Icarus 256 (2015) 13–21

Contents lists available at ScienceDirect

Icarus

journal homepage: www.elsevier .com/locate / icarus

Physical abrasion of mafic minerals and basalt grains: Applicationto martian aeolian deposits

http://dx.doi.org/10.1016/j.icarus.2015.04.0200019-1035/� 2015 Elsevier Inc. All rights reserved.

⇑ Corresponding author at: School of Environmental Science, University of Ulster,Coleraine, UK.

C. Cornwall a,b,⇑, J.L. Bandfield a,c, T.N. Titus d, B.C. Schreiber a, D.R. Montgomery a

a Department of Earth and Space Sciences, University of Washington, Seattle, WA 98195, USAb School of Environmental Science, University of Ulster, Coleraine, UKc Space Science Institute, Boulder, CO 80301, USAd United States Geological Survey, Flagstaff, AZ 86001, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 26 January 2015Revised 7 April 2015Accepted 13 April 2015Available online 18 April 2015

Keywords:Aeolian processesMars, surfaceMineralogyMars

Sediment maturity, or the mineralogical and physical characterization of sedimentary deposits, has beenused to identify sediment sources, transport medium and distance, weathering processes, and paleoen-vironments on Earth. Mature terrestrial sands are dominated by quartz, which is abundant in sourcelithologies on Earth and is physically and chemically stable under a wide range of conditions.Immature sands, such as those rich in feldspars or mafic minerals, are composed of grains that are easilyphysically weathered and highly susceptible to chemical weathering. On Mars, which is predominatelymafic in composition, terrestrial standards of sediment maturity are not applicable. In addition, the mar-tian climate today is cold and dry and sediments are likely to be heavily influenced by physical weath-ering rather than chemical weathering. Due to these large differences in weathering processes andcomposition, martian sediments require an alternate maturity index. This paper reports the results ofabrasion tests conducted on a variety of mafic materials and results suggest that mature martian sedi-ments may be composed of well sorted, well rounded, spherical polycrystalline materials, such as basalt.Volcanic glass is also likely to persist in a mechanical weathering environment while more fragile andchemically altered products are likely to be winnowed away. A modified sediment maturity index is pro-posed that can be used in future studies to constrain sediment source, paleoclimate, mechanisms for sed-iment production, and surface evolution. This maturity index may also provide insights into erosional andsediment transport systems and preservation processes of layered deposits.

� 2015 Elsevier Inc. All rights reserved.

1. Introduction

Mineralogical and physical characterization of sediment grainson Earth has been an important area of study that has significantlyincreased understanding of sediment source, transport mediumand distance, weathering processes, and paleoenvironments (e.g.,Bagnold, 1941; Folk, 1951, 1954; Suttner and Dutta, 1986; Weltjeand von Eynatten, 2004; Garcia et al., 2004). The term ‘sedimentmaturity’ has been commonly applied to clastic deposits on Earthand describes composition as well as grain texture (Folk, 1951;Boggs, 2006). Sediment maturity refers to the degree to which sed-iment has been modified by physical and chemical processes. Ingeneral, ‘‘mature’’ sediment grains are those that are well rounded,well sorted (consisting of similar sizes), and composed of mineralsthat are resistant to aqueous and physical weathering (Folk, 1951;

Pettijohn, 1975). The majority of mature terrestrial sands are dom-inated by quartz, which is abundant in source lithologies on Earthand is physically and chemically stable under a wide range ofconditions.

On Mars, however, terrestrial standards of sediment maturityare not applicable due to the martian surface having a predomi-nantly mafic composition (e.g., Christensen et al., 2000;Bandfield, 2002; Bibring et al., 2005). The presence of mafic miner-als indicates sediment immaturity under terrestrial conditionsbecause many common mafic minerals are highly susceptible tochemical weathering, softer than quartz, and may include a widerassemblage of minerals that contain cleavage planes, making themless durable during physical transport. The martian climate todayis cold, dry and dominated by wind-blown (aeolian) activity.Significant aqueous activity on the martian surface is thought tohave ended early in the planet’s history with geologic evidencerelating to liquid water on the surface dating to over 3 billion yearsago (e.g., Carr and Head, 2010). Thus, modern martian sediments

14 C. Cornwall et al. / Icarus 256 (2015) 13–21

are likely to have been largely shaped by physical weatheringrelated to aeolian processes.

Martian sediments require an alternate maturity index thattakes into account the absence of significant chemical weatheringand the presence of a primarily mafic surface composition(Grotzinger et al., 2011; Titus et al., 2012; Fenton et al., 2013). Amodified sediment maturity index can be used to place constraintson sediment source, paleoclimate, mechanisms for sediment pro-duction, and surface evolution. A maturity index can also provideinsights into erosional and depositional landscapes (sedimenttransport systems) and preservation processes of layered deposits.This study explores the physical durability of a variety of geologicmaterials and investigates the changes in grain texture during aseries of abrasion tests. The results of these experiments lead tothe construction of a maturity index (intended for sand-sized clas-tic aeolian sediments) more appropriate for the martian environ-ment and climate.

