Heterogeneities in the Martian Mantle Through Time: Clues from Meteorites and Gusev Basalts

41st Lunar and Planetary Science Conference (2010) 1336.pdf

HETEROGENEITIES IN THE MARTIAN MANTLE THROUGH TIME: CLUES FROM METEORITES AND GUSEV BASALTS. Mariek E. Schmidt1 and Timothy J. McCoy2, 1Dept. of Earth Sciences, Brock Univer- sity (500 Glenridge Ave., St. Catharines, ON Canada L2S 3A1; [email protected]), 2Dept. of Mineral Sciences, Smithsonian Institution (P.O. Box 37021, Washington, D.C. 20013-7012; [email protected])

Introduction: The composition of Martian basalts basalt groups: 1) Adirondack with lower, but increas- reflect their mantle source at the time of extraction, ing normalized concentrations of the incompatible and thus, we can use geochemical data combined with elements K, P, and Ti, resulting in a steep pattern and radiometric dates and geomorphic age estimates to 2) Columbia Hills (CH) basalts (Backstay, Hum- interpret how the Martian mantle evolved with time. boldtPeak, and Irvine) with flatter incompatible ele- So far, Martian basalts are constrained to four time ment patterns (higher K/Ti). Differences in K, P, and spans from near present day to earliest Mars (Fig 1). 1) Ti were likely not caused by Mars surface processes Basaltic volcanism as recently as ~10 Ma [e.g., 1] and because these elements are not enriched in soil and uplift of the Tharsis bulge [2] indicate mantle melt dust and they do not correlate with other indices of extraction is ongoing. 2) Crystallization ages of sher- alteration, such as Si/Al [10]. Instead, the origin of gottitic meteorites range from 173 to 575 Ma [3, 4]. 3) differences among the Gusev basalts can be linked to The surface age estimates of basalts examined in differences in their mantle source. Gusev basalts have Gusev Crater by the Mars Exploration Rover Spirit are higher CaO/Al2O3 than the shergottites and are thought Early Hesperian (~3.7 Ga) [5]. And 4) early mantle to originate from a separate mantle source [e.g., 12]. depletion suggested by shergottite Sm-Nd isotopes likely occurred within the first 33 m.y. of Mars history [6]. The Wänke and Dreibus (1988) estimate (WD) of bulk Mars approximates the ~4.6 Ga mantle [7].

Fig 1. Timeline of crystallization ages of shergottites and the surface ages of Gusev basalts and most recent volcanism.

Comparing geochemical datasets: Of the Martian basalts, compositional datasets exist for the shergottites [8] and the Gusev Crater basalts [9, 10]. Although the Gusev data are more limited, the alpha particle x-ray spectrometer (APXS) analyses include minor elements whose behavior during mantle melting ranges from highly incompatible (K and P) to highly compatible (Cr and Ni). In order to evaluate patterns in elemental concentrations, we plot whole rock analyses for Martian basalts on a chondrite-normalized abundance diagram or Fig 2. C1 chondrite [13] normalized element abundance (spider) “spider diagram” (Fig 2) that lists elements in order of diagrams for A: basaltic and olivine-phyric shergottites [8] and B:

increasing compatibility during (terrestrial) mantle Gusev basalts [9, 10].

melting. All shergottites are depleted in K. Some sher- The origin of Gusev basalts: We focus on the mi- gottites are also depleted in light rare earth elements nor elements (K, P, Ti, Cr, Ni) to model igneous proc- (REE) relative to heavy REE, such as QUE 94201 esses based on known partition coefficients (D). For with steep patterns (Fig 2A) that probably result from example, Rayleigh fractional crystallization of Adiron- partial melting of a mantle reservoir that was depleted dack magma steadily increases incompatible element in light REE early in Mars history [6]. Other shergot- bulk concentrations (K; D " 0) and rapidly decreases tites, such as Los Angeles, are more light REE-rich and K compatible element concentrations (Ni; Dbulk >>1). have flatter patterns that likely result from fractional Ni crystallization and assimilation of a light REE-rich After 50% crystallization, the increase in K is too small (0.13-0.25 wt% K2O) and the decrease in Ni is reservoir, such as the crust [11]. ! too great (149-5 ppm) for Adirondack to be parental to Although not all elements are included, overlaying ! the Gusev basalt data in Fig 2B demonstrates two main CH basalts as suggested by [14]. 41st Lunar and Planetary Science Conference (2010) 1336.pdf

Our preferred model for the origin of the Gusev ba- averaging volumes of Noachian and Amazonian crust. salts involves two-stage batch melting of WD mantle In sum, we estimate the total crustal thickness added (Fig 3); the primary reservoir melts to produce a (K- since the Noachian is <1 km. rich) magma like the CH basalts, leaving behind a depleted (K-poor) mantle reservoir that then melts a sec- ond time to produce a more depleted (K-poorer) Adi- rondack-like magma. Partition coefficients reflect 1 GPa phase assemblage of [11].

