Low-temperature carbonate concretions in the Martian meteorite ALH84001: Evidence from stable isotopes and mineralogy

magnetite within the pyroxene; however, these ap-pear to have been formed on the fracture surfaces atthe same time as the magnetite and Fe-sulfide phas-es associated with the carbonate globule.

22. J. L. Kirschvink, in Biomagnetism: An Interdiscipli-nary Approach, S. Williamson, Ed. (Plenum, NewYork, 1983), pp. 501-532.

23. The intact sample was mounted at one end of aquartz-glass fiber 15 cm long and 1 mm in diameter,which had been cleaned in concentrated HCl. It wasattached to this fiber using cyanoacrylate cementthat had been passed through a 0.2-mm syringe filterto remove ferromagnetic impurities (Fig. 1A). Theopposite end of the fiber was attached to the vertexof a flat quartz-glass triangle, such that the open faceof the carbonate layer on the fracture surface wasaligned parallel to the triangle’s surface. The edge ofthe triangle opposite from the fiber attachment wasfused to an elongate hook, allowing the assembly tobe suspended vertically on a thin (;200-g test) nylonfishing line U-loop. This loop was wound on a Teflonspool attached to a small computer-controlled step-ping motor mounted on the ceiling of the clean-labfacility, and held directly above the room-tempera-ture access port of the superconducting magnetom-eter. With this assembly, the sample could be raisedand lowered smoothly from the sample loading po-sition to a pair of computer-controlled solenoids forthe demagnetization and rock magnetic experi-ments, and to the center of a three-axis supercon-ducting moment magnetometer (a 2G EnterprisesTmodel 570, with DC-biased SQuIDs). A horizontalarrowmarked on the quartz triangle was aligned withthe 1X direction of the superconducting momentmagnetometer; vertical down was 1Z, and the 1Ydirection formed the third axis of a right-handed or-thogonal coordinate system. We were able to obtainreplicate measurements to better than 1% intensity,and 0.5° in direction, on magnetic moments as weakas 10212 A m22, equivalent to the saturation rema-nence produced by ;20 picograms of single-do-main (SD) magnetite.

24. This separation was done using a 150-mm-thick di-amond-impregnated copper wafering saw. The flatsurface of the carbonate-bearing grain was firstglued to a thin Pyrex cover slip such that it wasparallel to the surface of the quartz-glass triangle.The magnetic moment of this new assembly wasindistinguishable from that measured prior to addi-tion of the cover slip and the additional cement, con-firming that they were both nonmagnetic. The othersurface of the Pyrex cover slip was then bound to thesurface of a cylindrical brass stub with a tempera-ture-sensitive adhesive that had been filtered in ace-tone to remove ferromagnetic contaminants. Theorientation of the quartz fiber was marked on thebrass surface. Shortly after we began our first cut,the bond between the cyanoacrylic cement and thecover slip gave way. This left a small notch in thesample at the boundary between the pyroxenegrains (Fig. 1A). By remeasuring the NRM of thesample after this step, we were able to calculate bydifference the NRM vector that had been held by thematerial removed in the cut. Next, we fixed the flatsurface of the carbonate-bearing layer directly to thebrass stub with the adhesive, which held properlyduring the remainder of the wafering process. Thislast cut was adjusted slightly so that most materialwas removed from the larger grain, leaving a 2.2-mgfragment of the small pyroxene grain on the brassstub (Fig. 1D). While it was still bound to the stub,we then used the cyanoacrylate to cement a sec-ond quartz-glass triangle and fiber assembly to thethis new fragment, with a relative orientation iden-tical to that of the first sample. It was then freedfrom the brass stub by heating briefly to 110°C, andwashed with filtered acetone to dissolve traces ofthe adhesive. The sawing procedure left a 1.6-mgfragment of the carbonate-bearing grain attachedto the larger pyroxene grain. After measurement ofthe NRM, we were able to break this free with anonmagnetic ceramic scalpel blade, and by re-measuring the NRM vector, were able to recover bydifference the NRM vector of this small chip. Thefinal weight of the pyroxene grain was 12.7 mg,

implying that a total of 3.4 mg of the sample waslost in both sawing operations.

25. J. L. Gooding, Icarus 99, 28 (1992).26. R. F. Butler, Paleomagnetism: Magnetic Domains to

Geologic Terranes (Blackwell, Boston, 1992).27. AF demagnetization was not continued to higher lev-

els for the small pyroxene grain because the intensitybecame weak and it had a linear decay toward theorigin. Thermal demagnetization experiments werenot done, because of the possibility of irreversiblemineralogical changes on some of the Fe-S miner-als, such as the iron monosulfides reported byMcKay et al. (1).

28. J. L. Kirschvink,Geophys. J. R. Astron. Soc. 62, 699(1980).

29. P. L. Mcfadden and M. W. McElhinny, Earth Planet.Sci. Lett. 87, 161 (1988).

30. C. K. Shearer et al.,Geochim. Cosmochim. Acta 60,2921 (1996).

31. M. J. Dekkers, Phys. Earth Planet. Inter. 52, 376(1988).

32. H. P. Johnson et al.,Geophys. J. R. Astron. Soc. 41,1 (1975).

33. S. Cisowski, Phys. Earth Planet. Inter. 26, 56 (1981).34. We thank D. McKay and E. K. Gibson for our sample

of ALH84001, P. Carpenter for assistance with theSEM, and G. R. Rossman for help with the delicatesawing operation. H.V. acknowledges financial sup-port from the U.S. National Research Council. Wemade extensive use of the software provided by C.Jones ([email protected]) for the analysisand presentation of paleomagnetic data. B. C. Mur-ray, J. Eiler, and D. A. Evans made helpful sugges-tions on the manuscript. This is contribution no.5897 from the Division of Geological and PlanetarySciences of the California Institute of Technology.

