The new alpha particle X-ray spectrometer (APXS) is part of the Athena payload of the two Mars Exploration Rovers (MER). The APXS sensor head is attached to the turret of the instrument deployment device (IDD) of the rover. The APXS is a very light-weight instrument for determining the major and minor elemental composition of Martian soils, rocks, and other geological materials at the MER landing sites. The sensor head has simply to be docked by the IDD on the surface of the selected sample. X-ray radiation, excited by alpha particles and X rays of the radioactive sources, is recorded by a high-resolution X-ray detector. The X-ray spectra show elements starting from sodium up to yttrium, depending on their concentrations. The backscattered alpha spectra, measured by a ring of detectors, provide additional data on carbon and oxygen. By means of a proper calibration, the elemental concentrations are derived. Together with data from the two other Athena instruments mounted on the IDD, the samples under investigation can be fully characterized. Key APXS objectives are the determination of the chemistry of crustal rocks and soils and the examination of water-related deposits, sediments, or evaporates. Using the rock abrasion tool attached to the IDD, issues of weathering can be addressed by measuring natural and abraded surfaces of rocks.
 The alpha particle X-ray spectrometer (APXS) permits the determination of the elemental chemical composition of rocks and soil by placing its sensor head against the sample, powering on the instrument, and commanding it to acquire spectra. It does not require any sample preparation and is thus well suited for in situ measurements of the surface constituents of objects in space (planets, comets and asteroids). Its working principle is based on the bombardment of the sample surface with alpha particles and X rays from radioactive sources (244Cm) and the measurement of the energy distribution of alpha particles, scattered by sample atoms in a backward direction, and of characteristic X rays, emitted by the sample atoms upon recombination of ionizations caused by the radiation from the sources, processes commonly referred to as “Rutherford backscattering” or RBS, “particle-induced X-ray emission” or PIXE, and “X-ray fluorescence” or XRF. One version of the instrument has been on board of the Sojourner rover of the NASA Mars Pathfinder (MPF) mission. This instrument and the history leading to its development have been described before [Rieder et al., 1997a]. Results of the measurements at the Pathfinder landing site (a total of five rock and six soil analyses) and data interpretations have been reported [Rieder et al., 1997b; Wänke et al., 2001; Brückner et al., 2003].
 The newly developed APXS, part of the Athena Payload of the two Mars Explorations Rovers, can surpass the performance of the previous MPF APXS as shown below. The new Athena APXS is mounted on the instrument deployment devices (IDD) of the two rovers together with the Mössbauer spectrometer (MB) providing data on the mineralogy [Klingelhöfer et al., 2003], the microscopic imager (MI) showing the texture of the measured surfaces [Herkenhoff et al., 2003], and the rock abrasion tool (RAT) removing surface layers [Gorevan et al., 2003].
 The key APXS objective is the determination of the elemental composition of rocks, soils, and sediments. Measurements at the planned MER landing sites, Gusev Crater and Terra Meridiani, will provide new insights into the geochemistry of the surface of Mars, not only because of the increased APXS' sensitivity, but also the complementary data sets of the companion two instruments on the IDD (MB and MI), and the Panoramic Camera (Pancam) and the Miniature Thermal Emission Spectrometer (Mini-TES) on the rover deck. An overview of all of the instruments is found in the work of Squyres et al. .
2. Science Objectives
 The science objectives of the Mars Exploration Rover (MER) missions are committed to the search for past and current water activities on the surface of Mars. This is a prerequisite for the search for life, extinct or extant, that will be addressed by future missions. The question of water activities is an intrinsic part of the issue of the habitability of Mars. There are many hints that Mars was once warmer and wetter, but today Mars appears to be a very dry place.
 This leads to the question: where has all of the water gone? Some might be lost to space with the almost complete loss of a once thicker atmosphere; some might be hiding from orbital instruments that use radiation that penetrates only micrometers into the surface. The Mars Odyssey spacecraft was the first mission to provide tools to look into the subsurface: with neutrons down to about 1 m, with gamma rays several tens of centimeters. The combined data set of neutrons and gamma ray measurements revealed that there are considerable amounts of hydrogen in the surface, as both methods can only detect elements (H), but no compounds (H2O). Therefore the term “water equivalent hydrogen abundance” is used. The high amounts of H suggested that water ice is buried under a rather dry layer in the polar regions south of −60° latitude [Boynton et al., 2002] and north of +60° [Boynton et al., 2003]. A very recent compilation of global neutron maps show that even in midlatitudes the water equivalent hydrogen content varies between 2 and 8 wt % [Feldman et al., 2003]. This water containing layer, consisting of mixed water ice and soil or a mixture of hydrated minerals with soil, seems to be covered by a rather dry layer of soil (about 1–2% water equivalent hydrogen). It can even be speculated that the upper 1 cm contains less that 1% water equivalent hydrogen.
