sector-zoned aegirine from the ilimaussaq alkaline intrusion, south

American Mineralogist, Volume 79, pages 340-352, 1994

Sector-zoned aegirine from the Ilimaussaq alkaline intrusion, South Greenland:Implications for hace-element behavior in pyroxene

C. K. SnunnnInstitute of Meteoritics, Department of Earth and Planetary Sciences, University of New Mexico,

Albuquerque, New Mexico 87131-1126, U.S.A.

L. M. LlnsnNThe Geological Survey of Greenland, Oster-Voldgade 10, DK-1350 Copenhagen I( Denmark

Ansrn-Lcr

Sector-zoned aegirine crystals from a peralkaline nepheline syenite in Ilimaussaq, SouthGreenland, have been analyzed for major and trace elements using combined electronmicroprobe and secondary-ion mass spectrometry techniques. Unlike in calcic clinopy-roxene, the faster growing basal sector of the sodic clinopyroxene is enriched in incom-patible elements such as Sr, REE, andZr relative to the slower growing prism sectors. Theprism sectors are enriched in Al and Ti relative to the basal sector, but the Al-Ti enrich-ment patterns in the prism sectors contrast with those in calcic clinopyroxene. Zr, REE,and Sr exhibit a positive correlation to Ca in the M2 site and a negative correlation to Nain the M2 site.

The contrasts we report between aegirine and augite demonstrate that site characteristicsand the ability of a site to accommodate a cation are important factors for controllingsector enrichments. The addition of an acmite component [Ca (M2) + Fe2+ (Ml) - 5u(M2) + Fe3+ (Ml)l to the pyroxene results in a shift of the optimum potential radius forcations accommodated in the Ml and a change in the ideal charge characteristic of theM2 site. Those cations with radii that deviate the farthest from the optimum Ml and M2site radii and ideal charge for a particular pyroxene composition have a greater probabilityof being retained in a faster growing basal sector than in the more slowly growing prismsector. These observations do not, however, eliminate a model in which adsorption anddesorption kinetics plays an important role in the differences among sectors in majorcomponents [i.e., Na/(Na + Ca) and Fe3+/(Fe2+ + Fe3+)]. This mechanism may thenultimately be responsible for the differences in crystal chemistry or geometric factor amongsector growth surfaces.

INrnoouctIoN

Chemical sector zoning occurs when different gowthsectors or face loci of a single crystal have significantlydifferent compositions (Hollister, 1970; Hollister andGancarz, l97l; Dowty,1976). These compositional dif-ferences are produced on surfaces of simultaneouslygrowing crystal faces under identical conditions (Hollisterand Gancarz, I 97 l). The existence of sector zoning clear-ly indicates disequilibrium crystallization (Leung, 1974;Shimizu, I 98 I ; Paquette and Reeder, 1990a, I 990b). Sec-tor zoning, therefore, provides a perspective into diffu-sional and interface kinetic mechanisms controlling trace-element behavior under disequilibrium conditions (Al-barede and Bottinga,1972; Shimizu, l98l).

Chemical sector zoning has been described in numer-ous synthetic compounds and naturally occurring min-erals (see Dowty, 1976, and kung, 1974, fot summary).Perhaps the best documented sector zoning in silicates isin pyroxene (Strong, 1969; Hollister and Gancan, l97l;Bence and Papike, 1971, 1972; Thompson, 1972; Fer-guson, 1973; Nakamura, 1973; Nakamura and Coombs,

0003-004x/94l0304-0340$02.00 340

1973; lrung, 1974; Downes,1974; Dowty, 1976,1977;Harkins and Hollister, 1977; Carpenter, 1980; Shimizu,198 l;Larsen, l98l;Kouchi et al., 1983). Several modelshave been designed to explain chemical sector zoning inpyroxene. The model proposed by Hollister (1970) andHollister and Gancarz ( I 9 7 I ) is based on the concept thatlocal charge balance is maintained by charge-compensa-tory cation substitutions into completed sites on thegrowth surfaces. Differences in the atomic configurationon growth surfaces and the nature of the prominent sub-stitutions are important variables. Nakamura (1973),Dowty (1976,1977), and Shimizu (1981) suggested thatthe adsorption and desorption of cations into partiallycompleted sites on the growth surfaces is an importantmechanism for the development of sector zoning. Growthsurfaces with favorable protosite configurations are there-by preferentially enriched. Adsorption of elements intoprotosites is proportional to the charge to size ratio ofthecation. Therefore, in contrast to the model of Hollister(1970) and Hollister and Gancarz (1971), coupled cationsubstitutions do not necessarily have to compensate for

