K and Cl Concentrations on the Martian Surface Determined by the Mars Odyssey Gamma Ray Spectrometer: Implications for Bulk Halogen Abundances in Mars G

GEOPHYSICAL RESEARCH LETTERS, VOL. 37, L12204, doi:10.1029/2010GL043528, 2010 Click Here for Full Article

K and Cl concentrations on the Martian surface determined by the Mars Odyssey Gamma Ray Spectrometer: Implications for bulk halogen abundances in Mars G. Jeffrey Taylor,1 William V. Boynton,2 Scott M. McLennan,3 and Linda M. V. Martel1 Received 12 April 2010; revised 17 May 2010; accepted 21 May 2010; published 30 June 2010.

[1] Orbital gamma ray spectrometry shows that the Martian of the Martian surface. As described below, although K and surface has a mean Cl/K ratio of 1.3 ± 0.2, indistinguishable Cl are not correlated, the average Cl/K is 1.3, close to the CI from the ratio in CI chondrites (1.28). Although Cl and K chondrite value of 1.28 [Palme and Jones, 2004], but very fractionate by magma degassing and aqueous processing, different from the Wänke‐Dreibus value of 0.08 (Table 1). during igneous partial melting both elements are highly We argue here that bulk silicate Mars has a CI chondrite Cl/K incompatible. Thus, if the surface Cl/K reflects the bulk ratio and that the low Cl value in the Wänke‐Dreibus model crustal value, then the mantle, hence primitive silicate is caused by Martian meteorites being depleted in Cl Mars, also has a roughly CI ratio. Martian meteorite data because of loss of this volatile during eruption or near‐ indicate that Cl/Br is also approximately chondritic, surface intrusion [also see McSween et al., 2001]. A roughly suggesting that elements that condensed in the nebula chondritic Cl/Br in Martian meteorites suggests that Cl/Br is between ∼1000 K (K and Cl) to ∼500 K (Br) are also chondritic and the small difference in the condensation uniformly depleted in Mars at about 0.6 × CI chondrite temperatures of Br and I suggests that all three halogens are concentrations. Mars clearly does not contain 0.6 × CI in chondritic proportions. levels of H2O, which would be ∼6 wt%, indicating that Mars was constructed by planetesimals rich in volatile 2. GRS Methods elements, but not in water. Citation: Taylor, G. J., W. V. Boynton, S. M. McLennan, and L. M. V. Martel (2010), K and Cl concentra- [4] The GRS determines elemental concentrations from tions on the Martian surface determined by the Mars Odyssey the intensities of gamma rays produced by neutron capture Gamma Ray Spectrometer: Implications for bulk halogen abun- and/or scattering reactions (Si, Fe, Ca, Al, Cl, H) or by dances in Mars, Geophys. Res. Lett., 37, L12204, doi:10.1029/ radioactive decay (K, Th). Details of data acquisition and 2010GL043528. reduction are given by Boynton et al. [2007, 2008]. The data we use here for Cl and K were obtained between June 8, 2002 and April 2, 2005. The GRS spatial resolution is 1. Introduction defined as the nadir‐centered region within which ∼50% of ∼ ∼ [2] The best estimate for the bulk composition of the the signal originates, which is 3.7° arc radius ( 220 km), silicate portion of Mars is that developed by Wänke and although this varies somewhat with the gamma ray energy. Dreibus in a series of papers [Dreibus and Wänke, 1984, The data were binned initially at 0.5° × 0.5°, smoothed 1987; Wänke and Dreibus, 1988, 1994]. This robust esti- using a boxcar filter over a much larger radius (5° for K, 10° mate is based on reasonable cosmochemical assumptions for Cl) and then rebinned to 5° × 5° grid points. The mea- and the compositions of Martian meteorites. However, there sured gamma rays are produced in the upper few tens of is one curious discrepancy. The estimated abundance of Cl centimeters of the Martian surface, much deeper than is much lower than other elements with similar condensation sensed by visible to thermal infrared spectral instruments temperatures, such as K (50% condensation temperature of (10–100 mm). We use only