American Mineralogist, Volume 65, pages 1252-1262, 1980

The calcite-araqonite eouilibrinm:effects of Sr substitution aid anion brientational disorder

Wrttrarvr D. CanrsoN'

Department of Earth and Space SciencesU niversity of Califu rnia

Los Angeles, California 90024

Abstract

Hydrothermal experiments have been performed in the systen cacor-srco. from 350oto 650"C over a range of I to 5 kbar. The results reconcile inconsistencies in previous deter-minations of low-temperature phase equilibria for binary Ca-Sr carbonates, and demonstratethat the small amounts of Sr in natural metamorphic aragonite extend the field of its stabilityto slightly lower pressures than in the unary CaCO, system.

The phase equilibrium data also provide evidence for the existence of two structurally dis-tinct rhombohedral carbonates at elevated pressures in the binary Ca-Sr system. Because ofthe exact correspondence of the present high-pressure results to previous X-ray determina-tions of one-atmosphere phase relations, one of these two polymorphs is identified as a calcitestructure having ordered CO3 groups, and the other as a calcite-like structure in which the'CO,

groups display rotational disordering about their triad axes. The preferential fractiona-tion of Sr into the disordered phase favors a model for its crystal structure in which the cat-ions occupy ninefold coordinated sites.

Experiments in the binary ca-Sr system identify a three-phase equilibrium at which thesetwo rhombohedral phases coexist with an orthorhombic aragonite solid solution. This three-phase equilibrium can be traced to a unary triple-point near 475"C and l0 kbar in the p111eCaCO3 system; aragonite, ordered calcite, and disordered calcite are all stable under theseconditions.

Because the calcite-aragonite boundary is strongly curved at temperatures below the triplepoint but nearly linear above it, the curvature is attributed to a gradual disordering of tleCO, groups with increasing temperatue. Differential solubility of Sr in the ordered and dis-ordered calcite phases simultaneously in equilibrium with Sr-bearing aragonite requires,however, that the transition itself be thermodynamically frst-order.

Introduction and previous workUnderstanding the geologic prooesses operating in

subduction zones requires accurate determination ofthe conditions under which metamorphic rocks of theglaucophane schist facies have originated. The occur-rence in these rocks of aragonite, the high-pr€ssurepolymorph of CaCOr, offers one means of estimatingtheir conditions of origin, provided first that the cal-cite-aragonite boundary can be precisely located ex-perimentally, and second that the effects of the addi-tional chemical components present in naturalaragonite can be assessed. In the last decade, a Ereat

deal ofresearch has been directed to these ends; a re-view of these studies reveals agreement on generalfeatures but also discrepancies concerning the de-tailed behavior of the system.

Early attempts to define the calcite-aragonite equi-librium boundary yielded inconsistent results; for athorough review, consult Boettcher and Wyllie(1968). Later research has produced three determina-tions which are in much better agreement. BothBoettcher and Wyllie (1968) and Goldsmith andNewton (1969), using piston-cylinder apparatus, lo-cated the equilibrium above 400"C. Johannes andPuhan (1971) compared results obtained in gas-pres-sure apparatus in the temperature range 200o to700"C with those from a variety of piston-cylindertechniques. If the piston-cylinder pressures are ad-

I Present address: Department of Geologicalv€rsity of Tcxas at Austin, P.O. Box 7909.78712.

w03 -ffi 4x / 80 / 1 | | 2- r252$02.M

Sciences, Uni-Austin, Texas

1252

justed to conform to the results from the gas-pressureapparatus, all three studies agree to within an experi-mental error of +500 bars. Subsequent work in therange 800o to llOOoc (Irving and Wyllie, 1973) andin the range 50" to 150"C (Crawford and Hoersch,1972) is also consistent with these determinations,(Fig. l), although one study (Zimmsrmsl, l97l)placed the equilibrium from l50o to 350oC atmarkedly higher pressures.

Although in all these investigations the calcite-aragonite boundary was depicted with one nearly lin-ear section below 350"C and another of steeper dPldI slope above 500oC, the nature of the region be-tween these linear sections is controversial. Initially,the change of slope was attributed to the transitionfrom calcite I to calcite II, first described byBridgman (1939). Boettcher and Wyllie (1968), whoequated the high-pressure calcite I--calcite II transi-tion of Bridgman (1939) with Boeke's (1912) high-temperature transition from B-calcite to c-calcite,showed a discontinuity in the calcite-aragoniteboundary slope; they postulated, in agreement withGoldsmith and Newton (1969), a first-order calcite I-calcite II transition as its explanation. Johannes andPuhan (1971) preferred a continuously curvedboundary and inferred a second-order calcite l-cal-cite II transition. Subsequent studies by Kondo el a/.(1972) and Cohen and Klement (1973), however,suggested that neither the calcite I-calcite II transi-tion nor the B-calcite-a-calcite transition can accountfor the change ofslope ofthe calcite-aragonite equi-librium boundary; neither transition curve intersects

CARLSON: CALCIT:E-ARAGONITE EQUILIBRI UM t253

the boundary in the appropriate temperature range.The most recent work (Mirwald,1976,1979a,b) out-lined stability fields at temperatures above 800"C fortwo additional CaCO, polymorphs, designals6 st1-cite IV and calcite V (equivalent to Boeke's B-calciteand a-calcite). Nevertheless, the changing slope ofthe calcite-aragonite boundary remained unex-plained, and the phase transition which it seems torequire remained unidentifed.

