3.1

Optical design description

Compared to ChemCam on Curiosity, SuperCam has to accommodate a larger spectral band due to the addition of the passive IR spectroscopy investigation. Most of the MU optics have their performance specified in the [245 nm; 2600 nm] spectral range, and several dichroic plates ensure the distribution oflight to the different subassemblies.

LIBS laser emission. The 1064 nm laser beam is expanded to 1 cm diameter by a compact Galilean telescope and injected into the main Schmidt-Cassegrain telescope.

Raman laser emission. The 532 nm beam follows a different optical path. The beam is also expanded to 1 cm diameter by a compact Galilean telescope, but it remains collimated and is со-aligned on the telescope axis via a 2-mirror periscopic system.

Scattered light from LIBS and Raman laser shots is captured to monitor the laser energy output with a visible photodiode (laser housekeeping photodiode). This photodiode is located at the entrance of the Red Galilean expander.

Figure 7:

Mast Unit spectral range (left) and lasers emission layout (right).

Telescope. It focuses the LIBS laser beam onto the target. When the beam exits the telescope, it has the size of the primary mirror, 110 mm diameter. The same Schmidt-Cassegrain telescope captures light from the target. Photons traverse the laser dichroic (transparent to the 245 - 2600 nm range, except for the 900-1200 and 340 - 380 nm windows).

Objective and beamsplitter. Part of the visible light from a 20 mrad cone around the analysis spot is focused on the RMI camera via a 3-lenses objective. The beamsplitter focuses the light below 900 nm into a 300 μm-core fiber for LIBS/Raman/VIS reflectance analyses. The last dichroic plate (#15) acts as a notch filter working in reflection: the light is reflected in direction of the MU-BU optical fiber, except for a narrow band at 532 nm to prevent exciting the silica molecules of the optical fiber that would add noise on our Raman measurements.

Longer wavelengths are diverted to the IR spectrometer.

Focus. The telescope focuses on targets from 1 m to infinity by moving the secondary mirror with respect to the primary mirror using a translation stage mechanism. Autofocus can be achieved using two different methods: (1) Measuring the scattered light from the continuous wavelength laser (CWL) at 852 nm on target with a visible photodiode, (2) Finding the best contrast on the RMI images. Autofocus with CWL is used at distances lower than 7 m typically, and autofocus with RMI can be used at the same distances, and up to infinity. The focus requirement for LIBS and RMI is ±0.4% of the distance, but is much looser for Raman and VISIR (±25%), allowing rapid scanning across uneven terrains.

Figure 8:

Mast Unit collection path. Telescope and dioptric objective (left). Beamsplitter (right).

The MU common optical path has 31 optical elements, all custom made, except for two of them (the imager filter and the CWL aspherical collimating lens). The optics specifications result from a trade-off between performance and robustness. For example, the MU temperature will cycle every sol from ~-37°C to ~-10°C, causing stress in the optical coatings if they are too thick. Caution was taken to limit the number of optical layers (<100) for the most complex dichroic plates, and all the coated optics were tested under thermal cycling (1500 times -40°C to +35°C, 5°C/min).

The optics located at the laser output see high fluence (0.7 J/cm² at 1064 nm and 0.4 J/cm² at 532 nm), and their coating was tested under these laser fluxes. Dedicated test campaigns were also run to assess the risk of laser induced contamination, and specify the maximum acceptable level for molecular contamination.

Apart from the optics procurement, the two main challenges for the MU optical design were:

3.2

LIBS-Raman laser

The laser⁷ is the heart of the LIBS and Raman experiments. The laser cavity is based on a Nd:YAG crystal to provide the red line at 1064 nm. The Nd:YAG crystal is pumped by a multi-color laser diodes stack emitting in the 795 - 805 nm spectral band to cope with the narrow spectral acceptance of the Nd:YAG crystal, and its variations with temperature. At the cavity output, a second harmonic generator (KTP crystal) multiplies the beam frequency by two to provide the line at 532 nm. The phase matching is achieved using a Pockels cell to rotate the beam polarization. The laser needs a pumping current up to 155 A and high voltage of 1200 V to reach its full performance. The LIBS energy is higher than 24 mJ, the Raman energy is higher than 12 mJ, with a pulse duration of 4 ns. The laser can be used with repetition rate in the 1-10 Hz range. The near field beam has a rectangular shape of dimensions 2.1 x 2.3 mm² at 1/e² at room temperature. The M2 factor is 1.3, and allows a good focus of the LIBS laser beam on target. The laser cavity is sealed (dry air, 1 bar pressure inside), and interfaced on the OBOX through a Titanium bracket.