31 January 1997; accepted 20 February 1997

Low-Temperature Carbonate Concretions in theMartian Meteorite ALH84001: Evidence from

Stable Isotopes and MineralogyJohn W. Valley, John M. Eiler, Colin M. Graham,

Everett K. Gibson, Christopher S. Romanek, Edward M. Stolper

The martian meteorite ALH84001 contains small, disk-shaped concretions of carbonatewith concentric chemical and mineralogical zonation. Oxygen isotope compositions ofthese concretions, measured by ion microprobe, range from d18O 5 19.5 to 120.5‰.Most of the core of one concretion is homogeneous (16.7 6 1.2‰) and over 5‰ higherin d18O than a second concretion. Orthopyroxene that hosts the secondary carbonatesis isotopically homogeneous (d18O 5 4.6 6 1.2‰). Secondary SiO2 has d18O 5 20.4‰.Carbon isotope ratios measured from the core of one concretion average d13C 5 46 68‰, consistentwith formation onMars. The isotopic variations andmineral compositionsoffer no evidence for high temperature (.650°C) carbonate precipitation and suggestnon-equilibrium processes at low temperatures (,;300°C).

Carbonate concretions in ALH84001 pro-vide information on the nature of the an-cient martian atmosphere and hydrosphere,and aspects of their composition and mor-phology have been proposed as evidence forprimitive life on Mars (1). The tempera-tures of carbonate formation are uncertain,but central to these questions. Mineralequilibria and the morphology of magnetiteinclusions in carbonate have been used toinfer temperatures of over 650°C (2, 3), inwhich case carbon-based life is unlikely. Onthe other hand, oxygen isotope composi-tions of carbonate and the magnetic prop-erties of millimeter-scale subdomains inALH84001 have been interpreted to indi-

cate that the minerals formed at tempera-tures of 0° to 110°C (4, 5), suggesting pre-cipitation during weathering or alteration.Such conditions would be permissive of lifeas we know it.

The ALH84001 meteorite is composeddominantly of igneous orthopyroxene(Mg0.70Fe0.27Ca0.03SiO3) with minor cli-nopyroxene, olivine, chromite, pyrite, apa-tite or whitlockite, maskelynite (shock-pro-duced feldspathic glass An31Ab63Or6), andSiO2 (2, 6, 7). The sample is highly frac-tured, probably from impacts while it wasstill on Mars (6, 7). Secondary carbonatesare precipitated in some of these fracturesand as disseminated patches in brecciatedorthopyroxene (2, 6, 7).

We analyzed two carbonate concretionsfor isotopic and chemical compositionsfrom a group of 14 that are approximatelyco-planar within an area of 2 mm2 (Fig.1A). Concretion #1 appears to be two con-cretions grown together, each of which hasconcentric core-to-rim chemical variationsfrom Ca0.15Mg0.45Fe0.40CO3 to nearly pureMgCO3 [Figs. 1C, 1D, 2, and figure 1 of

J. W. Valley, Department of Geology and Geophysics,University of Wisconsin, Madison, WI 53706, USA.J. M. Eiler and E. M. Stolper, Division of Geological andPlanetary Sciences, California Institute of Technology,Pasadena, CA 91125, USA.C. M. Graham, Department of Geology and Geophysics,Edinburgh, EH9 3JW, Scotland, UK.E. K. Gibson, NASA–Johnson Space Center, Houston,TX 77058, USA.C. S. Romanek, Savannah River Ecology Laboratory,University of Georgia, Drawer E, Aiken, SC 29802, USA.

REPORTS

http://www.sciencemag.org z SCIENCE z VOL. 275 z 14 MARCH 1997 1633

on

Janu

ary

22, 2

007

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

(2)]. This variation is not continuous;there is an abrupt compositional breakbetween the orange, continuously zoned,Fe-rich core (XMg # 0.7 in Fig. 2) and the;10-mm-wide mantle of nearly pure mag-nesite (MgCO3, Mg-rich in Fig. 1D).Thin, black, Fe-rich rims separate the corefrom the mantle and the mantle from or-thopyroxene. The Fe-rich character of therims derives, at least in part, from submi-croscopic grains of Fe oxide (magnetite)and sulfide that may also be trace constit-uents within the orange carbonate (1).The concretions are thin disks and the

concentric zonation is seen optically only atthe outer perimeter of the disk. These obser-vations suggest that the concretions grewradially outward from the center along pre-existing cracks. Some of the compositionalzones are offset or truncated by microfrac-tures, demonstrating that a second fracturingevent postdated the precipitation of carbon-ates (6, 7). A thin (1 to 10 mm) vein of SiO2cuts across concretion #1 (orange in Fig. 1E).SiO2 is also found as irregularly shaped grainswithin orthopyroxene.

Previous oxygen and carbon isotope anal-yses of carbonate in martian meteorites have

been made by chemical dissolution or ther-mal decomposition of whole-rock powders ormineral separates. For ALH84001, valuesvaried from d18O 5 29 to 126‰VSMOWand d13C 5 221 to 153‰PDB [Fig. 3 (4, 8,9)], but most values were d18O 5 17.5 62.7‰ (1 SD, n 5 7) and d13C 5 36 6 10‰(1 SD, n 5 9). Both low d18O and low d13C(,0‰) values have been reported (Fig.3), although it is unclear which compo-nents of the meteorite (or minor terrestrialcontaminants) these low values reflect,whether they accurately measure a singlecomponent of the meteorite, and to whatdegree the meteorite is isotopically heter-ogeneous on a microscale. The question ofhomogeneity is important vis-a-vis thethermal history of the meteorite. Chemi-cal and isotopic gradients diffuse awaymore quickly at higher temperatures and,given knowledge of appropriate diffusivi-ties, the presence or absence of such gra-dients can be used as quantitative con-straints on thermal history.

Ion microprobe–secondary ion mass-spectrometry (SIMS) permits in situ analy-sis of oxygen and carbon isotope ratios insmall samples (5 ng, 20- to 30-mm-diameterpits, 6 to 8 mm deep) of carbonate andsilicate minerals (10–14). Sample sizes are103 to 106 times smaller than previous bulkanalyses, allowing us to analyze several spotsin a single concretion, the largest of whichmeasures 250 mm in its greatest dimensionand is estimated to weigh less than 2 mg.These in situ measurements can be corre-lated with chemical, spatial, and texturalinformation.