 The landing sites selected for MER seem to have been exposed to liquid water in the past and to offer safe landings [Golombek et al., 2003]. At the Meridiani Planum site, the objective is to investigate what processes formed the hematite that covers 15–20% of the surface (at least two distinctive geological units are present). This site was detected by the Thermal Emission Spectrometer (TES) of the Mars Global Surveyor (MGS) spacecraft [Christensen et al., 2000]. There are several mechanisms to form hematite: precipitation from iron-rich water in a lake or hydrothermal water in the ground. Some geological processes exposed the once covered hematite deposits and may have altered them. The instruments of the IDD are very well suited to determine the chemistry of the different units at the site, the mineralogy of the hematite and the other unit(s), and the texture of the minerals and other samples. These data will provide the base of reconstructing the geological processes that once formed the hematite-rich surface.
 The second landing site is Gusev crater, which is a large flat-floored crater. Its southern rim is breached by Ma'adim Vallis, which is one of the largest valley networks on Mars that seem to be formed by running water [Irwin et al., 2002]. Gusev crater could have been filled with sediments carried by the water from Ma'adim Vallis. Eventually, the standing body of water exited the crater through a gap in its northern rim. At the Gusev site, fluvial sediments transported from southern highlands seem to be deposited in a lacustrine environment and can be investigated by the Athena instruments. The sediments may have preserved clues about their formation on early Mars.
 Geochemical investigations are an important part of characterizing a landing site, which includes chemistry, mineralogy, and texture of surface material. The major APXS objective is the measurement of rocks, soils, and water-related products, such as sediments, evaporates, and other water-lain deposits. Chemical analyses of the samples will be supported by mineralogical analyses and images of their texture, hence facilitating their classification (igneous, metamorphic, sedimentary, water depositional, etc.). Concentration variations of sulphur and chlorine can be used as a proxy for the occurrence of different salts.
 The rock abrasion tool (RAT) will be used to remove surface layers of rocks. Measurements before and after the grinding will provide invaluable data on the weathering processes of Martian rocks in the very dry climate and chemically different environment of Mars compared with Earth. This will address open issues on weathering of rocks that could not be solved during the Pathfinder mission [Brückner et al., 2003].
 The analysis of the ubiquitously distributed soil at the landing sites will be compared with the analysis of Mars Pathfinder and Viking 1 and 2 soils. The soil can be transported by large dust storms and local dust devils and, as a consequence, covers all of the surfaces depending on their roughness and angle. Vertical rock surfaces are covered the least, while horizontal surfaces the most, as observed during the Pathfinder mission. Since the APXS cannot distinguish between proper sample surface and airborne dust, a mixed signal from both components is measured. To minimize the soil component, mainly vertical surfaces of rocks were measured by the APXS during the Pathfinder mission, an approach also to be taken during the MER missions. On the basis of the known composition of the soil (determined by the APXS), the undisturbed, soil-free compositions of the samples have to be derived. For the Pathfinder rocks, sulphur was used as a monitor for the amount of adhering soil. A soil-free rock composition was calculated by using data from all rocks and extrapolating the elemental concentrations to a low abundance of sulphur [Brückner et al., 2003]. For MER, APXS measurements of three or more target sites of the same rock, but with different soil coverage, should be done to trace the interference of adhering soil and to derive a soil-free composition. Brushing the rock surface and grinding off a thin layer (RAT operations) will also reveal the amount of adhering soil. Since the RAT operations will be limited due to their long time and high power consumptions, the multilocational approach of APXS measurements will be very useful.
 Investigations of local soil inhomogeneities measured during the rover traverses may yield insight into local weathering or water-induced processes. Additional objectives are to find weathering products of rocks that contribute to the local soil chemistry. Soil chemistries can be modeled by using potential chemical end members. Different surface scenarios can be assumed, such as mixtures of different weathered rocks and other source materials.
 On the basis of orbital measurements, as mentioned, the surface seems to be layered: a dry layer on top of a “wetter” layer. A specific rover action could investigate the layering of the surface by using one wheel to dig a hole into the ground, while the other wheels are held fixed. Measurements with the IDD instruments should be performed before and after each digging step. The APXS data would reveal any change in chemistry as function of depth. Since hydrogen is per se “invisible” for X rays and alphas, water equivalent hydrogen abundance cannot be detected directly by the APXS. However, if a rather large change in water abundance is occurring, indirect evidence can help. One can look for extra oxygen from H2O as a “missing element” in the X-ray mode provided the sensor head was properly docked. Adding up all elements as oxides would show, how much H2O is needed to complete the sum to 100% (taking potential distance effects into consideration). The alpha mode could show the extra oxygen, provided it is detectable under the CO2 interferences.