Figure 9:

LIBS-Raman laser. Credit: Thales-LAS.

3.3

Remote Micro-Imager (RMI)

Figure 10:

Remote Micro-Imager. The 3D+ cube (left) and the cube with the NIR cutoff assembly (right). Credit: SuperCam.

The RMI⁸ is based on a CMV4000 CMOS sensor (2048 x 2048 pixels of 5.5 μm) with a Bayer filter, embedded in a cube containing readout and control electronics. A NIR cut-off filter is placed in front of the cube to retrieve the colors. The useful field of view is 18.8 mrad in diameter, with a spatial resolution better 80 μrad over 10 mrad field of view in diameter, and a scale of 10 μrad per pixel. The RMI has an auto-exposure capability, and a High Dynamic mode to improve contrast and signal-to-noise ratio in the darkest zones of the image.

3.4

InfraRed Spectrometer (IRS)

The IR spectrometer^9,10 is an independent device, linked to the telescope objective by a periscopic system. The entrance of the IRbox is a pinhole (400 μm in diameter, 0.18 numerical aperture). The spectral selection is done by an Acousto-Optical Tunable Filter (AOTF), manufactured by Gooch & Housego, which allows a very compact and robust spectrometer with no moving parts. A radio-frequency signal drives a transducer attached to the side of the AOTF. For each frequency of the piezo, a single wave number is selected and scattered by ±6.5° (2 polarizations). The main beam is rejected, while both polarizations are registered by two HgCdTe (MCT) photodiodes (cold redundancy), from Teledyne Judson Technologies. These MCT photodiodes are packaged with a triple-stage TEC. The detector TEC can lower the photodiode temperature to 70°C below that of the spectrometer. Spectra up to 256 spectels can be registered in the 1.3 – 2.6 μm spectral band, with a spectral resolution better than 32 cm^-1 and signal-to-noise ratio of 56 or better.

Figure 11:

InfraRed Spectrometer (IRS) CAD view. Credit: SuperCam.

Figure 12:

InfraRed Spectrometer (IRS) flight model picture. Credit: SuperCam.

In order to correct the data, the instrument function shall be carefully calibrated¹¹. This required a 10-day test campaign just before we delivered the MU to the Jet Propulsion Laboratory (JPL) in California.

4.1

Demux

The demux function is to split the incoming light collected by the MU in three spectral bands to pass to the three spectrometers:

The optical entrance is an AVIM connector. The beam is collimated and spectral bands are selected by a series of two dichroic plates. The VNIR path also includes a long-pass filter with a sharp transition at 534 nm in order to block the remaining 532 nm laser light from reaching the spectrometer optics. The beams are then focused on the demux outputs connected to fiber bundles to reorganize the beam circular shapes in linear shapes that fit the spectrometers entrance slit. The bundles comprise 19 fibers in a circular array at the input, and a linear array at the exit. The fibers, each with a 50 μm core, a map such that the center fiber in the circular array is at the center of the line, the six fibers surrounding the center are placed three on either side, and so forth.

Figure 14:

Optical demultiplexer (demux) on the left, and an example of fiber bundle on the right ((a) shows the demux output and (b) the spectrometer entrance). Credit: SuperCam

4.2

UV & VIO spectrometers

The UV and VIO spectrometers have a classic Czerny-Turner design, and benefit from a high heritage from ChemCam. They are equipped with 2400 lines per mm reflection gratings, and 42-10 CCDs from E2V (2048 x 515 pixels, 13.5 μm pitch).

Figure 15:

UV & VIO spectrometers. Credit: SuperCam.

4.3

Transmission spectrometer

The transmission spectrometer is a new development for SuperCam to be able to detect the very weak Raman signals. The key features are the high throughput, signal intensification and time-gating.

The entrance beam is collimated and split in two spectral bands before reaching a pair of volume-phase (VPH) gratings. One of the grating assemblies contains a combination of two gratings written in the same substrate with a 1.8° separation angle between them. The 3 spectral orders are then focused on the intensifier. After intensification, a set of relay lenses reimages the beam on the CCD (same reference than the reflection spectrometers).

