The Design and Engineering of Curiosity

by Emily Lakdawalla

  Figure 9.4. The ChemCam calibration target as seen by the right Mastcam on sol 838. Many laser shot points are visible. Image 0838MR0036830000500777E01. Credit: NASA/JPL-Caltech/MSSS.

  9.2.2 Using ChemCam

  9.2.2.1 Sun-safety

  The mast points some of the instruments (Mastcams and Navcams) at the Sun regularly for navigation and science, but the ChemCam instrument is sensitive to the Sun, so there are many restrictions on mast motion in order to protect ChemCam. Depending on the focus position of ChemCam, it is either “sun-safe” (in which case it can tolerate the Sun passing through its field of view) or “sun-unsafe” (in which case the Sun passing through the field of view could seriously damage the instrument). During tactical planning, engineers check mast pointing to make sure that ChemCam will always be sun-safe before approving a sequence.4

  Even when ChemCam is sun-safe, the Sun shining into its window can heat the instrument, so engineers have to plan sequences to make sure that the Sun will not shine directly into the ChemCam window for longer than 3 minutes. They do this by defining a cone spanning plus or minus 16° around the ChemCam boresight, and modeling how long the Sun stays within it. Obviously, ChemCam observations will always be sun-safe if the instrument is pointing below the horizon. Adding 4° to the 16° cone to account for the curvature of Mars’ horizon yields 20° below the horizon as an always sun-safe pointing direction for the mast that requires no further checking. Also, slews of the mast to move to a new pointing position happen quickly enough that they don’t need to be checked for ChemCam warming as long as ChemCam is sun-safe.

  When ChemCam is being used and so is in a sun-unsafe focal range, the Sun must never be permitted to shine into its window. This has to hold true even if a rover anomaly happens in the middle of a ChemCam observation, an anomaly that may take several sols to resolve. The rover planners define a “keepout cone” 17° away from the ChemCam boresight. Sweeping this cone across the sky along the path of the Sun during a given sol produces a “keepout band”, a region in which ChemCam must never be allowed to perform an observation.

  With onboard fault protection software, the rover checks to ensure that mast pointings will be sun-safe before performing them. The rover even checks sun-safety once per second as it drives, and will stop a drive if sun-safety will be violated. (This has never actually happened.)

  Targets located 2 meters away present a different kind of hazard to ChemCam. The problem is that the “antireflection” coating on the front window is good at allowing sunlight and LIBS-produced plasma illumination to pass through it, but is not quite as antireflective at the 1067-nanometer wavelength of the laser. Some laser light bounces back from the front window on every laser pulse. For two distance ranges, 1.20 to 1.36 meters and 1.942 to 2.217 meters, the reflected laser light can be focused by the primary mirror onto the secondary mirror, possibly causing some damage to the secondary mirror at the spot of the laser hit. The ChemCam team has stated that the instrument can perform about 100,000 shots in these ranges without risking serious damage. The shorter of the two distance ranges is never needed, because it is even closer to ChemCam than the calibration target; ChemCam is likely never to be so close to a rock, because it would have to be next to a nearly vertical cliff. (The artwork in Figure 9.1 notwithstanding, it is an unlikely situation for Curiosity to be so close to such a steep cliff.) The longer of the two distances covers ranges extremely close to the rover, and only a few hundred such shots have been made.

  9.2.2.2 Types of Observations

  The ChemCam team divides their operational history into three seasons. Season 1 covers sols 0 to 800. Season 2, when ChemCam had no autofocus capability, lasted from sols 801 to 980. Season 3, in which ChemCam uses its RMI for autofocus, is from 981 to the present.5 See section 9.2.1.1 above for a description of the different autofocus modes.

  LIBS observations. To perform a LIBS observation, the science team looks at Navcam and Mastcam images that were taken on a previous sol and selects a target. Navcam stereo images provide geometric information, while higher-resolution Mastcam images can be used to fine-tune target selection. The farthest target that ChemCam has attempted LIBS on was Mell, on sol 530, at a distance of 7.45 meters, but 90% of targets are much closer, within 4.5 meters; the team restricts quantitative analyses to targets within 5 meters.6 Before ChemCam can begin an observation, the autofocus and LIBS lasers may need preheating. The telescope autofocuses on the target and collects a “before” RMI image. The instrument collects a passive spectrum to be used later, for subtraction from the LIBS spectrum. Then the LIBS laser fires many laser pulses, usually 30 but occasionally some other number. It can collect a maximum of 150 spectra at 3 per second (referred to as a “burst”) before having to pause to transfer data. After the LIBS operation is complete, RMI takes an “after” image, then moves to the next target or returns the focus to the sun-safe position. The experimental result is a spectrum whose peaks imply the presence of different elements.