points in regions where H contents 1006 K cf 948 K for Cl). The standard Wänke‐Dreibus are low enough not to interfere in the determination of Cl model composition for moderately volatile and volatile concentrations. Hydrogen has a high cross‐section for cap- lithophile elements are given in Table 1, along with their 50% turing thermal neutrons, significantly affecting the neutron condensation temperatures [Lodders, 2003], and listed in flux in the upper ∼30 cm of the Martian surface. The GRS order of decreasing condensation temperature. The low Cl data are corrected for this effect, but the correction is carries over to the values for the other halogens, except for F. uncertain at polar latitudes where H dominates elemental [3] The Mars Odyssey Gamma Ray Spectrometer (GRS) signatures. Accordingly, we constrained our results using a has returned data that allow us to determine the K, Cl, and mask based on H concentration, corresponding to roughly other elemental concentrations in the upper few decimeters ±45° of latitude from the equator. (The concentration of H does not affect K data because its g‐rays result from radioactive decay, but we use data only within the H mask to 1Hawaii Institute of Geophysics and Planetology, University of Hawaii at Manoa, Honolulu, Hawaii, USA. compare with Cl concentrations.) 2Lunar and Planetary Laboratory, University of Arizona, Tucson, Arizona, USA. 3. Overview of K and Halogen Geochemical 3Department of Geosciences, State University of New York at Stony Brook, Stony Brook, New York, USA. Behavior [5] Potassium, Cl, Br, and I are all large‐ion lithophile Copyright 2010 by the American Geophysical Union. elements that behave incompatibly during partial melting, 0094‐8276/10/2010GL043528

L12204 1of5 L12204 TAYLOR ET AL.: MARS BULK COMPOSITION L12204

Table 1. Standard Wänke‐Dreibus Model Mars Bulk Composi- ferent behavior after partial melting, as summarized above. tion Compared to Model Described in This Paper Nevertheless, the mean Cl/K value, 1.27 ± 0.46 (2‐sigma uncertainty), is within the uncertainties of the CI‐chondritic 50% Condensation value, 1.28, and substantially greater than the mean value in Temperaturea Wänke‐Dreibus Wänke‐Dreibusb This Work This Work Martian meteorites, 0.08, determined from data assembled (K) (ppm) (Mars/CI) (ppm) (Mars/CI) by Meyer [2009]. Figure 2 shows the Cl/K ratio divided into K 1006 305 0.56 305 0.56 six groups. One group (largest area in Figure 2) is defined Na 958 3700 0.74 3700 0.74 by the chondritic ratio plus and minus one standard devia- Cl 948 38 0.054 390 0.56 tion of the global dataset (0.2). The other ranges are one Rb 800 1.06 0.46 1.06 0.46 standard deviation wide. Figure 2 shows that the entire Cs 799 0.07 0.37 0.07 0.37 F 734 32 0.55 32 0.55 surface of Mars has a substantially higher Cl/K ratio than Br 546 0.145 0.041 2.2 0.60 reported for bulk Martian meteorites. The ratio is every- I 535 0.032 0.074 0.26 0.60 where greater than 0.6 and a substantial portion of the surface

a is within one standard deviation of the chondritic ratio. Cl/K Lodders [2003]. bDreibus and Wänke [1984, 1987] and Wänke and Dreibus [1988, 1994]. is significantly higher than the chondritic range over Tharsis, CI chondrite values from Palme and Jones [2004]. west of Tharsis, and in Elysium, and lower in Utopia Planitia. Cl enrichment in volcanic regions could be caused by fairly hence preserving the mantle source rock Cl/K in the magma recent volcanic deposits, such as the high‐Cl materials in produced [Schilling et al., 1980; Deruelle et al., 1992; Pyle Medusae Fossae [Keller et al., 2006]. and Mather, 2009]. The fluorine ion has a much smaller ionic radius than the other halides and tends to behave 5. Cl‐Br Relationships somewhat less incompatibly during partial melting [Pyle and Mather, 2009]. Thus, Cl/K (and Br/K and I/K) in sur- [9] Cl/Br in Martian meteorites (180) is close to the CI face rocks would give a good approximation to the mantle