 All of the data will be used to understand the current state of the landing sites, to test hypotheses concerning their geologic and climatic evolution, and to study the role of water in modifying Martian surfaces.
3. Instrument Description
 Driven by “lessons learned,” the design of the instruments to fly on the Mars Exploration Rovers differs in several respects from the design of the Pathfinder instrument, the most important ones being the use of a superior X-ray detector; the coaxial arrangement of sources and detectors; a source holder with a quick lock; titanium foil coverage of alpha sources; preamplifiers for alpha and X-ray detectors in the sensor head; and protective doors with a calibration target.
 The measurement geometry of the new design has been slightly decreased: the sample diameter is now 38 mm (old 50 mm) and the working distance (mean distance from sample to sources/detectors) is about 30 mm (old 50 mm). Although the instrument now uses only 2/3 of the radioactive material (six instead of nine sources), its sensitivity in the RBS (alpha) mode is about the same as that for the Pathfinder instrument. Sensitivity in the X-ray mode, however, has increased significantly: comparing the peak areas, we find a factor of more than 10 improvement for the elements from Ca to Fe; for the light elements (Na, Mg, Al, Si) the increase is even larger (Figures 1 and 2). Considering the fact that the resolution of the X-ray detector is almost twice as good as the one of the Pathfinder instrument, it is now possible to get statistically meaningful data in a measurement time of as little as 10 min, ideal for “touch-and-go” operation on the Martian surface (Figure 3).
3.1. Sensor Head
Figure 4 shows a photograph and Figure 5 shows a cross-sectional drawing of the sensor head. It is packaged in a cylindrical enclosure 53 mm in diameter and 84 mm in length and terminates in an insulating flange of 68 × 68 mm. The front part facing the sample contains the X-ray detector, mounted on the axis of the instrument, a cylindrical source holder with six alpha sources, and six rectangular alpha detectors. The coaxial arrangement of sources and detectors for alpha particles and X rays assures that both detectors “see” the same intensity distribution across the sample.
 Use of a superior X-ray detector (silicon drift detector with 10 mm2 active area, a 5 μm thick Be window and an energy resolution of about 160 eV at 5.9 keV) permits high-resolution measurements compared to the Mars Pathfinder APXS. A description of the basic detector design is found in Lechner et al. . Advanced detector versions were provided by KETEK, Munich, Germany.
 The new X-ray detector makes the former proton mode obsolete. The proton detectors used in the MPF design have therefore been replaced with a second group of detectors, identical to the alpha detectors, but not exposed to alpha particles from the sample. These detectors measure the background contribution to alpha spectra due to cosmic radiation at the surface of Mars and high-energy gamma background of the Cm sources as well as the Mössbauer source. The field of view for the X rays is delineated by means of a collimator in front of the detector: the collimator is formed by two apertures made from Zr, one immediately in front of the detector and one in the central orifice of the source holder.
 The alpha sources (six instead of nine for MPF) are contained in a source holder that attaches to the sensor head with a spring loaded bayonet-style mechanism. This permits quick and easy exchange of the sources without the need to disassemble the sensor head. The sources are covered with 2.5 μm thick titanium foils, turning them into “quasi-closed” sources (hermetically sealed sources are under development, but were not available for this mission). The foils prevent contamination of samples with source material, emitted from the sources as a result of “recoil-sputtering,” and at the same time reduce the energy of the alpha particles from 5.80 to 5.17 MeV, thereby avoiding a resonance in the 12C(α, α′)12C reaction at ∼5.7 MeV (Figure 6). This measure, together with an optimized design of the source-collimator-detector geometry, significantly reduces the background signals from carbon and oxygen in the Martian CO2-atmosphere. Nevertheless, this background signal remains the limiting factor for the determination of carbon in the samples (Figure 7).
 Twenty layers of thin stainless steel foils are located in front of the alpha detectors. They contain precisely etched honeycomb orifices to form the collimator of the alpha detectors. These orifices form tubular channels oriented toward the sample center, separated by thin walls.
 The sensor aperture is closed with two doors that protect detectors and sources from contamination with dust. These doors will open and lock in the open position, when a “contact ring” at the front face of the sensor head is pressed against an obstacle, such as a rock, with a force of about 2.5 N. An axial motion of this ring turns the doors inward by an angle of ∼100°. This axial motion is accomplished by guiding the ring with four spring-loaded rods such that pressing anywhere at its circumference will result in an axial travel. Two of these rods terminate in a ratchet mechanism that locks the doors in the open position. Closing is accomplished by pressing against a special “release lever”. These actions will be accomplished by means of the instrument deployment device (IDD), thus avoiding the need for additional motors and gears in the sensor. The position of the doors is monitored by micro switches whose signal is used to dock the instrument against a surface with the IDD.