2. Sediment samples

The natural terrestrial samples used for this study are maficsands collected from South Point Beach, Hawaii; Hilo, Hawaii;and the Moses Lake dune field in Washington. The Hawaiian sedi-ments are predominantly composed of olivine (South Point) andvolcanic glass (Hilo; Moberly et al., 1965; Marsaglia, 1993). TheMoses Lake sediment is composed of granodiorite and basalt grainswith minor smectite alteration products (Bandfield et al., 2002).Natural sediment samples were collected and chosen for analysisbased on the composition and environment in which they wereweathered and deposited. The Hawaiian beach sand samples rangein size from 420 lm (South Point) to 560 lm (Hilo) and originatefrom a humid environment, where aqueous activity dominates.The olivine grains of South Point are thought to originate from lavaphenocrysts or crystal ejecta in tuffs from Mauna Loa lava flows(Moberly et al., 1965) ranging in age from <200 ka to approxi-mately 400 ka (Sharp et al., 1996). These olivine grains wereweathered out of basalt and tuff deposits and subsequently trans-ported to the beach primarily by fluvial activity. The volcanic glassgrains originate from Mauna Loa lava flows as well but wereproduced when lava came in contact with the ocean and rapidlycooled (Moberly et al., 1965). The Moses Lake dune field sand grainsizes average 237 lm and originate from a relatively arid environ-ment. The basaltic dune sand originated from the Quincy Basin,where sediment was deposited by one or more of the Missoulaor Channeled Scabland flood events that scoured the underlyingColumbia Basalts between 17,000 and 12,000 years ago (Bretzet al., 1956; Nummedal, 1978). These sediments were subse-quently reworked by aeolian processes. For the purposes of thisstudy, only weathering textures of basaltic grains were analyzedfrom the Moses Lake sediments. Grains of basalt were manuallyseparated from the granodiorite particles under a microscope.

The abrasion samples were used to determine physical durabil-ity of common materials on Mars as well as to provide a compar-ison to terrestrial quartz-rich sediments. The variety of maficminerals and volcanic materials were chosen based on remotesensing observations of Mars. The surface of Mars is predominantlybasaltic in composition and comprised of minerals rich in magne-sium and calcium (e.g., Bandfield, 2002). Therefore, the materialschosen for this study include olivine (forsterite), clinopyroxene(augite), plagioclase (labradorite), volcanic glass and fine-grainedbasalt from the Columbia River Basalt Group volcanic deposits.The varieties of silica-rich samples that were used to relate abra-sion results to familiar terrestrial materials include crystallinequartz, polycrystalline quartz, and microcrystalline chert.

Unaltered and compositionally pure mineral samples were orderedfrom the suppliers Ward’s Science and D.J. Minerals Inc. ColumbiaRiver Basalt Group rock outcrop samples were collected from local-ities near Moses Lake, Washington.

3. Analysis techniques

3.1. Natural samples

SEM images were used to investigate differences in grainweathering textures between humid (Hawaiian) and arid (MosesLake) climates and to determine the amount of influence aqueousprocesses have on shaping volcanic sediments (Marshall et al.,1987). In addition, to better constrain the effects of aqueousweathering, the basaltic grains from Moses Lake were also com-pared to the abraded Columbia River basalt sample to investigatedifferences in grain texture and to see if the physical weatheringtextures present on the Moses Lake basalt grains could be repli-cated using a Bond air mill.

3.2. Abrasion tests

In preparation for abrasion, the mafic minerals, silica-rich mate-rials, and Columbia River basalt were crushed, sifted and sortedinto approximate equidimensional grain shapes with sizesbetween 2 and 3 mm. Abrasion tests were conducted using a mod-ified Bond air mill following the design and convention ofNitkiewicz and Sterner (1988). The modified Bond air mill usedin this study has similar dimensions to the mill employed byNitkiewicz and Sterner (1988), Fig. 1, with inner walls composedof 320 lm silicon carbide grit mixed with an epoxy resin and cop-per tubing employed for the forced-air entrance. Compressed air isconnected to the air entrance of the mill through rubber tubingand, when turned on, causes mineral fragments within the cham-ber to rapidly roll around the interior cylindrical track and quicklybecome well rounded and spherical.

Each abrasion experiment consisted of 100 grains, following thegrain analysis method of Rosenfeld et al. (1953), and was con-ducted for each mineral and volcanic material separately. Thesesamples are hereafter referred to as the homogenous samples.The final abrasion experiment consisted of a mixture of 50% basaltgrains and the remaining 50% consisted of equal parts of othermafic minerals and materials including: olivine, pyroxene, labra-dorite, volcanic glass, and a Mg-rich phyllosilicate. A basaltic mix-ture was included in the abrasion tests to investigate potentialdifferences in abrasion rates and durability of minerals as theywere abraded together. This sample is hereafter referred to asthe heterogeneous sample.

Due to the small size of the Bond air mill, 10 separate batches of10 grains were run at a time to provide data for 100 grains for eachsample composition. To ensure consistency between batches,grains were weighed so that each batch only had a difference of±0.005 g, as larger differences tended to cause grains to abrademore rapidly or slowly than other batches, thus introducing incon-sistences in abrasion rates. Grains were abraded using a constantflow of compressed air at approximately 62 kPa to ensure thegrains were not being abraded at different rates than other batches.The airflow used was strong enough to roll the grains rapidlyaround the cylindrical track inside the air mill, following the con-vention described in Nitkiewicz and Sterner (1988). A fine meshwas glued over the air exit on the mill and prevented grains largerthan 400 lm from escaping. The air exit also ensured that signifi-cant amounts of dust did not accumulate in the mill preventingvariability in the abrasion effectiveness between samples.

Fig. 1. Figure adapted from Nitkiewicz and Sterner (1988) showing design anddimensions (in mm) of Bond air mill. The modified air mill used in this study hassimilar dimensions and is composed entirely of silicon carbide with copper tubingfor the air entrance but lacks guidepins.

C. Cornwall et al. / Icarus 256 (2015) 13–21 15

Each homogenous batch was run at increments of 15 min forthe first hour then 30-min increments for another hour and30 min. The homogeneous samples were primarily used to investi-gate relative rates of changes in grain morphology and size of eachseparate material. The heterogeneous batches were abraded untilall material was gone and were also used to rank the durabilityof each material in a basaltic mixture.