We made 20 analyses of d18O and 4 ofd13C in the meteorite. We also made over500 analyses of 12 carbonate standards toassess analytical precision, stability, andstandardization (12–14). We analyzed sev-en spots for d18O in concretion #1 and twoin #2, and four spots for d13C in concretion#1 (Table 1). All analyses are of orange,Fe-rich magnesites in the concretion cores,except analysis #4, which overlaps thewhite magnesite mantle. The averagechemical compositions of analyzed spotswere determined using a focused electronmicroprobe beam on the rims of ion probepits after analysis and are in good agreementwith defocused beam analyses from beforeion probe analysis (Table 1 and Fig. 2).

The average d13C value was 146 6 8‰.This composition is based on data only fromthe beginning of the four analyses becauseheterogeneities present at depths of 1 to 4mm were attributed to contamination orreduced carbon in the sample (13, 14). Theresultant total counts for these analyses arelow and less precise than would otherwisebe possible. All four analyses agree within 1SD of each analysis. The average value is at

C2C2

A

B C

D E

C1C1

C-2C-1

O

O

OO1 2

3

4

OpxOpx OpxOpx

O

O5

6

BSE

Mg Epoxy

MantleMantle

Si

SiO2

BSE

Fig. 1. (A) Photograph of freshly broken surface of ALH84001 showing orange carbonate concretionsand host orthopyroxene. The dark crystals are chromite. Each concretion has thin black rims, rich inmagnetite and sulfides, on either side of a white magnesite mantle (1, 6, 7). Concretions #1 and #2 wereanalyzed in situ for stable isotope ratio and chemical composition. FOV 5 1.5 mm. (B and C) Back-scattered electron (BSE) images of concretion #1 (C) and #2 (B). The locations of ion probe spots #1 to6 are shown (other analyses were made deeper in the sample after repolishing). (D and E) Images ofchemical composition of the polished surface of carbonate concretion #1 and host orthopyroxene (Opx)before ion probe analysis. Part of the concretion is covered by epoxy (toward top of picture). (D) showsMg and (E) shows Si. The rounded, composite shape of this concretion is best seen in the image of Mg(D), the white band is the magnesite mantle that rims the concretion boundary with orthopyroxene anddivides this concretion into two nearly round subconcretions that have apparently grown together. Theleft side of the concretion is nearly round and fully exposed, but only the lower half of the right side isexposed. In (B) to (E) hot colors (white or yellow) indicate the highest concentrations or average atomicnumbers (BSE), and blue or black indicates the lowest concentrations. Field of view is 5 256 mm.

SCIENCE z VOL. 275 z 14 MARCH 1997 z http://www.sciencemag.org1634

on

Janu

ary

22, 2

007

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

the upper end of the range of previouslymeasured values (Fig. 3). On Earth, valuesof d13C in carbonates are typically near 0‰,values above 120‰ are rare for any min-eral [in some cases forming in carbonates asa result of bacterial methanogenesis (15)],and we know of no values above 135‰(15, 16). Thus, the carbon isotope ratios ofcarbonate in ALH84001 are not consistentwith any known terrestrial materials andindicate that they are extraterrestrial, asconcluded previously based on dissolutionmeasurements (4) and the presence of frac-tures that are presumably of impact originthat cross-cut the carbonates displacingchemical zonation (1, 6, 7). These d13Cvalues, which are 40 to 50‰ higher thanaverage values on Earth, are consistent withprecipitation from or equilibration with areservoir having a d13C value similar to thatestimated for the atmosphere of Mars (Fig.3), which is thought to be 30 to 50‰ higherthan on Earth (17, 18).

In concretion #1, five of the six d18Ovalues of orange Fe-rich magnesite are indis-tinguishable within analytical precision at16.7 6 1.2‰ (Fig. 3). The sixth spot yieldeda d18O value of 12.2 6 1.1‰. A seventhspot (pit #4) overlapped onto the white mag-nesite mantle and yielded the highest d18Ovalue of 20.6 6 1.3‰. The two analyses ofconcretion #2 (9.5, 13.5‰) are within ana-lytical precision of the one low d18O spot inconcretion #1 (Fig. 3). The ion probe resultscompare favorably with most of the previousbulk analyses, but the spatial resolution ofthe ion probe documents heterogeneity thatcould only previously be inferred. No at-tempt was made to analyze the magnesitemantles because those in our specimen aresmaller than the size of the ion beam. How-ever, the relatively high d18O value of thespot that overlaps the magnesite mantle (#4,Table 1) suggests that the mantle has aneven higher d18O value, consistent with theresults of Romanek et al. (4), who suggestedthat the d18O of the magnesite is 9‰ higherthan that of the core of the concretions onthe basis of stepwise acid dissolution experi-ments. There is no correlation of d18O valuewith chemical composition for the eightanalyses of orange carbonate. Although thenumber of analyses is limited, our eight d18Ovalues have a bimodal distribution (9.5 to13.5‰ and 15.5 to 18.7‰).

On Earth, d18O values as high ($16‰)as those that we have measured inALH84001 carbonates are only found inrocks that have interacted with isotope res-ervoirs formed at low temperatures[,100°C (16)]. A simple interpretation (4,19, 20) would be that the d18O values ofaqueous fluids on Mars are related by pro-cesses of exchange to those of martian ig-neous rocks, in which case the elevated

d18O values in carbonate concretions re-flect low temperatures (,;100°C) of pre-cipitation. However, this reasoning cannotfirmly constrain the temperature of carbon-ate formation until the d18O of the martianatmosphere or hydrosphere is known (21),and we thus sought other evidence to de-termine the formation temperatures of car-bonates in ALH84001.

