Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

Geochimica et Cosmochimica Acta 87 (2012) 69–80

Signatures in magnetites formed by (Ca,Mg,Fe)CO3

thermal decomposition: Terrestrial and extraterrestrialimplications

Concepcion Jimenez-Lopez a,⇑, Carlos Rodriguez-Navarro b,⇑,Alejandro Rodriguez-Navarro b, Teresa Perez-Gonzalez a, Dennis A. Bazylinski c,

Howard V. Lauer Jr. d, Christopher S. Romanek e

a Departamento de Microbiologıa, Facultad de Ciencias, Universidad de Granada, Campus de Fuentenueva s/n, 18071 Granada, Spainb Departamento de Mineralogıa y Petrologıa, Facultad de Ciencias, Universidad de Granada, Campus de Fuentenueva s/n, 18071 Granada, Spain

c School of Life Sciences, University of Nevada at Las Vegas, Las Vegas, NV, USAd ESCG/Barrios Technologies, Houston, TX 77258, USA

e NASA Astrobiology Institute and Department of Earth and Environmental Sciences, University of Kentucky, Lexington, KY, USA

Received 29 June 2011; accepted in revised form 19 March 2012; available online 28 March 2012

Abstract

It has never been demonstrated whether magnetite synthesized through the heat-dependent decomposition of carbonateprecursors retains the chemical and structural features of the carbonates. In this study, synthetic (Ca,Mg,Fe)CO3 wasthermally decomposed by heating from 25 to 700 �C under 1 atm CO2, and by in situ exposure under vacuum to the electronbeam of a transmission electron microscope. In both cases, the decomposition of the carbonate was topotactic and resulted inporous pseudomorphs composed of oriented aggregates of magnetite nanocrystals. Both calcium and magnesium were incor-porated into nanophase magnetite, forming (Ca,Mg)-magnetites and (Ca,Mg)-ferrites when these elements were present in theparent material, thus preserving the chemical signature of the precursor. These results show that magnetites synthesized in thisway acquire a chemical and structural inheritance from their carbonate precursor that indicates how they were produced.These results are not only important in the determination of the origin of chemically-impure, oriented nanophase magnetitecrystals in general, but they also provide important insights into the origin of the large, euhedral, chemically-pure, [111]-elongated magnetites found within Ca-, Mg- and Fe-rich carbonates of the Martian meteorite ALH84001. Based on ourexperimental results, the chemically-pure magnetites within ALH84001 cannot be genetically related to the Ca-, Mg- andFe-rich carbonate matrix within which they are embedded, and an alternative explanation for their occurrence is warranted.� 2012 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

Nanometer-sized magnetite (Fe3O4) crystals having avariety of morphologies, size distributions and chemical

0016-7037/$ - see front matter � 2012 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.gca.2012.03.028

⇑ Corresponding authors. Tel.: +34 958 241000�20426; fax: +34958 249486 (C. Jimenez-Lopez), tel.: +34 958 246616; fax: +34 958243368 (C. Rodriguez-Navarro).

E-mail addresses: cjl@ugr.es (C. Jimenez-Lopez), carlosrn@ugr.es (C. Rodriguez-Navarro).

compositions have been recovered from modern and an-cient environments (Thomas-Keprta et al., 2000). Magne-tite is found on Earth in igneous, metamorphic andsedimentary rocks as well as in extraterrestrial materialsincluding meteorites and interplanetary dust particles.Whereas magnetites formed at high temperature are clearlyabiotic, those formed in low-temperature environments e.g.,Precambian stromatolites, deep-sea and lake sediments andsoils, may have an abiogenic or biogenic origin (Thomas-Keprta et al., 2000). Biogenic magnetite is known to form

70 C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80

through biologically-induced (BIM; Frankel and Bazylin-ski, 2003) or biologically-controlled mineralization (BCM;Bazylinski and Frankel, 2003). The origin of many low-temperature terrestrial magnetites is unknown, and thereis still great controversy over how to distinguish inorgani-cally- and biologically-produced sedimentary magnetites(Jimenez-Lopez et al., 2010). BCM magnetites have beenproposed as magnetofossils by many authors because theysimultaneously meet all the criteria summarized in theMAB package defined by Thomas-Keprta et al. (2000).BIM magnetites have not been generally accepted up tonow as magnetofossils, with the exception of the extraordi-nary tabular single-domain magnetite crystals induced byGeobacter metallireducens GS-15 (Vali et al., 2004). The cri-teria contained in the MAB package, especially morphol-ogy, crystal-size and size distribution (i.e., Arato et al.,2005) have been widely used for many years to recognizebiotic origin of natural terrestrial magnetites. However,many discussions have risen when trying to apply those cri-teria to extraterrestrial magnetites, and the debate still goeson (Thomas-Keprta et al., 2009). These problems must beresolved to recognize mineral biomarkers in terrestrialand extraterrestrial materials (Jimenez-Lopez et al., 2010).

In the laboratory, magnetite can be synthesized abioti-cally as a primary or secondary mineral phase. Secondarymagnetite forms through the transformation of Fe-bearingminerals at low or high temperature. At low temperature(<100 �C), magnetite can be formed in solution by recrys-tallization of ferrihydrite, green rust and FeOOH under an-oxic conditions (Ishikawa et al., 1998; Zachara et al., 2002).Alternatively, a solid-state transformation mechanism isproposed for the thermal decomposition of siderite(FeCO3), or other Fe-rich carbonates, at higher tempera-tures in the absence of oxygen. Such a mechanism may beresponsible for magnetite formation in natural environ-ments resulting from the contact of a carbonate precursorwith a magmatic fluid or from a meteoritic impact onsedimentary carbonates (Golden et al., 2001, 2004;Thomas-Keprta et al., 2009; Bell, 2007). The decompositionmechanism has been explored in some detail with distinctprecursor minerals and under different environmental con-ditions (Golden et al., 2001, 2004; Thomas-Keprta et al.,2009), including thermal shock under conditions mimickingmeteoritic impact (Bell, 2007). The thermal mechanism ofmagnetite formation has not been extensively explored withregard to the origin of terrestrial magnetites. Nevertheless,it has largely been debated in the context of the subset ofmagnetites with unusual chemical and physical propertiesthat are intimately associated with carbonate disks in theALH84001 Martian meteorite, related by some to a bacte-rial origin (McKay et al., 1996) and by others to an exclu-sively inorganic origin through the thermal decompositionof the carbonate matrix in which they are embedded (Gold-en et al., 2000, 2001, 2004; Brearley, 2003; Treiman, 2003;Bell, 2007). Golden et al. (2001) reported that magnetitecrystals produced by the thermal decomposition of a mix-ture of pyrite, siderite, magnesite and ankerite were chemi-cally-pure, single-domain and defect-free, similar to the[111]-elongated magnetite crystals observed in ALH84001.To bolster their results, Golden et al. (2006) thermally

decomposed Copper Lake siderite in vacuo (closed system)whereas Bell (2007) decomposed this Copper Lake sideriteby shocking samples to pressures up to 49 GPa. They ob-tained magnetites with compositions ranging from puremagnetite [when pyrite was present, (Golden et al., 2006)]to Mg-bearing magnetite [610 wt%, (Golden et al., 2006)and 620 wt%, (Bell, 2007)] and claimed that these magne-tites displayed the same range of sizes, compositions andmorphologies as inorganic magnetites synthesized by Gold-en et al. (2004). Although a chemically diverse suite ofmagnetites were produced in these studies, no reactionproducts contained Ca. On the one hand, this is not surpris-ing as Golden et al. (2006) did not report the presence of Cain their starting material. On the other hand, Bell (2007) re-ported 11–14 wt% CaCO3 in their siderite target, yet no Cawas observed in the solid phase product of the shock reac-tion. It is unclear why Ca was detected in the same startingmaterial in one study but not the other but it is clear that acomplex interaction of materials and processes influencesthe nature of solids produced by thermal decomposition.