3.3. Shape analysis

Measurements of grain size and shape characteristics weremade using a CAMSIZER by Retsch Technology. CAMSIZER employsdynamic digital image processing described in the InternationalOrganization for Standardization ISO 13322-2 (www.iso.org).Measurements involve the projection of grain silhouettes in trans-mitted light as they free fall past the detector from a feeding chute.A high-resolution and low-resolution camera record the projectedshadows, allowing for high accuracy size measurements of parti-cles ranging from 30 lm to 30 mm in diameter with a maximuminaccuracy of 0.1 lm for each grain (e.g., Lo Castro andAndronico, 2009; Jerolmack et al., 2011). The particles are scannedin 64 directions and 60 frames are captured every second,

providing a precise and detailed analysis of a wide range of particleshapes using CAMSIZER analysis software (e.g., Lo Castro et al.,2009; Miller and Henderson, 2010; Patchigolla and Wilkinson,2010; Jerolmack et al., 2011). Particle shape parameters that arecollected in this study include size, roundness, and sphericity.

Particle shape measurements were collected on the naturalsamples and each homogenous grain sample before abrasion andafter each abrasion run. Grain size is calculated from the numberof pixels a grain silhouette occupies in images captured by the highand low-resolution cameras inside CAMSIZER. Grain roundness iscomputed using the formula:

4pA=P2;

where A is the area of the grain silhouette and P is the perimeter ofthe grain. Roundness values close to 1 indicate that a grain is wellrounded. Sphericity measurements are calculated by dividing thewidth by length of the grain (minimum extension divided by themaximum extension). Values close to 1 indicate that a grain shapeis highly spherical.

In addition, the amount of material being abraded during theabrasion tests was recorded for the homogenous samples.Sediment samples were weighed prior to abrasion and in-betweeneach abrasion run and recorded as the percentage of material lostto equalize between variances in mineral properties and small dif-ferences in sample weights.

3.4. Grain surface textures

Details of grain surface textures and the effects of chemical and/or physical weathering were examined for the natural sedimentsas well as a few abraded samples using a scanning electron micro-scope (JEOL – JSM5200). SEM images were taken of the Hawaiiansamples, Moses Lake dune field sediment, as well as the abradedcrystalline olivine, volcanic glass and Columbia River Basalt fromthe homogenous samples. Grains were mounted on a specimenstub using carbon tape and sputter-coated with gold to preventthe accumulation of electrostatic charge on the grain surface dur-ing imaging. For the Hawaiian and Moses Lake samples, a seriesof images were captured at a variety of magnifications to accentu-ate evidence of chemical and physical weathering on the grain sur-faces (Krinsley and Doornkamp, 1973; Margolis and Krinsley,1974; Marshall et al., 1987). A series of images at various magnifi-cations were also taken for the abraded samples to explore the tex-tures created using the Bond air mill. Comparison of grain surfacetextures of nautral and abraded samples provides a means todetermine how much of an influence aqueous weathering has ongrain morphology and how grain textures might differ in the aridmartian environment.

4. Results

4.1. Abrasion and shape analysis

Grain shape and rounding changed most dramatically withinthe first 15 min of abrasion for all samples. The rate of material losswas also the greatest during the first 15 min (Fig. 2) as angulargrain edges were chipped off and smoothed. Once grain edgesbecame more smoothed, the rate of material loss significantlydecreased and grains became increasingly rounded (Fig. 3). Thehomogenous sample of olivine obtained the highest degree ofsphericity and grain rounding of all the minerals after two hoursof abrasion. It was also the first mineral sample to be eroded toparticles <400 lm (Fig. 2). Augite and labradorite were also quicklyabraded but became flat and oblong and never achieved highsphericity like the olivine grains (Fig. 3). The most durable

Fig. 2. Rate of material loss vs abrasion time in minutes. Colors correspond toColumbia River Basalt, red; polycrystalline quartz, light gray; microcrystalline chert,blue; volcanic glass, black; crystalline quartz, dark gray; labradorite, violet; augite,orange; and olivine, green. Varieties of quartz are plotted as square symbols withdashed lines; volcanic materials are circles with solid lines; minerals containingcleavage planes are triangles with dot-dashed lines; olivine is a diamond symbolwith a solid line. (For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

Fig. 3. Graphs of grain shape changes during homogenous abrasion tests includinggrain size (A), sphericity (B), and roundness (C) vs abrasion time. Colors representolivine (green), Columbia River Basalt (red), volcanic glass (black), augite (orange),and labradorite (violet). Volcanic materials are plotted as circle symbols with solidlines; minerals containing cleavage planes are triangles with dot-dashed lines;olivine is represented as a diamond symbol with a solid line. Solid points joined bylines in (A) represent the 50th percentile value of the sample. Solid points joined bylines in (B) and (C) represent average values of shape parameters. Hollow pointsrepresent the minimum value of the sample and correspond to the shape and colorof the solid points. Capped lines above solid points show the maximum value foreach mineral sample. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)

16 C. Cornwall et al. / Icarus 256 (2015) 13–21

materials of the homogenous samples were the volcanic glass andthe Columbia River Basalt (Fig. 2). Of the mafic materials, the vol-canic glass and basalt were the slowest to decrease in grain sizeand increase in sphericity and roundness (Fig. 3). After 2 h and30 min, the volcanic glass and basalt grains had the lowest valueof roundness.

Of the silica-rich materials, polycrystalline quartz was the mostdurable and mono-crystalline quartz was the least durable. Aftertwo hours of abrasion, mono-crystalline quartz grains had a highersphericity and similar rounding values to polycrystalline quartzand chert (Fig. 4). Chert grains were more durable than crystallinequartz but less durable than polycrystalline quartz and had ahigher sphericity compared to polycrystalline quartz after 2 h ofabrasion. Grain shape characteristics of silica-rich samples werecompared to olivine, volcanic glass, and basalt and shape analysisresults suggest that the basalt grain morphologies evolve similarlyto polycrystalline quartz grains. Grain shapes for volcanic glassduring abrasion evolved differently than the basalt grains andthe morphology more closely matched mono-crystalline quartzin regard to sphericity and rounding. The greatest differencebetween mono-crystalline quartz and volcanic glass was grain size,where mono-crystalline quartz grains were smaller after 2 h ofabrasion. Olivine grain shapes evolved more rapidly and did notabrade in a manner similar to any kind of silica-rich material.