Ten analyses of d18O in orthopyroxenedemonstrate that it is homogeneous (4.6 61.2‰). One analysis of SiO2 from an irreg-ular 40- by 50-mm grain enclosed by pyrox-ene yielded a d18O value of 20.4 6 0.9‰(13). It is not certain if this SiO2 grain isrelated to maskelynite (or its precursor pla-gioclase) or to other glasses predating car-bonate (6, 7), or is related to the post-concretion SiO2 precipitating event thatformed the vein in Fig. 1E.

There is no textural evidence to indicateequilibration of SiO2 and other minerals, butif SiO2 is assumed to be in isotopic equilib-rium with orthopyroxene, then the measuredd18O values indicate that the temperature ofSiO2 precipitation was 144° 6 14°C (22,23). Equilibration of SiO2 and the relativelyhomogeneous core of concretion #1 wouldindicate a temperature of 125° to 205°C(24–27). Equilibration temperatures impliedby coexisting carbonate and orthopyroxeneare similarly low, but variable: ;170°C forthe dominant carbonate in concretion #1and ;310°C for the average carbonate inconcretion #2. These temperatures onlyhave meaning if the various phases are inisotopic equilibrium, yet the heterogeneityamong carbonates demonstrates a lack ofisotopic equilibration at the scale of less than1 mm and the contact between the magne-

Fig. 2. Compositions of carbonates in concretions #1 and #2 in the Ca-Mg-Fe system measured byelectron microprobe. Circles are analyses of 5-mm spots (beam focused to 1 mm), open squares areanalyses of 10 by 10 mm areas, and solid squares are average compositions of 20-mm-diameter spotsanalyzed for oxygen isotope ratio by ion microprobe. Carbonates are not stable above the lines shownfor 527° and 727°C (29).

Fig. 3. Values of d18O and d13C measured in the martian meteorite ALH84001 in situ by ionmicroprobe (solid symbols, this study), and in powders by acid dissolution, step-heating, or fluori-nation [open symbols (4, 9, 44)]. Values shown in light type are suspected of contamination orfractionation. Samples are: carbonate (circles), silicates (squares), and suspected organic matter(triangles). The vertical lines at d18O ; 7‰ represent the value for carbonate or quartz in equilibriumwith orthopyroxene at high temperature (.600°C). The d13C estimated for the atmosphere on Marsand Earth are also shown (16–18).

REPORTS

http://www.sciencemag.org z SCIENCE z VOL. 275 z 14 MARCH 1997 1635

on

Janu

ary

22, 2

007

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

site mantle and orange carbonate may con-tain a disequilibrium isotopic gradient overdistances of ,10 mm [and possibly ,1 mm(4, 7)]. In any case, the intermineral oxygenisotope fractionations are not consistentwith equilibration at the high temperaturespreviously inferred from the major elementchemistry of the carbonates (.650°C).

Two opposed hypotheses can be exam-ined to determine if they explain the geo-chemical and mineralogical features asso-ciated with carbonate concretions inALH84001: (i) Precipitation of carbonateat high temperatures possibly over a shorttime period. This might occur during ameteorite impact and subsequent hydro-thermal activity in which case a range ofequilibration temperatures could be repre-sented by different carbonates (2). (ii)Low-temperature nonequilibrium precipi-tation, such that isotopic and chemicaldisequilibrium during crystallization willbe maintained because rates of reactionsand solid-state diffusion are low.

The evidence favoring a high temperatureorigin of the carbonate concretions comesfrom major element analysis of mineral com-position and mineral equilibria (2, 6), andfrom textures in nanometer-scale magnetitegrains within the carbonate (3). Harvey andMcSween (2) estimated that the carbonatesprecipitated at temperatures of over 650°C onthe basis of their analyses of carbonates in thecalcite-magnesite- siderite system. This esti-mate assumes that: (i) the minerals were inchemical equilibrium at the time of forma-tion, (ii) equilibrium compositions have beenpreserved, and (iii) that the analyses representhomogeneous phases. These assumptions canbe applied to other minerals in ALH84001 asa test of the hypothesized high-temperatureorigin for carbonate.

Most of our analyses of carbonate in

ALH84001 fall within the two phase region(solvus) of ankerite-magnesite immiscibilityat 700°C (Fig. 2). Our analyses and previousstudies (2, 6, 7) show that these more Ca-rich compositions are common in the coresof carbonate concretions. If the previouslyreported compositions in the range of XCa 50.2 to 0.3 (2) represent homogeneous phases,equilibrium is only possible if carbonates pre-cipitated at temperatures of .;1000°C (28,29). These very high temperatures are incon-sistent with several other features of theconcretions: First, pyrite is a primary mineralin the meteorite (6). The equilibria, pyrite5pyrrhotite 1 S, sets an upper temperaturelimit for pyrite of 750°C (30). Second, mag-nesite, forsterite-rich olivine, and enstatite-rich orthopyroxene occur as touching phasesin ALH84001 [see figure 2 of (2)]. Theequilibria enstatite 1 magnesite 5 forster-ite 1 CO2 would be at 550°C (31) in thepresence of nearly pure CO2 at P 5 0.2 GPa(equivalent to a depth of over 50 km onMars). This temperature is an upper limit.Shallower depths or aqueous fluids couldreduce the stability of magnesite 1 SiO2 to amaximum of 300°C or less (31). Finally,several other metamorphic reactions wouldoccur under conditions of high temperatureequilibration and a number of metamorphicminerals that have not been reported inALH84001 would be expected along grainboundaries (for example, wollastonite,zoisite, scapolite, grossular, diopside, tremo-lite, anthophyllite, and talc). Thus, if equi-librium is assumed, there is no self-consistentinterpretation, and this suggests either thatthe carbonates are metastable or the analysesrepresent mixtures of fine-grained phases.