chondrite ratio (199) (meteorite data cited by Brückner et al. values if the halogens were not lost by degassing during [2008], Banin et al. [1992], and Dreibus et al. [1992, 1994, magma ascent and eruption. Note that K is not lost via 1996, 2003a, 2003b]. If reflective of the interior, it implies magma degassing. that Br, like Cl, is much higher in Mars than the estimate [6] Loss of halogens during magma degassing depends on made by Wänke and Dreibus. This assumes, of course, Cl magma composition and H2O concentration (see summary and Br are lost from magmas by the same amount during by Aiuppa et al. [2009]), but fractionation relative to each degassing. (Rocks analyzed in Gusev crater do not appear to other is probably less affected and is related to their ionic provide additional data on Cl and Br in igneous rocks. radii. Experiments by Bureau et al. [2000] show that fluid‐ According to Gellert et al. [2006], even the abraded surfaces magma partition coefficients for the halogens increase with of these rocks have high S and Cl, suggesting either con- ionic radius: F (Df/m = 0.1, ionic radius = 1.33 Å), Cl (4.5, tamination by debris generated during grinding or that the 1.81 Å), Br (14, 1.96 Å), I (78, 2.2 Å). (Ionic radii are for rocks have experienced aqueous alteration. A compilation of 6‐fold coordination and taken from Shannon [1976]; data all abraded rocks in Gusev shows Cl up to almost 2 wt%.) for F are from Webster [1992].) On the other hand, F is retained by lavas, so its concentration in volcanic rocks 6. Revised Halogens in Bulk Silicate Mars more accurately reflects its abundance in the mantle. Thus, the behavior of F in magmatic systems leads to a more [10] A central issue is whether the mean Cl/K value accurate estimate of its concentration in bulk silicate Mars determined for the upper few decimeters reflects the bulk (Table 1). composition of the planet. Taylor et al. [2006] addressed [7] Once on the surface, K and the halogens behave quite differently. K is retained in feldspars and residual igneous glass until sufficient weathering takes place to begin to dissolve those phases, whereas the Cl, Br, and I are present in highly soluble salts. Thus, surface aqueous processes continue to fractionate K from Cl and the other large‐ion halogens. The geochemistry of halogens on the Martian surface is complicated [e.g., Marion et al., 2009] and although Cl and Br behave similarly, they can still fractionate significantly from each other [e.g., Clark et al., 2005; Haskin et al., 2005]. Extracting the bulk crustal Cl, Br, and I from surface analyses is not yet possible with existing data from the Mars Exploration Rovers, so below we use Cl/Br and Cl/I ratios in Martian meteorites to evaluate Br and I in the mantle. This approach assumes that we can ignore the differences in fluid/magma partition coefficients. Figure 1. Mars Odyssey Cl and K concentrations for 5 × 5 degree grid points. Filled square represents the mean of the GRS data. Solid line is the CI chondritic ratio. Martian me- 4. Results: K and Cl Concentrations teorites plot along the bottom of the diagram and have sub- [8] GRS data show that Cl and K are not correlated stantially lower Cl/K (mean of 0.08), reflecting loss by (Figure 1), nor should they be considering their vastly dif- degassing of magmas at low pressure.

2of5 L12204 TAYLOR ET AL.: MARS BULK COMPOSITION L12204