The abrasion results for the heterogeneous sample differ fromthat of the homogenous samples. Grain shapes evolved at a slowerrate and the materials, especially olivine, lasted longer during abra-sion (Figs. 5 and 6). The Mg-rich phyllosilicate rapidly broke downinto thin sheets and fine-grained particles after the first 15 min ofabrasion and was eventually abraded into particles <400 lm. After75 min of abrasion, the phyllosilicate material was completelygone from the air mill. After another 120 min of abrasion, augitewas also abraded to dust while a few larger grains of labradoritepersisted for much longer. Olivine grains persisted significantlybeyond 2 h of abrasion (Fig. 6) and outlasted the labradorite by60 min. The basalt and volcanic glass grains are the most durablematerials but the basalt significantly outlasted the volcanic glassby 3 h of abrasion. After 7 h of abrasion, both the basalt and thevolcanic glass grains had achieved high values of rounding andsphericity (Fig. 6).

4.2. Grain texture

SEM images of Hawaiian grain surfaces show more evidence ofaqueous weathering textures than the Moses Lake dune field sed-iments and both samples displayed abundant physical weatheringtextures. The olivine grains from South Point Beach, Hawaii showextensive chemical pitting on the grain surfaces, (Fig. 7A) as well

Fig. 4. (A) Grain size vs% material passing a particular sieve size, (B) sphericity (b/l)vs sample volume percent, and (C) rounding (SPHT) vs sample volume percent ofcrystalline quartz (light blue squares), microcrystalline chert (medium bluetriangles), polycrystalline quartz (dark blue circles), Columbia River Basalt (redcircles), volcanic glass (black triangles), and olivine (green diamonds) after twohours of abrasion. (For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

Fig. 5. Heterogeneous abrasion results showing percent of material lost vs time inminutes. Heterogenous abrasion sample consisted of 50% Columbia River Basalt(red) and remaining parts equal portions of volcanic glass (black), olivine (green),labradorite (violet), augite (orange), and phyllosilicate (brown). Volcanic materialsare plotted as circle symbols with solid lines; minerals containing cleavage planesare triangles with dot-dashed lines; olivine is displayed as a diamond symbol with asolid line. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

C. Cornwall et al. / Icarus 256 (2015) 13–21 17

as scratches (Fig. 7B) and chipping from abrasion. The surface ofthe Hilo, Hawaii volcanic glass sample is covered in microscopicgas vesicles and displays similar chemical pitting and scratchingtextures as identified on the olivine grain but there is more evi-dence of chipping. Chipping textures were primarily observedinside depressions of vesicles (Fig. 7C). The surfaces of the basaltgrains from the Moses Lake dune field are not as smooth as thesand grains from Hawaii. Chemical pitting was identified on thebasalt grains but scratches and chipping are not as obvious. Thebasalt grains also display a number of microscopic fractures, typi-cally radiating out from a central point (Fig. 7D). Microscopic frac-tures are not present on any of the Hawaiian samples that wereanalyzed.

The abraded homogenous samples of olivine, volcanic glass, andbasalt show similarities in grain surface textures due to physicalabrasion to that of the Hawaiian and Moses Lake samples. The sur-face of an un-abraded Columbia River basalt grain is shown inFig. 8A. Fig. 8B and C shows the textural differences between theabraded basalt and the natural sample from Moses Lake, respec-tively. The abraded samples lack evidence of chemical weatheringinfluences and, instead, show extensive signs of chipping andscratching on all grain surfaces (Fig. 8B). These surface texturesare far more pervasive on the abraded samples than the naturalsamples. The most notable difference between the sediment sam-ples is the overall smoothness of the grains. The abraded grains arerougher than the natural samples, which have sharp edges anddebris fragments clinging to the surface (Fig. 8B). The radiatingfractures observed on the Moses Lake basaltic sand grains werenot present on any of the abraded basalt grains.

5. Discussion

5.1. Abrasion and shape analysis

The rapid changes in grain size and morphology within the first15 min of abrasion are due to the rounding and chipping of jaggededges and, for some minerals, cleaving along planes of weakness.This process creates multiple small grains and the rate of materiallost declines as grains attain a more physically stable, sphericalform.

Olivine was the least durable material abraded of the homoge-nous samples. It has a similar Mohs hardness to quartz with iden-tical fracture habits but may be more susceptible to abrasion dueto differences in crystalline structure and density (Willetts,1983). Minerals that possess cleavage planes, such as labradoriteand augite, were moderately durable and abraded into thin, elon-gate, tabular grains. A comparison between different forms ofquartz showed discrepancies in abrasion rates. Crystalline quartz(from a single crystal), as used in this experiment, was found toabrade more rapidly than polycrystalline quartz. Polycrystallinequartz is composed of multiple quartz crystal fragments of variablegrain sizes and, in the case of chert, the grains are composed ofmicroscopic crystal fragments. Previous studies have shown thatmono-crystalline structures are less durable during physicaltransport than polycrystalline forms due to the formation of

Fig. 6. Minerals in a heterogeneous basaltic mixture at various stages of abrasion (45 min, 225 min and 540 min). Minerals shown here include Columbia River Basalt (CRB),olivine (ol), labradorite (Plag), augite (CPX), volcanic glass (VG), and Phyllosilicate (PS) and are from the same experiment batch. These images correlate with results shown inFig. 5. We considered grains after 45 min of abrasion to represent an immature texture and composition; 225 min represents mature grains; and 540 min representssupermature grains where only basalt and volcanic glass remain with highly spherical shapes.