SHEARER AND LARSEN: SECTOR-ZONED AEGIRINE 341

charge balancing within a single growth step or layer, andpartition coefficients for all elements approach one. Ko-uchi et al. (1983) concluded that sector enrichment wasrelated to the effect of supercooling on the crystal growthmechanism (i.e., spiral gfowth, two-dimensional nucle-ation growth, layer growth, and crystallographic orienta-tion of growth steps) and to surface kinetics. Changingcrystal gowth mechanisms should result in a changingsite (or protosite) configuration of the growth surfaces.The importance of such mechanisms has been docu-mented in carbonates (Paquette and Reeder, 1990a,1990b; Reeder, 1992). Compositional differences amonggrowth sectors has also been attributed to the interplaybetween growth rates ofgrowth surfaces and the relativediffusion rates of cationic species in the melt (Downes,197 4; Larsen, 1 98 1).

Previous studies of minor- and trace-element incor-poration into growth sectors of pyroxene have focusedupon clinopyroxene in which the M2 site is primarilyoccupied by Caz+ (Shimizu, 198 1; Kouchi et al., 1983).This study focuses upon sector-zoned pyroxene with sub-stantial Na+ in the M2 site. This difference in M2 occu-pancy influences the character ofsector enrichments andprovides additional insight into the mechanisms in-volved in crystal growth, the formation of chemical sectorzoning, and trace-element substitution.

OccunnrucE oF sEcroR-zoNED AEGIRTNE

Sector zoning in aegirine has been described from al-kaline plutonic and hydrothermal rocks (Larsen, 1976,198 l; Ranlov and Dymek, 199 1). The feature may bemore common in such rocks than generally realized be-cause it may be difficult to identify. The material used inthis investigation comes from the Ilimaussaq alkaline in-trusion in South Greenland, from one of the samples(GGU 91969) that was also used by Larsen (1981). TheIlimaussaq intrusion consists of an early intrusion of au-gite syenite followed by a layered complex of peralkaline(agpaitic) nepheline syenites: sodalite foyaite, naujaite,kakortokite, and lujavrite (Ussing, I 9 12; Ferguson, 1964,I 970; Larsen, I 976). Sector zoning is frequently observedin aegirine from the sodalite foyaite and naujaite. Sectorzoning is less common or absent in the other rock types(Larsen, l98l). Aegirine from the sodalite foyaite wasselected for this study. The mineral assemblage in thesodalite foyaite consists ofalkali feldspar, sodalite, neph-eline, hedenbergite, fayalite, titanomagnetite, katophor-ite, eudialyte, aenigmatite, aegirine, arfvedsonite, anal-cime, and rinkite (Larsen, l98l). The crystallizationsequence among these minerals has been summarized byLarsen (1976,1977).

The aegirine in the sodalite foyaite forms subhedral toeuhedral prismatic crystals 0.1-10 mm long. The crystalsare elongated in the c direction and are somewhat platyparallel to {100}. The dominating forms are {100} and{ I 10}. The form under which the basal sector grew hasnot been identified with certainty. This sector has beendenoted by Larsen (1981) by the zone symbol [001], a

designation proposed by Harkins and Hollister (1977).Boundaries between sectors are usually not straight butpossess curvatures indicative ofdifferences in the relativegrowth rates of the sectors. The relative growth rates ofthe various sectors as deduced from morphology by Lar-sen (1981 ) a re [ 001 ] > {010 } > { l l 0 } > {100 } .

ANllvrrc,Lr- TECHNIQUES

After optical observation and preliminary microbeamanalysis of several aegirine grains from the sodalite foy-aite, a single sector-zoned aegirine crystal from sampleGGU 91969 was analyzed in detail with electron micro-probe (EMP) analysis and secondary ion mass spectrom-etry (SIMS). Sectors within the crystal were distinguishedbased on the criteria of Larsen ( 198 I ) for similar aegirinecrystals in the Ilimaussaq intrusion. A photograph of thesector-zoned aegirine crystal, a reconstruction ofthe sec-tor zoning, and EMP-SIMS points are shown in Figure l.