Thomas-Keprta et al. (2009) challenged these results bythermally decomposing Roxbury siderite, which is consid-ered to be a terrestrial analoge of the Fe-rich componentof the carbonate disks in ALH84001 meteorite. They ob-served the formation of Mg- and Mn-magnetites and pro-posed that a CaCO3 component separated from the(Ca,Mg,Fe)CO3 solid solution prior to the formation ofthe magnetite. However, this phase, or CaO, was not ob-served in their experimental products.

To resolve discrepancies observed in these studiesinvolving Ca incorporation in magnetites, and to shed lighton signatures that may help recognize magnetite formationby the thermal decomposition of a carbonate matrix in nat-ural environments, we studied the chemical compositionand microtextural features of magnetites resulting fromthe decomposition (both thermally from 25 to 700 �C,and in situ, under the electron beam of a transmission elec-tron microscope) of a synthetic (Ca,Mg,Fe)CO3 (character-ized in Romanek et al. (2009), Fig. 1). In particular, the fateof calcium in the resulting magnetite was monitored, andthe structural relationship(s) between the precursor andproduct phases were determined.

2. METHODS

Synthetic mixed carbonate samples were provided byRomanek et al. (2009). They were precipitated in labora-tory experiments conducted at 25 and 70 �C in free-driftexperiments from solutions containing NaHCO3,Fe(ClO4)2, Mg(ClO4)2 and Ca(ClO4)2. The resulting solidswere identified by XRD as Mg and/or Ca-substituted side-rites (samples FD15 and FD16) and Fe and/or Ca-substi-tuted magnesites (samples FD20 and FD21). Nomagnetite was detected in these samples by XRD and/orTEM. The FD carbonates have a chemical composition(determined by inductively coupled plasma optical emissionspectrometry; ICP-OES; Perkin-Elmer 4300DVS) listed inTable 1. A comparison of the chemistry of these syntheticcarbonates with natural terrestrial and extraterrestrial(Ca,Mg,Fe)CO3 is presented in Fig 1. Each sample was

Fig. 1. (Ca,Mg,Fe)CO3 ternary diagram representing the compo-sition of FD samples used in this study in comparison to othermixed cation carbonates found in terrestrial environments andALH84001. Light grey field contains natural occurrences of sideriteand natural magnesites, and dark grey field is siderite–magnesitesolid solution boundary at 250 �C (Romanek et al., 2009); dash-linecircumscribed field represents ALH84001 carbonates from Treimanand Romanek (1998).

C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80 71

separated into two aliquots: (1) a dry, powder sample and(2) sliced sections embedded in resin. In the latter case, sam-ples were fixed with glutaraldehyde and post-fixed with os-mium tetraoxide. Afterwards, the sample was dehydratedwith ethanol, and embedded in Embed 812. Ultrathin sec-tions (50–70 nm) were prepared with a microtome (ReicherUltracut S microtome, DIATOME diamond blade),mounted onto copper grids and carbon coated.

Both the dry powder sample and the sliced sections of(Ca,Mg,Fe)CO3 were maintained in anaerobic conditionsinside an anaerobic chamber (Coy Laboratory ProductsInc.) and each decomposed to magnetite as describedbelow.

2.1. Procedure 1

Mixed cation carbonates were decomposed by heatingthe dry, powder FD samples by differential scanning calo-rimetry (DSC). A sample was weighed in an alumina cruci-ble to ±2 lg. The sample was then placed in the center ofthe hot zone of a small volume SENSYS evo DSC (SeteramInstrumentation; www.Setatam.com) differential scanningcalorimeter. The instrument was sealed, evacuated and thenflushed with pure CO2 carrier gas for 10 min. The calorim-eter was evacuated and back flushed with CO2 two addi-tional times to obtain a low background for O2, which isespecially critical for the DSC analysis of redox-sensitiveelements (e.g., Fe). The calorimeter was then purged withCO2 at a rate 5 cm3 min�1. After CO2 purged for 30 min,the sample temperature was ramped up to 700 �C at a rateof 20 �C min�1. The sample was held at 700 �C for 10 minand then ramped down to room temperature. In a duplicaterun, once the sample reached the maximum T and was thenramped down to room temperature (after 30 min), the

process was repeated to ensure complete reaction and toobtain a curve for baseline subtraction. No microstructural,mineralogical or chemical differences were observed be-tween magnetite crystals subjected to one or two heatingevents. Results reported here correspond to DSC-magne-tites heated twice.

Sufficient material was produced in the reaction to deter-mine the mineralogy of DSC-samples using a single-crystalX-ray diffractometer equipped with an area detector (Bru-ker D8 SMART APEX, Germany). A frame (or 2-dimen-sional (2D) diffraction pattern) was collected under thefollowing operating conditions: Mo Ka, 50 kV, 30 mA,0.5 mm collimator diameter and 120 s exposure time. TheXRD equipment used for the analyses has a CCD areadetector that can register a 2D diffraction pattern contain-ing complete Debye–Scherrer rings. These patterns were la-ter integrated and converted into conventional 2theta scanfor mineral identification. XRD2DScan software (Rodri-guez-Navarro, 2006) was used to convert these 2D diffrac-tion patterns into regular 2h linear scans. This softwarewas also used for background subtraction and integrationof peaks in the 2h scans. The integration of the intensityof rings allows better particle statistics even in very smallsamples (of a few lg) and the detection of individual min-eral grains of minor phases. The detection limit for minorphases is generally established to be a few percent units(<5%) for conventional X-ray powder diffractometers butthe detection limit of XRD equipment with area detectoris much lower (Bhuvanesh and Reibenspies, 2003; He,2009). In order to accurately determine the d-spacings forthe main magnetite reflections, LaB6 was used as an inter-nal standard for angular calibration. The unit cell parame-ter (a) was determined using a line regression model:dhkl = a/(h2 + k2 + l2)0.5. For this calculation, four of thestrongest reflections of magnetite (e.g., 113, 004, 115 and044) were considered.

