Synchrotron x‐ray diffraction for sealed Mars Sample Return sample tubes

The joint NASA‐ESA Mars sample return campaign aims to return up to 31 sample tubes containing drilled sedimentary and igneous cores and regolith. The titanium alloy tubes will initially still be sealed when they are retrieved. Several types of measurement will be carried out on sealed samples in the pre‐basic characterization phase of scientific investigation. We show that powder x‐ray diffraction (XRD) analysis can be successfully carried out on sealed samples using an x‐ray source at the I12 beamline of Diamond Light Source synchrotron. Our experiment used an analog sample tube and a Martian regolith analog (Icelandic basaltic sand). The titanium walls of the tube analog give strong but few diffraction peaks, making identification of the major constituent mineral phases feasible. A more significant constraint on quantification of mineral phase abundances by this XRD technique is likely to be the grain size of the sample. This technique opens up the possibility of initial mineralogical analysis of samples returned from Jezero crater without opening the sample tubes and the potential changes to the sample that entails.


INTRODUCTION Mars Sample Return
NASA, ESA, and the UK are collaborating on the Mars sample return (MSR) campaign which aims to retrieve drill cores of Martian rock for terrestrial analysis, starting with the Mars2020 rover, Perseverance, which landed successfully in Jezero crater in February 2021 (Meyer et al., 2022).MSR aims to help determine past and present habitability, develop our understanding of the Martian climate and its history, the evolution of Mars as a geological system, and help prepare for human exploration (Mustard et al., 2013).The rover's instruments are being used to identify the most scientifically valuable and return-worthy Mars materials.It uses a tungsten carbide drill to extract rock cores or regolith and deposits the material in titanium sample tubes, which are then hermetically sealed to prevent contamination.Mars2020 plans to collect a wide variety of sample types including igneous rocks; eolian, fluvial, and lacustrine sedimentary rocks; and regolith (iMOST, 2018).
As of April 21, 2023, Perseverance has collected one atmospheric sample, two regolith samples, and 16 rock cores.The regolith samples are thought to be a mix of igneous and sedimentary grains.Of the drilled rock cores, eight are from rocks thought to be sedimentary and the rest from igneous rocks (NASA, 2022).The rover has collected pairs of samples and left one of each pair in a depot on the surface at Three Forks where they could be collected later, to serve as a backup in case the samples on the rover become inaccessible.Photos of the ends of the cores are available (NASA, 2022) and Perseverance has investigated their source rocks with its instrument suite, which will allow some conclusions to be drawn on the characteristics of the samples before they arrive on Earth.This is not easy, however, and even whether some samples are sedimentary or igneous in nature is difficult to determine from the limited data available (Farley et al., 2022).
Up to 31 sample tubes are planned to be returned to the Earth, where much more extensive and discoveryresponsive analysis can be done than allowed by a rover.It will collect approximately 10-15+ g of Martian rock or regolith per tube, gas from the Martian atmosphere, and witness tubes.Witness tubes are well understood sampling surfaces that track the accumulation of contamination during the mission.Once they have landed on Earth, the samples will be sent to a purpose-built Mars sample receiving facility (SRF) for continued curation, characterization, and investigation.Analyzing the contents in the tube is an important early step in the sample return workflow (Haltigin et al., 2022;Holt et al., 2019;Meyer et al., 2022;Mustard et al., 2013).

Containment
MSR missions are classified as Restricted Category V, necessitating extensive measures to protect against backward contamination of the Earth's environment (Rummel, 1989).Preventing the introduction of potentially harmful extraterrestrial material to the Earth is referred to as planetary protection and is one of the top priorities of a sample return mission from one of the planetary bodies classified as potentially being able to sustain life.This is due to the risk of any extant pathogens or other dangers, such as prions or toxins, being capable of rapid spread and harm.
Containment of samples for planetary protection is only one half of the contamination requirements for returned samples.It is vital to prevent terrestrial matter, particularly organics, minerals, metals, and atmospheric gases from interacting with the Mars samples to ensure accurate measurements without false positives.This is especially important due to the many investigations into features of the samples that are expected to be present only in amounts near the limits of detection for current instrumentation, like biochemical biosignatures.For these reasons, stringent two-way containment measures will be in place for returned Mars material until it is sterilized or assessed safe, at which time planetary protection-specific measures can be reduced or removed (Meyer et al., 2019;United Nations, 1966).
Returned sample science is being planned in conjunction with containment protocols so that the initial sample science phase can be completed safely, expediently, and with minimal damage to the science value of the returned Martian material.Maximizing the amount of information gathered from samples before their hermetically sealed tubes are opened is thus very helpful for the returned sample science process.