In addition to these unresolved problems in theunary system CaCOr, doubts persisted regarding theeffects of large-radius cations substituting for Ca innatural aragonites. Of these substituents, Sr is themost significant, as its oxide comprises approxi-mately I mole percent of typical metamorphic arago-nite (Coleman and Lee,1962). Because SrCO, belowgl2"C at atmospheric pressure has the aragonitestructure, it is possible that the small amounts ofSrCO, component in natural systems might stabilizethe orthorhombic carbonate to lower pressures thanin the pure CaCO, system.

To measure this effect, Froese and Winkler (1966)examined CaCO,-SrCO, equilibria at 500o-700oCand l-15 kbar. Chang (1965, l97l) also studied thissystem above 500"C, both at atmospheric pressure

and from l0 to 25 kbar as part ofthe ternary systemCaCO,-SrCO,-BaCOr. The application of these ex-perimental data in the temperature range relevant toblueschist metamorphism is, however, problematical.Despite reasonable agreement af high pressures andtemperatures, extrapolations of Froese and Winkler'sdata lead to contradictions with Chang's data at oneatmosphere, predicting a considerably more Ca-richorthorhombic phase than is observed. For instance,Chang (1965, Fig. 5) indicated a nearly constantcomposition (about 53 nole percent SrCOr) for or-thorhombic carbonate in equilibrium with calcite atone atmosphere over the temperature range 450"-600"C; the data of Froese and Win-kler (1966, Fig. 4)suggest a range of compositions (from perhaps 40 to55 mole percent SrCO.) at one atmosphere over thesame temperature interval. The discrepancies be-come more pronounced at lower temperatures, in theexperimentally uninvestigated range critical to blue-schist-facies metamorPhism.

Thus the detailed behavior of the calcite-aragoniteequilibrium is obscured by contradictory data re-garding the effect of Sr, especially at low temper-atures, and by disagreement about the nature and ef-fect ofthe phase transition in calcite near 500"C andl0 kbar. In this study, these problems were foundamenable to hydrothermal investigation in the sys-

30!

5tdtrlaaultrr

CoC0, PHASE RELATI0NS

v

fr.),8 KS.+tl\ L t t I / ",, n/' ,/-

f'rnl ^r'F{ti---.-r ---"" -o/(I)

u A -u'/ '

n L200 400 600 800 {000

TEMPERATURE ('C)1200

Fig. l. Phase relations in the unary system CaCO3. Dashed linesrepresent metastable equilibria. "A" : aragonite; "I" through'Y": calcite polymorphs. Data from Crawford and Hoersch, 1972rc+m; Johannes and hrhan, l97l (J+P); Irving and Wyllie, 1973

Q+|n; Mirwald, 1976 (M'l; Kondo et al., 1972 (KS+M);Bridgman, 1939 (.8).

t254 CARLSON: CALCITE-ARAGONITE EQUILIBRIUM

tem CaCOr-SrCOr, over the temperature range350"-650oC and HrO pressures of l-5 kbar.

Experimental techniques and equilibrium criteria

All experiments were performed in standard cold-seal hydrothermal apparatus, using filler rods and ex-ternal Cr-Al thermocouples. Temperature was con-trolled to within +5oC and pressure uncertainties areless than 20 bars. Reactants were sealed in silver cap-sules with 20 weight percent of an aqueous 0.075 MLirCO3 solution to catalyze reaction. (Comparisonruns using pure water con-firmed that the traceamounts of LirCO, did not displace the equilibria,but were somewhat effective in speeding reaction atlow temperatures.)

Di-fferent lsshniques were required to determinethe opposite boundaries of the miscibility gap. Onthe Sr-rich side, two types of starting materials wereemployed; one was a two-phase mixture of reagentCaCO, (calcite) and SrCO, (strontianite), the otherwas a homogeneous orthorhombic phase prepared(metastably) by precipitation from aqueous solutionsat 90oC. By choosing bulk compositions lying withinthe miscibility gap and running charges of both typesin separate capsules inside the same pressure vessel,the equilibrium compositions of the product phaseswere approached simultaneously from the low-Srand high-Sr sides, thereby bracketing the curves.Run durations increased from 7 days at 650o to 6ldays at 350'C. These runs were isobaricallyquenched by cooling to room temperature in a com-pressed air stream; temperatures below l00oC weretypically reached in 3-4 minutes. All such nrns crys-tallized two carbonates: a rhombohedral Ca-richphase and an orthorhombic Sr-rich phase. Composi-tions of orthorhombic run products were obtainedfrom the determinative curve of Froese and Winller(1966, Fig. 2), based upon the 132 interplanar spac-ing derived from X-ray powder di-ffraction data andusing Si metal as a standard. Multipte determinationsof known compositions showed the standard errors tobe approximately +l mole percent.

The compositions of the rhombohedral phases inthese runs could not, however, be used to bracket theCa-rich limb of the miscibility gap, owing to the dif-ficulty of preserving a single Sr-bearing rhombohe-dral phase during quenching. Similar problems werenoted by Froese and Winkler (1966, p. 559) andChang (1965, p. 358). Optical examination of theseruns reveals that most of the Ca-rich grains includetiny crystals with rectangular outlines, interpreted asquench crystals of strontianite. C.alcite soal4ining

these quench crystals appears as a single phase in X-ray diffraction patterns, because the strontianite com-prises so small a proportion of the total, but calcitelattice parameters derived from X-ray patterns ofsuch two-phase mixtures lead to incorrectly inferredoriginal compositions.