The microstructure and composition of individual crys-tals was determined using a Philips model CM200 transmis-sion electron microscope (TEM) (Philips, Eindhoven, TheNetherlands; 200 kV, 10–20 lA, windows for analyses:100 � 200 nm) equipped with an EDAX solid-state energydispersive X-ray detector with an ultrathin window(UTW). For these analyses, DSC-magnetites were embed-ded in resin and sliced using the same protocol describedabove for the FD carbonate samples. At least 20 individualcrystals were analyzed for each decomposed mixed cationcarbonate sample. Quantitative analytical electron micros-copy (AEM) and selected area electron diffraction (SAED)analyses were performed on the same spots of individualcrystals on thin edges in scanning TEM (STEM) mode(<400 counts s�1) using a 4 nm diameter beam, a100 � 20 nm scanning area for AEM–TEM and a ca.500 nm diameter circular area centered in the squared areafor SAED. For AEM–TEM analyses, a low backgroundcondenser aperture and an analytical Be sample holder wereemployed to improve spectrum quality. Muscovite, albite,biotite, spersertine, olivine, and titanite standards were usedto obtain k-factors (kx/kSi), allowing X-ray intensities to becorrected for element analysis (Champness et al., 1982). Thedetermined k-factor for Ca is 1.15, 1.35 for Fe and 1.07 for

Tab

le1

An

alyt

ical

and

tran

smis

sio

nel

ectr

on

mic

rosc

op

y(A

EM

–TE

M)

anal

yses

of

DS

C-m

agn

etit

esan

dT

EM

-mag

net

ites

.±

=st

and

ard

dev

iati

on

.(C

a,M

g,F

e)C

O3

pre

curs

ors

dat

aar

eta

ken

fro

mR

om

anek

etal

.(2

009)

(an

alys

esp

erfo

rmed

by

ind

uct

ivel

yco

up

led

pla

sma

op

tica

lem

issi

on

spec

tro

met

ry).

Sam

ple

Pre

curs

or

carb

on

ates

(mo

l%)

Dec

om

po

siti

on

T(�

C)

DS

C-m

agn

etit

es(w

t.%

)T

EM

-mag

net

ites

(wt.

%)

Ca

±0.

1M

g±

0.2

Fe

±0.

3X

RD

Ca(

n)

±3

Mg(

n)

±4

Fe(

n)

±3

Ca(

n)

±3

Mg(

n)

±4

Fe(

n)

±3

FD

1527

.413

.359

.3S

ider

iteb

497

3.7–

5.6

(4)

3–6.

8(4

)54

.9–5

7.9

(4)

7.1–

11.2

(8)

2.1–

4.3

(8)

56.2

–74.

3(8

)F

D16

012

.387

.7S

ider

iteb

490

nd

(11)

0.5–

8.7

(11)

a57

.8–6

3.3

(11)

nd

(10)

2.7–

8.2

(10)

48.8

–65.

2(1

0)F

D20

21.7

54.6

23.7

Mag

nes

iteb

577

0.3–

13.2

(8)

6.8–

33.4

(8)

21.9

–54.

3(8

)7.

5–16

.4(8

)24

–33

(8)

17.3

–50.

7(8

)F

D21

070

.929

.1M

agn

esit

eb56

0n

d(6

)1.

4–32

.9(6

)26

.8–6

8.5

(6)

nd

(9)

6.7–

41.4

(9)

12.6

–39.

8(9

)

n=

nu

mb

ero

fcr

ysta

lsan

alyz

ed.

nd

=n

on

det

ecte

d.

aM

ost

of

the

sam

ple

s(1

0/11

)ex

hib

ited

Mg

wt%

P4.

2%.

bX

RD

pat

tern

shif

ted

wit

hre

spec

tto

pu

reen

dm

emb

er,

see

Ro

man

eket

al.

(200

9)fo

rd

etai

ls.

72 C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80

Mg. Based on these k-factors, average errors for analyzedelements (two standard deviations), expressed as a percent-age of the atomic proportions, are 5 (Mg), 3 (Ca), and 3(Fe). Determination of the mineralogy of the crystals ana-lyzed by AEM–TEM was performed by calculating d-spac-ings from SAED patterns, using the same analyticalwindow. Measured d-spacings were compared to JCPD filesfrom the International Center for Diffraction Data (ICDD)for mineral identification.

2.2. Procedure 2

Sliced sections of FD samples were decomposed in vacuo

under the electron beam in a Philips model CM200TEMoperating at 200 kV (Eindhoven, The Netherlands; 10–20 lA, beam spot size: 100 nm). Selected area electron dif-fraction (SAED) was performed to determine the zone axisof each crystal. Measured d-spacings were compared toJCPD files from the ICDD for both mineral identificationand the determination of Miller Indexes of atomic planesparallel to the electron beam. The individual mixed cationcarbonate crystal, held in the determined orientation, wasthen kept under the electron beam for 5–20 min to decom-pose it into magnetite. Decomposition was only achievedwhen using a large condenser aperture (200 lm). Afterphase transformation, SAED analysis was performed todetermine the zone axis of the resulting product.

This procedure was performed on �8 crystals from eachmixed cation carbonate sample listed in Table 1 (with theexception of FD15). The resulting set of magnetite crystalsare referred to as TEM-magnetites. The chemical composi-tion of individual mixed cation carbonate precursors andthe resulting TEM-magnetite crystals were analyzed byAEM–TEM, as described above for the DSC magnetites.

In the present study, low-substituted magnetites wouldbe called (Ca,Mg)-magnetites while high-substitutedmagnetites would be addressed as (Ca,Mg)-ferrites, follow-ing the terminology in Orewczyk (1990) and Thomas-Keprta et al. (2009).

3. RESULTS AND DISCUSSION

3.1. Chemical signatures in magnetites formed by the

decomposition of synthetic (Ca,Mg,Fe)CO3

3.1.1. DSC-decomposition

Each mixed cation carbonate sample that was heated byDSC completely decomposed pseudomorphically into aporous aggregate of magnetite nanocrystals (Fig. 2) that re-tained the gross morphology of the precursor (Fig. 2c andd). None of the original carbonate matrix was detected byXRD in any case. Magnetite formed as a consequence ofthe decomposition of the precursor carbonate, in the ab-sence of oxygen, but in the presence of CO2. Note thatthe thermal decomposition of siderite in the absence of O2

results in the formation of wustite (FeO) and CO2. CO2 actsas an electron acceptor, thereby oxidizing the Fe2+ to Fe3+,during the reaction CO2! CO + 1/2O2 (Gallagher andWarne, 1981). As a result, part of the Fe2+ in wustite is

C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80 73

oxidized to Fe3+, resulting in magnetite as the final solidphase (Gallagher and Warne, 1981; Brearley, 2003). Be-cause of the molar volume differences between carbonateprecursor and product oxide, a significant porosity (ca.50%) was generated.

Non-isothermal DSC results displayed a maximumendothermal peak (representing carbonate decomposition)at 497 �C for DSC–FD15, 490 �C for DSC–FD16(Fig. 2a), 577 �C for DSC–FD20 (Fig. 2b) and 560 �C forDSC–FD21 (Table 1). Because sample size/morphology,mass, and heating rate were similar for the differentDSC–FD samples, variations in the decomposition temper-ature were mainly related to the carbonate composition,although other factors such as grain size and crystallinitymay also account for slight changes in peak characteristics.The higher the Ca and Mg content (FD15 vs. FD20) thehigher the decomposition temperature; conversely, thehigher the Fe content (FD15 vs. FD20 and FD16 vs.FD21) the lower the decomposition temperature (Table 1).This is consistent with differences in decomposition temper-ature for the pure end-member minerals (Hurst et al., 1993;Brearley, 2003).