Fig. 7. Chemical and physical weathering textures of terrestrial grains (Krinsley and Doornkamp, 1973; Marshall et al., 1987). (A) Aqueous pitting due to chemical weatheringon an olivine grain from South Point Beach, Hawaii. (B) scratches due to abrasion on the same olivine grain. (C) Gas vesicles and chipping on the surface of a volcanic glassgrain from Hilo, Hawaii. (D) Fracture on the surface of a basalt grain from Moses Lake.

18 C. Cornwall et al. / Icarus 256 (2015) 13–21

micro-fractures that easily propagate through a single crystal(Harrell and Blatt, 1978). In applying the Griffith ‘FractureTheory’, Brace (1961) describes how pre-existing flaws in a singlecrystalline structure result in macroscopic fractures that spreadunder stress. Fractures also form in polycrystalline materials butdo not propagate past individual grain boundaries, resulting inan increased resistance to physical stress (Brace, 1961).

The Columbia River basalt grains were the most durable mate-rial tested in this study with grain morphologies that evolved sim-ilarly to polycrystalline quartz during abrasion. The durability ofbasalt may be due to the fact that it is polycrystalline, being com-posed of a variety of mafic minerals of variable grain sizes, such asolivine, plagioclase and pyroxene. Hence, the mechanical durabilityof aeolian sediment on Mars may not be controlled by the physicalproperties of individual minerals but rather by the crystalline tex-ture of the materials.

The heterogeneous sample abrasion results were used as a basisfor a mafic sediment maturity index (Fig. 9). Minerals lasted longerduring abrasion most likely due to the presence of the phyllosili-cate mineral, which rapidly broke apart into thin sheets and fineparticles within the first 45 min, resulting in decreased effective-ness of abrasion inside the air mill. Augite became completely

abraded to dust after about 3 h and 15 min and the labradoritewas abraded to dust after about 4 h and 45 min. This differencein abrasion times of augite and labradorite is most likely due tothe difference in mineral hardness, where labradorite is harderthan augite on the Mohs scale. The Columbia River basalt grainsoutlasted the volcanic glass during abrasion by about 3 h and theolivine grains by approximately 5 h. This experiment suggests thatbasalt is the most resistant to abrasion and should be abundant inmartian aeolian sedimentary deposits. However, these experi-ments do not take into account ductile response to stress, whichhas been observed in some materials during transport and abra-sion. Tumbling experiments conducted on gypsum and quartz haveshown that quartz (albeit harder) abraded more rapidly thangypsum (Szynkiewicz et al., 2013). This difference in abrasion ismost likely due to the brittle nature of quartz. The materials in thisstudy demonstrate a predominantly brittle response to stress withthe exception of the phyllosilicate mineral in the heterogenousabrasion sample. The phyllosilicate mineral was the first materialto be abraded to dust and it is possible that the presence of hardermaterials and minerals aided in the abrasion of the softer phyllosil-icate. In the study conducted by Szynkiewicz et al. (2013), theyfound the quartz grains had significantly abraded the interior of

Fig. 8. Differences in grain textures of basalt grains. (A) Unabraded Columbia RiverBasalt grain. (B) Abraded Columbia River Basalt after �14 h of abrasion. (C) Basaltgrain from Moses Lake dune field.

Fig. 9. Textural and compositional maturity index for mafic sediments on Mars.Materials with high physical durability are polycrystalline basalt and volcanic glass.Materials indicating immature sediments contain minerals with cleavage planesthat weaken physical durability. Olivine may indicate a local or more regionalsource depending on the other minerals present in the sediment. Fig. 6 displaysexamples of immature, mature and supermature grains produced during abrasionexperiments.

C. Cornwall et al. / Icarus 256 (2015) 13–21 19

the quartz flask used for tumbling, resulting in an increase of thesample mass. In contrast, the gypsum sample mass slightlydecreased. This may suggest that in the presence of harder miner-als, elastic materials may still abrade more rapidly. Future studieswill need to be conducted to determine how durable elastic miner-als (such as gypsum) are in a mixture of hard, brittle materials.

5.2. Proposed martian sediment maturity index

Martian dune fields composed of mature sediments could becomposed of a mixture of basalt and volcanic glass grains (Figs. 6and 9) as well as chemically derived minerals such as sulfatesand salts that were weathered locally or in situ. Immature sedi-ments are likely to include individual pyroxene and plagioclasecrystals in the sediment and the source material is most likely local(Figs. 6 and 9). The morphological differences between the elon-gated, tabular grains of pyroxene and plagioclase to those of spher-ical grains of volcanic materials and minerals without cleavage

planes would result in sorting effects during aeolian transport.Some wind tunnel experiments have shown that grains with alower sphericity were more readily transported than sphericalgrains but only at low wind speeds (Williams, 1964; Willettset al., 1982). Particles having a low sphericity also had longerand flatter saltation trajectories than spherical grains but sphericalgrains exhibited more aeolian bed activity, whereas low sphericitygrains were less frequently dislodged (Willetts, 1983; Rice, 1991).Thus, aeolian transport is the most effective for spherical grains. Itis likely that minerals with cleavage planes will become segregatedfrom spherical grains and accumulate in different areas of a dunefield. However, it is improbable that grains with a low sphericitywill be transported long distances due to their susceptibility tomechanical erosion in combination with aeolian sorting effects.Therefore, it is unlikely that plagioclase and pyroxene minerals willbe present in mature aeolian sediments as individual crystals,which have traveled far from the sand source.