Consequently, the Ca-rich limb of the miscibilitygap was located by bracketing the maximum limit ofSr solubility in the rhombohedral carbonate. Theseexperiments used a series of increasingly Sr-richcompositions at each temperature and pressure, untilthe rhombohedral phase became saturated with Srand a separate orthorhombic Sr-rich phase appeared.The presence or absence of an equilibrium ortho-rhombic phase in the run products was determinedoptically, using the silver and manganese sulfate5faining solution described by Deer et al. (1966, p.500). This lsshnique pennits recogniliol of an ortho-rhombic phase present in amounts too small to be de-tected in X-ray powder diffraction patterns. Theequilibrium was reversed in the usual manner, by ob-taining at each temperature and pressure the limitingcompositions for which a two-phase assemblage con-verts to a single phase, and vice yersa. Run durationswere l0 days at 650o, increasing to 68 days at 450"C.These runs were quenched by first decompressing toone atmosphere, then cooling; in consequence,quench crystals were not observed to form.

Results

Figure 2 presents those data points which bracketthe equilibria on plots of pressure vs. composition, inseven isothermal sections. The results of this studyare shown as circles; open symbols denote composi-tions lying within the miscibility gap, and filled sym-bols denote compositions lyng in one-phase fields.Tables I and2 detail experimental conditions and re-sults for individual hydrothermal runs. Data fromfour other sourc€s are also included: (l) points locat-ing the calcite-aragonite reaction in the pure system,shown as squares, are from Johannes and Puhan(1971); (2) points indicating the pressure at which aphase transition was suspected to occur between tworhombohedral carbonates in the pure CaCO, system,shown as diamonds, are from Goldsmith and New-ton (1969); (3) points defining the Sr-rich limit of themiscibility gap, shown as rectangles in Figures 2Aand 2C, are from Froese and Winkler (1966)'z; and

2 These data represent slightly higher temperatures than thosewith which they are plotted, r'.e., 560otlOoC (instead of 550'C) inFig.2C, and 655otl0"C (instead of 650'C) in Fig. 2A.

14

12

10

8

6

q

CARLSON: CALCITE-ARAGON IT,E EQUI LI BRI UM

2

40 50 60SrCO.

N L

r255

14

?,tzJuJ10tr]Bafroffo

2

o5ulrlaaulrn

o5IrJrlaaUJr0-

8

6

4

10 20MOLE

,f0 20 30 40 50 60MOLE o/o SrCO.

10 ?0 30M.OLE O/O

A L"o 30 40 50 60o/o SrCO.

(e) T=450"C

UJ(laa[Jtr0-

B

6

4

0

a5UJrlaaIJr0-

oJUJt_laaUJff

0_

30o/o

40 20 30 40 50 60MOLE o/o SrCO3

,f0. 20MOLE

40 50 60SrCO"

12

40

n l r I | | |

"0 10 20 30o/o SrCO.MOLE

Fig. 2. Isothermal sections of the system CaCOr:51gOr. "Co" : ordered rhombohedral solid solution (ordered calcite); *Ca" :

disordered rhombohedral solid solution (disordered calcite); "A" : orthorhombic solid solution (aragonite). Symbol types designate

sources ofdata: squares, Johannes and Puhan (1971); diamonds, Goldsmith and Newton (1969); triangles, Chang (1965, l97l); circles,

this study. Open circles denote compositions lying within the miscibility gap, and filled circles denote compositions lying in one-phase

fields. Rectangles in Figures 2A and 2C are from Froese and Winkler (1966) and repres€nt data obtained at slightly different

temperatures (655'Cil0' in Fig. 2A; 560oC+10" in Fig. 2C).

(f) T=4000c14

12

10I

8

o

.{

aJUIrlaaUJtr0-

MOLE 7. SrCO.

1256

(4) data delirniting the miscibility gap, plotted justabove the composition axis and shown as triangles,are from Chang's (1965) one-atm work. The valuesreported by Chang (1971) at l0 kbar, shown as tri-angles in parentheses, are also plotted for reference.These last data were obtained in an opposed-anvilapparatus, and no details ofpressure catbration werepublished for them. Compared to the rest of the re-sults in Figure 2, the "10 kbar" values are clearly dis-crepant, especially at low temperatures. Con-sequent ly, they were not used to locate theequilibrium curves.

The boundaries of the miscibility gap are drawn tofall between the limits obtained in this study and toconnect with the data from other sources on the pres-sure and composition axes. On the Ca-rich side, allcurves are bracketed by a composition interval oftwo mole percent SrCOr. On the Sr-rich side, the un-

Table l. Runs bracketing Sr-rich limit of miscibility gap

BuIkDuratlof, composition

(days) (n1 Z src03)

CARLSON: CALCITE-ARAGONITE EQUILI BRI UM

orthorhonbicrun products(d Z srco3)

Ppt* Mlx*

Table 2. Runs bracketing Ca-rich linit of miscibility gap

Bulk Reactant ProductP Duratlon coaposltlon phases* phases*

(kbar) (days) (m1 Z SrCOj)

*AbbretLations for plusee :A = a"agonite soli.d. solutimC = qlctte eoLid soLuti,on

certainties depend upon reaction rates and run times,with allowance for error in determining compositions(+l mole percent). At 650"C, reaction is completeand the width of the brackets is merely the range ofdeterminative error; at lower temperatures, as thewider brackets show, the equilibrium is less tightlyconstrained because of sluggishness of the re-equili-bration.