The size of DSC-magnetite crystals ranged from 5 to40 nm and their morphologies ranged from hexagonal, torectangular, rhombic and irregular (Fig. 2i and g). Well-developed chains of magnetite crystals were detected in allsamples (Fig. 2g and e) with the exception of DSC–FD20.DSC-magnetites displayed a preferred crystallographic ori-entation with respect to each other, as revealed by bothbrightfield and darkfield TEM images (Fig. 2g; darkfieldimages not shown). Because the DSC–FD20 precursorwas composed of an aggregate of randomly oriented submi-crometer- to micrometer-sized carbonate crystals formingspherulites (Romanek et al., 2009), a preferred orientation(see the absence of Debye rings in SAED of Fig. 2h) oc-curred in small magnetite aggregates (ca. 200 nm in size;see squared section in Fig. 2h) corresponding to the individ-ual crystals in the precursor.

Bulk X-ray diffraction (XRD) analyses of the decom-posed solids revealed the presence of magnetite for samplesDSC–FD15 and DSC–FD21, and a mixture of magnetiteand a small percentage of hematite for samples DSC–FD16 and DSC–FD20 (possibly due to partial oxidationof Fe2+ during the analysis). No other minerals were de-tected by XRD (e.g., CaO, CaCO3, and/or MgO minerals).Furthermore, systematic shifts in the d-spacings of magne-tite reflections were detected, indicating that there is a mod-ification of the unit cell dimensions due to the incorporationof Ca and/or Mg into the magnetite structure. The presenceof Ca and Mg in magnetite was also confirmed by the detec-tion of these cations by AEM–TEM in sliced sections of themagnetite crystals, as discussed below. In the case of sampleDSC–FD16, XRD analyses reveal a significant contractionof the unit cell dimensions (a = 8.253 ± 0.005 A, comparedto a = 8.396 A for pure magnetite reported in the ICCDPDF 16–629). In sample DSC–FD20, there is only a slightcontraction in the dimensions of the unit cell(a = 8.347 ± 0.005 A). In contrast, sample DSC–FD15shows a significant expansion of the unit cell dimensions

(a = 8.495 ± 0.005 A). The contraction or expansion ofthe unit cell is due, respectively, to the incorporation ofsmaller Mg2+ (i.e., DSC–FD16) or larger Ca2+ (i.e.,DSC–FD15) ions into the magnetite structure, consistentwith results of the AEM–TEM analyses. No evidence ofoxidation of magnetite to maghemite (c-Fe2O3) was de-tected by XRD.

Selected area electron diffraction (SAED) and analyticaland transmission electron microscopy (AEM–TEM) wereperformed sequentially on the same selected areas whileilluminating one or several nanocrystals of each DSC–FDsamples. In all cases, SAED data yielded the typical d-spacings for magnetite, although these d-spacings were low-er than those of magnetite in DSC–FD16, DSC–FD20 andDSC–FD21 samples and slightly larger than those of mag-netite in DSC–FD15. These shifts could not be attributed tocation incorporation due to error associated with theSAED measurements. AEM–TEM analyses of the same se-lected area showed the presence of varying amounts of Caand Mg in the magnetite nanocrystals, forming (Ca,Mg)–magnetites and (Ca,Mg)–ferrrites. DSC–FD15 and DSC–FD20 crystals showed the presence of 66.8 wt% (11 wt%in precursor) and 33 wt% (48 wt% in precursor) Mg,respectively, and the presence of 65.6 wt% (26 wt% in pre-cursor) and 13.2 wt% (23 wt% in precursor) Ca, respec-tively (Table 1). Analyses of individual crystals fromDSC–FD16 and DSC–FD21 (no Ca in precursor) alsoshowed the presence of 69 wt% Mg (9 wt% in precursor)and 33 wt% Mg (64 wt% in precursor) Mg, respectively.Variability of Mg and Ca content between single crystalsis attributed to chemical zoning in the precursor material,since these carbonates were grown in closed, free-driftexperiments, in which the relative concentration of aqueousiron, calcium and magnesium varied throughout the time-course experiment (Romanek et al., 2009).

3.1.2. In situ TEM decomposition

After exposure to the electron beam for �20 min in theTEM, each mixed carbonate crystal decomposed to a pseu-domorph consisting of a porous aggregate of magnetitenanocrystals �5–10 nm in size (Fig. 3). As for the DSC-samples, SAED and AEM–TEM analyses were performedsequentially on a specified area of individual crystals.SAED analyses yielded typical magnetite d-spacings. As ob-served in DSC magnetites, the calculated d-spacings wereshifted to lower values in TEM–FD16, TEM–FD20 andTEM–FD21 and to higher values in TEM–FD15. AEM–TEM analyses of TEM–FD15 and TEM–FD20 crystalsshowed that they contained up to 4.3 and 33 wt% Mg,respectively, and up to 11.2 and 16.4 wt% Ca, respectively(Table 1), whereas TEM–FD16 and TEM–FD21 crystals(no Ca in precursor) contained up to 8.2 and 41 wt% Mg,respectively. As in the case of DSC magnetites, the variabil-ity of Mg and/or Ca within crystals of the same sample iscorrelated to the heterogeneous composition of the carbon-ate precursor. Although the Mg and/or Ca concentrationsfor some magnetites were close to the detection limit (±4and 3 wt%, respectively), most of the crystals exhibited con-centrations well above the detection limit.

Fig. 2. (a and b) DSC spectra of FD16 (a) and FD20 (b) samples. Both spectra show one endothermal peak, indicating the thermaldecomposition of the carbonate; (c and d) TEM micrograph of mixed carbonate crystals from FD16 (c) and FD20 (d). Inset: AEM–TEManalysis showing the presence of Mg and Ca; (e and f) TEM images of product pseudomorphs showing porous aggregates of magnetitecrystals that preserve the morphology of the precursor [DSC–FD16 (e) and DSC–FD20 (f)]. (g, h, i, and j) TEM images of DSC–FD16 (g andi) and DSC–FD20 (h and j) magnetite crystals. The arrows indicate magnetite crystals with facing corners. SAED patterns (d-spacingcorresponding to magnetite) and AEM–TEM analyses (showing Mg and Ca) are included. AEM–TEM spectra: Mg is marked with a solidarrow and Ca is marked with a dotted arrow. Squares correspond to areas analyzed by AEM–TEM. SAED patterns correspond to a ca.500 nm diameter circular area centered in the squared areas.

74 C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80

Fig. 3. Brightfield TEM images of mixed-cation carbonate crystals.(a) carbonate crystals FD21 prior to e-beam induced decomposi-tion; (b) detail of the crystal in the center of (a) (arrow) beforedecomposition, observed along the [010]carbonate zone axis (SAEDpattern in inset); (c) carbonate crystal in (b) after exposure to theelectron beam, now transformed into a mesoporous pseudomorphof magnetite nanocrystals (with size 65 nm) showing a preferredcrystallographic orientation as shown by the discrete SAED spots(inset) for the �222 (d222 = 2.42 A), 400 (d400 = 2.09 A) and �140(d440 = 1.48 A) reflections, corresponding to the [110]magnetite zoneaxis.

C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80 75

3.1.3. Chemical signatures observed in DSC- and TEM-

magnetites

Considered collectively, XRD, SAED and AEM–TEMresults show that the decomposition of the mixed carbonateresults in the formation of (Ca,Mg)-substituted magnetiteto (Ca,Mg)–ferrites as a solid solution, and not a mixtureof separate phases (CaO + MgO + Fe3O4 or CaO +(Mg,Fe)3O4). Moreover, these data support the contentionthat spinodal decomposition of the substituted magnetitesinto pure end-members does not occur, nor does the re-car-bonation of the oxides. The formation of a Mg-substitutedmagnetite (solid solution) from the thermal decompositionof Mg-substituted siderites is consistent with results re-ported by other researchers (Gallagher and Warne, 1981;Gallagher et al., 1981; Dubrawski, 1991; Gotor et al.,2000; Cohn, 2006; Isambert et al., 2006; Bell, 2007;Thomas-Keprta et al., 2009).