Figure 2. Map showing the distribution of increments in the Cl/K ratio on Mars, inside the H‐mask (see text). The largest region represents the CI chondritic ratio ± one standard deviation of the population. (The standard deviation is 0.2.) Other increments are one standard deviation wide. The large area occupied by the Cl/K values within one standard deviation of the CI chondritic value suggests that Cl/K at the Martian surface is close to chondritic. this issue with respect to K and Th. Two factors argue that melt inclusions in some nakhlites and chassinites. Most the surface reflects the mean crustal composition. First, the notable is MIL 03346, which contains amphiboles with up Martian crust was bombarded continuously during most of to 8 wt% Cl [Sautter et al., 2006]. In addition, melt inclu- its formation. This not only mixed previously‐formed sur- sions in the Chassigny dunite contain Ti‐rich biotite with face and subsurface rocks, but also formed a regolith that 0.4 wt% Cl [Johnson et al., 1991]. Using a biotite/melt could be reworked continuously as the crust grew. Second, distribution coefficient of 1.5 for Cl [Icenhower and isotopic data for Martian meteorites indicate that the crust London, 1997], Filiberto and Treiman [2009] calculate a has not been recycled into the mantle [Borg et al., 2003]. Cl concentration of 0.3 wt% in the magma from which the Thus, what went into the crust as it was constructed by biotite crystallized. Other investigators have reported evi- magmatism stayed in the crust. Of course, aqueous pro- dence for Cl‐rich fluids exsolving from magmas giving rise cesses redistributed materials and fractionated K from Cl, to the MIL 03346 nakhlite [McCubbin et al., 2009] and but these processes did not necessarily transport most of the Chassigny [McCubbin and Nekvasil, 2008]. While we can- crustal inventory of Cl into the uppermost surface layers. not pin down the total magmatic Cl or the K/Cl from these Thus, we believe it is reasonable to conclude that bulk Mars data, the meteorite evidence supports the idea that Martian has close to a chondritic Cl/K ratio. Using the Wänke‐ magmas contained much higher levels of Cl than bulk Dreibus value for K (305 ppm (Table 1)), we estimate a bulk meteorite analyses suggest. Cl abundance of 390 ppm (Table 1). [13] The correlation between Cl and Br in Martian me- [11] We can check the conclusion that Mars has chondritic teorites [Brückner et al., 2008] suggests that they are in Cl/K. Assuming that the bulk crust has a chondritic Cl/K chondritic proportions, though the mean Cl/Br (180) is ratio, we estimated the K and Cl abundances in bulk silicate slightly less than chondritic (199). As discussed in section 3, Mars. If the crust is 57 km thick [Wieczorek and Zuber, Br is lost more readily from degassing magmas than is Cl. 2004] and the mantle extends to a depth of 1760 km, the Thus, the roughly chondritic Cl/Br might have been lower crust occupies 4.6 wt% of silicate Mars. If that mass of crust than observed (because Br was higher), so use of the represents the average amount of partial melting of the chondritic ratio or the mean for the Martian meteorites is mantle over time and the distribution coefficients for K and conservative. If anything, the Br concentration might be Cl during mantle partial melting are 0.001, use of the higher compared to Cl and other moderately volatile standard equations for equilibrium partial melting indicates elements. that the current mean mantle concentration is 170 ppm K [14] We estimate iodine in bulk silicate Mars by assuming and 220 ppm Cl. With 4.6% partial melting and the crustal it has the same relationship to Cl as does Br. That is, if Cl/Br mass assumed, half the total inventory of K and Cl is in the is roughly chondritic, Cl/I probably is as well, simply crust and half is still in the mantle [also see Taylor et al., because the condensation temperatures of Br and I do not 2006]. Thus, the primitive mantle (same as bulk silicate differ significantly (Table 1). However, meteorite data are Mars) has 340 ppm K and 440 ppm Cl. These values are in insufficient to establish that firmly. accord with our revised estimate for bulk silicate Mars (Table 1). 7. Discussion [12] Martian meteorites provide evidence for magmas being much richer in Cl before eruption. Filiberto and [15] Our revised estimate of the abundances of moderately Treiman [2009] succinctly summarize the evidence. The volatile and volatile lithophile elements in bulk silicate Mars key evidence is the presence of Cl‐bearing amphibole inside is about 0.6 × CI chondrites (Table 1) and indicates that