Also shown in Figures 2A through 2D are curvesconnecting the data points of Goldsmith and Newtonfor the calcite I-calcite 1[ 6ansition (now recognizedas a transition to some phase other than calcite II) toisothermal invariant points along a horizontal line.This line marks a three-phase equilibrium, represent-ing the reaction of Sr-bearing aragonite with Sr-bear-ing calcite to produce a second Sr-bearing rhom-bohedral phase. Because this second rhombohedralphase cannot be preserved on quenching, and thuscannot be confrmed in the run products, the locationof these features is interpretive, based on the follow-ing rationale. The existence in the binary system of atwo-phase field for rhombohedral carbonates ofslightly different compositions was originally sug-gested by the abrupt change of slope and apparentslight offset of the Ca-rich limb of the miscibility gapshown in Figures 2,{ and 2B. To confirm the pres-

T( "c )

650 1 10

550 2 LO

550 4 10

c+A cc c+A

c+A cc c+A

c+A cc c.+A

c+A cc c+A

c+A cc c+A

c c+A

I6181618L6I8

I41610I2L214

c+A cc c+A

c+A cc c+A

c+A cc c+A

c+A cc c+A

c+A cc c+A

c+A cc c+A

c+A c

10L2

IOL2

10L2

68

d

500 2 L4

600 3 L4

600 4 L4

550 2 24

550 4 24

500 2 40

500 4 40

I

("c)P

(kbar)

650650650650550

600600600600500

550550550550550

500500500500500

450450450450450

400400400400400

350350350350350

I4

L4

L4

343434

34

39393 93939

12

45

I23

5

I2345

I2345

4 040404040

3030303030

3030303030

25

2525

5 5 . r 5 5 . 95 4 . 8 5 6 . 15 1 . 3 5 2 . L46.5 46 .344.8 44 .O

5 4 . 0 5 4 . 25 0 . 3 5 0 . 64 7 , 0 4 6 , 940.9 43 .83 7 . 0 4 0 . 5

4 9 , 8 s r - . 34 7 . 0 5 1 , 642.3 46 .53 8 . 1 4 3 , 034.7 39 ,4

49.6 52.244.2 47 ,A4 2 . 5 4 2 . 53 5 , 5 3 8 . 7

2020202020

2020202020

2020202020

5 0 . 946.74 0 . 6

27.4

52.246,3

3 7 . 43 0 . 0

61616I6r

5 2 . 5 5 3 . 846.3 48 .63 7 . 9 4 3 . 83 1 . 3 3 7 , 125 .3 32 .6

5 6 . 3 5 8 . 24 5 . 2 5 0 . 13 6 . 0 4 7 , 62 9 . 4 4 0 . 023. I 32 .1

12

45

*Signi | iee t1pe of reactf f i ts:

Mr = tieture of reagent CaCO3 (cal,eitd Md "eagent

S"CO3( qtymtianite )

Ppt = lorcgeneoas ortlnrhambie @precipitdte (arcgonite

eolid soTutim)

CA RLS ON : CA LCITE-A RAGON I TE EQUI LI BRI U M 1257

Fig. 3. Isobaric sections of the system CaCOr-SrCOr. Symbolsand field labels as for Figure 2.

ence of such a two-phase region, additional experi-ments were performed at2 and 4 kbar; the results aregiven in two isobaric sections in Figure 3. (Note thatthe boundaries of the two-phase fields are con-strained to intersect for zero percent SrCO, at Gold-smith and Newton's transition points; these inter-sections, not plotted, occur near 890o for Figure 3Aand 810"C for Figure 38.) The distinct increase insolubility of SrCO, in the rhombohedral phase be-tween 550o and 560'C at 4 kbar (Fig. 38) is the pri-mary basis for including a field for the coexistence oftwo rhombohedral phases in Figures 2 and 3.

The surface dsfining the Sr-rich limit of the mis-cibility gap is contoured in P-T projection in Figure4, allowing comparison with other available data.Heavy lines indicate the results of the present study,

350 400 450 500 550 600 650TEMPERATURE ('C)

Fig. 4. Isoplethal contourc on Sr-rich limit of miscibility gap.Heavy lines from this study. Light lincs from Froese atrd lVhkler(1966). Composit ions in rectangles from Chang (1965).Discrepancies at high temperatures and pressures result fromdi-fferent choices of the calcite-aragonite boundary curve (seetext).

constrained at the end-member calcite-aragoniteboundary (the zero percent contour) by the data ofJohannes and Puhan (1971). Light lines are fromFigure 4 of Froese and Winkler (1966), *6s ulilizedClark's (1957) work to locate the zero percent con-tour. Chang's (1965) values for the composition oforthorhombic carbonate in equilibrium with calciteat one atm appear enclosed io lsctangles just abovethe composition axis and are expressed in mole per-cent SrCO..

Discussion

Effect of Sr on calcite-aragonite stability relations

The compatibility of Chang's (1965) determinationof the Sr-rich limit of the miscibility gap with thedata of the present study is apparent in Figure 4. Incontrast, agreement with Froese and Winkler's(1966) results occurs only at the high temperatures atwhich their experiments were performed. Significantdiscrepancies are evident when their values a1s lin-

early extrapolated to lower temperatures. It seems,therefore, that the disagreement between Chang'sdata and Froese and Winkler's wotk is merely the re-sult of an unwananted linear extrapolation of thecontours. With that difficulty resolved, it may be con-fidently asserted that, in the temperature range ofblueschist metamorphism, even large amounts ofSrCO, component have a negligible effect on thepressure at which the calcite-aragonite reaction takesplace. This same cbnclusion was reached by Froeseand Winkler; their inference is correct, despite thediference between observed and extrapolated values,because the contours in Figure 4 curve upward ratherthan downward with decreasing temperature.