One of the most interesting results of this study is thatCa was incorporated in magnetite as a result of the carbon-ate mineral decomposition. These results are intriguing be-cause Mg and/or Mn are known to co-precipitate inmagnetite (ionic radius of 72 and 83 pm, respectively; Shan-non, 1976), but the incorporation of Ca is more difficult be-cause the ionic radius of Ca is relatively large compared tothat of Fe2+ [ionic radius of 100 pm (Shannon, 1976) and78 pm (Bloss, 2000), respectively]. Whereas the substitutionof calcium in magnetite is not widely recognized in naturalsettings, it may occur by thermal decomposition mecha-nism. Vidyasagar et al. (1984) demonstrated that a varietyof complex metal oxides, including calcium ferrite (Ca-Fe2O4) and brownmillerite (Ca2Fe2O5), can be producedby the thermal decomposition of CaCO3–FeCO3 solid solu-tions under air or oxygen at >400 �C. Calcium magnetitehas been synthesized at high temperature from CaO– or cal-cite–hematite melts (Hrynkiewicz et al., 1971; Orewczykand Jasienska, 1987; Orewczyk et al., 1995; Orewczyk,2000), and low-temperature Ca–magnetite is documentedin the remediation of acid mine drainage waters (Wanget al., 1996; Morgan et al., 2003).

In contrast, other researchers have not observed theincorporation of Ca in magnetite during the thermaldecomposition of mixed-cation siderite (Gallagher andWarne, 1981; Gallagher et al., 1981; Ware and French,1984; Dubrawski, 1991; Hurst et al., 1993; Isambert andValet, 2003; Cohn, 2006; Isambert et al., 2006; Bell, 2007;Thomas-Keprta et al., 2009). Thomas-Keprta et al. (2009)suggest that the decomposition of an ankerite–dolomite so-lid solution would follow a sequential pathway in which thesolid solution precursor separates first into calcite and a sid-erite–magnesite phase prior to the nucleation and growth ofthe magnetite. This proposed mechanism would yield eitherCaCO3 (or CaO) in the resulting products, but these phaseswere not observed in this study or those obtained by others(Gallagher and Warne, 1981; Dubrawski, 1991; Cohn,2006; Isambert et al., 2006; Thomas-Keprta et al., 2009).The failure to detect Ca in magnetite is most likely due tothe low concentration of Ca in the precursor solids usedin previous decomposition studies [maximum 1.3 wt%(Gallagher and Warne, 1981); trace amounts (Gallagheret al., 1981; Cohn, 2006); maximum 1.8 wt% (Isambert

et al., 2006; shock pressures 8.4–25.9 GPa); 1.8 wt%(Thomas-Keprta et al., 2009)] compared to the presentstudy (up to 25.5 wt% in FD15 and 22.8 wt% in FD20).

The effect of temperature on the chemistry of the result-ing solid may also be critical. High temperatures couldfacilitate the spinodal separation of CaO from a Mg-magnetite (or Mg-ferrite) over time, as reported for thedecomposition of double carbonates (Spinolo and

76 C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80

Anselmi-Tamburini, 1989). Nevertheless, the results in thepresent study suggest that this process does not occur.Firstly, no CaO, calcite or other Ca-bearing mineral phasewas detected by bulk XRD of the DSC-magnetites. Sec-ondly, our results support the contention that the thermalhistory of the samples does not affect the chemical signatureinherited from the precursor. In fact, TEM-magnetites(and/or ferrites) and DSC-magnetites (and/or ferrites) showthe same chemical signatures, although they had differentthermal histories [TEM-magnetites, analyzed directly afterdecomposition, were exposed to electron-beam heating(Egerton and Malac, 2004) at a T corresponding to thedecomposition of siderite/impure-siderite in vacuum, i.e.,�385–450 �C (Brearley, 2003 and references therein),whereas DSC-magnetites were heated to 700 �C (highertemperatures and time after decomposition compared toTEM-magnetites)]. Therefore, higher temperatures and/orlonger heating times do not cause the spinoidal separationof the Ca and/or Mg-magnetites (and/or (Ca,Mg)–ferrites).The main difference between these two sets of samples(TEM- and DSC-) is the crystal size (�5 nm for TEM-magnetites and �40 nm for DSC-magnetites). These differ-ences are caused by the coarsening of the oxide nanocrys-tals formed after decomposition [as previously reportedfor siderite (Angermann et al., 2010) and other carbonates(Rodriguez-Navarro et al., 2009)].

The formation of metastable solid solutions (rather thanspinoidal separation), as a result of a thermal decomposi-tion process, is also described for other double carbonates,such as dolomite (Spinolo and Anselmi-Tamburini, 1989)and impure siderite (Angermann et al., 2010). In fact, thethermal decomposition of double carbonates has been pro-posed as an efficient method for the low-T synthesis of(metastable) mixed metal oxides (e.g., (Mg,Ca)O (Spinoloand Anselmi-Tamburini, 1989) and (Mn,Zn,Fe)3O4 (Anger-mann et al., 2010)). The resulting mixed oxides form via dif-fusionless, topotactic replacement of the parent carbonate(Spinolo and Anselmi-Tamburini, 1989), thus enabling theincorporation of metals cations of dissimilar radii in theoxide. According to the model proposed by Rodriguez-Navarro et al. (2009), the topotactic transformation occursas a consequence of CO2 loss, accompanied by limited atomdisplacement, and shrinkage along specific [hkl] directionsin the transformed carbonate. During this process, newtransient structures appear as well as empty spaces (vacan-cies) between cations (Rodriguez-Navarro et al., 2009). Cal-cium fills these spaces as the transient structure evolves inthe inverse spinel oxide. Such metastable solid solutionspersist over several hundred degrees above the decomposi-tion temperature (Spinolo and Anselmi-Tamburini, 1989;Angermann et al., 2010). This explains why magnetites inthe DSC and TEM experiments include Ca and/or Mg de-spite their different thermal history and the different relativeconcentration of cations in the precursor samples.

3.2. Topotaxy between the precursor (Ca,Mg,Fe)CO3 and

the resulting magnetite

An additional experiment was performed with TEM–FD16, TEM–FD20 and TEM–FD21 to determine whether

the transformation of the mixed carbonate to magnetitewas topotactic (signifying a three-dimensional lattice conti-nuity) between the reactant and product phases (Bloss,2000). This study was performed to determine whether ornot there was a genetic structural relationship between theprecursor carbonate and the resulting magnetite to explainthe pseudomorphic relationship between reactant and prod-uct observed in DSC-magnetites as well as the preferredcrystallographic orientation among DSC-magnetites. Ulti-mately, this study provides a mechanistic model for theincorporation of Ca and/or Mg in magnetite.

