope ratio measurements. These measurements serve only as a means of evaluating the limitations imposed by spectral noise—without taking steps to limit fractionation and actively measuring the gas temperature, repeatable isotope measurements are not possible. By fitting each peak to a Gaussian profile and comparing the amplitudes and their uncertainties to those in the HITRAN database [23], a 50 s scan of this region provides ${\delta }^{18}\mathrm{O}$ and ${\delta }^{17}\mathrm{O}$ ratios with 1 $\sigma$ precisions limited by spectral noise to no less than $\pm 0.02‰$ and $\pm 0.1$‰, respectively. The final achievable precision would be subject to repeatability of temperature determination and sample fractionation, but these are outside the scope of this work. Although ${\mathrm{HD}}^{16}\mathrm{O}$ is detectable in the wider-range scan, the signal-to-noise ratio is poor—use of a different spectral window would be advisable if H/D ratio measurements are desired. Likewise, if the goal were to measure very low concentrations of ${\mathrm{H}}_2^{16}\mathrm{O}$, there are much stronger absorption lines available that are outside the dynamic range of this instrument at the water vapor partial pressure investigated.

Table 1. Water Spectrum Data

View Table | View all tables in this article

 

Fig. 4. Water vapor isotope ratio scan. The data shown are from a typical water vapor scan using the CubeSat ring-down spectrometer in the $7183.1–7185.0{\mathrm{cm}}^{-1}$ range. The numbered absorption peaks are identified in Table 1. Peaks 3, 4, and 5 provide the most convenient measurements for oxygen isotopes in this region. Spectral noise limits isotope ratio measurements to precisions on the order of $\pm 0.02\%$ and $\pm 0.1\%$ for ${\delta }^{18}\mathrm{O}$ and ${\delta }^{17}\mathrm{O}$ ratios, respectively, with a scan time of approximately 50 s. Total cell pressure was 1 mbar for this measurement, with a water vapor partial pressure of 45 μbar.

Download Full Size | PDF

C. Evaluation of Design Goals

A performance comparison for this instrument and other instruments intended for spaceflight or laboratory use is shown in Table 2. The goal of improving on the performance of the MSL TLS while reducing size has clearly been achieved; the noise-equivalent absorption coefficient and sampled volume are both decreased by factors of nearly 200. Although a parallel effort at JPL to miniaturize the Herriott cell design of the MSL instrument has produced a spectrometer compatible with CubeSat size constraints, this CRDS approach yielded more substantial improvements in both performance and volume.

Table 2. Comparison with Other Spaceflight and Laboratory Instruments

View Table | View all tables in this article

In comparison to other early-stage planetary science instrument concepts—a wavelength-modulated (WM) integrated cavity output spectroscopy (ICOS) instrument developed at JPL [25] and a near-IR CRDS instrument developed at Princeton in collaboration with Goddard Space Flight Center and the Lehmann group [26]—the CubeSat CRDS instrument lags in terms of noise-equivalent absorption coefficient, but is the only instrument small enough to meet the stated design criteria. The Princeton instrument could likely be adapted to occupy a similar volume, but uses components available only in narrow wavelength ranges—this allows detection of methane at 1.65 μm, but these lines are 140 times weaker than those of the 3.27 μm band [26]. Adapting the CubeSat CRDS design for 3.27 μm is achievable and should allow roughly equivalent performance for methane detection. As such, the Princeton instrument is effective for methane measurements but is not as easily adapted to targets in other wavelength regions. The JPL WM-ICOS instrument also performs well, but has not been demonstrated with a cavity design appropriate for a CubeSat mission; the performance of ICOS instruments is highly dependent upon cavity mode spacing, which in turn is a function of cavity length and mirror diameter. As expected, instruments intended for laboratory use or terrestrial field applications typically outperform specialized instruments meant for planetary science, as they do not face the same design constraints; more points of comparison are available in Ref. [29].

The CubeSat CRDS instrument is sufficiently small for most potential CubeSat missions, although some minor modifications would likely be necessary for the case of a spherical Venus probe. Venus presents the greatest challenge for a CubeSat probe due to the very high temperatures and pressures as one approaches the surface—these necessitate a sturdy pressure vessel and thick insulation to ensure the safety of the instrument during its operating window. It should be noted that while the gas giants obviously reach higher atmospheric pressures than Venus, data transmission becomes impossible due to atmospheric interference long before such pressures are reached, and as such, they are not a limiting factor as in the case of Venus. For the example cases cited in the design criteria above, we assume the pressure vessel occupies either a 20 cm diameter sphere or a 30 cm long, hemisphere-capped cylinder of 10 cm diameter; in each case, we also assume a 2.5 cm thick multilayer thermal insulation blanket similar to that used for the Pioneer Venus probe [30], which limits the space available for the instrument payload. In the case of a spherical pressure vessel, it would be necessary to reduce the spacing between the vacuum windows and the cavity mirrors and use retroreflectors to couple the cavity to the laser and detector; these changes should not significantly impact performance. The diameter of the optical cell should, of course, be reduced in future designs now that it has been confirmed that reducing mirror diameter does not impact performance.

As intended, the basic design of the CubeSat CRDS instrument is compatible with commercially available equipment in the target range of 1–5 μm. Mirrors of equal quality are available throughout this range, with some regions boasting significantly higher reflectivity that should be taken advantage of when possible; likewise, high-quality crystalline mirrors should be utilized as they continue to develop in the mid-IR region [31]. Both mirrors and back-reflection isolation components do need to be designed for specific wavelengths; while truly off-the-shelf components would be preferable, these are still readily available on short timeframes. Adequate lasers and detectors are available as well; although mid-IR components do not yet match their near-IR counterparts (particularly when cryogenic cooling is unavailable), interband cascade lasers and HgCdTe detectors do not dramatically decrease performance in comparison to the near-IR (see Appendix A and Fig. 5). As with higher-performance mirrors, it is reasonable to take advantage of SOAs for triggering when available.

 

Fig. 5. Allan deviation comparison for the CubeSat ring-down instrument operating at near- and mid-IR wavelengths. The upper red trace shows the performance of the spectrometer following its retrofit for operation at 3.27 μm; the lower black trace shows performance at the intended operating wavelength of 1.39 μm. At an integration time of 1 s, the mid-IR performance lags behind the near-IR by a factor of 4. Note that the adjacent-resonance etalon cancellation technique could not be used following the mid-IR retrofit due to back-reflection isolation issues; this increased noise levels but reduced the minimum time to acquire a datum point relative to the near-IR implementation.

Download Full Size | PDF

4. CONCLUSION

This prototype instrument successfully demonstrates that cavity-based approaches can be used to improve upon the performance of the MSL-era TLS with a compact, simple design that is compatible with CubeSat-based probe designs. When operated at 1.39 μm, the noise-equivalent absorption coefficient of $3.7\times {10}^{-9}{\mathrm{cm}}^{-1}{\mathrm{Hz}}^{-1/2}$ improves upon TLS by a factor of 192. This performance could be readily increased through the use of a longer cavity or alternative triggering mechanisms for some applications, though these options will not be compatible with all mission profiles. The probed gas volume is also reduced by a factor of 172, substantially reducing sample collection and handling requirements. Although there are many applications for water vapor isotope analysis in planetary science, the design of the spectrometer should be adaptable to many targets throughout the near- and mid-IR with comparable performance (see Appendix A).