The image intensifier is the MX-10130 from ITT Exelis, biased and gated using a newly-designed high voltage power supply (HVPS) located below the opto-mechanical part of the spectrometer. The gain of the intensifier can be adjusted by choosing the value of high voltage of the microchannel plate. The timing of the high voltage is controlled in order to time-gate the signal. Time-gating is synchronized with the CCD readout. These parameters shall be adapted to the technique as the LIBS produces a signal far brighter than Raman spectroscopy. Whereas LIBS and Raman signals are in the range of nanoseconds, fluorescence is optimal after microseconds.

Figure 16:

Visible and near-infrared transmission spectrometer (left). Three grating orders projected on the CCD (right). Credit: SuperCam.

5.1

Mast Unit

Telescope focus. Both autofocus modes (CWL and RMI) were extensively tested in thermal environment and using rocks of different structures and albedo. Robustness and repeatability were verified. Optical focus offsets between the different techniques were characterized; they will be automatically corrected on-board.

The CWL autofocus is optimized when sun illumination is weak or the target is in shadow. The scan range is large and thus tolerant to uncertainty of the seed distance. RMI autofocus takes more time to run, but is better suited to longer target distances. The scan is very small and requires a good knowledge of the seed distance, compatible with the rover capability.

Figure 17:

Autofocus curves for the same target at four distances. (Top) CWL autofocus curves. At 2 m, the measurement is repeated and ended up with a different gain, but the same focus to within 1 motor step. (Bottom) RMI autofocus curves. Credit: Maurice et al. (2021).

LIBS laser emission. The telescope equipped with the newly-designed primary mirror (without nickel layer) is almost diffraction-limited down to -40°C. Combined with more than 12 mJ deposited on the target, the irradiance is well above the minimum requirement of 1 GW/cm². At short distances, it is recommended to lower the laser pumping current in order to avoid the spectrometers saturation.

Figure 18:

LIBS point spread function at different distances and T = -25°C. The theoretical 1^st Airy ring is indicated, as well as the irradiance on target. The minimum requirement is 1 GW/cm².

Raman laser emission. The irradiance on target varies with temperature and distance mostly due to the laser beam divergence, but also to the energy provided by the laser. Most of the time the irradiance will be in the range of 15 – 30 kW/mm² (where the science team expects 30 kW/mm²). It can drop to 10 kW/mm² at the extreme case of 7-meter range and OBOX temperatures lower than -31°C. Firing the Raman laser with longer bursts (up to 200 shots) and co-adding spectra on the CCD sensor will help increasing the signal to noise ratio.

Substantial effort was invested to optimize the Raman laser beam alignment on the telescope line of sight. This includes late design changes, careful alignment procedure, and verification tests at each step of the integration and environmental tests. The last verification was made at the JPL Spacecraft Assembly Facility in December 2019, just before Perseverance left Pasadena for integration to the rocket stack.

Imaging. The RMI field of view (18.8 mrad) and flat field were accurately characterized for correction on-ground. The spatial resolution, driven by the telescope optical quality, is defined as the frequency at which the modulation transfer function is equal to 20%. It was measured as a function of temperature, distance and position in the field of view.

Figure 19:

Alignment of the Raman laser (in green) with respect to the telescope line of sight (black circle that represents the 0.7 mrad field of view) at different temperatures and distances, and during a simulated Mars cooling cycle.

Figure 20:

RMI resolution. (Left) Contrast always at the center of the field of view for different temperatures. (Right) Contrast at the center of the field of view and at 5 mrad off the center. The requirement is to be above the dotted line in red. Credit: Maurice et al. (2021).

The capability to record images from short distances (even lower than 1.56 m corresponding to the SuperCam calibration targets on the rover deck), and up to infinity was demonstrated.

Figure 21:

Series of RMI images acquired at-40°C during the MU thermal tests. The same rock is placed at different distances. Its size is ~7.3 cm (L) x ~7.2 cm (H). The dark crystal at the center of the first image measures ~8 mm (L) by ~7 mm (H). LIBS pits can be seen. The last is an image of a collimator reticule that mimics a target at infinity. Credit: SuperCam.

Infrared spectrometer. The IRS was fully characterized at subsystem level at LESIA facilities (Meudon, France). Spectral and radiometric performance were assessed in thermal environment for both photodiodes.

Figure 22:

The IRS spectral resolution (nominal photodiode) measured for three temperatures of the spectrometer. The requirement is to be below the red dotted line at 32 cm^-1. Credit: Maurice et al. (2021).

The measurements were used to feed the radiometric model and predict the signal to noise expected on Mars, depending on the thermal environment properties and hour of day.