The rates of fluid-hosted metamorphicreactions are known to be rapid at hightemperatures, such that if carbonate compo-sitions represent super-solvus temperatures,

the above-discussed reactions could be ex-pected to occur at grain boundaries over thetime scales of minutes to hours (32). Studiesof high-temperature carbonate precipitatedduring terrestrial impact events confirm thisexpectation, showing them to be intimatelyassociated with shock-produced glass and,where carbonate touches silicate in hostlithologies, the expected high-temperaturemetamorphic minerals occur [for example,clinopyroxene and larnite (33)]. The car-bonates in ALH84001 show no similar reac-tion relations with adjacent host phases.

If temperatures were high, then exchangeby diffusion would cause homogenization ofisotopic and chemical gradients (34). Thesharpest gradients documented in ALH84001are in Mg/Fe ratios between the nearly puremagnesite of the white concretion mantlesand the orange Fe-rich magnesite of the cores[occurring over a length scale of ;1 mm (7)].The d18O value of pit #4 (20.6‰, Table 1)suggests that there may also be large gradientsin d18O across magnesite mantles (Table 1).A gradient of 9‰ was inferred by dissolutionexperiments (4). At 1000°C, experimentaldata suggest that gradients in cation compo-sition would homogenize in 1 to 100 days (35)and the inferred differences in d18O wouldhomogenize in 10 to 2500 min (36). Howev-er, at low temperatures (;100°C), composi-tional contrasts will be preserved in the ab-sence of recrystallization. On Earth, gradientsof 5 to 13‰ over 10 to 400 mm have onlybeen documented from diagenetic and low-temperature (from ,80° to 400°C) alterationenvironments (10, 37, 38). For .650°C ig-neous or metamorphic processes, measurablediffusion gradients are typically less than 2‰(10, 39). In addition, if temperatures werehigh, then fluids forming high d18O carbon-ates must also have had elevated d18O valuesand would be expected to have exchangedwith the orthopyroxene along the boundary ofcracks causing heterogeneity that is not ob-served. It is evident from these considerationsthat the mineralogy and isotopic values of thecarbonate concretions and their relations tothe host orthopyroxene cannot be reconciledwith high-temperature equilibration.

The alternative hypothesis is that carbon-ates precipitated from a fluid at low tempera-tures. Such conditions are common on Earth,frequently forming minerals that are notequilibrated, either isotopically or chemically(37, 38). Fine-scale oxygen isotope heteroge-neity is readily formed and easily preserved atlow temperatures where diffusion is slow, butrequires a complex process of precipitationand rapid cooling if formed at high tempera-tures. Delicate mineral overgrowths and reac-tion textures such as observed in ALH84001carbonates (6, 7) are common in low temper-ature (#300°C) environments on Earth, andin many cases, sharp contrasts in chemical

Table 1.Chemical compositionmeasured by electronmicroprobe and stable isotope ratio measured byion microprobe of 20-mm-diameter spots in carbonate concretions #1 and #2 and silicate minerals fromALH84001.

Analysis spot XMg* XCa XFe XMnd18O(‰)

61 SD(‰)†

1 concretion 1 0.598 0.084 0.312 0.006 15.9 1.32 concretion 1 0.522 0.121 0.343 0.014 17.2 1.03 concretion 1 0.546 0.110 0.335 0.009 12.2 1.14 concretion 1 0.747 0.062 0.188 0.003 20.6 1.312 concretion 1 0.627 0.071 0.299 0.003 15.9 1.214 concretion 1 0.507 0.127 0.350 0.015 15.5 1.215 concretion 1 0.586 0.085 0.295 0.034 18.7 1.35 concretion 2 0.578 0.094 0.317 0.011 9.5 1.16 concretion 2 0.566 0.104 0.323 0.008 13.5 1.2SiO2 20.4 0.9

d18O of orthopyroxene: 2.6, 3.1, 5.6, 4.9, 3.2, 6.4, 5.6, 4.7, 5.6, 4.0‰; average 5 4.6 6 1.2‰

d13C of carbonate: 49, 36, 45, 55‰; average 5 46 6 8‰

*X5molar fraction of cations in carbonate. †Values of 1 standard deviation (SD) dictated by counting statistics forcarbonate analyses #1 to 15. For multiple analyses (N . 1), external precision about the mean is reported.

SCIENCE z VOL. 275 z 14 MARCH 1997 z http://www.sciencemag.org1636

on

Janu

ary

22, 2

007

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

and isotopic composition (37, 38) are pre-served at the same scale as that seen in thecarbonate concretions.

There are well-documented situationswhere low-temperature (from ,100° to500°C) carbonate veins cut minerals with-out visible reaction, including orthopyrox-ene-bearing lithologies (40) and samplesshattered by meteorite impact (41). It hasbeen suggested that nearly pure CO2 andhigh temperatures are necessary in order toprevent hydration reactions involving py-roxene at the time that carbonates wereprecipitated (2). However, carbonates canprecipitate over a wide range of tempera-tures and from water-rich fluids withoutproducing hydrated minerals (40).

Low-temperature carbonate mineralscommonly form in apparent violation ofequilibrium thermodynamics, includingaragonite, and high-magnesium calcitesand calcian dolomites (which are insidethe solvus). If these minerals were assumedto form in stable equilibrium, one woulderroneously conclude that the Earth’soceans exist at temperatures over 500°Cand surface pressures of at least severalthousand atmospheres. The kinetic pro-cesses that cause apparent disequilibriumare not well understood and can be eitherbiogenic or abiogenic (42). The scale ofobservation is another factor; some car-bonates are actually complex submicro-scopic mixtures that require high-resolu-tion transmission electron microscopy forproper characterization (43). It is notknown if all of the carbonates reportedfrom ALH84001 are truly homogeneoussingle phases at the submicrometer-scale,and textures characteristic of low-temper-ature growth may yet be found (43).

Thus, there are many well-studied andcommon terrestrial analogs for low-tempera-ture processes that could form carbonate con-cretions, such as described in ALH84001. Allof the chemical, mineralogical and isotopicevidence that we have presented are consis-tent with such a model. In contrast: (i) theisotopic data that we present are not consis-tent with high-temperature equilibration inALH84001; (ii) there is no mineral, chemical,or phase equilibria evidence that indicates aself-consistent high-temperature (.650°C)genesis for the carbonate concretions inALH84001; (iii) and well-studied high-tem-perature terrestrial analogs have featuresclearly inconsistent with those observed inALH84001.