Fig 4. Mass balance model of the evolution of the Martian crust and mantle since the formation of the planet (time=0) to the present day. Fig 3. Two-stage batch melting of a WD mantle at F=0.03 and 0.10 plotted on a C1 chondrite [13] normalized K-P-Ti-Cr-Ni abundance In order for PM to persist to the Hesperian, as indi- diagram. cated by the CH basalts, a high degree of mantle partial It is unlikely the same parcel of mantle melted to melting (F>0.05) must account for crust formation. form all Gusev basalts. Instead, the CH basalts origi- Thermal models [2] suggest that in order for mantle nated by melting a primary mantle reservoir, while convection to support the Tharsis uplift today, ~50% of Adirondack basalts came from a reservoir that had radioactive elements (including K) presently reside in previously been partially melted. Importantly, primary the Martian mantle. If all heat-producing elements are WD mantle must have existed at least until the Hespe- in PM domains, then PM makes up at least 10-15% of rian (~3.7 Ga), indicating early light REE depletion of the present mantle volume. A possible solution is 40- the shergottite mantle source was not uniform. 50 km crust was created by ~10% partial melting of The Evolution of the Martian Mantle: To ad- the mantle, leaving ~20% primary WD mantle. The dress how volumes of primitive and depleted reservoirs Martian crust likely formed by inhomogeneous, multi- in the Martian mantle may have varied with time, we stage partial melting of the mantle as suggested for the present a mass balance model of coupled crustal Gusev basalts. However, our model demonstrates that growth and mantle depletion (Fig 4). Assuming no at high melt fractions (F>0.05), heterogeneities in the crustal recycling, our model involves three reservoirs: mantle formed by incomplete melt extraction, may the crust and two mantle reservoirs - a K-rich, primi- persist over the history of Mars. tive mantle (PM) of WD composition [7] and a K- References: Lucchitta (1987) Science 235, 565-567. poor, depleted mantle (DM). A volume of fertile PM [2] Kiefer (2003) Meteoritics & Planet. Sci. 38, 1815- partially melts at melt fraction F (F=0.03, 0.05, and 1832. [3] Jones, J.H. (1985) Geochim. et Cosmochim. 0.10) to generate volumes of DM and crust, where the 50, 969-977. [4] Borg et al. (2001) LPS XXXII, #1144. total volume of the crust and the mantle is constant. A [5] Greeley et al. (2005) JGR 110, E05008. [6] Borg significant unknown is the thickness of the crust be- et al. (1997) Geochim. et Cosmochim. 61, 4915-4931. cause as it thickens, a greater portion of the mantle is [7] Wänke and Dreibus (1988) Philos. Trans. Royal involved in its production. We use a mean crustal Soc. London 325, 545-557. [8] Meyer (2009) thickness of 40-50 km based on Mars Orbiter Laser http://curator.jsc.nasa.gov/antmet/mmc/index.cfm. [9] Altimeter (MOLA) topography and gravity models Gellert et al. (2006) JGR, 111, E02S05. [10] Ming et [15]. It is impossible to constrain crustal growth rates al. (2008) JGR 113, E12S39. [11] Rubin et al. (2000) because we lack absolute ages or information about the Geology 28, 1011-1014. [12] Monders (2007) Meteor- subsurface. Instead, we assume the bulk of the Mar- itics & Planet. Sci. 42, 131-148. [13] Sun and tian crust was built by the Noachian (within first 1 Ga). McDonough, (1989) Geol. Soc. London Special Pub. The volume of Amazonian age crust was determined 42, 313-345. [14] McSween et al. (2006) JGR 111, using surface area estimates of Amazonian volcanic E09S91. [15] Neumann et al. (2004) JGR 109, rocks [16] and assuming 5 km associated underlying E08002. [16] Tanaka et al. (1992) Mars, UA Press, crust. We estimated the volume of Hesperian crust by 345-382.