 The inner surfaces of the doors carry an internal calibration target (a thin Au layer over Kapton) that permits us to verify and monitor proper performance of the instrument after landing on Mars. In the alpha spectrum, peaks of Au and C are produced that are used for energy and efficiency calibration. In the X-ray spectrum, several Au peaks of different energy are obtained that provide information on performance and on possible contamination of the beryllium entrance window of the X-ray detector.
 The rear part of the sensor head contains a stack of five printed circuit boards (see block diagram in Figure 8). It holds hybrid preamplifiers for the alpha and X-ray detectors. The X-ray channel contains an additional amplifier to increase the signal amplitudes. Signals, ∼200 mV full scale for X ray and alpha, are routed to the main electronics differentially, thus effectively suppressing common mode noise. The signal and power connection to the main electronics is done by a shielded flexible printed circuit cable. These measures greatly reduce the sensitivity of the instrument to external sources of noise (e.g., switching power supplies). The sensor head also contains the generator for the bias voltage of the X-ray detector (−120 V; low power charge pump cascade) and a temperature sensor for gain correction.
3.2. Alpha Sources
 The curium alpha sources are specifically designed for APXS measurements. In order to minimize the energy spread of the alpha particles, interelement compounds of curium and silicon (curium silicides) were made from semiconductor-grade silicon. Six sources provide a total activity of 30 mCi (1.1 GBq). Even this source strength leads to rather long counting times to obtain good counting statistics for alpha spectra. To avoid any recoil sputtering contamination of the inside of the instrument and the samples, selected pinhole-free very thin Ti foils (2.5 μm) are placed in front of the Cm sources (Figure 6).
 The daughter decay product of curium-244 is plutonium-240, which provides strong X-ray radiation of about 14 and 18 keV (Figure 9).
3.3. Main Electronics
 As shown in the block diagram (Figure 8) the main electronics consists of the analog signal conditioning segments (six pole Gaussian filter amplifiers, threshold discriminators and peak detectors), an analog multiplexer, a 16 bit analog-to-digital converter and an 8 bit microcontroller.
 Control logic determines the presence of a relevant signal and generates an interrupt in the microcontroller. To avoid additional noise in the analog signal chain, the microcontroller is kept in idle mode until the analog signal is processed and buffered in the peak detectors.
 Selection of the appropriate multiplexer input, conversion of the signal amplitude to a digital number and registration of the signal by incrementing the number of counts in the corresponding amplitude channel of the respective detector is then handled by the microcontroller. Conversion time is typically 200 μsec. A digital temperature compensation routine that minimizes the influence of temperature changes during long measurements adds another 100 μsec. For a mean count rate of 100 Hz, a total dead time of below 5% is achieved.
 The microcontroller is equipped with a watchdog circuit that performs a soft reset in case of an abnormal program flow. The data are stored in 32 kilobyte SRAM that are buffered by a battery located on the main electronics board.
 The interface for commanding the instrument and transfer of data consists of an RS 422 serial link. Power is provided to the instrument directly from the board battery (nominally 28 V); voltages required by the electronics (+5 V digital, ±5 V analog and ±12 V analog) are generated by its own power converter and filters. For specifications see Tables 1 and 2.
Table 1. APXS Sensor Head, Including Door Mechanics, Microswitches, an X-Ray Channel, Two Alpha Channels, and Tantalum Shielding
Power +12 V, mA
Power −12 V, mA
Table 2. APXS Main Electronics
Power +5 V, mA
Power −5 V, mA
 The X-ray spectrum is divided into 512 channels. The lower threshold is fixed at ∼850 eV. This is sufficient to detect Na at 1040 eV. The upper energy limit is about 16 keV. The spectral range includes the K lines up to Zr and the L and M lines of higher Z elements. It also contains elastic scattered Pu lines at 14.3 keV and 12.6 keV, as well as inelastic scattered peaks. The alpha and background spectra use 256 channels and range up to about 6 MeV.
3.4. Design Qualification
 The instruments were tested according to the rigid requirements of the MER lander project. Their performance over the expected Martian day and night temperature ranges was tested in a vacuum chamber using external radioactive sources. During Martian night temperatures the instrument provides best energy resolution of 160 eV. During day time temperatures (less than −10°C) short term measurements (touch and go) can be carried out with degraded resolution of about 200 eV.
4.1. Calibration Prelaunch
 During the calibration campaign in the Mainz laboratory many geostandards were measured. These standards are powdered geological samples, whose elemental concentrations are certified by qualified institutions. The two flight instruments containing their flight sources were calibrated using 11 validated samples (eight geostandards and three meteorites) (Figures 10 and 11) and a set of oxide and metal standards.