REFERENCES AND NOTES___________________________

1. D. S. McKay et al., Science 273, 924 (1996).2. R. P. Harvey and H. Y. McSween Jr., Nature 382, 49

(1996).3. J. P. Bradley, R. P. Harvey, H. Y. McSween Jr.,

Geochim. Cosmochim. Acta 60, 5149 (1996).

4. C. S. Romanek et al., Nature 372, 655 (1994).5. J. L. Kirchvink, A. T. Maine, H. Vali, Science 275,

1629 (1997).6. D. W. Mittlefehldt, Meteoritics 29, 214 (1994).7. A. H. Treiman, ibid. 30, 294 (1995).8. d18O 5 1000 [(18O/16O-sample)/(18O/16O-VSMOW )

21], where VSMOW is the international standardocean water. d13C 5 1000 [(13C/12C-sample)/(13C/12C-PDB) 2 1] where PDB is the Pee Dee belemnitestandard.

9. M. M. Grady, I. P. Wright, C. Douglas, C. T. Pillinger,Meteoritics 29, 469 (1994); A. J. Jull, C. J. Eastoe, S.Xue, G. F. Herzog, ibid. 30, 311 (1995); L. A. Leshin,S. Epstein, E. M. Stolper, Geochim. Cosmochim.Acta 60, 2635 (1996).

10. J. W. Valley, C. M. Graham, B. Harte, J. M. Eiler, P.D. Kinny, Rev. Econ. Geol. 7, in press.

11. J. M. Eiler, C. Graham, J. W. Valley, Chem. Geol. inpress.

12. J. M. Eiler, J. W. Valley, C. M. Graham, Lunar Planet.Sci., in press.

13. The sample, ALH84001,200, was cast in epoxy andpolished. The sample was repolished twice duringthe analysis of carbonates (after analysis #6 and be-fore #12) to remove 5 to 10 mm from its surface andreveal freshmaterial for analysis. Stable isotope anal-ysis wasmadewith a Cameca ims 4f ionmicro probeat the University of Edinburgh using a high-energyoffset (350 6 25 eV for O, 250 6 25 eV for C). Thespot size was 20 mm for O and 30 to 40 mm for C by6 to 8mmdeep (volume5 ;2000mm3,mass' 5 ngof mineral for O). Analyses of standards had preci-sions (based on counting statistics) of 61‰ (1 SD)for 18O/16O and 61.5‰ for 13C/12C. Sample anal-yses were comparable in precision for O, but C anal-yses of sample were truncated because of hetero-geneities at depths of 1 to 4 mm (14). At the begin-ning of each sample d13C measurement, countrates, and isotope ratios were steady, but after 17 to56 cycles (at depths of 1 to 4 mm), count ratesincreased and then decreased, and measured iso-tope ratios decreased by 48 6 2‰. Since thesesample analyses were bracketed by good analysesof carbonate standards, these changes must resultfrom real differences in the sample at depth. Possibleexplanations include: contamination by epoxy or pol-ishing materials, differences in chemical compositionwhich affect count rate and aSIMS (asims 5 Rsims/Rtrue, where R 5 18O/16O or 13C/12C), or the pres-ence of reduced carbon phases inherent to the sam-ple with drastically lower d13C. Only data from theinitial, steady part of each depth profile were used tocalculate d13C of the sample. Other analytical detailsare reviewed by (10). The 24 analyses were preced-ed by over 500 analyses of 12 homogeneous car-bonate standards, nine of which are in the Ca-Mg-Fesystem including: two calcites, two aragonites, Fe-dolomite, three Mg-bearing siderites (XFe 5 0.96,0.86, 0.71), magnesite, rhodochrosite, strontianite,and witherite (12). The matrix correction (aSIMS) wasfound to vary nonlinearly across the Ca-Mg join, inthe Ca-Mg-Fe system but to correlate well with XFe(for O, a 5 0.968 at XFe 5 1 and 0.927 at XFe 5 0; forC, a 5 0.956 at XFe 5 1 and 0.932 at XFe 5 0). Thecarbonate analyses in Table 1 are standardizedagainst a Mg-rich siderite (XFe 5 0.71) and the effectof solid solution is calibrated by the other eight Ca-Mg-Fe standards. Within the limited compositionrange of the orange carbonates we analyzed inALH84001, differences in chemistry correspond tocalibration differences of 61‰. Small amounts ofsubmicroscopic Fe oxides and sulfides may bepresent in the analyzed pits, especially in pit #4, butthis is unlikely to have a significant effect on thesedata. The d18O analyses of orthopyroxene are cali-brated against conventional analyses of this meteor-ite (44) as well as against ion probe analyses of astandard enstatite yielding the same value within un-certainty. The analysis of SiO2 is standardized by ionprobe data for Amelia albite (11). Chemical compo-sitions were measured by electron microprobe inEdinburgh and in Madison using wavelength-disper-sive spectrometers (WDS). Normal standardizationand correction procedures were employed. Thedepth of x-ray excitation for quantitative chemical

analysis is 5 mm such that the electron probe datarepresent nearly the same volume of sample as the 6to 8 mm deep ion probe pits. Images of chemicalcomposition (Fig. 1) were made in Madison withstage scans and WDS.

14. J. W. Valley, J. M. Eiler, C. M. Graham, E. K. Gibson,Jr., C. S. Romanek, Lunar Planet. Sci., in press.

15. J. Veizer, W. T. Holser, C. K. Wilgus, Geochim. Cos-mochim. Acta 44, 579 (1980); T. F. Anderson andM.A. Aruthur, in Stable Isotopes, in Sedimentary Geol-ogy, SEPM Short Course 10, 1 (1983); M. A. Aruthuret al., Annu. Rev. Earth Planet. Sci. 15, 47 (1987); E.M. Galimov, Geochim. Cosmochim. Acta 55, 1697(1991).