Change in slope of the calcite-aragonite equilibriumboundary

The controversy regarding the precise location ofthe calcite-aragonite reaction curve between 350oand 500'C focuses upon the manner in which thecalcite transition near l0 kbar and 500"C affects theequilibrium between the orthorhombic and therhombohedral phases. The calcite-aragonite equilib-rium curves published by Boettcher and Wyllie(1968) and by Goldsmith and Newton (1969) bothconsist of two intersecting lines with distinctly differ-ent slopes. Both teams of investigators perforfired ex-periments down to 400'C and then extended theircurves through the point determined by Crawfordand Fyfe (1964) at 100"C. In the absence ofdata be-tween l00o and 400"C, all the points could be fitted

M.OLE'/. SrCOg

Contours on Sr-rich limit ofmiscibility gop

noIa;ebJJo

t258 CARLSO N : CALCI TE-ARAGON ITE EQ U I LI BRI U M

to a pair of straight lines whose intersection produceda discontinuity of slope, indicative of a first-orderphase transition in one of the two minerals. An alter-native model was invoked by Johannes and Puhan(1971) who, with gas-apparatus data extending downto 200"C, found the calcite-aragonite equilibriumboundary to be continuously curved. Because theirboundary exhibits no break in slope, they inferredthat the phase transition is of second order. They fur-ther noted that the uncertainties in their data aresmall enough to preclude fitting the data with twostraight lines.

Because this portion of the calcite-aragoniteboundary is quite signifi€nl petrologically, it is espe-cially unfortunate that direct evidence for the natureof this unquenchable phase transition is difrcult toobtain. Nearly all the inferences drawn so far haverelied upon extrapolation either from the low-tem-perature high-pressure studies of Bridgman (1939),Jamieson (1957), Davis (1964), Wang (1968), andMerrill and Bassett (1975), or from the low-pressurehigh-temperature experiments of Boeke (1912),Smyth and Adams (1923), and Eitel (1923). As notedbefore, however, the transitions observed in these P-Zregions do not appear to be the cause ofthe chang-ing slope of the calcite-aragonite equilibrium; thedata of Kondo et al. (1972) demonstrate this for thefirst group of studies,3 and the work of Cohen andKlement (1973) and Mirwald (1976,l979a,b) does sofor the second group.

In contrast to these interpretations based on ex-trapolations, Goldsmith and Newton (1969) tried toobtain direct evidence of the nature of this transitionat pressures and temperatures near the change inslope of the calcite-aragonite boundary. They uti-lized differential thermal analysis, sought twinning insingle crystals cycled through the transition, and at-tempted to demonstrate di-fferential solubility ofMgCO. in the two phases-all methods intended todetect a discontinuity in some physical property atthe transition. Although their findings were sugges-

3 It would seem prudent, however, to regard these results (basedon ultrasonic velocity measurements) as tentative, because of thelikelihood that partial invcrsion of calcite to aragonite occurs inexpcriments run at these relatively high temperatures. Thc kineticsdata ofDavis and Adams (1965) show appreciable conversion inonly a few minutes near 450oC and 8 kbar, and reaction rates willincrease exponentially with temperature. Additionally, 5ingls sem-ples were cycled through multiple runs, prolonging the exposureto high temperature. In the absence of characterizations of thephase(s) present at the end ofeach run, this conclusion should per-haps be accepted with reservation.

tive of a first-order phase transition, no lsshniqueproduced an unequivocal result.

The present study presents observations at hightemperatures and pressures which should help to ex-plain the change in slope of the calcite-aragoniteboundary. I suggest that the phase change which al-ters the slope of the boundary is correlative with theorder-disorder transition observed in binary Ca-Srcarbonates by Chang (1965), a transformation dis-tinct from both the calcite I-II and the calcite IV-V(B-a) transitions. Chang noted the gradual dis-appearance of ll3-type reflections in X-ray diffrac-tion patterns taken at high temperature at atmo-spheric pressure. The I 13 reflection is the strongest ofa class of reflections whose presenoe implies orienta-tional ordering of the CO, groups in a rhombohedralcarbonate structure; their absence results in a halvingof the c-translation distance and a change of spacegroup from R3c to R3zr. The disordering under dis-cussion here refers to the angular orientation of thetrigonal planar CO. groups about their three-foldaxes. In the ordered structure of calcite I, the CO,groups of successive layers have opposite orienta-tions, while the observed symmetry of the high+em-perature forms requires the orientation of successiveCO, groups to be, at least on the average, the same.As pointed out by Megaw (1973), this may be reason-ably accomplished by one of two means: either theCO, groups retain their two possible calcite I orienta-tions but with a random sequence of alternations, orthe CO, groups may be rotated (either plus or mingg30') to a new orientation halfway between the twopossible orientations of calcite I. In either case, theorientational disorder makes an additional contribu-tion to the molar entropy of the crystal making thedisordered polymorph the higher entropy form.

Chang's (1965) experiments at atmospheric pres-sure show that the intensity of the l13 reflection for acomposition of 44 mole percent SrCO, diminishes tozero at 705"C, thus marking the completion of theorder-disorder transition and producing an equilib-rium among three carbonate polymorphs. Similarly,the phase diagrams in Figures 2 and,3 of my studylocate a phase change between two rhombohedralcarbonates; they likewise depict three phases simulta-neously stable in the binary system, but at elevatedpressures. The isobaric sections of Figure 3, whichare based on phase equilibrium data, are topologi-cally identical to the one-atn phase diagram ofChang (1965, Fig. 5), which is based on the loss ofthose X-ray peaks which imply rotational ordering ofthe CO, groups in the carbonate structure. Because

CARLSON: CALCIT:E-A RAGON IT:E EQUI LI BRI UM

of this precise equivalence, the two fields for rhom-bohedral carbonate in this investigation have beenidentified as ordered and disordered calcite.