Prior to SIMS and electron microprobe analysis, thecrystal was photographed to document analysis locationsprecisely. In addition, backscattered electron imaging andpreliminary microprobe analyses documented both sec-tor and concentric zoning. To provide combined SIMSand EMP analyses of identical points, it was decided toperform SIMS analyses prior to detailed EMP analysis.SIMS analysis craters are relatively easy to locate for sub-sequent EMP analysis.

The SIMS analyses were obtained with a Cameca IMS3f ion probe located at the Woods Hole OceanographicInstitute. Rare earth element analyses were carried outwith a primary O ion beam, which was 25-30 pm indiameter. Moderate energy filtering was used to removemolecular ion interferences (Shimizu and Hart, l98l).The data for a second set of trace elements (Zt, Ti, Cr,V, Sr, Y) were obtained with a primary O- ion beam 5-8 pm in diameter and with more stringent energy filtering(90 V offset). The secondary ion intensities were mea-sured by an electron multiplier coupled with pulse-count-ing circuitry. All concentrations were calculated from em-pirical relationships between intensity ratios relative toSi (as CelSi) and concentrations (i.e., working curves) ob-tained from well-documented standards (Irving and Frey,1984; Shimizu, unpublished data). Accuracy and preci-sion for REE data are l5o/o or less and <l0o/o for Y, Ti,V. Zr. and Sr.

Following SIMS analysis, the points adjacent to SIMSspots were analyzed for major and minor elements usinga JEOL 733 electron microprobe equipped with an Ox-ford-Link system at the University of New Mexico. Min-eral analyses were obtained using mineral and oxide stan-dards, an acceleration voltage of 15 KeV, and a beamdiameter of 2-5 pm. Measured X-ray intensities werereduced according to the method of Bence and Albee(1968).

Cnvsrnr, cHEMrsrRY oF PYRoxENE

The pyroxene group minerals crystallize in a variety ofspace groups, but for our purposes only the C2/c for au-

342

a.

SHEARER AND LARSEN: SECTOR-ZONED AEGIRINE

(1 10)Fig. l. (a) Photograph ofone ofthe aegirine crysrals used in

this study. The aegirine is approximately 6 mm in length. (b)Idealized section (dashed line), which approximates the orien-tation of a. (c) Identified sectors and EMP-SIMS analyses points.Location ofanalyses in Table I are designated by number and

b.

(1 11)

(1 10)b

x-section(010)

of crystal

open point. (d) Na/(Na + Ca) zoning in crystal. Patterns repre-senting Na,/(Na + Ca) values are as follows: dotted : <0.80,horizontal lines : 0.85-0.90. white : 0.90-0.95. and inclinedlines : 0.95-0.98.

c.

{

(100)

(110) l,.'E

o a17o oo looll

gite and aegirine are important. The pyroxene structurecan be described in terms of alternating tetrahedral andoctahedral layers that lie parallel to the (100) plane. Thetetrahedral layer is composed of infinite chains of corner-sharing tetrahedra running parallel to c. The octahedrallayer is composed of the sixfold- to eightfold-coordinatedsites. Ml octahedra share edges to form infinite chainsparallel to c. The M2 polyhedra are either sixfold oreightfold coordinated, depending on their occupancy. Thesite is eightfold coordinated when it contains Ca or Nabut is approximately sixfold coordinated when contain-ing smaller cations. Shearer et al. (1989) and McKay(1989) reviewed the impact of Ml and M2 site charac-teristics on trace-element behavior.

Parameters describing Ml and M2 site characteristicsfor diopside, hedenbergite, and acmite can be calculatedbased on the methods used by Smyth and Bish (1988),Caporuscio and Smyth (1990), and Smyth (1992). Usingthe data of Smyth and Bish (1988), we calculated thecation site radii, site electrostatic potential, and optimalcation charge for Ml and M2 sites in hedenbergite andacmite (Table l). The data in Table I for diopside werecalculated by Caporuscio and Smyth (1990). Over thecompositional range from hedenberyite to acmite, the M2site radius varies by <lol0. The optimal cation chargevaries by approximately 54o/o. The M I site over the samecompositional range shows a different relationship. TheMl site radius varies by 5ol0, whereas the optimal chargevaiesby 23o/o.