Returned Sample Science
Once the returned Martian samples have landed on Earth and been collected and brought to the SRF, the Earth-based part of MSR science can begin.Sample processing and analysis is divided into five stages: hardware de-integration, pre-basic characterization, basic characterization (BC), preliminary examination (PE)/ sample safety assessment protocol (SSAP), and sterilization and release (Meyer et al., 2019;Tait et al., 2022).
Pre-basic characterization is carried out on the samples while they are still in their sealed sample tubes.The aim of Pre-BC is to gather information to inform how tubes are opened and the material analyzed in BC and PE, as well as fleshing out information in the sample's catalog entry.Currently, the measurement methods include x-ray computed tomography (CT), which will provide volumetric scans of the samples inside the tubes, providing information on structure, texture, orientation, and hinting at the different mineral phases present, as well as measurements of sample magnetism via a magnetometer.Pre-BC also includes measurements that must be carried out before tube opening, such as dust photography and removal, and checking of the sample tube seal integrity (Tait et al., 2022).
Basic characterization is the next stage of sample analysis.Its purpose is to inform PE and the SSAP, as well as how best to go about the subsampling process, and to provide information to populate the sample catalog with, and so will be carried out uniformly on every sample (Meyer et al., 2019;Tait et al., 2022).
The aim of PE is to provide sufficient information to allow principal investigators and consortia to design and propose scientific investigations of each sample, and to let the sample allocation committee best decide how to allocate the limited sample mass (Meyer et al., 2019;Tait et al., 2022).PE is currently the first phase for which XRD measurements are planned, which is the subject of this study.We show that XRD analysis could be done in the Pre-BC phase instead if several technical challenges are overcome.
X-ray diffraction analysis makes use of Bragg's law (Bragg & Bragg, 1913) to measure the lattice spacing of crystalline materials by passing x-rays through a sample and capturing the resulting diffraction pattern.The peaks of the diffracted x-ray spectrum correspond to specific lattice spacings, and thus specific crystalline materials, with their relative intensity correlating to the relative grain abundance of that phase in the beam path, after taking into account instrument parameters and diffraction efficiency.This makes XRD a useful method for phase identification and quantification of minerals.Other information can also be gathered with high quality data, including magnitude of lattice strain and crystallite size.
X-ray diffraction can only be carried out at the later PE stage of sample characterization due to limitations with conventional commercial x-ray diffractometers.The x-ray tube sources in commercial diffractometers lack the output intensity and wavelength control needed to penetrate the ~13 mm diameter rock cores that Perseverance is collecting.The penetration of, for example, the 8.04 keV Cu k a x-rays commonly used by commercial diffractometers through 13 mm of typical basalt is essentially zero as calculated through the Beer-Lambert law (Swinehart, 1962), using NIST's XCOM photon cross section database (Berger et al., 2010) to find the total attenuation coefficient of the basalt as 50.15 cm 2 g À1 , assuming a typical density of 2.9 g cm À3 .Many powder x-ray diffractometers utilize a Bragg-Brentano reflection diffraction geometry intended for small, flat powder samples, whereas a transmission geometry would be required for full quantitative phase analysis of an intact core.Furthermore, powdering the sample to achieve very small crystallite sizes (<10 lm) is essential before conducting bulk powder XRD in such scanners due to their small sampled volume, therefore requiring sample processing before XRD analysis.In situ XRD on the Martian surface by the CheMin instrument on the Mars Science Laboratory Curiosity rover, however, has demonstrated that analysis of a few tens of milligrams of sample with particle sizes <150 lm in diameter results in bulk mineralogy of this grain size with a detection limit of 1 wt% for crystalline phases with abundances >12 wt%, unit-cell parameters of major phases, and allows for quantification of the x-ray amorphous component (Blake et al., 2012;Rampe, Blake, et al., 2020).Sample tube opening and subsampling are therefore required for conventional powder XRD, hence its current placement in the PE phase.

Aims
We aim to show that powder XRD analysis can be successfully carried out on at least some returned Mars samples in unopened sample tubes and thus could be included in the Pre-BC phase of returned sample science.XRD as an analytical technique meshes particularly well with concurrent CT analysis, as data from one are useful for the other.CT can be used to identify volumes of interest from sample structure, texture, or attenuation, the mineralogy of which can then be identified with XRD.These mineralogical data can be fed back into analysis of the tomogram, allowing greater geological understanding of the sample.Pre-BC XRD would provide mineralogical data much earlier in the sample science process, improving decision-making around sample science, curation, and handling in all later phases.
While a conventional x-ray tube cannot provide an appropriate x-ray beam, a synchrotron source is capable of much higher intensities and precise wavelength selectivity.We investigate the feasibility of powder XRD measurements of MSR samples in sample tubes using such a source.We also determine whether the diffraction signal from the sample tube walls prevents sample phase identification and quantification, as minerals that have most of their stronger diffraction peaks overlapping with tube wall peaks will be more difficult to analyze.