3of5 L12204 TAYLOR ET AL.: MARS BULK COMPOSITION L12204 these elements are not fractionated from each other during tedbyJ.F.Bell,pp.103–124, doi:10.1017/CBO9780511536076.006, planetary accretion. This also applies to the abundances of Cambridge Univ. Press, Cambridge, U. K. Brückner, J., G. Dreibus, R. Gellery, S. W. Squyres, H. Wänke, A. Yen, volatile chalcophile elements, significant amounts of which and J. Zipfel (2008), Mars exploration rovers: chemical composition are in the core. Thus, assuming a roughly 0.6 × CI initial by APXS, in The Martian Surface: Composition, Mineralogy, and Phys- concentration of volatile chalcophile and siderophile ele- ical Properties,editedbyJ.F.Bell,pp.58–102, doi:10.1017/ ments could be a useful starting point for geochemical CBO9780511536076.005, Cambridge Univ. Press, Cambridge, U. K. Bureau, H., H. Keppler, and N. Metrich (2000), Volcanic degassing of bro- models of core formation. mine and iodine: Experimental fluid/melt partitioning data and applica- [16] The uniform 0.6 × CI fractionation pattern suggests tions to stratospheric chemistry, Earth Planet. Sci. Lett., 183,51–60, that Mars accreted substantially from planetesimals formed doi:10.1016/S0012-821X(00)00258-2. Chambers, J. E. (2004), Planetary accretion in the inner solar system, Earth on average at relatively low temperature, perhaps at greater Planet. Sci. Lett., 223, 241–252, doi:10.1016/j.epsl.2004.04.031. distances from the Sun than the present orbit of Mars, Clark, B. C., et al. (2005), Chemistry and mineralogy of outcrops at consistent with accretion models [e.g., Chambers, 2004; Meridiani Planum, Earth Planet. Sci. Lett., 240,73–94, doi:10.1016/ ’ j.epsl.2005.09.040. O Brien et al., 2006]. What does this imply about H2O Deruelle, B., G. Dreibus, and A. Jambon (1992), Iodine abundances in abundance in bulk Mars? CI chondrites have about 10 wt% oceanic basalts: Implications for earth dynamics, Earth Planet. Sci. Lett., – H2O contained in hydrous silicates, salts, and organic 108, 217 227, doi:10.1016/0012-821X(92)90024-P. compounds [Kaplan, 1971]. It is clearly unrealistic for the Dreibus, G., and H. Wänke (1984), Accretion of the Earth and the inner planets, in Proceedings of the 27th International Geological Congress, Martian bulk water content to be 0.6 of this level (6 wt%). vol. 11, pp. 1–20, VNU Sci., Utrecht, Netherlands. On the other hand, studies of Martian meteorites indicate Dreibus, G., and H. Wänke (1987), Volatiles on Earth and Mars: A com- that the water contents of Martian magmas [e.g., McSween parison, Icarus, 71, 225–240, doi:10.1016/0019-1035(87)90148-5. Dreibus, G., K. H. Jochum, H. Palme, B. Spettel, F. Wlotzka, and H. Wänke et al., 2001; McCubbin et al., 2010] are too high to be (1992), LEW 88516: A meteorite compositionally close to the “Martian consistent with Wänke and Dreibus’s [1988] estimate of mantle,”, Meteoritics, 27, 216. only 39 ppm in bulk silicate Mars. Their estimate is tied to Dreibus, G., A. Burghele, K. P. Jochum, B. Spettel, F. Wlotzka, and their Cl abundance. Using the same approach but with our H. Wänke (1994), Chemical and mineral composition of ALH 84001: A Martian orthopyroxenite, Meteoritics, 29, 461. higher Cl concentration, we estimate 390 ppm of H2Oin Dreibus, G., B. Spettel, F. Wlotzka, L. Schultz, H. W. Weber, K. P. Jochum, bulk Mars, still substantially less than 6 wt%, indicating and H. Wänke (1996), QUE94201: An unusual Martian basalt, Meteorit. decoupling of volatile lithophile elements and water. Planet. Sci., 31, A39. Dreibus, G., W. Huisl, B. Spettel, and R. Haubold (2003a), Comparison of McCubbin et al. [2010] estimate from hydrous amphibole in the chemistry of Y‐000593 and Y‐000749 with other nakhlites, Lunar melt inclusions in Chassigny that the mantle source region Planet. Sci., XXXIV, Abstract 1586. for the Chassigny magma contained between 130 and