The P-T conditions for the binary three-phaseequilibrium observed in this study connect smoothlywith Chang's conditions for the coexistence of threephases (705'C at one bar) and the triple-point regionof the calcite-aragonite curve (500"C at l0 kbar),also strongly suggesting that the transitions areequivalent. To see this more graphically, the set ofpoints at which the transition occurs in the binarysystem may be projected onto the pressure-temper-ature plane, producing a line with negative slopewhich intersects the contours of the Sr-rich limit ofthe miscibility gap. Such a projection is made in Fig-ure 5. Significantly, the points of intersection divideeach of the contours into a highly curved portion atlower temperatures and a nearly linear portion athigher temperatures. The slope of each contour is theratio of the entropy change to the volume change forthe reaction of the two-phase assemblage to a singleorthorhombic phase, for the composition appropriateto each contour: dP/af : A,S/AV. Thus wherearagonite solid solution is in equilibrium with the or-dered calcite solid solution, AS/AV is gradually in-creasing with temperature; above the calcite transi-tion temperature, where aragonite solid solution is inequilibrium with the disordered rhombohedralphase , A,S / L,V is nearly constant.

Thus the curvature of the contours in the binarysystem and, by extension, the much less apparentcurvature of the calcite-aragonite boundary in theunary system, may be regarded as evidence of a grad-

TEMPERATURE ('C)

Fig. 5. Projection of order-disorder transition boundary ontopressure-temperature plane. The intersections with the contoursof the Sr-rich side of the miscibility gap mark the change fromincreasing slopes to constant slopes of the contoun.

ual orientational disordering taking place in therhombohedral carbonates. This requires that muchof the disordering occur in the unary system over aninterval, rather than taking place entirely at the tran-sition point. In support of this viewpoint, Ramanspectroscopy of calcite by Salje and Viswanathan(1976) identified an ionic rotational mode correlatedto the coupling of oppositely oriented CO, groups,which weakens and eventually disappears at hightemperatures. Their calculated calcite-aragoniteequilibrium based upon the premise of anion dis-order reproduces, with excellent agreement, the cur-vature of the boundary determined by Johannes andPuhan (1971).

A further important feature of the present data isthe evidence for a discontinuity in the disorderingwhen the transition point is actually reached. Thedata in Figure 3 display an abrupt increase in the sol-ubility of SrCOr in the rhombohedral phase between550" and 560"C at 4 kbar. In view of the smoothcurves evident above and below this small temper-ature interval, the offset is taken to represent the or-der-disorder transition. If the transition were of sec-ond order, no offset in the curve would occur; only itsslope would be discontinuous. The observed effectcan be reconciled only with a first-order phase transi-tion. Evidence for the same phenomenon near 630"Cat2kbar (Fig.3) is equivocal; the data are permissiveof a small offset of the curve, but do not require it.Chang (1965, p. 357) described attempts to locate atwo-phase region at 775" and 825oC and one atm,s5ing a high-temperature X-ray diffractometer. Em-ploying compositions spaced at intervals of 2 molepercent SrCOr, he found no evidence for the coexis-tence of two rhombohedral forms. This is not surpris-ing, considering that the two-phase field spans acompositional range of only about 4 mole percent at4 kbar (Fig. 3B), decreasing to a range of 2 mole per-cent or less at 2 kbar (Fig. 3A).

The observed fractionation of Sr preferentiallyinto the disordered phase serves to constrain possiblemodels for its crystal structure. For NaNOr, which isisomorphous with calcite, Megaw (1973) discussedthe disordering models introduced above. She favorsthe second model, involving 30o rotations of the an-ionic groups, for tle high-temperature structure be-cause it produces more nearly optimal interatomicdistances, and because it agrees with extant thermalexpansion and X-ray diffuse scattering data. She alsopointed out that in the first model the randomizedcalcite I orientations led to coordination of the Naion (or Ca ion in CaCO') by 12 equidistant "half-ox-

oJUJtrlaaUJtrn

Effect of order-disorder tronsition oncurvoture of contours on Sr-rich

ooL

aiSlrlJo

1260 CARLSON: CALCITE-AMGON ITE EQUILIBRIUM

ygens," of which only six could represent actuallypresent neighbors of one cation. The second model,involving rotation of the NO, groups (or CO, groupsin CaCOr) to new halfway positions, resulted in eachcation fsing surrounded by eighteen "half-oxygens"or nine true neighbors. In an isothermal section suchas Figure 2, it is evident that despite its higher en-tropy the disordered phase is denser than the orderedphase. Furthermore, Sr is a larger cation than Ca andis preferentially fractionated into the disorderedstructure. Both observations tend to support the sec-ond structural model, in which the cations are morehighly coordinated, and which is therefore morelikely to be the denser structure.

One alternative possibility, namely that the transi-tion which inflects the calcite-aragonite equilibriumcurve involves Bridgman's calcite II, has also beeninvestigated experimentally. Merrill and Bassett's(1975) structure determination of calcite II demon-strates that, despite a smaller cell volume, calcite IIhas a greater average Ca-O bond lenglh than calciteI. In consequence, calcite II is expected to accommo-date substitution of the larger Sr ion for Ca morereadily than calcite I. This is compatible with the ob-served distribution of Sr between the two rhombohe-dral phases seen in Figures 3A, and 3B. Thus the cal-cite II structure is a potential candidate for thedisordered calcite phase encountered in the presentstudy, the results of Kondo et al. (1972) notwith-standing. However, if calcite II were actually thephase involved, then the disordered rhombohedralphases of Chang (1965) would be expected to havethe calcite II structure, in view of the apparent corre-spondence between the CaCO, triple point and thebinary three-phase equilibrium.