 To perform the calibration the instruments are placed into a “Mars simulation chamber,” where temperature and atmosphere can be adjusted. The APXS sensor head looking downward is mounted in the upper part of the chamber. The housing of the sensor head is in direct contact with a two-stage Peltier cooler that lowers the sensor temperature to about −35°C. At this temperature level, the best energy resolution is achieved.
 The simulation chamber can be divided into two sections by closing a gate valve. A sample can be inserted in the lower part while the upper part remains under vacuum or low pressure. After having opened the gate valve, the sample is lifted upward to the opening of the sensor head, where a hard stop provides a reproducible sample position.
 The calibration was carried out under Martian like atmospheric conditions, i.e., a pressure of 10 mbar CO2. To guarantee an identical long-term atmospheric pressure, a continuous gas flow was maintained by a pressure feedback controlled valve system.
 Some of the structural elements of the calibration chamber consist of glass, which allowed optical inspection of the instrument and the sample. Furthermore, light sensitivity of the detectors could be tested by strong illuminations, a problem that occurred during the Mars Pathfinder mission for the sensor head. The new APXS sensor head was found to be insensitive to stray light. Shining strong light directly into the opening of the instrument, a situation prevented normally by the sample body covering the opening, was found to make the alpha spectra slightly noisier, but not the X-ray spectra.
4.2. Calibration Postlaunch
 After delivery of the flight sensor heads, the calibration program was continued with a nearly identical laboratory sensor head. The number of geostandards was augmented to extend the concentration ranges of the elements and to vary considerably the type of composition matrix. These data are used to test theoretical calculations on expected line intensities as a function of elemental composition. These investigations will eventually reduce the errors of the concentrations of an unknown sample introduced by the effect of different composition of the standards. A preliminary outcome is provided in the chapter on data evaluation; the complete calibration will be described elsewhere.
4.3. Calibration on Mars
 A check of the performance of the instrument after the landing on Mars will be done making use of the internal calibration target on the doors and the Compositional Calibration Target (CCT) that is mounted on the rover chassis in reach of the IDD.
 The X-ray spectrum of the internal calibration target (Figure 12) shows gold, nickel, and copper lines. The target consists of a set of thin layers of gold, Kapton (carbon), and nickel on top of the copper-beryllium body of the doors. Energy calibration, FWHM, and linearity can be checked by evaluation of the copper and gold lines and comparison with prelaunch data. Contamination of the beryllium entrance window of the X-ray detector will be noticeable by an intensity reduction of the low-energy M lines of gold compared to the L lines. The corresponding alpha spectrum of the calibration target is seen in Figure 13. Energy calibration can be checked with the position of the gold peak (a peak and not a step because of the small thickness of the Au layer). The carbon step provides an additional check of consistency.
 An X-ray spectrum of a duplicate of the flight CCT is shown in Figure 14. The CTT consist of a magnetite plate. This target was designed for the needs of the MB, but, it can also be used by the APXS to check its FWHM usually determined for the 6.4 keV line of iron. As the target is mounted on the outside of the rover, it will eventually be covered with dust, but the line shape of the Fe line will not be affected.
5. Operation of the Instrument
 The command interface to the rover is kept very simple. The basic operation consists of two commands: (1) APXS_COMMENCE, start a new measurement and (2) APXS_END, stop the measurement, transmit all data, and power off the instrument.
 During operation the instrument firmware controls all the necessary functions. No Rover interaction or computer wakeup during night is needed for long term measurements. The instrument performs a temperature calibration every 30 s to avoid gain shift with temperature. The internal memory can store up to 12 sets of spectra. During long term measurements, the processor stops the acquisition after a predefined time, stores the spectrum, and starts a new cycle. All spectra contain a unique identification tag.
 The temperatures of the sensor head and the electronics slice inside the warm box of the rover are sampled together with the spectra in 30 s intervals. A logbook in the memory contains the history of commands that were sent to the instrument. All the data are stored in a battery buffered SRAM (32 kilobyte) to avoid data loss. The complete content of the SRAM is transmitted to the Rover computer for downlink.
5.2. Operation on Mars
 There are two ways of APXS operation: long and short measurements. Long-term spectra can be taken during the night to allow for excellent counting statistics of the X-ray mode and for suitable statistics of the alpha mode. To search for trace elements, 2–4 hours are sufficient for the X-ray mode. The search for carbon requires at least 8 hours for the alpha mode as the alpha sensitivity is low. X-ray and alpha mode always operate together.