Sample Tube Analog
In order to test XRD analysis of MSR samples, an analog of the sample tubes with the same composition and dimensions for the purposes of XRD was required.As up-to-date design specifications were not publicly available at the time of manufacture, earlier design information was used.The final sample tube design seen in Figure 1 consists of the tube opening at the top in which the hermetic seal is placed after sampling, the white alumina-coated central sample-containing section, and the bottom golden section where the rover interfaces mechanically with the tube (NASA/JPL-Caltech, 2020).The analog sample tube can be seen in Figure 2.Only the sample-containing section of the tube was reproduced in the analog design.Both tubes were made out of Ti-6Al-4V titanium alloy (5.5%-6.75%Al, 3.5%-4.5%V, <1% other elements, remainder Ti) (Redmond et al., 2020).The analog tube's inner diameter was designed to be 13.45AE 0.05 mm, with a wall thickness of 0.7 AE 0.1 mm, where the precision was restricted by machining equipment capabilities.These dimensions were designed to match (Redmond et al., 2020) as closely as practically feasible, which specified a minimum inner diameter of 13.40 mm and a wall thickness of 0.7-0.729mm.The tube analog used for these experiments had an average wall thickness of 515-550 AE 5 lm.The effect of this difference in wall thickness is discussed further in "Discussion" Section.The thickness of the interior titanium nitride coating was unknown, but presumed to be several microns as typical of TiN coatings (BryCoat Inc., 2023;Eltropuls, 2023;Wallwork, 2023), which would contribute negligible diffraction signal compared to the uncertainty in wall thickness of the analogs.The alumina coating thickness is likewise unknown, but these are often amorphous or part-amorphous, preventing or reducing diffraction by the material (Murray et al., 2016).

Sample Analog
A regolith sample analog was prepared for the monochromatic diffraction experiment, a fluvial volcanic sand from Þ orisj€ okull, Iceland, collected as a Mars analog as part of the NASA SAND-E analogue mission (Figure S1) (Ewing et al., 2020).The sediments collected by SAND-E are sourced from basaltic glaciovolcanoes and are composed primarily of volcanic glass, plagioclase, pyroxene, olivine, and likely some palagonite (Bedford et al., 2022), as determined using an Olympus Vanta portable x-ray fluorescence spectrometer.The specific sample used for this experiment has a mix of grain sizes ranging between 2.8 mm and 70 lm, that is, very fine gravel to very fine sand, though fine to very fine sand predominated.These grain sizes are a high priority for sampling because fine-grained sediments are thought to have high biosignature preservation potential (Cady et al., 2003;Westall et al., 2015).Very fine-grained sedimentary rock has been observed by Perseverance at Jezero crater in an area called the Hogwallow Flats (Kivrak, 2022), so our analog is representative of Mars material in this aspect.
As x-ray CT scanning is part of Pre-BC, a CT scan of the sample inside the titanium sample tube analog was carried out at the University of Manchester.The volumetric data were used to try to provide morphological context to the phases identified through XRD, as distinct mineralogical phases can often be identified by their difference in x-ray attenuation.The Nikon XT H 320 "Custom Bay" cone beam scanner was used.A cross-sectional slice of the resulting scan is shown in Figure 3.For this scan, the tungsten target x-ray tube was supplied with 100 kV and 7.5 W. 3141 projections were taken, with 500 ms exposure times.A 30 dB gain was applied to the detector, and four frames were taken and averaged per projection.The reconstructed volume had a voxel size of (9.2346 lm) 3 and had a radial beam hardening correction applied.The volume data are in 16bit greyscale.The scan lacked the spatial resolution required to determine the size of the finer grains and so comprehensive grain size analysis was not possible in this case, but is a useful possibility for larger grained samples or higher resolution instruments.