Given the subtlety of the features distinguishingthe calcite II structure from a disordered rhombohe-dral structure, it is conceivable that these details ofthe X-ray diffraction pattern might previously havebeen overlooked in binary Ca-Sr carbonates, as theywere in the first two X-ray characterizations of calciteII. To explore this possibility, high-lsmnerature X-ray powder diffraction patterns of binary Ca-Sr car-bonates were obtained using a Norelco difractome-ter with a platinum strip heater. These experimentswere designed to observe, ifpossible, the presence ofdffiaction maxima corresponding to the Oll, 221,214, and 104 peaks reported as new reflections forcalcite II by Merrill and Bassett (1975). Under a CO,atmosphere, metastable orthorhombic coprecipitateswere observed to undergo transition at temperaturesconsistent with Chang's (1965, Fig. 5) phase diagram.

Diffraction patterns were obtained for the composi-tion CaruSr* at75o" and 800oC, and for the compo-sitions CaroSrro and CaroSrro at 900oC. In addition, arhombohedral starting material of compositionCar"Sr* (synthesized at 925"C in CO, and quenchedin water) was examined at 800"C. None of the exper-irnents showed any increase of intensity above back-ground in the vicinity of the peaks noted in the cal-cite II difraction pattern, although even low-intensity peaks of the disordered rhombohedralstructure were easily detected. The implication, de-spite the strong similarity in the behavior of the l13-type peaks reported by Jamieson (1957) and Davis(1964) for calcite II and by Chang (1965) for dis-ordered binary carbonates, is that the two phases arein fact distinct entities. This result lends credence tothe fnding of Kondo et al. (1972) that calcite II is no-where stable with respect to aragonite.

Finally, the relationshiF between the order-dis-order transition examined here and the calcite l-cal-cite IV transition of Mirwald (1976, l979a,b) de-serves attention. Although the location of the order-disorder transition in the binary system seems well-determined, its P-Tpath in the unary CaCO, systemcan only be guessed at. The location ofthe transitionidentified by Goldsmith and Newton (1969) isadopted here, but because of ambiguities in that de-termination, the boundary may depart significantlyfrom that trajectory (the diamonds in Figs. 2Athrough 2D) as it moves away from the triple-pointregion. It must, however, have a negative d,P/dTslope, and the recent discovery of a possible phasechange in CaCO, at 725"C and 40 bars (Mirwald,1979b) provides a tempting candidate for the low-pressure location of the order-disorder transition.Mirwald, however, interpreted his data as indicatingthe onset of disordering near 725" at 40 bars, withcompletion near 800oC. In his view, the transition ofcalcite I to calcite IV is an order-disorder transitionwhich intersects the calcite-aragonite boundary near800"C, not near its change of slope. Possibly morethan one disordered structure is involved; thus thecalcite I-IV phase change and the order-disordertransition of this study may intersect in a triple pointat elevated pressures. But experimental con-frmationof such speculation will be extremely difficult to ob-tain, owing to the subtlety of the distinctions amongthe phases. For the present, it must suffice to say thatthe variety of possible disorder mechanisms (e.g.,Megaw, 1973; Strlmme., 1975) allows for a greatcomplexity of polymorphic transitions a1 high tem-peratures; even this comparatively simFle unary sys-

CARLSON: CALCITE-ARAGONITE EQUILIBRIUM

65ldr.laaultr0_

"'il sss -lssTEMPERATURE ( 'C)

Fig. 6. P-T-X diagram of phase relations in the system CaCO3-SrCO3. Surfaces bounding the two-phase field for ordered anddisordered calcite are indicated by shading. Data on rear face from Johannes and Puhan (1971), Mirwald (1976), and after Goldsmithand Newton (1969); data on bottom face after Chang (1965). Field labels as for Figs. I and 2.

tem of fundamental geologic significance cannot beregarded as fully determined.

gnmmsry

Phase relations in the binary system CaCOr-SrCO, are summarized in Figure 6, a rectangulardiagram with axes of pressure, temperature, andmole percent SrCOr. The bottom face of the prismpresents Chang's (1965) one-atmosphere data, modi-fied to show a two-phase field for the coexistence ofordered and disordered rhombohedral carbonate.The width of that field is slightly exaggerated forclarity. The rear face displays the calcite-aragonitereaction determined in gas-pressure apparatus by Jo-hannes and Puhan (1971), the order-disorder transi-tion consistent with Goldsmith and Newton (1969),and the stability field of calcite IV after Mirwald(1976). The curved lines in the interior of the prismreproduce Figures 2A, 2C, 2E, and, 2G. The twoshaded surfaces are those which enclose the two-phase field for rhombohedral carbonates.

The data of this study demonstrate that the smallamounts of Sr found in natural aragonites are nearlyinconsequential with regard to pressure estinates oforigin for metamorphic rocks of the glaucophane-schist facies. The results further imply that thechange of slope of the calcite-aragonite equilibriumboundary between 350o and 500'C is due to anion

orientational disordering in calcite, premonitory to afirst-order phase transition. This in turn suggests thatthe calcite-aragonite reaction is best represented bythe continuously curved boundary of Johannes andPuhan (1971), modified to show a small discontinuityin slope in the vicinity of the triple point inferred byBoettcher and Wyllie (1968) and Goldsmith andNewton (1969).