Dreibus, G., R. Haubold, W. Huisl, and B. Spettel (2003b), Comparison of the chemistry of Yamato 980459 with DaG 476 and SaU 005, in Interna- 250 ppm H2O, also substantially higher than estimated by tional Symposium—Evolution of Solar System Materials: A New Perspec- Wänke and Dreibus [1988]. It appears that except for a small tive from Antarctic Meteorites, pp. 19–200, Natl. Inst. Polar Res., Tokyo. percentage of planetesimals, the raw materials for Mars may Filiberto, J., and A. H. Treiman (2009), The effect of chlorine on the liquidus have accreted before temperatures dropped to the water of basalt: First results and implications for basalt genesis on Mars and Earth, Chem. Geol., 263,60–68, doi:10.1016/j.chemgeo.2008.08.025. condensation T, about 180 K. Alternatively, water may have Gellert, R., et al. (2006), Alpha particle X‐ray spectrometer (APXS): Results been lost preferentially by impact heating during accretion from Gusev crater and calibration report, J. Geophys. Res., 111, E02S05, [e.g., Bond et al., 2010], or it might have reacted with metallic doi:10.1029/2005JE002555. Haskin, L. A., et al. (2005), Water alteration of rocks and soils on Mars at iron during accretion, accounting for the high FeO in Mars the Spirit rover site in Gusev Crater, Nature, 436,66–69, doi:10.1038/ [Wänke and Dreibus, 1994]. nature03640. Icenhower, J. P., and D. London (1997), Partitioning of chlorine and fluorine between biotite and granitic melt: Experimental calibration at 200 MPa H O, Contrib. Mineral. Petrol., 127,17–29, doi:10.1007/ [17] Acknowledgments. This work was supported by NASA and the 2 Mars Odyssey project at JPL. G.J.T., L.M.V.M., and S.M.M. were also s004100050262. supported by grants from the NASA Mars Data Analysis Program. We Johnson, M. C., M. J. Rutherford, and P. C. Hess (1991), Chassigny petro- genesis—Melt compositions, intensive parameters, and water contents of thank two anonymous reviewers for their useful comments, which – improved the manuscript. We are grateful to the entire Odyssey GRS team Martian (questionable) magmas, Geochim. Cosmochim. Acta, 55, 349 for assistance with producing these unique data and for continuing discus- 366, doi:10.1016/0016-7037(91)90423-3. sions about their significance. Kaplan, I. R. (1971), Hydrogen (1), in Handbook of Elemental Abundances in Meteorites, Ser. Extraterr. Chem., vol. 1, edited by B. Mason, pp. 21– 27, Gordon and Breach, New York. Keller, J., et al. (2006), Global distribution of chlorine measured by Mars GRS, J. Geophys. Res., 111, E03S08, doi:10.1029/2006JE002679, References [printed 112(E3), 2007]. Aiuppa, A., D. R. Baker, and J. D. Webster (2009), Halogens in volcanic Lodders, K. (2003), Solar system abundances and condensation tempera- systems, Chem. Geol., 263,1–18, doi:10.1016/j.chemgeo.2008.10.005. tures of the elements, Astrophys. J., 591,1220–1247, doi:10.1086/ Banin, A., B. C. Clark, and H. Wänke (1992), Surface chemistry and min- 375492. eralogy, in Mars, edited by H. H. Kieffer et al., pp. 594–625, Univ. of Marion, G. M., D. C. Catling, and J. S. Kargel (2009), Br/Cl partitioning in Ariz. Press, Tucson. chloride minerals in the Burns formation on Mars, Icarus, 200, 436–445, Bond, J. C., D. S. Lauretta, and D. P. O’Brien (2010), Making the Earth: doi:10.1016/j.icarus.2008.12.004. Combining dynamics and chemistry in the solar system, Icarus, 205, McCubbin, F. M., and H. Nekvasil (2008), Maskelynite‐hosted apatite in 321–337, doi:10.1016/j.icarus.2009.07.037. the Chassigny meteorite: Insights into late‐stage magmatic volatile evo- Borg, L. E., L. E. Nyquist, H. Weissman, C.