X-Ray Diffraction Experiment Design
The monochromatic synchrotron diffraction experiments were performed at the I12 beamline (Diamond Light Source, 2020; Drakopoulos et al., 2015) at Diamond Light Source, Didcot, United Kingdom, in Experimental Hutch 1.
The measurements used a flat-panel Pilatus 2M CdTe detector (Donath, 2020) in a transmission configuration with the sample, with an active area of 253.7 9 288.8 mm and 172 lm pixel pitch.An Si (111) double crystal bent Laue monochromator was used to produce a beam energy of 54.427 AE 0.002 keV as measured using (Hart et al., 2013).Measurements were taken of three different sample configurations: the basaltic sand inside the analog sample tube, the sand in a negligible diffracting polyether ether ketone (PEEK) polymer tube (to ensure correct sample shape for comparison to the later subtracted diffraction pattern), and the empty sample tube (for subtraction from the sample-in-tube pattern).The optimal sample to detector distance was unknown, so several different distances were tested in preliminary scans.1200 mm was chosen as the sample to detector distance for the final measurements based on wanting to maximize the number of diffraction rings that impinge on the detector area while also maintaining a clear separation between them.The CeO 2 standard's diffracted ring pattern at this energy and sample to detector distance had to be taken into account at the same time to ensure robust geometric calibration.A beam energy of 54 keV was required in conjunction with this distance to ensure detection of the majority of the diffraction peaks produced while maintaining clear peak separation.Broadening of diffraction peaks due to the length of the sample in the beam was a concern, which was why a large detector area and high beam energy and thus low diffraction angle are desirable, so as to minimize the relative width of peaks on the detector.Other sources of broadening, including imperfect x-ray monochromation, sample strain, and crystallite size-related broadening, were also considered.The I-12 beamline provides a significantly narrower bandwidth than typical laboratory diffractometer x-ray sources (~0.5 keV full width half maximum compared to ~2.35 keV for a Cu k-a 1 source (H€ olzer et al., 1997)).No significant lattice strain was expected as the sediment was loosely packed.Crystallite size-related broadening was possible but would also be anticipated with Martian regolith.Test measurements of the analog showed clear, narrow peaks in the selected experimental configuration.The experimental setup can be seen in Figure S2.
The beam profile was a 0.5 9 0.5 mm square, and a 30 s exposure time was used.The sample analog and sample tube were continuously rotated about the vertical (long) axis at a rate of 10°s À1 for bulk diffraction measurement.Twelve measurements were taken for each sample and tube configuration, through different locations on the side of the tube.The points on the tube were all located on side of the tube's center in a 4 9 3 grid (horizontal and vertical, respectively) with 1 mm grid spacing, spanning from the center of the tube to 1.5 mm away from the right inner edge of the tube wall.The volume of tube interior sampled per measurement ranged from 0.623 cm 3 at the center to 0.292 cm 3 at the furthest point horizontally.
The effects of crystallite and grain size on phase identification and quantification were also taken into consideration.Crystallite size is not equal to grain size, as a grain may be comprised of multiple smaller crystallites.Basalts typically have crystallite sizes of a few hundred microns or less.Olivine and augite commonly form larger crystals in basalt (Blatt & Tracy, 1996).
Counting statistics are a possible issue, since if a mineral phase is comprised of only a few large grains, only a "spotty" diffraction pattern will form on the detector instead of full Debye rings, decreasing the signal to noise ratio and increasing the risk of mineral quantification error due to intensity clipping at individual spots.The synchrotron experiment measured a much larger sample volume than a conventional diffractometer, however, as conventional measurements typically measure <0.1 cm 3 for rock samples.Some "spottiness" is apparent on detector images for the synchrotron experiments.The "spottiness" of rings can, however, be used to estimate crystallite size in two-dimensional (2-D) diffraction, depending on experimental parameters (Bramble et al., 2015), though this was not attempted in this work.
X-ray microabsorption will have a negligible effect on phase quantification despite the comparatively large grain size due to the high x-ray energy used.Brindley's criterion (Brindley, 1945) was calculated to determine this.For this estimation, a typical basaltic composition was assumed for the largest grains present in the sample analog using l 9 D, where l is the linear attenuation coefficient and D is the particle diameter, and found to be ~0.004.This is in the regime Brindley described as having negligible effect from microabsorption.
Optimization of the diffraction geometry and the measurement of the beam energy were accomplished using a CeO 2 calibration standard and the DAWN software package (Basham et al., 2015;Filik et al., 2017).DAWN was also used to convert the 2-D detector data into 1-D 2-theta diffraction patterns.
The absorbed radiation dose of a sample is an important figure for MSR science, as x-rays of sufficient intensity and energy can alter the sample's properties, especially organic chemistry which is of specific interest to Mars science.An upper limit for the x-ray dosage absorbed by the sample was therefore calculated.No measurement of the fraction of x-ray absorption was taken so it was assumed (unrealistically) that all x-rays were absorbed.It was found that one of the measurements described in this paper would transfer a maximum of ~6.3 9 10 À3 J to the sample, discounting absorption by the sample tube.If these measurements were performed in a vertical line scan along the center of a typical 60 mm long, 12.5 g returned Mars sample, ~0.76 J maximum would be transferred in total, equivalent to a maximum dose of ~61 Gy.If all this energy was absorbed as heat, assuming a lower end heat capacity for a basalt of 400 J kg À1 K À1 (Xiaoqing et al., 2018), this would result in a sample temperature increase of 0.15 K.This upper limit is much lower than required to avoid effects on amino acid content (Friedrich et al., 2016).
In order to determine the modal mineralogy of the sample analog, a reference powder diffraction measurement was taken on a conventional Bruker D8 Advance lab diffractometer at the University of Leicester.This used a Cu K-a x-ray source and Bragg-Brentano diffraction geometry for XRD of ground powder of the regolith analog sediment.4417 angular points were measured, with a 30 s exposure time.The basaltic sand was powdered in a ball mill for this measurement.