  Depth profiling. Most of the time, the ChemCam team averages the data from many LIBS shots at each point to improve the signal-to-noise ratio of LIBS data. However, by analyzing shots individually, it is possible to perform depth profiles for elements. Each LIBS shot ablates approximately 1 micrometer of material, and ChemCam has occasionally found composition to change at that scale. For example, in the rock Bathurst Inlet, lithium, rubidium, sodium, and potassium concentrations decrease with depth, possibly “due to aqueous alteration processes (i.e. frost deposition, followed by melt and evaporation or sublimation) that have preferentially mobilized the alkalis.”7 Another rock had a thin layer of manganese on the surface, thin enough for ChemCam to penetrate through.8 ChemCam has performed depth profiles of up to 1000 shots.9

  Rasters. ChemCam commonly performs several observations in a “raster” or array. Rasters can be of any size, but 1-by-5, 1-by-10, and 3-by-3 arrays are the most common. It takes about 30 minutes to perform a 1-by-5 raster, 40 for a 3-by-3, and an hour for a 1-by-20. The most common spacing between points is 2 milliradians (giving approximately 1-millimeter point-to-point spacing at locations close to the rover) but the spacing may be wider or narrower.10 The pointing accuracy of the mast is such that ChemCam has been able to perform rasters within drill holes (Figure 9.5). RMI images are taken at the beginning and end of the raster observation. When necessary for large rasters, additional images are taken in the middle. During season 2 (sols 801 to 980), when ChemCam had no autofocus capability, there were few rasters (see section 9.2.1.1). Unfortunately, this coincided with virtually all of the time spent at the first major field site after arrival at Mount Sharp, Pahrump Hills.

  Figure 9.5. Example raster data within a drill hole, Telegraph Peak. Top left: MAHLI image 0911MH0004750000303057R00 of the drill hole under nighttime illumination. A tiny white vein is visible in the drill hole wall. Bottom left: ChemCam RMI image CRM_479251063_CCAM03921 of the drill hole; red square shows the region of a 4x4 raster targeting the vein and the area around it. Right: Zoom in on the vein and the raster of ChemCam measurements showing how hydrogen, calcium, and iron content vary among the different observations. Within the vein there is an increase in calcium and hydrogen and a decrease in iron, thought to indicate that the vein is composed of a hydrated calcium sulfate, likely bassanite. This set of observations was performed during ChemCam Season 2. Courtesy William Rapin.

  Passive mode. As a part of every observation, the spectrometers first gather spectral information using reflected sunlight without firing the LIBS laser. Such “passive” observations can also be gathered without taking any LIBS data, useful on distant targets. Passive spectra have helped the team identify iron oxidation states, diagnosing the mineralogical shift from less-oxidized magnetite to more-oxidized hematite in bedrock along the rover’s traverse after it reached the Bagnold dune field. The team also identified the presence of iron sulfates from their 430-nanometer absorption feature in passive spectra taken of rocks around Pahrump Hills. ChemCam passive sky o
bservations investigate the abundance of water vapor and oxygen in the atmosphere, useful for comparison to REMS and SAM measurements of local humidity and oxygen abundance.11

  Blind targeting. The rapid pace of drive campaigns means that there is often not time to receive the Navcam data needed for targeting ChemCam images before the rover drives away. Beginning on sol 318, to gather some ChemCam data during drives, ChemCam performed blind-targeted measurements of a patch of ground directly to the right of the rover at the end of drives, at a distance that would be 3 meters away if the ground were level. Initially, they performed only single-point analyses, but they added blind line scans with multiple shot points beginning on sol 386.12 Blind targeting was performed only during ChemCam season 1 (until sol 801), because the new autofocus algorithm requires a distance seed derived on Earth from analysis of Navcam images.13 Blind targeting was eventually replaced by AEGIS targeting.