As demonstrated by Dowty (1976), the growth sectors{ 100}, { I l0}, {0 l0}, and { I I I } have different surface siteconfigurations. Surfaces ofleast bonding parallel to { I I I }expose only M2 sites or Ml sites with four of the sixoctahedral bonds directed toward the crystal (4/6 site us-ing the terminology of Nakamura,1973) (Dowty, 1976).Both the { I l0} and {010} surfaces expose Ml sites, withfour of the six octahedral bonds and four of the eightcubic bonds directed toward the crystal. The {100} sur-face has both Ml and M2 sites simultaneously exposed,with three of the six octahedral bonds and four of theeight cubic bonds directed toward the crystal. The (100)surface has four times the number of Ml and M2 siteswith advantageous configurations (Dowty, 1976) than the{Tll} surface. Hollister (1970), Hollister and Gancarz(197 l), Kouchi et al. (1983), Paquette and Reeder (1990a,1990b), and Reeder (1992) noted that the surface config-uration perpendicular to the growth direction was not theonly important structural variable in cation substitution.They suggested that different growth mechanisms may beimportant in exposing different site configurations. Forexample, in pyroxene the dislocation step to {100} hasboth tetrahedral and Ml sites simultaneously exposed onthe growth surface, whereas Ml and M2 are simulta-neously exposed on the {100} surface (Hollister, 1970;Hollister and Gancarz, l97l). Kouchi et al. (1983) illus-trated that the style of the growth mechanism may beinfluenced by the degree ofsupercooling.


Radius (pm)Site energy (kcaumol)Potential (V)ldeal charge

Radius (pm)Site energy (kcal/mol)Potential (V)ldeal charge

TABLE 1. Characteristics of M1 and M2 sites in clinopyroxene

Diopside" Hedenbergite- Acmite"

Ml site67.7 73.1

-1284 1232-27.86 -26.7

2.40 2.30

ll2 site109 .8 111 .1

-957 -960-20.75 -20.80

1.79 1 .79

343

62.s-2424

-35.033.02

1 1 1 . 9-308

-13.41 . 1 6

'Cameron et al. (1973), Smyth and Bish (1988)... Clark et al. (1969), Smyth and Bish (1988).

ANar,vrrc.u, RESULTS

Major- and minor-element microprobe analyses of thepyroxene grain analyzed are summarized in Table 2. Cat-ion formulas including Fe2+ and Fe3* were calculated as-suming four cations and six O atoms in the formula (e.g.,Robinson, 1980). The pyroxene analyzed for this studyhas limited pyroxene quadrilateral components. The onlyquadrilateral component observed is CaFe'?+SirOu, whichranges from 2 to 260/0. Larsen (1981) observed that thiscomponent reached 0o/o in other sector-zoned aegirinecrystals from the same intrusion where no Fe2* remainedafter recalculation. The nonquadrilateral components in-volve Na substitution in the M2 site and Al, Ti, and Fe3*substitution in the Ml site. The dominant other com-ponents in these pyroxenes are acmite and jadeite. Theacmite component ranges from approximately 7 5 Io 90o/oand reflects the coupled substitution Ca'?+ (M2) * Fe2+(Ml) - Na'* (M2) + Fe3+ (Ml). The jadeite componentranges from 2 to l0o/o and reflects the coupled substitu-tion ca2+ (M2) + Fe'+ (M1) * Na'+ (M2) + Al3+ (Ml).Tio* (<2o/o) is most likely incorporated by the followingcoupled substitutions: (Fe,Al)3* (Ml) - 0.5Fe2* (Ml) +0.5 (Zr,Ti)a* (Ml) (Larsen, l98l; Ranlov and Dymek,l99l) and Ca'z* (M2) * 0.5 Fe2+ (Ml) * Na'+ (M2) +0.5Ti4* (Ml). Other potential coupled substitutions in-volving Ti and Al are unlikely because aegirine has a verylow content of I4lAl (Deer et al., 1978; Nielsen, 1979;Larsen, 1981; Czamanske and Atkin, 1985; Ranlsv andDymek, l99l).

Major-element concentric zoning in individual sectorsis dominated by the variability in acmite and hedenberg-ite components. The sector and concentric zoning illus-trated in Figure ld depicts Na/(Na * Ca) variation in theaegirine crystal. Within the individual sectors the sodiccomponents increase from core [Na/(Na + Ca) : 0.78]to rim [Na/(Na + Ca) : 0.97].