Fig. 3a and b show mixed carbonate crystals (TEM–FD21) prior to decomposition by the electron beam. TheSAED pattern shows that this crystal was a single phase,not a mixture of various carbonate phases. The crystal isoriented along the [010] zone axis (SAED in inset inFig. 3b). After irradiation with the electron beam for20 min, using a large condenser-lens aperture, a porouspseudomorph comprised of nanometer-sized magnetitecrystals was produced (Fig. 3c). The pseudomorph had atypically mottled texture that is characteristic of beam-damaged carbonates (Wenk et al., 1983). The SAED pat-tern of the magnetite pseudomorph shows diffuse Debyerings where maxima are clearly observed, indicating thatthe magnetite nanocrystals display a highly orientedarrangement, consistent with the bright field image of thepseudomorph presented in Fig. 3c. In fact, the SAED inFig. 3c shows that the [110] zone axis of magnetite crystalsis parallel to the [010] zone axis of the precursor carbonate(Fig. 3b). A second systematic orientation relationship isthe ½�441�carb//[110]mag. It follows that two of the periodicbond chains (PBC) of the carbonate precursor (i.e., thoseparallel to h�441i and [010] directions) transform into the<110> direction in the oxide product. These topotactic ori-entation relationships, observed in all samples studied (seealso Fig. 4), are identical to those found in other carbonatessuch as calcite, whose thermal decomposition is also topo-tactic (Rodriguez-Navarro et al., 2009). Interestingly, otherresearchers such as Thomas-Keprta et al. (2009) and Bell(2007) have shown SAED patterns of magnetite nanocrys-tal produced upon thermal decomposition of Fe-rich car-bonates with similar features as those presented here,demonstrating a preferred orientation of the magnetitecrystals [see Fig. 6b in Thomas-Keprta et al. (2009) andFig. 9 in Bell (2007)]. Nevertheless, these authors do notcomment on the fact that their observations indicate atopotactic decomposition of the precursor carbonate.

Rodriguez-Navarro et al. (2009) reported that oxidenanocrystals (about 5 nm in size), formed during the earlystages of in situ carbonate (calcite) decomposition in theTEM, aggregated in an oriented fashion following the col-lapse of the mesoporous structure, similar to that shown inFig. 3c. This transformation is associated with a release ofmechanical stress from the mismatch between the structureof the reactant and product phase during the topotacticdecomposition. After this diffusionless coarsening process,sintering associated with increased temperature and heatingtime contributes to the final growth of the crystals to pro-duce the grain sizes observed in DSC experiments whenheating continues above the decomposition temperature

Fig. 4. TEM of carbonate crystals FD16 (a) and a FD20 (e) before electron beam decomposition with their corresponding AEM–TEManalysis (inset) and SAED patterns [(b) and (f) for FD16 and FD20, respectively]. TEM of the carbonate (pseudomorphs) transformed intomagnetite nanocrystals after electron-beam heating: TEM–FD16 (c) and TEM–FD20 (g). AEM–TEM analyses (inset) and SAED patterns[(d) and (h) for TEM–FD16 and TEM–FD20, respectively]. Squared areas correspond to those sequentially analyzed by SAED and AEM–TEM. SAED patterns correspond to a ca. 500 nm diameter circular area (centered in the squared areas) and show that the magnetitenanocrystals display a preferred crystallographic orientation with the precursor carbonate (i.e., absence of Debye rings). Solid and dashedarrows in AEM–TEM spectra indicate the peaks for Mg and Ca, respectively.

C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80 77

of carbonates (Rodriguez-Navarro et al., 2009; Angermannet al., 2010). Most important, the coarsening process doesnot modify the crystallographic relationship between parentand product phases, nor the chemical signature (i.e., Caand/or Mg incorporation), thus preserving the original

topotactic relationship and the chemical fingerprint undertemperature conditions spanning from very low-T (�382–450 �C, TEM) up to high-T (700 �C, DSC). In fact, ther-mally-activated decomposition of carbonates such as dolo-mite, ankerite, siderite, magnesite, ottavite, smithsonite,

78 C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80

cerusite, and calcite (Rodriguez-Navarro et al., 2009 andreferences herein) has been shown to be topotactic andnot dependent on heating rate, peak T or heating time,and pCO2 (Rodriguez-Navarro et al., 2009). Thus, a similarcrystallographic relationship is expected between reactant/product regardless of the thermal history of the sample.

3.3. Implications for terrestrial and extraterrestrial

magnetites

The experiments from the present study demonstrate thatmagnetite produced by the thermal decomposition of(Fe,Mg,Ca)CO3 preserves a legacy of the geochemistryand structural attributes of the phase from which it formed.These signatures could be important in recognizing the ori-gin of some terrestrial and extraterrestrial nanomagnetitesformed through this inorganic pathway. In this context,the chemical and structural signatures observed in this studycould provide insights, for example, in recognizing the ori-gin of nanophase (Mg,Ca)-magnetites, that may result fromthe thermal decomposition of (Fe,Mg,Ca)CO3 under a CO2

rich atmosphere [as likely existed on Early Earth (Ohmotoet al., 2004) and possibly on Early Mars], following eventssuch as meteoritic impacts or metasomatism.

In the case of magnetites of the ALH84001 Martianmeteorite, our results suggest that aggregates of anhedral,fine-grained magnetites observed by Barber and Scott(2003), typically near fracture zones, are consistent withthe partial decomposition of ALH84001 carbonates, sincethese authors show a crystallographic orientation with the

Fig. 5. Orientational relationships between siderite and magnetite structspace group, showing the stacking of alternate layers of Fe2+ and CO2�

3 irhombohedral cleavage plane (right) shows the different PBCs (Periodmagnetite, Fd3 m space group, showing the orientation of the [110] and [�1carbonate (10�14) planes, and PBC directions, and the magnetite crystals fo<�441>carbonate // [110]magnetite and [010]carbonate // [110]magnetite; (d) examthermal decomposition of the siderite precursor in the DSC and observe

surrounding carbonate similar to that observed in the pres-ent study. This thermal decomposition probably occurredat a temperature not intense enough to fully decomposethe carbonate and to coarsen the magnetites to produce lar-ger euhedral crystals. On the other hand, the textural andchemical analyses presented in this study show that chemi-cally pure, euhedral magnetite crystals ranging in size be-tween 30 and 50 nm in ALH84001 carbonate disks arenot consistent with a thermal decomposition event. It isimportant to note that the chemical purity of these magne-tites is still being debated. Some authors claim that theycontain up to 3% Mg (Golden et al., 2006; Bell, 2007) whileothers defend the chemical purity of these magnetites(Thomas-Keprta et al., 2000). Nevertheless, and in thecontext of structural relationships, Barber and Scott(2002) concluded that there was topotaxy between euhedralmagnetite nanocrystals and Ca,Mg,Fe-carbonates inALH84001. Note, however that their conclusions werebased on the orientation relationship defined by McTigueand Wenk (1985) in calcite. The latter have been recentlyreinterpreted by Rodriguez-Navarro et al. (2009) as fol-lows: during a thermal decomposition event, and uponthe devolatilization of CO2, the CO2�

3 groups in the PBCdirections of siderite (Fig. 5a) convert to O2� groups, whichsubsequently rearrange along with metal cations along the[110] direction for the product oxide (Fig. 5b). In otherwords, magnetites align along the [110] directions, whichare parallel to the original PBC direction of the carbonateprecursor (Fig. 5c and d). Such an orientation relationshipbetween the carbonate precursor and the chemically-pure,

ures. (a) structure (left) of the hexagonal unit cell of siderite, R�3cons along the c-axis. The projection of the structure on the (10 �14)ic Bond Chains) existing in this carbonate; (b) the structure of11] directions; (c) structural/orientational relationships between thermed upon thermal decomposition of the carbonate precursor, with

ple of magnetite crystals aligned // to [110]magnetite formed upond in the TEM.