Phase Identification and Quantification
Before analysis of the diffraction data, the empty titanium tube diffraction spectra were first scaled in intensity to match their corresponding sample-in-tube spectra, and then subtracted from the sample spectra in order to try to remove Ti diffraction peaks and restore any overlapping sample peaks.
Phase identification and preliminary quantification were carried out using the QualX 2.0 program (Altomare et al., 2015).QualX uses a stick-peak analysis method where the peaks' intensities and positions are used to search-match the pattern to a crystallographic database and perform semiquantitative analysis to estimate the relative abundances of the identified phases.The Crystallography Open Database (COD) (Downs & Hall-Wallace, 2003;Gra zulis et al., 2009Gra zulis et al., , 2012Gra zulis et al., , 2015;;Merkys et al., 2016Merkys et al., , 2023;;Quir os et al., 2018;Vaitkus et al., 2021) was used for this.Match accuracy was determined by Figure of Merit (FoM), calculated from the weighted differences of intensity and 2h between experimental and database peaks.FoM can range from 0 to 1, with 1 indicating a perfect match with no differences.The weights are user-determined and range from 0 to 1.For these experiments, the intensity and 2h weights were set equally at 0.5, favoring neither.The "number of phases" weight, which influences how many different phases are expected and thus whether the algorithm tries to fit the pattern with a single phase or multiple, was set at 0.25, favoring the matching toward multiple phases.The final phase selection from the ranked list of matches was done by plausibility of the presence of the mineral and FoM.
QualX has limitations in how it carries out phase identification and quantification on a diffraction pattern that became apparent during analysis.It often identified multiple similar members of a single solid solution series with similar FoM.This is a limitation of the searchmatch method used by QualX, as in reality, the basaltic sediment likely contains a continuous spread of members with slightly different cation compositions, but the search-match algorithm is preferential toward accounting for a set of diffraction peaks using a single chemical formula that best fits it instead of a mixture of many slight variations.The entries in the COD are also limited by what has been measured and uploaded, and do not account for all possible members of solid solution series or other minerals with various amounts of impurities.This also sometimes resulted in chemical formulae with improbable elemental compositions being identified, as they were the best fit for the spread of solution members.When this occurred, the highest FoM phase was thus selected as the representative phase of the group of similar minerals in this analysis, and small amounts of improbable impurities were ignored when selecting phases.The exact elemental compositions that QualX identified were therefore not regarded as accurate and instead taken as an identification of that solid solution series or mineral group in the measurement.
Rietveld refinement was carried out on select spectra to more accurately determine relative mineral abundances using the QualX quantities as an initial guess, more precisely identify mineral chemistry through unit-cell parameters, and confirm the accuracy of the QualX phase identifications.Spectra were selected for refinement to confirm the presence or absence of the spread of phases identified through QualX in the 12 measurements taken of the sample analog.Rietveld refinement refines a theoretical diffraction pattern compared to the measured pattern using a least squares approach (Rietveld, 1969).Refinement was done using the MAUD refinement program (Lutterotti et al., 1999;Lutterotti & Bortolotti, 2003).The weighted profile R-factor, R wp , was used to evaluate the quality of the results, along with visual analysis of the fit.R wp is calculated by dividing the square root of the residuals between the fit and the data by the weighted intensities, and ranges from 0 to 1, with 1 representing a perfect fit of the model to the experimental data (Toby, 2006).
While QualX cannot reliably identify the precise mineral chemistry, this can be estimated for some minerals using mathematical relationships between composition and unit-cell parameters developed by Morrison et al. (2018).As Rietveld refinement estimates these parameters, compositions of solid-solution series members were calculated where possible.
While it is possible to quantify the amorphous phase fraction via XRD, this requires adding reference standards to the sample (Rampe, Bristow, et al., 2020).This work therefore concerns itself with only the crystalline fraction of the basaltic sand sample.

QualX Analysis
QualX was used to identify the mineral phases from the conventional lab diffractometer pattern and semiquantitatively estimate their relative abundances to use as a starting parameter for Rietveld refinement.The identified phases were andesine at approximately 45 wt %, diopside at approximately 30 wt%, and forsterite at approximately 25 wt%.

MAUD Analysis
Rietveld refinement using the MAUD program was carried out to more accurately determine the quantity of the major mineral constituents identified as well as the unit-cell parameters (see Figure 5a).The fit for this combination of minerals achieved an R wp of 1.7, with visual inspection supporting a closely matching fit.The measured relative quantities were 48.2% diopside, 43.3 AE 1.4 wt% andesine, and 8.6 AE 0.3% forsterite.MAUD calculates estimated standard deviations of quantity relative to a standard phase which is not refined and thus cannot calculate an uncertainty for this phase.This is why one phase quantity in each refinement has no stated uncertainty.