As observed by Larsen (1981), the individual sectorsshow specific individual enrichment schemes (Table 3).The hedenbergite component (CaFe'z*SirOu) is enrichedin the basal sector [001] relative to the prism sectors.Larsen (198 l) concluded that the order ofCa enrichmentin the prism sectors is {l l0} > {100} > {010}. Obvi-

344

TABLE 2. Selected analyses of pyroxene grain in Fig. 1

SHEARER AND LARSEN: SECTOR.ZONED AEGIRINE

{100} [001] t l 1 0 )Analysis

no 341 7

sio,Tio,ZrO"Al2o3Fe.O"FeOMnOMgoCaONaro

Total

50.450.630.520.96

28.O73.380 .190.023.44

11.4199.07

1.9800.0202.000o.o240.0180.8280.1 110.0070.0000.9880.1440.8681.O12

35.813.22.90.21 .21 . 15.8

41206.51 . 54.08.5

3825

51.620.790.151.07

31.410.190.220.001.00

13.0399.29

1.9920.0082.0000.0340.0230.9060.0130.0070.0000.9830.0410.9741 . 0 1 5

d - I

1 . 40.30.020.20.78.1

64229.123.3c . c

1 131

50.080.371 . 1 80.66

24.905.210.330.025.70

10.0998.54

51.611.300.401.04

27.932.980.240.002.26

12.29100.05

1.9910.0092.0000.0390.0380.8100.0960.0080.0000.9880.0920.9171.009

15.44.20.90.080.50.96.6

1 00567.21 . 55.18.2

2922

52.00o.520.171.72

30.130.810.400.011 . 1 3

12.9799.86

1.9930.0072.0000.0710.0150.8680.0260.0110.0000.9910.0460.9631.009

8.01 . 90.40.020.52.3

12.63804

7.21 . 73.2

25.8't288

51.031 . 1 60.411.22

26.423.740.240.003.56

11.5799.35

1.9860.0142.000o.o420.034o.7740.1220.0080.0000.9800.1 480.8721.020

35.5't6.7

3.2o.21 .51 .89.7

80008.42.44.2

11.23061

51.07o.421 . 1 30.75

25.006.240.380.045.28

10.64100.95

2.0060.0002.0060.0350.0130.7210.2000.0120.0030.9840.2200.8151.035

Trace elements (SIMS) (ppm)

Formula based on tour cations and six O atomsSiAIT,.,AITiFe3*Fe,*MnMgM1,. ,

NAM2,.,

CeNdSmEuDyErYbTi

CrSr

Zr

2.OO20.0002.OO20.0210.0110.748o.1740.0130.0000.9780.245o.7771.022

37.415.33.70.32.02.3

1 1 . 12500

5.91 . 64.8

19.28763

81,324.09.31 . 08.26.1

12.32906

6.31 .57

28.18332

Note: ZrO2: Zr trom the ion probe and is not included in calculation of formula. Feros and FeO are calculated from Fe,., and pyroxene stoichiometry.Points are located in Fig. 1c.

ously, the enrichment in the sodic components is the re-verse. Larsen (198 l) observed that the sector enrichmentschemes for Na and Fe3+ were slightly different, with Nahaving {010} > {100} > {l l0} > [001] and Fe3* having{100} > {l l0} ? {010} > [001]. This difference is at-tributed to the jadeite component enrichment in the {010}sector.

Selected trace-element data for the pyroxene are pre-sented in Table 2. The pyroxene shows high but variableabundances of LREE (Ce 5.7-82.8 ppm), HREE (4.5-26.1 ppm), Y (4.1-89.9 ppm),Zr (700-10680 ppm), andTi (2898-11893 ppm). Trace elements such as Sr (2.5-8.3 ppm), Cr (1.4-2.4 ppm), and Y (4.2-11.6 ppm) areIower in abundance and show much less variability. Onthe basis ofour previous analysis ofpyroxene site accom-modation for trace elements, it is anticipated that theREE, Y, and Sr would be accommodated in the M2 siteof pyroxene, whereas Al, Zr, V, Cr, and Ti would bebetter accommodated in the Ml site. (Shearer et al., 1989).To understand better the relationships between relative

trace-element enrichments of individual sectors and ma-jor-element components, we calculated correlation coef-ficients among trace and major elements (Table 4).