C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80 79

euhedral [111]-elongated magnetite crystals has not beendescribed in ALH84001. Furthermore, the absence ofporosity and the presence of carbonates that fully surroundthe euhedral, >50 nm magnetites are inconsistent with thetopotactic decomposition of the carbonate into magnetite.In fact, whereas some degree of sintering (at temperatureshigher than those corresponding to the ferroan carbonatedecomposition temperature) is required to reach those sizes(>50 nm) for the magnetite crystals, the surroundingALH84001carbonates are intact (Barber and Scott, 2002;Thomas-Keprta et al., 2009). These critical points werenot addressed by Barber and Scott (2002). Alternatively,the relationships determined by Barber and Scott (2002)may indicate an epitaxial overgrowth of the ferrous carbon-ate around pre-existing magnetite crystals, as was suggestedby Thomas-Keprta et al. (2009), and, thus, a genetic rela-tionship between these two phases is not guaranteed. Final-ly, if thermal decomposition of the Ca,Mg,Fe-ALH84001carbonates was the origin of the euhedral, [111]-elongated,chemically-pure subset of ALH84001 magnetites, otherchemically-pure oxides (MgO and CaO) should have beendetected, in contrast with findings of the present studyand those of Thomas-Keprta et al. (2009). Furthermore,no CaO (or CaCO3), that could be indicative of the spinoi-dal separation of an intermediate mixed oxide, has beenfound associated with magnetite in ALH84001. With re-spect to MgO, Barber and Scott (2002, 2003) describe ori-ented “foamy” MgO nanocrystals (ca. 3 nm in size)associated with nanopores and in contact with partiallydecomposed magnesite, but none in association with ferro-an carbonates.

The chemical and structural fingerprints of magnetiteproduced through thermal decomposition of a precursorphase lead to the conclusion that large, chemically-pure,euhedral, [111]-elongated ALH84001 magnetites are allo-chthonous with respect to the carbonate matrix in whichthey are embedded, supporting the conclusions ofThomas-Keprta et al. (2009). Therefore, these magnetitescould not be produced by the thermal decomposition ofthe carbonate matrix in which they are embedded, and analternative mechanism for their formation must be sought.

ACKNOWLEDGEMENTS

Financial funding was provided by Grants CGL2007-63859,CGL2010-18274 and MAT2009-11332 from the Spanish Ministryof Science and Innovation (MCI) and GREIB (BIO103) fromJunta de Andalucia. This research was also funded in part by aGrant to C.S.R. through the NASA Astrobiology Institute.D.A.B. is supported by US National Science Foundation GrantEAR-0920718 and A.R.N. by Grant CTM2007-65713. Thanks goto Johnson Space Center (NASA) for the DSC analyses, and tothe CIC personal from the University of Granada for technicalassistance. We also thank A. Mucci and H. Vali for valuable com-ments that greatly improved this manuscript.

REFERENCES

Angermann A., Hartmann E. and Topfer J. (2010) Mixed-metalcarbonates as precursors for the synthesis of nanocrystallineMn–Zn ferrites. J. Magn. Magn. Mater. 322, 3455–3459.

Arato B., Szanyi Z., Flies C., Schuler D., Frankel R. B., Buseck P.R. and Posfai M. (2005) Crystal-size and shape distributions ofmagnetite from uncultured magnetotactic bacteria as a poten-tial biomarker. Am. Mineral. 90, 1233–1241.

Barber D. J. and Scott E. R. D. (2002) Origin of supposedlybiogenic magnetite in the Martian meteorite Allan Hills 84001.Proc. Natl. Acad. Sci. USA 99, 6556–6561.

Barber D. J. and Scott E. R. D. (2003) Transmission electronmicroscopy of minerals in the martian meteorite Allan Hills84001. Meteorit. Planet. Sci. 38(6), 831–848.

Bazylinski D. A. and Frankel R. B. (2003) Biologically controlledmineralization in prokaryotes. Rev. Mineral. Geochem. 54, 95–

114.

Bell M. S. (2007) Experimental shock decomposition of siderite andthe origin of magnetite in Martian meteorite ALH84001.Meteorit. Planet. Sci. 42, 935–949.

Bloss D. (2000) In Crystallography and Crystal Chemistry (eds.MSA). Washington, DC.

Bhuvanesh N. S. P. and Reibenspies J. H. (2003) A novel approachto micro-sample X-ray powder diffraction using nylon loops. J.

Appl. Crystallogr. 36, 1480–1481.

Brearley A. J. (2003) Magnetite in ALH 84001: an origin by shock-induced thermal decomposition of iron carbonate. Meteorit.

Planet. Sci. 38, 849–870.

Champness P. E., Cliff G. and Lorimer G. W. (1982) Quantitativeanalytical electron microscopy of metals and minerals. Ultra-

microscopy 8, 121–132.

Cohn A. (2006) Formation of magnetite nanoparticles by thermaldecomposition of iron bearing carbonates: implications for theevidence of fossil life on Mars. In National Nanotechnology

Infrastructure Network, Research Experience for Undergradu-

ates, Accomplishments Program (Cornell, NY, USA). pp. 58–59.

Dubrawski J. V. (1991) Thermal decomposition of some siderite–magnesite minerals using DSC. J. Therm. Anal. 37, 1213–1221.

Egerton R. F. and Malac P. L. M. (2004) Radiation damage in theTEM and SEM. Micron 35, 399–409.

Frankel R. B. and Bazylinski D. A. (2003) Biologically inducedmineralization by bacteria. Rev. Mineral. Geochem. 54, 217–

237.

Gallagher P. K. and Warne S. S. J. (1981) Thermomagnetometryand thermal decomposition of siderite. Thermochim. Acta 43,

253–267.

Gallagher P. K., West K. W. and Warne S. S. J. (1981) Use ofMossbauer effect to study the thermal decomposition ofsiderite. Thermochim. Acta 50, 41–47.

Golden D. C., Morris R. V., Yang S. V. and Lofgren G. E. (2000)An experimental study on kinetically-driven precipitation ofcalcium–magnesium–iron carbonates from solution: implica-tions for the low-temperature formation of carbonates inmartian meteorite Allan Hills 84001. Meteorit. Planet. Sci. 35,

457–465.

Golden D. C., Ming D. W., Schwandt C. S., Lauer H. V., Socki R.A., Morris R. V., Lofgren G. E. and McKay G. A. (2001) Asimple inorganic process for formation of carbonates, magne-tite and sulfides in Martian meteorite ALH84001. Am. Mineral.

86, 370–375.

Golden D. C., Ming D. W., Morris R. V., Brearley A., Lauer H.V., Treiman A. H., Zolensky M. E., Schwandt C. S., Lofgren G.E. and McKay G. A. (2004) Evidence for exclusively inorganicformation of magnetite in Martian meteorite ALH84001. Am.

Mineral. 89, 681–695.