Figure 23:

Signal-to-noise simulations for the IRS, as a function of local time (HH:MM), for a thermal environment close to Jezero crater and a start of the instrument at 13:00. This is a worst hot case. Four SNR simulations are shown, as the instrument is warming up, for 86 spectels acquired in 80 s (signal and dark). The requirement is to be above the dotted line at SNR = 56. Credit: Maurice et al. (2021).

Before leaving France, MU underwent a calibration campaign¹¹ for the IR channel, to measure the radiometric and linearity responses of the instrument over its nominal range of operations, in terms of instrument detector temperatures and spectral range. The derived instrument transfer function will be used to calculate the expected instrumental signal-to-noise ratio for typical observation scenarios of mineral mixtures expected to be found in Jezero crater, and ultimately to retrieve the spectral properties of the regions of interest observed by the rover.

5.2

Body Unit

Spectral characteristics. Spectral ranges, spectral resolution, and spectral registration (wavelength versus pixels) were characterized over BU expected temperature range on Mars. Spectral resolution requirement is easily met and remains acceptable over a large temperature range from-40°C to +35°C. Spectral registration also varies with temperature: a drift of ~1 pixel over 35°C was observed, whereas 0.2 pixels stability is required. Data could be corrected on-ground thanks to regular spectral calibration using LIBS technique on the SuperCam Titanium target.

Figure 24:

BU spectrometers spectral resolution over the 245-853 nm spectral range, and at different steps of the qualification tests. Credit: SuperCam.

Transmission spectrometer gain and timing. The intensifier gain is a function of the high voltage power for the microchannel plate. The characterization was done, so that we can command the gain depending on the investigation (LIBS or TRR/L). Also, the intensifier gate timing shall be adjusted to optimize the signal on the detector. This is done by acquiring a series of spectra while increasing the delay of the shutter. These on-grounds characterizations are mandatory to be able to adjust parameters value in-flight.

Figure 25:

Left: characterization of the intensifier gain versus the high voltage command. Right: optimization of the intensifier gate timing to get the maximum of signal. The timing depends on the target distance. Credit: Wiens et al. (2020).

5.3

End-to-end tests on Perseverance

All SuperCam functions and techniques were exercised, and performance checked at instrument level. The data were retrieved from several test campaigns at JPL facilities, either at room temperature in ATLO (Assembly, Tests and Launch Operations), or in the huge thermal chamber.

Figure 26:

Perseverance rover installed in a thermal chamber at JPL (left). Perseverance tests in ATLO at JPL (right). Credit: NASA/JPL-Caltech.

Instrument Response Function. As for the infrared spectrometer, the instrument response function (IRF) is needed for the other spectroscopic techniques to invert the data and get calibrated spectra. ATLO tests were the first and only opportunity to measure the SuperCam IRF for the BU spectrometers, and thus for the LIBS and TRR/L. The team set up absolute calibrated sources on the ATLO floor, and recorded a series of spectra with Perseverance mast pointing at these sources. Preliminary IRF curves were obtained.

Figure 27:

Preliminary instrument optical response functions for the transmission (left) and UV (right) spectrometers. The units are digital numbers (DN) per photon incident at the telescope aperture. Data were taken with the instrument mounted in the rover. For the reflection (UV, VIO) spectrometers, several read-out configurations (number of CCD rows) were tested to avoid saturation. For the transmission spectrometer (left), the intensifier gain depends on the techniques: 2500 for LIBS and 3200 for Raman spectroscopy. The response curves of the transmission spectrometer clearly show the three different (green, orange, red) diffraction orders. Credit: Wiens et al. (2020).

SuperCam techniques on Perseverance at cold. JPL proceeded to a two-week set of rover thermal tests under Martian pressure. The chamber was equipped with a solar simulator, and SuperCam team installed two sets of targets located at 2.6 m and 4.5 m from the rover mast. We were able to run the two autofocus modes, acquire RMI images, LIBS, TRR-L and VISIR spectra, and record sounds. The instrument behaved extremely well and end-to-end performance was demonstrated.

Figure 28:

RMI images taken during STT of a composite target (62% Ilmenite, 38% Hematite) at 2.6 m, using 19 ms integration time. (Left) Before LIBS, 30 shots. (Middle) After LIBS. (Right) Difference the two images to highlight the laser pit. This level of detail is expected of most RMI images taken in the rover’s arm work zone and slightly beyond. Credit: Wiens et al. (2020).

LIBS, TRR/L and VISIR spectra are shown in Wiens et al.⁵