The presence of olivine in an aeolian deposit depends on multi-ple factors including availability, transport distance, and crystallinestructure of the mineral. Source material may be olivine rich orolivine poor basalts, affecting the amount of olivine available foraeolian transport. Olivine also has a higher density than most maficminerals and volcanic materials that may result in selective trans-port of less dense particles especially at low wind speeds (Willetts,1983). Aeolian sorting effects due to density differences may limitthe distance olivine grains travel from the source material. Lastly,mono-crystalline olivine is likely to be less durable during physicaltransport than polycrystalline olivine. If martian olivine grains arepolycrystalline, it is likely that they may be transported longer dis-tances from the source material. Therefore, interpreting the com-position of aeolian deposits on Mars for a maturity indexrequires knowledge of the local bedrock composition as well asthe geologic history (e.g. fluvial processes) of an area to accountfor changes in olivine abundance during aeolian transport anddeposition.

5.3. Olivine in martian sediments

Remote sensing and lander observations have shown that thepresence of olivine in sediments is highly variable and is stronglydependent on the local weathering environment/history. In general,the northern plains are predominantly composed of high-silica

20 C. Cornwall et al. / Icarus 256 (2015) 13–21

phases and depleted in pyroxene and olivine, while the southernhighlands are composed of plagioclase and high-calcium pyroxenewith an enrichment of low-calcium pyroxene and olivine relativeto younger crust (Bandfield et al., 2000; Mustard et al., 2005;Rogers and Christensen, 2007; Koeppen and Hamilton, 2008;Poulet et al., 2009). Dune fields located in the northern plains areoften enriched in gypsum (Langevin et al., 2005; Horgan et al.,2009; Masse et al., 2010) and low-albedo (dark) aeolian sedimentsin the southern highlands are typically composed of mafic mineralsseemingly unaltered by aqueous activity (Jaumann et al., 2006).These sediments are commonly composed of plagioclase and highcalcium pyroxene with variable amounts of olivine and low calciumpyroxene (Tirsch et al., 2011). In a few cases of southern highlandintercrater dune fields, enrichment in olivine spectral signatureswas found in otherwise olivine-poor basaltic regions (Tirsch et al.,2011). These areas of olivine enrichment might be due to mechanicalsegregation, where aeolian sorting has preferentially transportedand deposited olivine similar to volcanic sands in Iceland(Mangold et al., 2011). Other areas on Mars, for example NiliFossae, Argyre Planitia, Gusev Crater, Isidis Planitia, and Gangesand Eos Chasmas have high-thermal inertia surfaces rich in olivinebut the surrounding sediments are olivine poor (Hamilton andChristensen, 2005; Yen et al., 2005; Ruff et al., 2006; Bandfield andRogers, 2008; Edwards et al., 2008). These sediments could haveexperienced chemical alteration by acidic fluids that preferentiallyweathered olivine crystals (Tosca et al., 2004; Hurowitz et al.,2006; Stopar et al., 2006; Bandfield et al., 2011) or this observationcould be an indication of differing mineral durability during aeoliantransport. Olivine has been observed to be present in young, small,active-looking dunes in the southern highlands (Tirsch et al., 2011)and is generally found in association with pyroxene (Poulet et al.,2007). The presence of olivine in geologically young aeolian depositsalong with mechanically fragile pyroxene may suggest that olivine islargely mono-crystalline on Mars and mechanically weathers morequickly than other mafic sediments, consistent with the abrasionresults in this study.

5.4. Weathering textures

The natural sediment samples have smoother grain surfacesthan the mechanically abraded grain samples (Fig. 8). This suggeststhat even in arid environments, such as Moses Lake, aqueous pro-cesses have a considerable impact on grain surface texture. Grainson Mars have probably been predominantly shaped by mechanicalweathering with little aqueous interaction and may have roughersurface textures, similar to those created from the abrasion tests.In addition, immature deposits may be composed of grains thatare more angular with rougher surface textures compared toimmature terrestrial sediments. Alternatively, martian grainsmay have been subject to some aqueous weathering dependingon their location and age of the deposit. Aqueous influences maybe found in aeolian sediments that have been previously trans-ported by fluvial processes early in Mars’ history or inactive dunefields that have been stabilized by geochemical cements (Gardinet al., 2011) or seasonal ice (Bourke, 2005). Polar aeolian depositsmay have significantly different grain textures compared to equa-torial aeolian deposits, due to possible chemical reactions fromannual cycles of frost deposition and sublimation. Fracturingobserved on the surfaces of Moses Lake basalt sands could not bereplicated by abrasion alone. The presence of such fractures sug-gests that an additional physical process is present, such as cyclesof heating and cooling. It is possible that these additional processesare also active on Mars, affecting the rate of physical weathering ofmartian basaltic grains, especially in the polar regions.

6. Conclusion

Abrasion of a variety of mafic materials indicates that maturemartian sediments are likely composed of very well sorted, wellrounded, spherical poly-crystalline basalt grains (Figs. 6 and 9).Any volcanic glass present is also likely to persist in this environ-ment (Figs. 6 and 9) while chemically altered products are likelyto be weathered locally or in situ. Immature sediments are likelyto contain minerals with cleavage planes and be poorly sorted,sub rounded and angular (Figs. 6 and 9). The presence and relativeabundance of olivine may be used as an index mineral, indicatinglocal and regional sources as well as sediment maturity and rela-tive deposit age.

As grain surface textures in arid environments on Earth appearto be heavily influenced by aqueous interaction, grain surfaces arelikely to have rougher textures because physical weathering pro-cesses have dominated the surface of Mars for the past 3 billionyears.

Acknowledgments

Special thanks to Dave McDougall for help involving equipmenttraining and sample preparation in the laboratory, Dr. Hanson Fongfor SEM sample preparation, and Mike Harrell for donating somemineral samples and use of his modified Bond air mill which madethis study possible. This research was partially funded by aUniversity of Washington departmental research grant in thedepartment of Earth and Space Sciences. Any use of trade, firm,or product names is for descriptive purposes only and does notimply endorsement by the U.S. Government.