QualX Analysis
Identified minerals are grouped by name instead of specific formula due to the inaccuracies discussed in "Phase Identification and Quantification" Section.An example analyzed spectrum of measurement 316 can be seen in Figure 4.A summary of identified minerals in the 12 measurements taken in this experiment can be seen in Table 1.Although there was a fair amount of variation in identified minerals between the 12 measurements, conclusions can be drawn on what the likely bulk mineralogy of the basaltic sand sample is.Plagioclase feldspar was identified in every measurement, though solid solution chemistry varied.The most common was anorthite, with occasional detections of labradorite, andesine, and albite.In four measurements, a second member of the plagioclase series was detected concurrently with anorthite, including labradorite and albite.Diopside was detected in all 12 measurements.The olivine forsterite was identified in 10 of 12 measurements.The measurements without it instead had muscovite or microcline.Both minerals were only identified once each, however, though with moderately higher FoM than forsterite's mean FoM (0.754 and 0.772 compared to 0.725, respectively).Iron-rich minerals were identified in four measurements: pyrite was identified in three and hematite once.A small fraction of compounds with high atomic number elements would match the highly attenuating (bright) grains present in x-ray CT data in Figure 3.
It was concluded that pyroxene, plagioclase feldspar, and olivine are present in the sample.Hematite is known to be present in the analog but was only detected once, suggesting it is near the limit of detection for this identification method.Microcline, muscovite, and w€ ustite are unlikely to be present in the sample and are thus likely false detections.The main three constituent minerals add up to 97%, so ~3% iron mineral content is plausible.

MAUD Analysis
Rietveld refinement using the MAUD program was carried out on measurement 311 to confirm the presence and quantity of the major mineral constituents identified, as only those three minerals were identified using QualX for this measurement.The fit for this combination of minerals achieved an R wp of 2.9, with visual inspection supporting a closely matching fit.The measured relative quantities were 41.1 wt% labradorite, 44.4 AE 2.9 wt% diopside, and 14.6 AE 1.4 wt% forsterite.The minerals' compositions were then estimated using the unit-cell parameters.The plagioclase phase formula was calculated to have 0.47 Ca and 0.53 Na atoms per unit cell (andesine composition).The olivine was found to be Fo 72 (chrysolite).The system of equations for the composition for pyroxene did not come to a solution for this measurement or any of the others.
A refinement of measurement 316 was also carried out in order to try to verify the presence and quantity of pyrite.Refinement was of poor quality when using the anorthite species QualX identified for this measurement but was much improved when the labradorite species from measurement 311 was used instead.The fit of the pattern can be seen in Figure 5b; with an R wp of 2.9 again.Visual inspection confirmed the fit to still be quite closely matching the data.Rietveld refinement found no pyrite to be present.The plagioclase phase formula was calculated to have a somewhat different composition with 0.64 Ca and 0.36 Na atoms per unit cell (labradorite composition).The olivine was found to also have a slightly different elemental ratio, with Fo 77 (chrysolite composition).

QualX Analysis
As seen in Table 2, diopside and anorthite were identified in every measurement, with comparatively high FoM.Forsterite and pyrite were identified in seven measurements at moderately lower FoM.Albite, labradorite, and andesine were identified in conjunction with anorthite in several measurements with FoMs around 0.76.Iron-rich minerals were identified in nine of the 12 patterns; magnetite was identified in two at 6.5 wt % on average with a high FoM of 0.78, the same as anorthite's, which are the highest mean FoMs of any mineral for this series of measurements.This was not found in the PEEK patterns.Microcline feldspar was identified once, in conjunction with the plagioclase  feldspar anorthite.Enstatite was also identified once, in conjunction with another pyroxene, diopside.
It was concluded that pyroxene, plagioclase feldspar, and olivine are present in the sample, while the other sporadically identified minerals are thought to be less likely, though plausible in low quantities.

MAUD Analysis
Rietveld refinement was carried out on measurement 238 (seen in Figure 6) to verify and quantify the plagioclase, pyroxene, and olivine phases, as only those three minerals were identified using QualX for this measurement.The empty tube analog diffraction spectrum was then scaled to the same intensity using an average of the heights of several of the Ti peaks where they had little overlap from sample diffraction.This spectrum was subtracted from measurement 238 to try to remove the Ti diffraction signal and was then also refined (Figure 6b).A good fit was achieved for this combination of minerals, with an R wp of 3.2, backed up by visual inspection of the fit in MAUD.The measured relative abundances were 36.1 wt% anorthite, 46.4 AE 5.6 wt% diopside, and 17.5 AE 4.2 wt% forsterite for this refinement.
The plagioclase phase formula was calculated to have 0.56 Ca atoms and 0.44 Na (labradorite composition).The olivine was found to be Fo 80 (chrysolite composition).
A refinement of measurement 245 was also carried out in order to try to verify the presence and quantity of albite and pyrite.A good fit was achieved, with an R wp of 2.9.Visual inspection confirmed the quality of the fit again.Anorthite quantity was measured at 14 wt%, diopside at 47 AE 52 wt%, albite at 26 AE 41 wt%, forsterite at 14.5 AE 7.1 wt%, and pyrite at 0.09 AE 0.61 wt%.The plagioclase phase formula again gave an unphysical result for the same reason as measurement 316.The olivine was found to be Fo 86 (chrysolite composition).Figure 7 shows the MAUD refinement of this measurement with the contributions of the individual phases.