‐Y. Shih, and Y. Reese (2003), lution in Martian magmas, Am. Mineral., 93, 676–684, doi:10.2138/ The age of Dar al Gani 476 and the differentiation history of the Martian am.2008.2558. meteorites inferred from their radiogenic isotopic systematics, Geochim. McCubbin, F. M., N. J. Tosca, A. Smirnov, H. Nekvasil, A. Steele, M. Fries, Cosmochim. Acta, 67, 3519–3536, doi:10.1016/S0016-7037(03)00094-2. and D. H. Lindsley (2009), Hydrothermal jarosite and hematite in a Boynton, W. V., et al. (2007), Concentration of H, Si, Cl, K, Fe, and Th in pyroxene‐hosted melt inclustion in Martian meteorite Miller Range the low and mid latitude regions of Mars, J. Geophys. Res., 112, E12S99, (MIL) 03346: Implications for magmatic‐hydrothermal fluids on Mars, doi:10.1029/2007JE002887. Geochim. Cosmochim. Acta, 73, 4907–4917, doi:10.1016/j. Boynton, W. V., G. J. Taylor, S. Karunatillake, R. C. Reedy, and J. M. Keller gca.2009.05.031. (2008), Elemental abundances determined by Mars Odyssey GRS, in The McCubbin, F. M., A. Smirnov, H. Nekvasil, J. Wang, E. Hauri, and D. H. Martian Surface: Composition, Mineralogy, and Physical Properties, edi- Lindsley (2010), Hydrous magmatism on Mars: A source of water for the

4of5 L12204 TAYLOR ET AL.: MARS BULK COMPOSITION L12204

surface and subsurface during the Amazonian, Earth Planet. Sci. Lett., Sect. A Cryst. Phys. Diffr. Theor. Gen. Crystallogr., 32, 751–767, 292, 132–138, doi:10.1016/j.epsl.2010.01.028. doi:10.1107/S0567739476001551. McSween, H. Y., Jr., T. L. Grove, R. C. Lentz, J. C. Dann, A. H. Holsheid, Taylor, G. J., et al. (2006), Bulk composition and early differentiation of L. R. Riciputi, and J. G. Ryan (2001), Geochemical evidence for mag- Mars, J. Geophys. Res., 111, E03S10, doi:10.1029/2005JE002645, matic water within Mars from pyroxenes in the Shergotty meteorite, [printed 112(E3), 2007]. Nature, 409, 487–490, doi:10.1038/35054011. Wänke, H., and G. Dreibus (1988), Chemical composition and accretion Meyer, C., Jr. (2009), The Mars meteorite compendium, NASA Johnson history of the terrestrial planets, Philos. Trans. R. Soc. London, Ser. A, Space Cent., Houston, Tex. (Available at http://curator.jsc.nasa.gov/ 325, 545–557, doi:10.1098/rsta.1988.0067. antmet/mmc/index.cfm). Wänke, H., and G. Dreibus (1994), Chemistry and accretion history of Mars, O’Brien, D. P., A. Morbidelli, and H. F. Levison (2006), Terrestrial planet Philos.Trans.R.Soc.London,Ser.A, 349,285–293, doi:10.1098/ formationwithstrongdynamicalfriction,Icarus, 184,39–58, rsta.1994.0132. doi:10.1016/j.icarus.2006.04.005. Webster, J. D. (1992), Water solubility and chlorine partitioning in Cl‐rich Palme, H., and A. Jones (2004), Solar system abundances of the elements, granitic systems: Effects of melt composition at 2 kbar and 800°C, Geo- in Treatise on Geochemistry,vol.1,Meteorites, Comets, and Planets, chim. Cosmochim. Acta, 56,679–687, doi:10.1016/0016-7037(92) edited by A. Davis, pp. 41–61, Elsevier‐Pergamon, Oxford, U. K. 90089-2. Pyle, D. M., and T. A. Mather (2009), Halogens in igneous processes and Wieczorek, M. A., and M. T. Zuber (2004), The thickness of the Martian their fluxes to the atmosphere and oceans from volcanic activity: A review, crust: Improved constraints from geoid‐to‐topography ratios, J. Geophys. Chem. Geol., 263, 110–121, doi:10.1016/j.chemgeo.2008.11.013. Res., 109, E01009, doi:10.1029/2003JE002153. Sautter, V., A. Jambon, and O. Boudouma (2006), Cl‐amphibole in the nakhlite MIL 03346: Evidence for sediment contamination in a Martian W. V. Boynton, Lunar and Planetary Laboratory, University of Arizona, meteorite, Earth Planet. Sci. Lett., 252,45–55, doi:10.1016/j. 1629 East University Blvd., Tucson, AZ 85721, USA. epsl.2006.09.024. L. M. V. Martel and G. J. Taylor, Hawaii Institute of Geophysics and Schilling, B. G., M. B. Bergeron, and R. Evans (1980), Halogens in the Planetology, University of Hawaii at Manoa, 1680 East‐West Rd., mantle beneath the North Atlantic, Philos. Trans. R. Soc. London, Ser. – Honolulu, HI 96822, USA. ([email protected]) A, 297, 147 178, doi:10.1098/rsta.1980.0208. S. M. McLennan, Department of Geosciences, State University of New Shannon, R. D. (1976), Revised effective ionic radii and systematic studies York at Stony Brook, Stony Brook, NY 11794‐2100, USA. of interatomic distances in halides and chalcogenides, Acta Crystallogr.,

5of5