The chondrite normalized patterns for the REE are en-riched in both LREE (Ce) and HREE (Yb) and depletedin MREE (Sm, Eu, DV) Gig. 2). All have negative Euanomalies. This pattern is unusual compared with quad-rilateral pyroxenes (Shearer et al., 1989, l99l; McKay,1989). Although we anticipated that MREE-oxide inter-ferences would result in minor false enrichments in theHREE pattern, the U-shaped REE pattern appears to bereal. This pattern is typical of Na- and Fe-enriched py-roxenes that crystallized from a peralkaline magma (Lar-sen, 1979; Mahood and Stimac, 1990) (Fig. 2a). We in-terpreted the U-shaped pattern of the aegirine asrepresenting the interplay of pyroxene partition coeffi-cients (KIREE > KdTREE) and the LREE-enriched characterwith a negative Eu anomaly of the Ilimaussaq peralkalinemagma. A typical whole rock pattern for these rock typesis illustrated in Figure 2d.


Tleu 3. Sector enrichments and relative growth rates for the llimaussaq aegirine and augite

345

Relative enrichment in aegirine Relative enrichment in augite

AITiFe3*Fe2*tr6

MnCaNaLREEHREEZr

5rGrowth rate

0} ? {010)0) > {0101

> [001]> [001]> [001]? { 100 )> {010 }: t 010 )

{ 0 1 0 } > t 1 1 0 } > { 1 0 0 }{ 1 1 0 } > { 0 1 0 } > { 1 0 0 }{100} >[001] >[ 0 0 1 ] :[001] '[001] >{010} -[001] >>[001] >[001] >>[001] >>[001] :[ 0 0 1 ] :[001] >relative growth rate in aegirine[001 ] > { 010 } > { 110 } > { 100 )

1 0 0 1 > 1 1 1 0 t0 ) > { 100 )0 l > { 1 0 0 } > { 0 1 0 }

{ 1 0 0 } > { 0 1 0 } > ( 1 1 0 } > { 1 1 1 } .{ 1 0 0 } > { 0 1 0 } > { 1 1 0 } > { 1 1 1 } -

{ 1 0 0 ) > { 1 1 1 } "{ 1 0 0 ) - { 1 1 1 } ' -{ 1 0 0 } > t l 1 1 } - -{ 1 0 0 } > { 1 1 1 } - .{ 100 } > {T11 } t{ 100 } > (T11 } t{ 1 0 0 } > { 1 1 1 } - -{ 100 } > t I 11 } t{ 1 0 0 } > { 1 1 1 } . '{ 1 0 0 } > { 1 1 1 } . 't 1 0 0 ) > { 1 1 1 ) - -relative growth rate in augite{T11 } > { 100 } . 'i T l 1 ) > { 100 } , { 110 } , { 010 }+

0 0 ) > { 1 1 0 } > [ 0 0 1 ]0 0 ) > { 1 1 0 }00 ) > { 110 }00 ) > { 110 )00 ) > { 110 }0 l : {100}0 } : { 1 0 0 } : { 0 1 0 }0 l : {100}

. Hollister and Gancarz (1971).' . Shimizu (1981).t Predicted based on the observations made by Shimizu (1981).+ Leung (1974).

LREE and MREE are highly correlated with the cationsubstitutions in the M2 site (Table 4). Ce, Nd, Sm, Eu,and Dy have a high positive correlation with Ca and ahigh negative correlation with Na and Na/(Na + Ca).That is, an increase in the sodic components is correlatedwith a decrease in REE. The HREE such as Er and Ybappear not to be strongly correlated with major-elementcomponents in the M2 site. In addition, the REE aresubstantially enriched in the basal sector relative to theprism sectors (Tables 2 and 3 and Figs. 2 and 3). Thiscontrasts with the observations of Shimizu ( 198 I ) for sec-tor-zoned augite (Table 3). In the prism sectors of theaegirine, the REE are slightly more enriched in {100}re lat iveto { l l0} .

Based on observations made on calcium clinopyroxenefrom basalts and eclogites, the REE concentration in cli-nopyroxene is dependent upon the nature of the REEsubstitution mechanism, the characteristics ofthe M2 site(i.e., optimal cation size and charge for the site), the con-centration of REE in the magma, and the relative enrich-ment schemes in sector-zoned pyroxenes (McKay, 1989;Shearer et al., 1989; Caporuscio and Smyth, 1990). Therelative importance of each of these variables in the ob-served correlation between REE and major M2 site com-ponents can be evaluated. Assuming charge balance, REEmay be substituted into the M2 site of pyroxene throughtwo possible coupled substitutions: 2Ca (M2) * Na (M2)+ REE (M2) and Ca (M2) + Si (T) * REE (M2) + Al(T). If this mechanism solely controlled REE behavior,Ca and REE would show a negative correlation. Unlikefor quadrilateral pyroxenes, in which Ca substitution intothe M2 site props open the site and thereby increases theability for REE to substitute into that site (Shearer et al.,1989; McKay, 1989), it is anticipated that substitutionsof the type Na * Ca will have a limited effect on the size