DISCUSSION
It should be noted that uncertainties calculated by Rietveld refinement generally tend to be underestimated (Prince, 1993).The estimated standard deviations only measure the precision of the quantification and do not account for systematic errors.
It is possible that a significant fraction or even the majority of the iron-rich minerals known to be present in the sample are amorphous and thus undetectable, as measured in (Pandey et al., 2023) similar Icelandic basalts.
It was observed that the difference in tube analog wall thickness from the flight tubes only had a small effect on the size of the 2h regions affected by the Ti peaks.The thinner walls reduced peak width by approximately 24%, or 0.01°of the full width half maximum, but the doubling of the peaks due to the separation of the walls in the beam path still predominated (see Figure 8) and so this effect was not significant.
The titanium tube QualX analysis had the supposed pyrite peaks again overlapping with other phases, so the identification is less reliable than the FoM would indicate.The three reliably identified minerals' mean quantities add up to very close to 100%, but with the uncertainties, this could allow for a few percent of other materials, such as the iron-rich minerals, which again would explain the high attenuation grains visible in CT in Figure 3.
As can be seen in the difference plot in the bottom of Figure 6b, the titanium wall peak subtraction was suboptimal.After the peak subtraction, parts of the affected areas were negative in some measurements despite the intensity of the empty sample tube measurements having been scaled to match the tube-with-sample measurements.The regions of the subtractions and a small buffer around them were thus excluded in the Ti tube MAUD analyses (311 and 316) due to provoking worse fitting around the regions and having a large impact on residuals, making them inaccurate.The diffraction peaks caused by the tube walls can be more clearly seen in the empty tube reference measurement in Figure 8.
In the refinement of measurement 245 of the Ti tube measurements, it is unlikely for albite and anorthite to coexist, and a single member of the plagioclase solidsolution series is instead expected.Despite the large errors calculated in measurement 245, the quantities of the two titanium tube refinements had differences of <~1%, which makes the errors suspect.It is currently unknown why these large errors were calculated; it is suspected to be caused by parameters which had not fully converged by the end of the refinement cycles, despite main parameters such as phase abundance and background already having converged.Rietveld refinement errors are uncertainties of the fitting process and require for both the model to be correct and the refinement to be fully minimized to be accurate (Saville et al., 2021).
Significant differences in relative quantity of the three main mineral phases were found when comparing both synchrotron measurements to the reference measurement, as can be seen in Table 3.The mean difference was 5.4%, with a maximum detected difference of 8.4%.The cause of this discrepancy is currently unknown, though it is suspected to be a variation in crystallite size between mineral phases in the unpowdered basaltic sand.As the differences between the PEEK and titanium tube measurements are significantly smaller, the discrepancy is unlikely to be due to the titanium tube and instead caused by the sample material.The Rietveld refinements of the two experiments quantifying the three main identified minerals, 238 and 316, were in good agreement.The differences in quantity of diopside, anorthite, and forsterite can be seen in Table 4.The difference in diopside quantity was within the smaller MAUD-calculated error of the two.Forsterite was just over its calculated error.The smaller errors from measurement 238 are shown, as the PEEK refinement had much larger errors suspected to be caused by the lack of full convergence in the refinement mentioned previously.