Golden D. C., Ming D. W., Lauer Jr. H. V., Morris R. V., TreimanA. H. and McKay G. A. (2006) Formation of “chemicallypure” magnetite from Mg–Fe–carbonates: implications for theexclusively inorganic origin of magnetite and sulfides in martian

80 C. Jimenez-Lopez et al. / Geochimica et Cosmochimica Acta 87 (2012) 69–80

meteorite ALH84001. Lunar Planet. Sci. XXXVII. LunarPlanet. Inst., Houston. # 1199 (abstr.).

Gotor F. J., Macias M., Ortega A. and Criado J. M. (2000)Comparative study of the kinetics of the thermal decompositionof synthetic and natural siderite samples. Phys. Chem. Miner.

27, 495–503.

He B. B. (2009) Two-Dimensional X-Ray Diffraction. John Wileyand Sons, Hoboken, NJ.

Hurst H. J., Levy J. H. and Patterson J. H. (1993) Sideritedecomposition in retorting atmospheres. Fuel 72, 885–890.

Hrynkiewicz A. Z., Kulgawczuk D. S., Mazanek E. S., PustowkaA. J., Sawicki J. A. and Wyderko M. E. (1971) Mossbauer effectstudy of calcium magnetites. Phys. Stat. Sol. B 43, 401–405.

Isambert A. and Valet J. P. (2003) Stable Mn-magnetite derivedfrom Mn-siderite by heating in air. Earth Planet. Sci. Lett. 108,

EMP2-1–EMP2-9.

Isambert A., De Resseguierm T., Gloter A., Reynard B., Guyot F.and Valet J. P. (2006) Magnetite-like nanocrystals formed bylaser-driven shocks in siderite. Earth Planet. Sci. Lett. 243, 820–

827.

Ishikawa T., Kondo Y., Yasukawa A. and Kandori K. (1998)Formation of magnetite in the presence of ferric oxyhydroxides.Corros. Sci. 40, 1239–1251.

Jimenez-Lopez C., Romanek C. S. and Bazylinski D. A. (2010)Magnetite as a prokaryotic biomarker: a review. J. Geophys.

Res. 115, G00G03. http://dx.doi.org/10.1029/2009JG001152.

McTigue J. W. and Wenk H. R. (1985) Microstructures andorientation relationships in the dry-state aragonite–calciteand calcite–lime phase transitions. Am. Mineral. 70,

1253–1261.

McKay D. S., Gibson, Jr., E. K., Thomas-Keprta K. L., Vali H.,Romanek C. S., Clemett S. J., Chillier X. D., Maechling C. R.and Zare R. N. (1996) Search for past life on Mars: Possiblerelic biogenic activity in martian meteorite ALH84001. Science

273, 924–930.

Morgan B. E., Lahav O., Hearne G. R. and Loewenthal R. E.(2003) A seeded ambient temperature ferrite process fortreatment of AMD waters: magnetite formation in the presenceand absence of calcium ions under steady state operation.Water SA 29, 117–124.

Ohmoto H., Watanabe Y. and Kumazawa K. (2004) Evidencefrom massive siderite beds for a CO2-rich atmosphere before�1.8 billions years ago. Nature 429, 395–399.

Orewczyk J. (1990) Application of thermal analysis for theinvestigation of calcium ferrites. J. Therm. Anal. 36, 2153–2156.

Orewczyk J. (2000) Model studies of doped oxides Reductionprocess of magnetite doubly doped with calcium and magne-sium. J. Therm. Anal. Calorim. 60, 265–269.

Orewczyk J. and Jasienska S. (1987) Studies of calciomagnetitephase by means of thermal analysis. J. Therm. Anal. 32, 1711–

1714.

Orewczyk J., Jasienska S., Ciesla W. and Durak J. (1995) Effect ofthermal conditions on formation of monocrystalline calcio-magnetites. J. Therm. Anal. 45, 947–953.

Rodriguez-Navarro A. (2006) XRD2DScan: new software forpolycrystalline materials characterization using two-dimen-sional X-ray diffraction. J. Appl. Crystallogr. 39, 905–909.

Rodriguez-Navarro C., Ruiz-Agudo E., Luque A., Rodriguez-Navarro A. B. and Ortega-Huertas M. (2009) Thermal decom-position of calcite: mechanisms of formation and texturalevolution of CaO nanocrystals. Am. Mineral. 94, 578–593.

Romanek C. S., Jimenez-Lopez C., Navarro A. R., Sanchez-Roman M., Sahai N. and Coleman M. (2009) Inorganicsynthesis of Fe–Ca–Mg carbonates at low temperature. Geo-

chim. Cosmochim. Acta 73, 5361–5376.

Shannon R. D. (1976) Revised effective ionic radii and systematicstudies of interatomic distances in halides and chalcogenides.Acta Crystallogr. A32, 751–767.

Spinolo G. and Anselmi-Tamburini U. (1989) Nonequilibrium (Ca,Mg)O solid solutions produced by chemical decomposition. J.

Phys. Chem. 93, 6837–6843.

Thomas-Keprta K. L., Bazylinski D. A., Kirschvink J. L., ClemettS. J., McKay D. S., Wentworth S. J., Vali H., Gibson, Jr., E. K.and Romanek C. S. (2000) Elongated prismatic magnetitecrystals in ALH84001 carbonate globules: potential Martianmagetofossils. Geochim. Cosmochim. Acta 64, 4049–4081.

Thomas-Keprta K. L., Clemett S. J., McKay D. S., Gibson E. K.and Wentworth S. J. (2009) Origins of magnetite nanocrystalsin Martian meteorite ALH84001. Geochim. Cosmochim. Acta

73, 6631–6677.

Treiman A. H. (2003) Submicron magnetite grains and carboncompounds in Martian meteorite ALH84001: inorganic, abioticformation by shock and thermal metamorphism. Astrobiology

3, 369–392.

Treiman A. H. and Romanek C. S. (1998) Bulk and stable isotopiccompositions of carbonate minerals in Martian meteorite AllanHills 84001: no proof of high formation temperature. Meteorit.

Planet. Sci. 33, 737–742.

Vali H., Weiss B., Li Y.-L., Sears S. K., Kim S. S., Kirschvink J. L.and Zhang C. L. (2004) Formation of tabular single-domainmagnetite induced by Geobacter metallireducens GS-15. Proc.

Natl. Acad. Sci. USA 101, 16121–16126.

Vidyasagar K., Gopalakrishnan J. and Rao C. N. R. (1984) Aconvenient route for the synthesis of complex metal oxidesemploying solid-solution precursors. Inorg. Chem. 23, 1206–

1210.

Wang W., Xu Z. and Finch J. (1996) Fundamental study of anambient temperature ferrite process in the treatment of acidmine drainage. Environ. Sci. Technol. 30, 2604–2608.

Ware S. S. J. and French D. H. (1984) The application ofsimultaneous DTA and TG to some aspects of oil shalemineralogy. Thermochim. Acta 76, 179–200.

Wenk H. R., Barber D. J. and Reeder R. J. (1983) Microstructuresin carbonates. In Carbonates. Mineralogy and Chemistry (ed. R.J.Reeder). Rev. Mineral. 11, 301–367.

Zachara J. M., Kukkadapu R. K., Fredrickson J. K., Gorby Y. A.and Smith S. C. (2002) Biomineralization of poorly crystallineFe(III) oxides by dissimilatory metal reducing bacteria(DMRB). Geomicrobiol. J. 19, 179–207.

Associate editor: Alfonso Mucci