 To make an APXS measurement, the sensor head has to be “docked” at the selected sample by the IDD. The nominal APXS docking procedure is the following: (1) use images taken by the rover's front cameras: Navcams (navigation cameras) and/or Hazcams (hazard avoidance cameras) to calculate the IDD positioning; (2) open the doors by pressing the APXS contact ring against the CCT or any other solid surface of the rover; (3) position the APXS contact ring up against a selected sample area with a positional accuracy of 10 mm and 10° on a target not previously contacted, or 4 mm and 3° for a target previously contacted by any one of the IDD instruments; and (4) after data acquisition, the APXS doors are closed by rotating the turret past a roller until the release lever is actuated. The docking of the sensor head to the sample will be monitored by the rover's front cameras.
 During the day, “touch-and-go” operations can be performed as long as the temperatures are below −10°C to allow for good energy resolution of the X-ray detector. The lower the temperature, the better is the resolution of the X-ray detector: an optimum resolution is obtained at temperatures below −25°C. Therefore measurements should occur early morning or late afternoon on Mars, while the time around noon should be avoided. Midday temperatures can reach zero centigrade or higher, while during the night the temperatures fall below −60°C. The shortest measurement time for touch-and-go should be at least 15 min to allow for several minutes of warm up of the sensor head and 10 min counting time. Longer times of 30–60 min are good compromises between long night measurements and very short “touches.” No statistically meaningful alpha spectra are expected during touch-and-go measurements.
 When the rover has been moved to a new place and stereo images have been taken of the work volume of the IDD by Navcam and Hazcam, decisions will be made after the next downlink to earth. The rationale for IDD measurements are: first do touch-and-go and after downlink, decide on staying (more investigations) or going. A very short measurement of a soil target or a nearby rock with the APXS will be performed. If time permits, a short Mössbauer measurement (about an hour) can follow. A microscopic imager picture could precede or follow the sequence depending on time of day and resulting shadow of the turret on the target. After the next downlink, data will be analyzed. If the chemistry of the target is comparable to previous measurements within the error limits of the analysis, no further measurements are necessary. If the soil chemistry changed outside of the analytical errors of key elements, longer measurements should be planned for the following night. Key elements, such as sulfur, chlorine, and potassium, are those whose concentration is expected to be highly variable.
 If the target is a rock, detailed preparations have to be done. To position the APXS on a rock, preferably on its vertical side, requires IDD operations based on stereo images of the target. Vertical positioning has the advantage of lower dust contamination as compared to a top horizontal rock surface. With respect to chemistry, a similar rationale holds for the rock as for the soil. If the chemistry is outside of the analytical errors, a long measurement should be planned for the following night. There is no need to move the IDD after the touch, as MB measurements should be done, too. Since the MB needs to measure during different temperature regimes, the MB measurements will start during the day and continue into the night. Later in that night, the rover would wake up and bring the APXS in position. If a search for carbon is planned, 8–12 hours of the night should be reserved for the APXS to obtain reasonable alpha spectra, otherwise 2–4 hours are sufficient for high-quality X-ray spectra.
 Touch-and-go APXS measurements of the soils are a quick and easy way to monitor the chemistry of the rover traverse. The APXS team will analyze the spectra immediately after their downlink to Earth and report to the rover science team on the chemical composition of the most recent sample. A decision will be made whether to proceed with the traverse or to stay and make more measurements. The normal operation of the APXS will be done in conjunction with MB and MI and the Hazcams. Instruments on the turret are used one after the other. If time is limited, only short APXS measurements may be done. If longer measurements are required during the day, the APXS can accumulate spectra without interferences, when the panorama cameras or the Mini-TES is operating, as tests have confirmed.
5.3. APXS and Tools
 When RAT operations on a rock are planned to investigate possible weathering rinds for example, APXS, MB, and MI measurements have to be performed before and after the RAT actions. The APXS will make a sufficiently long measurement (preferentially 2 hours) on the undisturbed selected target area before any RAT action takes place. The inner diameter of the APXS contact ring is 38 mm, while its outer diameter is 49 mm (the outer diameter of the APXS tube is 53 mm). A borehole with a diameter of 45 mm is ground by the RAT producing a flat abraded area [Gorevan et al., 2003]. Therefore the APXS will not fit into the borehole to touch the abraded area; it will be positioned only at the outer surface of the rock. The intensity of the measured signal will decrease with increasing distance of sample to APXS. However, any change in chemistry will be easily detectable independent of the distance (the APXS is calibrated for effects of distance).
 If the rover is used to dig a hole into the ground with a wheel, the APXS should monitor the chemistry before and after each digging step. Provided the diameter of the hole is wide enough (more than 60 mm), the APXS can be inserted into it to make measurements.