References

Bagnold, R.A., 1941. The physics of blown sand and desert dunes. Prog. Phys. Geogr.18, 91–96.

Bandfield, J.L., 2002. Global mineral distributions on Mars. J. Geophys. Res. 107.http://dx.doi.org/10.1029/2001JE001510.

Bandfield, J.L., Rogers, A.D., 2008. Olivine dissolution by acidic fluids in ArgyrePlanitia, Mars: Evidence for a widespread process? Geol. Soc. Am. 36, 579–582.

Bandfield, J.L., Hamilton, V.E., Christensen, P.R., 2000. A global view of martiansurface compositions from MGS-TES. Science 287, 1626–1630.

Bandfield, J.L., Edgett, K.S., Christensen, P.R., 2002. Spectroscopic study of the MosesLake dune field, Washington: Determination of compositional distributions andsource lithologies. J. Geophys. Res. 107. http://dx.doi.org/10.1029/2000JE001469.

Bandfield, J.L., Rogers, A.D., Edwards, C.S., 2011. The role of aqueous alteration in theformation of martian soils. Icarus 211, 157–171.

Bibring, J.-P. et al., 2005. Mars surface diversity as revealed by the OMEGA/MarsExpress observations. Science 307, 1576–1581.

Boggs Jr., S., 2006. Sedimentology and Stratigraphy. Pearson Education, ISBN 0-13-154728-3.

Bourke, M.C., 2005. Niveo-aeolian and Denivation Deposits on Mars, Drylands:Linking landscape processes to sedimentary environments. In: BGRG/BSRGInternational Conference. London.

Brace, W.F., 1961. Dependence of fracture strength of rocks on grain size. In: 4thSymposium on Rock Mechanics. Penn. State Univ., pp. 99–103.

Bretz, J.H., Smith, H.T.U., Neff, G.E., 1956. Channeled Scabland of Washington: Newdata and interpretations. Geol. Soc. Am. Bull. 67, 957–1049.

Carr, M.H., Head, J.W., 2010. Acquisition and history of water on Mars. In: Cabrol,N.A., Grin, E.A. (Eds.), Lakes on Mars. Elsevier, pp. 31–67.

Christensen, P.R. et al., 2000. Identification of a basaltic component on the martiansurface from Thermal Emission Spectrometer data. J. Geophys. Res. 105, 9609–9621.

Edwards, C.S., Chistensen, P.R., Hamilton, V.E., 2008. Evidence for extensive olivine-rich basalt bedrock outcrops in Ganges and Eos Chasmas, Mars. J. Geophys. Res.113, E11003.

Fenton, L.K. et al., 2013. Summary of the Third International Planetary DunesWorkshop: Remote sensing and imaging analysis of planetary dunes, Flagstaff,Arizona, USA, June 12–15, 2012. Aeolian Res. 8, 29–38.

Folk, R.L., 1951. Stages of textural maturity in sedimentary rocks. J. Sed. Res. 21,127–130.

Folk, R.L., 1954. The distinction between grain size and mineral composition insedimentary-rock nomenclature. J. Geol. 62, 344–359.

C. Cornwall et al. / Icarus 256 (2015) 13–21 21

Garcia, D. et al., 2004. Geochemical variability induced by entrainment sorting:Quantified signals for provenance analysis. Sed. Geol. 171, 113–128.

Gardin, E. et al., 2011. High albedo dune features suggest past dune migration andpossible geochemical cementation of aeolian sediments on Mars. Icarus 212,590–596.

Grotzinger, J. et al., 2011. Mars sedimentary geology: Key concepts and outstandingquestions. Astrobiology 11. http://dx.doi.org/10.1089/ast.2010.0571.

Hamilton, V.E., Christensen, P.R., 2005. Evidence for extensive, olivine-rich bedrockin Nili Fossae, Mars. Geology 33, 433–436.

Harrell, J., Blatt, H., 1978. Polycrystallinity: Effect on the durability of detrital quartz.J. Sed. Pet. 48, 25–30.

Horgan, B.H. et al., 2009. Distribution of hydrated minerals in the north polar regionof Mars. J. Geophys. Res. 114, E01005.

Hurowitz, J.A. et al., 2006. In situ and experimental evidence for acidic weatheringof rocks and soils on Mars. J. Geophys. Res. 111, E02S19. http://dx.doi.org/10.1029/2005JE002515.

Jaumann, R. et al., 2006. Dark materials in martian craters. Lunar Planet. Sci., XXVII,Abs. #1735.

Jerolmack, D.J., Reitz, M.D., Martin, R.L., 2011. Sorting out abrasion in a gypsumdune field. J. Geophys. Res. 116. http://dx.doi.org/10.1029/2010JF001821.

Koeppen, W.C., Hamilton, V.E., 2008. Global distribution, composition, andabundance of olivine on the surface of Mars from thermal infrared data. J.Geophys. Res. 113, E05001.

Krinsley, D.H., Doornkamp, J.C., 1973. Atlas of Quartz Sand Surface Textures.Cambridge Univ. Press, London.

Langevin, Y. et al., 2005. Sulfates in the north polar region of Mars detected byOMEGA/Mars Express. Science 307, 1584–1586.

Lo Castro, M.D., Andronico, D., 2009. Grain size distributions of volcanic particles byCAMSIZER. Conferenza Rittmann, Nicolosi (Ct), Italy, 11–13 June 2009.

Lo Castro, M.D., Andronico, D., Nunnari, G., Spata, A., Torrisi, A., 2009. Shapemeasurements of volcanic particles by CAMSIZER. Conferenza Rittmann,Nicolosi (Ct), Italy, 11–13 June 2009.