  AEGIS targeting. AEGIS stands for Automated Exploration for Gathering Increased Science.14 It is a set of artificial-intelligence algorithms to enable a rover to autonomously select and/or refine observation targeting, first used on the Opportunity Mars Exploration Rover. For Curiosity, AEGIS permits ChemCam target selection without waiting for instructions from the ground. It runs on the rover’s main computer, acquiring Navcam images, identifying potential targets, filtering them based on criteria supplied by the ChemCam team, and then ranking the targets. The ChemCam team can adjust the target selection criteria each time an AEGIS sequence is planned, prioritizing outcrop, dark or light rocks, or other kinds of targets. On Mars, the whole process takes only 4 to 8 minutes and can be performed immediately after the end of a drive. AEGIS can also refine the pointing of ChemCam LIBS shots, running on RMI images to select bright veins or grains for targeting. Adding AEGIS capability to Curiosity began in summer 2015; uplink was spread out over many weeks. The code was installed into flight software on sol 1141. Checkouts were complete in February 2016, on sol 1237. AEGIS was used for routine operations for the first time on sol 1343.15 The Navcam mode has been used more often than the RMI mode.

  Long-distance imaging. The RMI is Curiosity’s highest-resolution camera, and is often pointed at very distant targets. For instance, the team used the RMI in a long-distance campaign to monitor sand ripple motion on the backs of the Bagnold dunes, to study the upper part of Peace Vallis where it enters the crater from the northwestern rim, and to get a look at possible future science locations on Mount Sharp (Figure 9.6). Long-distance imaging has dramatically better quality in season 3, after sol 981, now that the RMI can autofocus on distant targets.

  Figure 9.6. ChemCam RMI long-distance observation of Mount Sharp, sol 1283. Credit: NASA/JPL-Caltech/LANL/MSSS/James Sorenson.

  Dusting. The rapid expansion of air around the superheated plasma generated by ChemCam shots can very effectively remove very fine Martian dust from rock surfaces (Figure 9.7). A typical 30-shot observation blasts aside dust for a 6-to-9-millimeter diameter around the shot point.16 Although ChemCam hasn’t yet been used for this purpose deliberately its accidental use as a remote dust removal tool improves the value of Mastcam multispectral imaging on rocks, and the MAHLI team likes to image LIBS pits to see the color of dusted-off rocks.

  Figure 9.7. Two ChemCam raster targets at Windjana, “Stephen” (top) and “Neil” (bottom), were dusted off by ChemCam, revealing the dark color of the rock beneath the bright dust. MAHLI image 0627MH0001900010203555C00, NASA/JPL-Caltech/MSSS.

  9.2.2.3 Calibration

  Transforming ChemCam spectra into elemental abundances requires comparing the data to ChemCam measurements of samples of known composition. Initially, the ChemCam team’s calibration library contained 66 samples with compositions expected to be found on Mars that had been shot with the flight model of ChemCam under Earth ambient conditions. The original library performed well for common Mars materials like dust and basalt, but relatively poorly for more unusual materials like calcium sulfate veins and feldspars.17 Two laboratory copies of ChemCam (one each at Los Alamos and CNES) operate under Mars-like temperature and pressure conditions. The Los Alamos laboratory expanded the data set to 450 different types of rocks, covering a wider range of compositions than in the initial 66. The team began testing a new calibration model on ChemCam data during the mission’s second conjunction period, beginning on sol 1004, and delivered recalibrated versions of earlier major-element data to the Planetary Data System on December 3, 2015.

  9.2.3 Anomalies

  The main anomaly encountered by ChemCam on Mars was the loss of the autofocus laser on sol 801 (section 9.2.1.1), which resulted in no autofocus capability until sol 983, when the new RMI-based autofocus algorithm was put into use. One drawback is that relying on the RMI means that autofocus doesn’t work at night or even, sometimes, in areas of deep shadow. ChemCam rarely performed observations at night before the anomaly because of the high energy cost of heating the mast actuators at night, so that restriction has had little effect, but the issue of rover shadowing has occasionally affected target selection.18 The new method also scans a shorter range of possible focal distances than the old method, so it requires a good-quality distance seed, meaning that shooting in the blind is no longer possible. On the plus side, laser autofocus didn’t work at distances longer than 18 meters, so the new autofocus method performs much better at imaging at infinity.