5.4. Gamma Ray Background
 On the IDD turret, the APXS sensor head and the accompanying Mössbauer sensor head (MB) are positioned very close together. The preferred position on the turret, APXS opposite to the MB, was taken by the microscopic imager to prevent radiation damage of the optics. Therefore the gamma ray background that is produced by the Co57 sources of the MB sensor head was carefully investigated. These sources, one reference source with a strength of ∼10 mCi and a main source of about 150 mCi, are located only 10 cm away from the APXS detectors. Early measurements revealed that the Co57 source increased the background in the APXS X-ray spectrum by a factor of about 10 mainly due to Compton scattered gamma rays of the 136 keV Co57 line. The reference and the main source contributed equally to the APXS background because of their relative positions. The reference source was closer to the APXS detectors compared to the main MB source and illuminated the APXS sensor head from the rear. By attaching 0.5 mm thick Ta shielding plates, weighing only 25 g each, on both instruments, the additional background produced by MB could be reduced to a level comparable to the normal APXS background, so just doubling the overall background. Tests with a massive iron sample revealed that the additional Co57 sources did not introduce additional elemental signals beside a general background.
 Several APXS background peaks were found in the X-ray spectra that could be attributed to different local materials used in the sensor head: Zr: collimator structure, Cu: mounting screw of the X-ray detector housing, and Ti: elastically scattered signal from the Ti foils that cover the sources. These sample-independent backgrounds are taken into account for data evaluation. No other background peaks were found.
6. Data Evaluation
6.1. Spectra Deconvolution
 The deconvolution of the measured spectra is performed to derive the peak areas of the characteristic element peaks including statistical standard deviations. A theoretical model spectrum containing the superposition of all elemental peaks is calculated by using a spectrometer response function. These peak shape parameters are subject to a least square fit routine that varies them until the deviation of the calculated from the measured data points is minimized.
 A special fitting routine was developed that contains the following features: (1) Gaussian shaped peaks with up to five characteristic peaks per element. The ratios of the different shell lines can be varied within limits to obtain physically meaningful fits of overlapping lines (e.g., a Ni Kβ line is only fitted if there is a Kα line); (2) energy resolution (full width at half maximum) as a function of energy, electrical noise, and Fano factor; (3) escape peaks of all major peaks as a function of energy; (4) a background model, including a constant background and a step function on the low-energy side of the peaks; (5) exponential tailing on the low-energy side of the peaks; and (6) peak position as a variable parameter for fitting within small limits. Since the measured spectrum represents the spectrometer response, the instrument performance can be checked by the spectrum evaluation.
6.2. Elemental Abundances
 The derived peak areas have to be converted to elemental abundances of the sample. There are two ways to go that can be combined in the end: experimentally derived calibration and theoretical modeling.
 A set of geostandards was measured whose elemental concentrations were very well known and spanned the range of expected Martian concentrations. The peak areas of the X-ray spectra were compared with their elemental concentrations. Eleven geological samples were measured in the flight configuration: both flight sensor heads were equipped with flight sources and main electronics boards, identical to the flight versions, which were already delivered to the flight rovers.
 The second way is to calculate the X-ray signal strength for all elements of the measured geostandards by using a theoretical model, such as an extended fundamental parameter method. The physical processes of X-ray excitation in the sample by alpha particles and monoenergetic X rays are modeled, including the stopping or attenuation of incoming radiation, the magnitude of atomic shell ionization cross sections, and the attenuation of outgoing X rays including secondary fluorescence effects. The development of this software tool is presently under way.
 Measurements were done to investigate the magnitude of PIXE and XRF excitations that are needed for the theoretical modeling. The contribution of XRF was determined by shielding the alpha particles of the curium source with a thin aluminum foil (Figure 15). The X-ray spectrum of the Cm source was determined to get the excitation spectrum for XRF (Figure 9).
 A basic correction method has been formulated and can be applied to the measured data (Figures 16 and 17). It turned out that for higher Z elements, where the XRF is the dominant part, the attenuation coefficient for the incoming and outgoing radiation has to be taken into account (Figure 18).
 A simple theoretical model describes the measured intensity:
attenuation coefficient of the incoming radiation (14.3 keV);
attenuation coefficient of the elemental characteristic energy.
The attenuation coefficients were calculated according to the experimentally determined composition.
 For Fe, the uncorrected and corrected calibration curve is shown in Figures 16 and 17. This simple matrix correction works well for high Z elements, where X-ray radiation is the dominant excitation process.
6.3. Sampling Depth
 The different depths that are sampled by the X-ray mode and the alpha mode were investigated. Model calculations were made for a rock matrix of andesitic composition (sample SSK1), similar to the composition of Martian rocks found at the Pathfinder landing site.
 The Figure 19 shows the energy of backscattered alpha particles after scattering from Na to Fe atoms in the matrix. With a spectral cut-off energy of 0.5 MeV the corresponding maximum depths in the matrix are 2 μm for C, 3.5 μm for O, 6 μm for Si and 8 μm for Fe.