Mangold, N. et al., 2011. Segregation of olivine grains in volcanic sands in Icelandand implications for Mars. Earth Planet. Sci. Lett. 310, 233–243.

Margolis, S.V., Krinsley, D.H., 1974. Processes of formation and environmentaloccurrence of microfeatures on detrital quartz grains. Am. J. Sci 274, 449–464.

Marsaglia, K.M., 1993. Basaltic island sand provenance. In: Johnsson, M.J., Basu, A.(Eds.), Processes Controlling the Composition of Clastic Sediments, SpecialPaper 284. Geological Society of America, Boulder.

Marshall, J.R., Whalley, W.B., Krinsley, D.H., 1987. The origin of some ‘‘chemical’’textures on quartz-grain surfaces: Interpretation and environmentalimplications. In: Marshall, J.R. (Ed.), Clastic Particles: Scanning ElectronMicroscopy and Shape Analysis of Sedimentary and Volcanic Clasts. VanNostrand Reinhold, New York, pp. 248–253.

Masse, M. et al., 2010. Martian polar and circum-polar sulfate-bearing deposits:Sublimation tills derived from the North Polar Cap. Icarus 209, 434–451.

Miller, N.A., Henderson, J.J., 2010. Quantifying sand particle shape complexity usinga dynamic, digital imaging technique. Agron. J. 102, 1407–1414.

Moberly, R.L., Baver, D., Morrison, A., 1965. Source and variation of Hawaiian littoralsand. J. Sed. Res. 35, 589–598.

Mustard, J.F. et al., 2005. Olivine and pyroxene diversity in the crust of Mars. Science307, 1594–1597.

Nitkiewicz, A.M., Sterner, S.M., 1988. An improved Bond air mill for the preparationof spherical single crystals. Am. Mineral. 73, 662–666.

Nummedal, D., 1978. Field trip stop descriptions. In: Baker, V.R., Nummedal, D.(Eds.), The Channeled Scabland: A Guide to the Geomorphology of the ColumbiaBasin, Washington, A Comparative Planetary Geology Field Conference in theColumbia Basin, June 5–8, 1978. NASA, Washington, D.C., pp. 131–168.

Patchigolla, K., Wilkinson, D., 2010. Crystal shape characterization of dry samplesusing microscopic and dynamic image analysis. Part. Part. Syst. Charact. 26,171–178.

Pettijohn, F.J., 1975. Sedimentary Rocks. Harper and Row, New York.Poulet, F. et al., 2007. Martian surface mineralogy from Observatoire pour la

Mineralogie, les Glaces et l’Activite on board the Mars Express spacecraft(OMEGA/MEx): Global mineral maps. J. Geophys. Res. 112, E08S02. http://dx.doi.org/10.1029/2006JE002840.

Poulet, F. et al., 2009. Quantitative compositional analysis of martian mafic regionsusing the MEx/OMEGA reflectance data. 2. Petrological implications. Icarus 201,84–101.

Rice, M.A., 1991. Grain shape effects on aeolian sediment transport. Acta Mech.Suppl. 1, 159–166.

Rogers, A.D., Christensen, P.R., 2007. Surface mineralogy of martian low-albedoregions from MGS TES data: Implications for crustal evolution and surfacealteration. J. Geophys. Res. 112, E01003.

Rosenfeld, M.A., Jacobsen, L., Ferm, J.C., 1953. A comparison of sieve and thin-section technique for size analysis. J. Geol. 61, 114–132.

Ruff, S.W. et al., 2006. The rocks of Gusev Crater as viewed by the Mini-TESinstrument. J. Geophys. Res. 111, E12S18. http://dx.doi.org/10.1029/2006JE002747.

Sharp, W.D. et al., 1996. The 40Ar/39Ar and K/Ar dating of lavas from Hilo 1-km corehold, Hawaii Scientific Drilling Project. J. Geophys. Res. 101, 11607–11616.

Stopar, J.D. et al., 2006. Kinetic model of olivine dissolution and extent of aqueousalteration on Mars. Geochim. Cosmochim. Acta 70, 6136–6152.

Suttner, L.J., Dutta, P.K., 1986. Alluvial sandstone composition and paleoclimate, I.Framework mineralogy. J. Sed. Pet. 56, 329–345.

Szynkiewicz, A. et al., 2013. The polar sulfur cycle in the Werenskioldbreen,Spitsbergen: Possible implications for understanding the deposition of sulfateminerals in the North Polar Region of Mars. Geochim. Cosmochim. Acta 106,326–343.

Tirsch, D. et al., 2011. Dark aeolian sediments in martian craters: Composition andsources. J. Geophys. Res. 113, E03002.

Titus, T.N., Hayward, R.K., Bourke, M.C., 2012. Interdisciplinary research producesresults in the understanding of planetary dunes. Eos, Trans. Am. Geophys. Union91 (32), 281.

Tosca, N.J. et al., 2004. Acid-sulfate weathering of synthetic martian basalt: The acidfog model revisited. J. Geophys. Res. 109, E05003. http://dx.doi.org/10.1029/2003JE002218.

Weltje, G.J., von Eynatten, H., 2004. Quantitative provenance analysis of sediments:Review and outlook. Sed. Geol. 171, 1–11.

Willetts, B.B., 1983. Transport by wind of granular materials of different grainshapes and densities. Sedimentology 30, 669–679.

Willetts, B.B., Rice, M.A., Swaine, S.E., 1982. Shape effects in aeolian grain transport.Sedimentology 29, 409–417.

Williams, G., 1964. Some aspects of the eolian saltation load. Sedimentology 3, 257–287.

Yen, A.S. et al., 2005. An integrated view of the chemistry and mineralogy of martiansoils. Nature 436, 49–54.