CONCLUSIONS AND FUTURE WORK
We have demonstrated that monochromatic synchrotron powder diffraction is capable of useful mineral phase identification and quantification for returned MSR samples inside their sample tubes.While contamination and planetary protection control measures would need to be taken to carry out synchrotron XRD in the pre-basic characterization phase, these would likely not need to be overly complex due to the straightforward nature of the experimental method, only requiring an isolator box with a pair of simple windows for the x-rays and perhaps a motorized sample stage inside the box.
While peak broadening caused by the size of the sample in the beam path was a concern, we have shown that this did not cause sufficient overlap between sample diffraction peaks to prevent successful phase identification and quantification for the common Martian basalts, and likely does not for many of the other current and anticipated samples' mineralogies.
The titanium walls of the MSR sample tube analog give strong but few diffraction peaks (see Figure 8) which do not interfere enough with a typical Martian basaltic mineralogy's diffraction pattern enough to make it unidentifiable.Subtraction of peaks using diffraction data from an empty tube served to prevent the detection of titanium alloy in the initial phase identification, but the regions encompassing the Ti peaks had to be excluded for Rietveld refinement anyway to optimize fitting.The measurements of the empty tube were useful in identifying the size and position of the regions, which changed depending on what part of the tube the beam was penetrating due to the differences in separation of the tube walls.Despite small differences in the dimensions of the analog compared to the flight tubes, a very similar impact due to the walls is expected for the sealed Mars sample tubes currently anticipated to arrive on Earth in 2032-2033.
While measuring through the titanium sample tube analog did not have a large impact on the accuracy of phase quantification, it is likely that the sample being in a natural, unpowdered state did, as evidenced by the up to 8.4 wt% differences in relative quantity found when comparing both of the synchrotron experiments' data to the reference pattern.In the future, a repeat of the synchrotron measurements with powdered sample may aid in understanding the source of the phase quantity discrepancies, whether it is crystallite sizes varying with mineralogy or something else.It is important to understand the sources of uncertainty in measurements of returned Mars samples before they arrive.Investigation of QualX analysis with different weights for phase matching would also be useful.

FIGURE 1 .
FIGURE 1. Mars2020 sample tube.Image courtesy NASA/ JPL-Caltech.The tubes are ~13.9cm long.(Color figure can be viewed at wileyonlinelibrary.com) FIGURE 3. Horizontal slice of an x-ray computed tomographic (CT) scan of the Icelandic basaltic sand sample inside an analog perseverance sample tube.A 100 kV x-ray beam with 7.5 W of power was used with a custom Nikon XT H 320 scanner.At least three distinct mineral phases are visible in the grains based on differences in grayscale.The bright spots are composed of material enriched in heavier elements, here iron-rich compounds.(Color figure can be viewed at wileyonlinelibrary.com) FIGURE 4. QualX phase identification and quantification of diffraction measurement 316, of the basaltic sand sample in a PEEK polymer tube with the same inner diameter as the perseverance sample tubes.All peaks detected in the diffraction pattern are shown by the black lines.The peaks of the identified diopside species are displayed in orange.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 5 .
FIGURE 5. (a) The MAUD Rietveld refinement of the XRD pattern of crushed powder of the regolith analog taken using conventional lab diffractometer.Data points are black circles.The fit (red) has a weighted percentile residual of 1.7.The pattern is visibly noisier than the synchrotron measurements, though peak broadening is somewhat less.(b) The refinement of diffraction measurement 316, of the basaltic sand sample in a PEEK polymer tube.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 6 .
FIGURE 6. MAUD Rietveld refinements of diffraction measurement 238, of the basaltic sand sample in the Perseverance sample tube analog.In both figures, the regions around the tube wall peaks had to be ignored for the refinement.(a) The unprocessed spectrum.(b) The same pattern with the titanium diffraction peaks caused by the tube walls subtracted using a pattern from an empty tube analog.The fitting differences in the difference plot at approximately 5.2°, 6°, 8.5°, and just below 10°are due to remnants of the subtracted tube wall peaks.(Color figure can be viewed at wileyonlinelibrary.com) FIGURE 8. MAUD Rietveld refinement of the diffraction pattern of the empty titanium sample tube analog.The doubling of the peaks caused by diffraction from two walls is clearly visible, and broadens the 2-theta regions affected by the sample tube signal which interferes with analysis of the sample inside.(Color figure can be viewed at wileyonlinelibrary.com)

TABLE 1 .
Summary of minerals identified from the 12 diffraction measurements taken of the basaltic sand in a non-diffracting PEEK tube using QualX.Mean quantity for plagioclase in each measurement is 44.1 AE 7.3 wt%.The second column indicates in how many of the 12 diffraction spectra of the sample taken each mineral was identified to be present by QualX.Mean abundance is the average of the abundances in those measurements where the phase was detected.Standard deviations were not calculated when minerals had too few occurrences for them to be statistically meaningful. Note:

TABLE 2 .
Summary of minerals identified from the 12 diffraction measurements taken of the basaltic sand inside the sample tube analog using QualX.Mean quantity for plagioclase in each measurement is 44.1 AE 10.1 wt%.The second column indicates in how many of the 12 diffraction spectra of the sample taken each mineral was identified to be present by QualX.Mean abundance is the average of the abundances in those measurements where the phase was detected.Standard deviations were not calculated when minerals had too few occurrences for them to be statistically meaningful. Note:

TABLE 3 .
Quantity differences between MSR analog XRD measurements and the reference measurement.Note: The larger of the errors from the two compared refinements are shown.

TABLE 4 .
Differences in mean quantity of the main three identified minerals with (meas.311) and without sample tube analog (238) from MAUD Rietveld refinement.