16. J. Hoefs, Stable Isotope Geochemistry (Springer,Berlin, ed. 4, 1997), p. 201.

17. R. H. Carr, M. M. Grady, I. P. Wright, C. T. Pillinger,Nature 314, 248 (1985).

18. C. P. Harzmetz, I. P. Wright, C. T. Pillinger, inWorkshopon theMars Surface and Atmosphere Through Time, R.M. Haberle et al., Eds. (Tech. Rep. 92-02, Lunar andPlanetary Institute, Houston, TX, 1992), p. 67.

19. R. N. Clayton and T. K. Mayeda, Geochim. Cosmo-chim. Acta 52, 925 (1988).

20. I. P. Wright, M. M. Grady, C. T. Pillinger, ibid. 56, 817(1992).

21. iiii, J. Geophy. Res. 95, 14,789 (1990); B. M.Jakosky, Icarus 94, 14 (1991); Geophy. Res. Lett.20, 1591 (1993); S. M. Clifford, J. Geophys. Res. 98,10,973 (1993).

22. H. Chiba, T. Chacko, R. N. Clayton, J. R. Goldsmith,Geochim. Cosmochim. Acta 53, 1985 (1989).

23. J. M. Rosenbaum, T. K. Kyser, D. Walker, ibid. 58,2653 (1994).

24. J. R. O’Neil, R. N. Clayton, T. K. Mayeda, J. Chem.Phys. 51, 5547 (1969).

25. R. N. Clayton, J. R. O’Neil, T. K. Mayeda, J. Geo-phys. Res. 77, 3057 (1972).

26. Z. D. Sharp and D. L. Kirschner, Geochim. Cosmo-chim. Acta 58, 4491 (1994).

27. Isotope fractionations are calculated for calcitewhich approximates the value for intermediate mag-nesite-siderite solid solutions, and for quartz whichapproximates silica.

28. L. M. Anovitz and E. J. Essene, J. Petrol. 28, 389(1987); P. L. McSwiggen, Phys. Chem. Miner. 20, 42(1993).

29. P. M. Davidson, Am. Miner. 79, 332 (1994).30. J. R. Craig and S. D. Scott, in Sulfide Phase Equilibria,

P. R. Ribbe, Ed. (vol. 1, Reviews in Mineralogy Miner-alogical Society of America, Washington, DC 1974).

31. W. Johannes, Am. J. Sci. 267, 1083 (1969); R. G.Berman, J. Petrol. 29, 445 (1988).

32. B. J. Wood and J. V. Walther, Science 222, 413(1983).

33. I. Martinez, P. Agrinier, U. Scharer, M. Javoy, EarthPlanet. Sci. Lett. 121, 559 (1994).

34. J. Crank, The Mathematics of Diffusion (ClarendonPress, Oxford, 1975).

35. J. R. Farver and R. A. Yund, Contr. Min. Petrol. 123,77 (1996); D. J. Cherniak, Eos 76, F683 (1995); W.G. Minarik and E. B. Watson, Earth Planet. Sci. Lett.133, 423 (1995).

36. T. F. Anderson, J. Geophys. Res. 76, 3918 (1969); A.K. Kronenberg, R. A. Yund, B. J. Giletti, Phys. Chem.Miner. 11, 101 (1984); J. R. Farver, Earth Planet. Sci.Lett. 121, 575 (1994).

37. J. W. Valley and C. M. Graham, Contrib. Miner.Petrol. 109, 38 (1991); J. W. Valley and C. M. Gra-ham, Science 259, 1729 (1993); J. W. Valley and C.M. Graham, Contrib. Miner. Petrol. 124, 225 (1996);C. M. Graham, J. W. Valley, B. L. Winter, Geochim.Cosmochim. Acta 60, 5101 (1996).

38. R. L. Hervig, L. B. Williams, I. K. Kirkland, F. J. Long-staffe,Geochim. Cosmochim. Acta 59, 2537 (1995).

39. J. M. Eiler, J. W. Valley, L. P. Baumgartner, ibid. 57,2571 (1993); J. M. Eiler, J. W. Valley, C. M. Graham,L. P. Baumgartner, Am. Miner. 80, 757 (1995).

40. J. Morrison and J. W. Valley, Geology 16, 513(1988); J. Geol. 99, 559 (1991).

41. J. W. Valley, S. C. Komor, K. Baker, A. W. A. Jeffrey,I. R. Kaplan, A. Råheim, in Crystalline Bedrock, A.Boden and K. G. Eriksson, Eds. (Springer-Verlag,

REPORTS

http://www.sciencemag.org z SCIENCE z VOL. 275 z 14 MARCH 1997 1637

on

Janu

ary

22, 2

007

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

New York, 1988), p. 156.42. R. J. Reeder, Ed., Carbonates: Mineralogy and Chem-

istry, vol. 11, Reviews in Mineralogy (Mineralogical So-ciety of America, Washington, DC, 1983); J. W. Morse,in ibid., p. 227; I. Barnes and J. R. O’Neil, Geochim.Cosmochim. Acta 35, 699 (1971); T. F. Anderson andM. A. Arthur, in Stable Isotopes in Sedimentary Geology(Short Course 10, Society of Economic Petrologists andMineralogists, Dallas, TX, 1983), p. 1-1; J. M. Hayes,Marine Geol. 113, 111 (1993); I. D. Clark and B. Lauriol,

Chem. Geol. 102, 217 (1992); T. McConnaughey,Geochim. Cosmochim. Acta 53, 163 (1988).

43. R. J. Reeder, inMinerals and Reactions at the AtomicScale, P. R. Buseck, Ed., Rev. Miner. 27, 381 (1992).

44. R. N. Clayton and T. K. Mayeda, Geochim. Cosmo-chim. Acta 60, 1999 (1996).

45. J. Craven aided in all aspects of the ion microprobeanalysis in Edinburgh and M. Spicuzza analyzed car-bonate standards in Madison. J. Muhl measured ionprobe pit depth and volume by optical interferometer. B.