AcknowledgmentsThis work was supported by an NSF Graduate Fellowship, by

NSF grant F,Ak 77-1.0649/Ernst, and by the Department of Earthand Space Sciences at UCLA. Dr. G. E. Brown graciously madeavailable the high-temperature X-ray facilities at Stanford Univer-sity, and G. A. Waychunas and K. D. Keefer were most helpfulduring the course of those experiments. I am especially indebtedto W. G. Ernst, W. A. Dollase, A. L. Boettcher, J. L. Rosenfeld,and W. M. Thomas for critical rcading of the manuscript, and forcontinual discussion and encouragement.

ReferencesBoeke, H. E. (1912) Die Schmelzerscheinungen und die umkehr-

bare Umwandlung des Calciumcarbonats. Neues lahrb. Min-eral. Geol., I,9l-212.

Boettcher, A. L. and P. J. Wyllie (1968) The calcite-aragonitetransition measured in the system CaO-CO2-H2O. J. Geol., 76,3l+330.

Bridgrnan, P. W. (1939) The high-pressure behavior of miscella-neous minerals. Am. J. Sci.,237,7-18.

Chang, L. L. Y. (1965) Subsolidus phase relations in the systemsBaCO3-SrCO3, SrCO3-CaCO., and BaCO3-CaCO3. f. Geol.,73,346-368.

t262 CARLSON: CALCITE-ARAGONITE EQUILIBRIUM

- (1971) Subsolidus phase relations in the aragonite-typecarbonates: I. The system CaCO:-SrCOr-BaCOy Am. Mineral.,56,1660-1673.

Clark, S. P. (1957) A note on calcite-aragonite equilibrium. lzr.Mineral., 42, 564-566.

Cohcn, L. H. and W. Klement (1973) Determination of high-tem-p€rature transition in calcite to 5 kbar by differential thermalanalysis in hydrostatic apparatus. J. Geol.,8I,72+727.

Colemart" R. G. and D. E. Lee (1962) Metamorphic aragonite inthe glaucophane schists of Cazadero, California. Am J. Sci.,260,577-595.

Crawford, W. A. and W. S. Fyfe (1964) Calcite-aragonite equilib-rium at l00oC. Scr'ence, 144,1569-1570.

- and A. L. Hoersch (1972) Catate-aragonire equilibriumfrom 50oC to 150"C. Am. Mineral., 57,995-99E.

Davis, B. L. (1964) X-ray difraction data on lrpe high-pressurephases of calcium carbonate. Science, 145, 489491.

- and L. H. Adams (1965) Kinetics of the calcite-aragonitetransformation. I. Geophys. Res., 70, 433-441.

Deer, W. A., R. A. Howie and J. Zussman (1966) An Introductionto the Rock Formrng Minerals. Longmans, London.

Eitel, W. (1923) Uber das binare System CaCO3-Ca2SiOa utrd denSpurrit. Neues Jahrb. Mineral. BeiL Band,48,63-74.

Froese, E. and H. G. F. Winkler (1966) The system CaCO3-SrCO3at higb pressures and 500oC to 700"C. Can. Mineral., & 551-566.

Goldsmith, J. R. and R. C. Newton (1969) P-T-X rclations in thesystem CaCO3-MgCO3 s1 high lsnp6laturcs and pressures. ArtuJ. Sci., 267,160-190.

Irving, A. J. and P. J. Wyllie (1973) Melting relationships in CaG-CO2 and MgO-CO2 to 36 kilobars with comrnents on CO2 inthe rnantle. Earth Planet. Sci. Lett.,20,220-225.

Jamieson, J. C. (1957) Introductory studies ofhigh-pressure poly-morphism to 24,000 bars by X-ray diffraction with some com-ments on calcile II. f. Geol., 65,33+343.

Johannes, W. and D. Puhan (1971) The calcite-aragonite transi-tion, reinvestigated. Contrib. M ineral. Petrol., 3 I, 28-38.

Kondo, S., K. Suito and S. Matsushima (1972) Ultrasonic obser-vation ofcalcite I-II inversion to 700"C. J. Phys. Earth,20,245-250.

Megaw, H. D. (1973) Crystal Sttttctures: A Working Approach.'W.B. Saunders, Philadelphia.

Merrill, L. and W. A. Bassett (1975) The crysral structure ofCaCO3 flI), a high-pressure metastable phase of calcium car-bnate. Acta Crystallogr., B 3 I, 343-349.

Mirwal4 P. W. (1976) A differential thermal analysis study of thehigh-tenpel31gre polymorphism of calcite at high pressure.Contrib. Mineral. Petrol., 59, 33-40.

- (1979a) Determination ef a high-lsmperature transition ofcalcite at 800'C and one bar C02 pressurc. Nanes fahrb. Min-eral. Monatsh., 309-315.

- (1979b) The electrical conductivity ofcalcite between 300and l200oC at a CO2 pressure of 40 ban. Phys. Chem. Minerals,4, 29r-297.

Salje, E. and K. Viswanathan (1976) The phase diagram calcite-aragonite as derived from the crystallographic properties. Corr-trib. Mineral. Petrol., 55,55-67.

Smyth, F. H. and L. H. Adams (1923) The system elcium oxide-carbon dioxide. J. Am. Chem. 5oc.,45, 1167-1184.

Strlmme, K. O. (1975) On the crystal structures of the high+em-perature forms of strontium and barium carbonate atrd structur-ally related compounds. Acta Chem. Scand., A29, 105-110.

Wang, C. (1968) Ultrasonic study of phase transition in calcite to20 kilobars and 180'C. J. Geophys. Res.,73,3937-394.

Zimmermar\ H. D. (1971) Equilibrium conditions of the calcite/aragonite reaction between 150"C and 350'C. Nature Phys. Sci.,231,203-2U.

Manuscript received, January 28, 1980;

acceptedfor publication, May 7, 1980.