  9.3 APXS: ALPHA PARTICLE X-RAY SPECTROMETER

  APXS measures the elemental composition of rocks and soils by emitting alpha particles and X-rays at a surface and counting the X-rays that return. Curiosity’s APXS is the third in a line of similar instruments carried to Mars on Pathfinder and the Mars Exploration Rovers, with improvements that make it more sensitive, faster, and able to operate over a wider range of the Martian day. 19 The APXS team uses measured elemental abundances to group observed targets into classes of similar composition, often comparing rocks across rover landing sites. The principal investigator is Ralf Gellert of the University of Guelph, Canada, and APXS was provided to the mission by the Canadian Space Agency.

  APXS is located on the turret and has to be deployed to close proximity or in contact with a rock or soil (Figure 9.8). Because it requires the arm and several hours for a high-quality measurement, it gets used less frequently than the remote sensing instruments. It sees heavy use at sample sites and significantly less frequent use at stops during traverses. APXS measurements also assist the team in selecting drill targets. By analyzing drill tailings, APXS can help the CheMin team constrain the composition of the component of the drilled rock that is amorphous and therefore not accessible to CheMin mineralogical analysis. APXS measurements of potassium in drill tailings combined with SAM measurements of argon gas evolved from a sample have been used to measure the exposure ages of outcrops through potassium-argon dating.20 Elements that APXS can detect are listed in Table 9.2.

  Figure 9.8. APXS deployed onto the John Klein drill target. Mosaic of four Mastcam images taken on sol 168. Credit: NASA/JPL-Caltech/MSSS.

  9.3.1 How APXS works

  APXS has three main elements: a sensor head located on the end of the robotic arm; an electronics unit located in the front left corner of the body; and a calibration target that is mounted below the MAHLI calibration target attached to the shoulder azimuth actuator. Just like MAHLI, the APXS sensor head is separated from the turret by a set of three springy wire rope assemblies to isolate the instrument from vibrations caused by drilling and CHIMRA sample processing. Parts of APXS are shown in Figure 9.9.

  Figure 9.9. Parts of APXS. Top: flight hardware (NASA/University of Guelph). Middle: photo taken during initial turret checkout (MAHLI image 0032ML0000620000100855E01, NASA/JPL-Caltech/MSSS). Lower left: photo showing location of calibration target (NASA/JPL-Caltech release PIA14255). Lower right: MAHLI photo of the APXS calibration target taken during initial checkout on sol 34 (0034MH0000480010100038E01, NASA/JPL-Caltech/MSSS).

  The sensor head contains six curium-244 sources. Three of them are
covered in a titanium foil and emit both alpha particles and X-rays, while the other three are more thoroughly sealed and emit only X-rays. A cutaway drawing of the internal workings of the sensor head is shown in Figure 9.10. When the alpha particles impinge on atoms in the upper tens of micrometers of the target, they cause the atoms to eject inner-shell electrons, which emit X-ray photons as they fall back to their ground state, a process called particle-induced X-ray emission (PIXE). Impinging X-rays can have the same effect, in X-ray-induced fluorescence (XRF). Particle-induced X-ray emission is a more efficient process for small-mass atoms (sodium to titanium), while X-ray-induced fluorescence is more effective for larger atoms (chromium to strontium). APXS uses a silicon drift detector to detect and count these emitted X-rays. The half-life of curium-244 is 18.1 years, more than the anticipated lifetime of Curiosity’s power supply, so degradation of the curium sources should not have a significant impact on APXS use over the course of the mission.

  Figure 9.10. Cross-section of the APXS sensor head. University of Guelph.

  APXS “sees” deeper into a target for heavy elements (to depths of 50 microns or greater) than for light elements (for which APXS may only be measuring the topmost 5 microns). The energy of X-ray emission depends on the element, so by measuring the amount of X-ray emission with respect to wavelength, APXS can detect the elements that are present (see Figure 9.11).21 Employing an analysis derived from observations of samples of known composition on Earth, the APXS team can convert these X-ray spectra into elemental abundances. Using the dust removal tool to brush off a measurement site improves APXS’s ability to sense the abundances of lighter elements in the rock rather than the dust. For a list of all spots brushed for APXS analysis, see Table 5.​2. The brush can still leave as much as 30% of the original dust covering behind.22 The most dust-free targets that APXS examines are dumped drill fines.