Hess mounted and carefully polished the sample. J.Fournelle, T. Patterson, S. Burgess, and P. Hill assistedin electron microprobe analysis. M. Diman drafted thefigures and B. Barker aided in photography. G. Ross-man and R. Reeder donated carbonates as potentialstandards. We thank three anonymous reviewers forprompt, helpful comments. This research was support-ed by NSF, NASA, DOE and NERC.

10 February 1997; accepted 21 February 1997

Vacuum Squeezing of Solids: MacroscopicQuantum States Driven by Light Pulses

G. A. Garrett, A. G. Rojo, A. K. Sood,* J. F. Whitaker, R. Merlin†

Femtosecond laser pulses and coherent two-phonon Raman scattering were used toexcite KTaO3 into a squeezed state, nearly periodic in time, in which the variance of theatomic displacements dips below the standard quantum limit for half of a cycle. Thisnonclassical state involves a continuum of transverse acoustic modes that leads tooscillations in the refractive index associated with the frequency of a van Hove singularityin the phonon density of states.

Squeezing refers to a class of quantum me-chanical states of the electromagnetic fieldand, more generally, of harmonic oscillatorsfor which the fluctuations in two conjugatevariables oscillate out of phase and becomealternatively squeezed below the values forthe vacuum state for some fraction of a cycle(1). Thus, a squeezed electromagnetic fieldprovides a way for experimental measure-ments to overcome the standard quantumlimit for noise imposed by vacuum fluctua-tions. As such, the generation of squeezedlight with various nonlinear processes hasattracted much attention as a means of re-ducing noise in optical interferometry andlight-communication networks (1).

Following the work on photons (1), avariety of intriguing proposals were put for-ward dealing with squeezed states of otherbosons—particularly those associated withatomic vibrations in molecular (2) and con-densed-matter systems (phonons) (3)—aswell as polaritons (4). In addition, squeezedphonons were considered in variational ap-proaches to the ground state of stronglycorrelated electron-phonon problems (5).Here, we report an experimental demon-stration of phonon squeezing in a macro-scopic system (6). We have generated asqueezed mechanical state by exciting a

crystal, KTaO3, with an ultrafast pulse oflight. The measurements were performedwith the standard pump-probe setup (Fig.1). Second-order coupling of the photonswith the lattice vibrations [specifically,transverse acoustic (TA) modes] amountsto an impulsive change in the phonon fre-quency that gives rise to squeezing; thismechanism is closely related to that used togenerate two-photon coherent states inquantum optics (7). We monitored thesqueezed state by measuring the transmis-sion of a second (probe) pulse that is sensi-tive to changes in the refractive index aris-ing from the modulations in the meansquare displacement of the atomic posi-tions. Our state comprises a continuum ofmodes, but the probe transmission is domi-nated by a single frequency associated witha van Hove singularity in the phonon den-sity of states.

The Hamiltonian relevant to our prob-lem is H 5 Sq(Hq 1 Uq), where Hq 5 (Pq

2

1 Vq2 Qq

2 )/2 is the harmonic contributionto the lattice energy and (8, 9)

Uq 5 214

3 ~q)F2Qq2 (1)

Here, Qq is the amplitude of the phonon offrequency Vq and wave vector q, Pq is theassociated canonical momentum, F is themagnitude of the electric field, 3 5Sij3ijeiej,3ij(q) is the second-order polariz-ability tensor associated with Raman scat-tering (RS), and e 5 F/F is a unit vector(for clarity, we omit the phonon branchindex). Equation 1 describes an effectiveinteraction between two phonons of oppo-site momenta and two photons and reflectsthe quadratic term in an expansion of theelectronic susceptibility in powers of atomicdisplacements (10).

The generation of the squeezed state isbest understood at temperature T 5 0. LetE denote the pump field, and consider theassumption, valid in our experiments, thatthe period of the relevant phonons is largecompared with both the time it takes forthe pulse to cross the sample and theoptical pulse width t0, that is, we ignorethe dependence of the field on positionand approximate F2 5 E2(t) 5 (4pI0/nRc)d(t) in Eq. 1 (I0 is the integratedintensity of the pulse, nR is the refractiveindex, c is the speed of the light, and d isthe Dirac delta function). Then, if cq

2 isthe wave function (the ground state) of agiven mode at t5 02 immediately before thepulse strikes, integration of the Schrod-inger equation gives the wave function att 5 01

cq1 5 expSijqVqQq

2

\Dcq

2 (2)

where jq 5 (pI03/2cnRVq) and \ isPlanck’s constant divided by 2p. It followsthat ^Qq(t)& 5 0 (the brackets denote ex-pectation value). We use the equation ofmotion for Qq

2 and the initial conditionsfrom Eq. 2 to obtain the variance

G. A. Garrett, A. K. Sood, R. Merlin, Center for UltrafastOptical Science, University of Michigan, Ann Arbor, MI48109–2099, and Department of Physics, University ofMichigan, Ann Arbor, MI 48109–1120, USA.A. G. Rojo, Department of Physics, University of Michi-gan, Ann Arbor, MI 48109–1120, USA.J. F. Whitaker, Center for Ultrafast Optical Science, Uni-versity of Michigan, Ann Arbor, MI 48109–2099, USA.

*On leave from Department of Physics, Indian Institute ofScience, Bangalore 560 012, India.†To whom correspondence should be addressed.

Fig. 1. Schematic snapshotdiagram of the experiment(not to scale). The strongerpump pulse drives the sam-ple into an excited time-varying state, which per-turbs the weaker probepulse that follows behind.Here, the signal of interest isthe transmitted intensity of the probe beam as a function of the time delay t, as measured by the relativedistance between the two pulses.

SCIENCE z VOL. 275 z 14 MARCH 1997 z http://www.sciencemag.org1638

on

Janu

ary

22, 2

007

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om