Figure 20 shows from which depth which percentage of the total cumulative intensity of X rays originates. This plot includes X rays generated by particle-induced X-ray emission and X-ray fluorescence. For Na, 50% of the total intensity originates in a layer of 0.8 μm and 95% from a layer of 3.5 μm. For Si the respective numbers are 2 and 6.5 μm, for Ca 4 and 14.5 μm and for Fe 22.5 and 97 μm (not shown).
 The large increase in sampling depth for Fe in comparison with Ca is the consequence of the rapidly increasing contribution of X-ray fluorescence to the total output with rising Z: for Ca the ratio of X rays produced by X-ray fluorescence versus particle-induced X rays is 10:90, whereas in the case of Fe this ratio is 90:10. Where particle-induced X rays predominate, the sampling depth is limited by the range of alpha particles in the matrix. For X-ray fluorescence the limiting factor becomes the absorption of the characteristic X rays in the sample matrix; the much larger penetration of Pu-L X rays is playing only a minor role.
6.4. Data Products
 Raw and evaluated data will be delivered to the NASA Planetary Data System. Since our spectra are small (X ray 1 kilobyte, alpha 0.5 kilobyte), they will be provided in ASCII format. Since the X-ray data are better quality compared to the alpha data, the major focus will be on the evaluation of the X-ray data. The X-ray products will be: raw spectra (counts in one column), energy calibrated X-ray spectra (energy and counts in two columns), preliminary oxide concentrations based on measured concentration calibration curves simply corrected for matrix effects (ASCII table), and after mission end elemental concentrations resulting from iterative applied theoretical corrections (ASCII table). The alpha products will be: raw spectra (counts in one column), gain-shift-corrected spectra if necessary (counts in one column), flags for detection of carbon, and after mission end concentration of carbon and oxygen using the concentration results of the other elements from the X-ray mode.
 The new alpha particle X-ray spectrometer (APXS) is a small instrument to determine in situ the chemistry of various samples at landing sites on Mars. The instrument uses two different methods for analysis: X-ray spectroscopy (X-ray mode) and alpha backscattering (alpha mode). In the X-ray mode, there is dual excitation of X rays by curium sources, alpha-induced X-ray emission (PIXE) and X-ray fluorescence (XRF). Elements from sodium up to zinc and higher in Z can be detected in the X-ray mode and their concentrations can be derived. Owing to a new type of X-ray detector, the sensitivity is very much increased compared to the Mars Pathfinder APXS, which leads to measurement times as short as 10 min for major and minor elements, while 2 hours are needed in general for trace elements. In the alpha mode, low Z elements, such as carbon and oxygen, can be detected in addition to the main elements of the X-ray mode. However, there is a reduced sensitivity for low Z elements measured in a CO2 atmosphere. So, the main application of the new APXS is X-ray spectroscopy, where it is unsurpassed. Both APXS modes cannot detect water directly. Here, indirect methods have to be applied and data from other instruments should be used. Owing to the type of radiation (X rays and alphas), the average sampling depth of an investigated surface is between 1 and 20 μm, depending on the element. Even thin weathering rinds can falsify the measurements: the concentration of the surface of a rock does not correspond to its bulk composition. Here, the rock abrasion tool (RAT) comes into play, because it can provide a fresh surface by grinding. Making measurements before and after the RAT actions will provide a depth profile of the investigated rock. In addition, even thin evaporates resulting from former water activities can easily be analyzed by the APXS. The ubiquitous soil expected at the landing sites will cover almost all surfaces to a varying degree and produce a mixed signal of sample and soil for the APXS, which has to be corrected for. To obtain a good performance of the X-ray detector, its operating temperature has to be below −10°C, which leads to the requirement of excluding the hours around noon for APXS measurements. Early morning and late afternoon are very well suited as well as the night, where long-term measurements will be carried out because the alpha mode needs at least 8 hours to obtain decent counting statistics. As the APXS is mounted on the instrument deployment device together with the Mössbauer spectrometer and the microscopic imager, coordinated planning of the usage of the three instruments is very efficient, since the interpretation of the AXPS data benefits, if data on mineralogy and surface texture are also available for the same rock or soil. Because of its short measurement time, the APXS should be docked to as many samples as possible (touch-and-go) providing a “chemical log” of the rover traverse. When chemistry did change, longer and more detailed measurements will be needed for thorough characterization. The APXS provides the means for a thorough geochemical exploration of the MER landing sites leading to more insight into the potential traces of past and current water.
 We are thankful to R. Haubold, W. Huisl, and B. Spettel for analytical support. This work was supported by the Max-Planck-Society and the German Space Agency (DLR) contracts 50 QM 0014 and 50 QM 0005.