of the M2 site. The similarity of mean M2 site size forhedenbergite (Cameron et al., 1973) and acmite (Clark etal., 1969) illustrate this point (Table l). Therefore, thisminor change in the size of the M2 site has a limitedefect on the site's ability to accommodate cations suchas the REE and does not account for the observed REE-Na-Ca correlations. Similar correlations among REE, Na,and Ca were obseryed by Caporuscio and Smyth (1990)in mantle eclogite pyroxenes along the diopside-jadeitesolid solution. They attributed this correlation to dra-matic differences in optimal charge characteristics be-tween the M2 sites in diopside and jadeite (approximately22o/o).The differences in the calculated optimal charge forthe M2 sites in hedenbergite (1.79) and acmite (1.16) istherefore a dominant variable influencing the trace ele-ment substitution into the M2 site. Although the coupledsubstitution involving Na in M2 and smaller trivalentcations in Ml does not afect the size characteristics ofthe M2 site, it does affect the size characteristics of theMl site (Cameron and Papike,l98l; Table I of this pa-per). That has implications for trace-element substitu-tions into that site.

An additional variable that may be of importance isthe magma composition. Crystallization of exotic REE-bearing phases (Larsen, 1977, 1979) would result in de-creasing REE and Ca and increasing Na during pyroxenegrowth.

Zra+ ls most easily accommodated in the Ml site ofclinopyroxene. The sector enrichment documented byLarsen (1981) and this study is [001] >> {100} > {010}> {l l0}. The sequence of sector enrichment for Zr isopposite to that of Ti. The enrichment of the basal sectorcompared with the prism sector contrasts with the obser-vations made in augite (Shimizu, 198 l). Zr shows a pos-itive correlation to Ca in the M2 and a negative correla-

346 SHEARER AND LARSEN: SECTOR-ZONED AEGIRINE

Ce -0.681Nd 0.580Sm -0.645Eu -0.686Dy -0.619Er -0.121Yb 0.024Ti 1.000v 0 .158Cr -0.238Sr -0.486Y -0.355Zr -0.733

-0.793 -0 8920.684 -0 786

-0 743 -0.8450.808 -0 895

-0 750 -0.850-0 220 -0.358

0 017 0 .1030.466 0 4370.134 0 120

-0 416 0.484-0 469 0.502

0.375 -0.486-0.736 -0.767

TABLE 4. Trace-element and maior cation correlation coefiicientsfor the aegirine

Ti (M1) Al (M1) Ca (M2) Na (M2) Na/(Na + Ca)

[0011 {100 }At

0.1

1000

{110}

Averagesodslite toyaite(Larcen,1979)

Ce tu fb Ce A.r Yt

Fig.2. Chondrite-normalized REE patterns for various sec-tors in the aegirine crystal shown in Fig. l. (a) The [001] sectorand a pattern ofhedenbergite separate (dashed line) from foyaite(sample 91943: Larsen, 1979). (b) The {100} sector. (c) The {1l0}sector. (d) A whole-rock REE pattern for an undersaturated rocktype from the Ilimaussaq alkaline intrusion (Larsen, 1979).

DrscussroN

Role of diffusional vs. surface kinetic processes

When crystals grow in a magma, there are at least fourprocesses that must be considered in a complete descrip-tion of trace-element substitutions into structural sites.These are reactions occurring at the crystal-melt inter-face, the transport of species to that interface, the remov-al of latent heat generated during crystallization at theinterface, and the bulk flow of magma. The relative im-portance and interplay among these processes may influ-ence the relative enrichments observed in sector-zonedpyroxenes. For silicate materials, the interface processesand diffusion are usually considered the most important(Kirkpatrick, 1974,19'75,1981). As pointed out by Kirk-patrick (198 l), the removal of heat from the interface isof little significance because thermal diffusivities (10 '-

l0-3 cm'zls) are several orders of magnitude larger thanthe diffusion coemcients (> 10 u cm'zls). Although mostcalculations of composite gradients adjacent to crystalgrowth surfaces have assumed a static system, bulk flowmay be important in natural magmatic systems. Shear,resulting from density or flow driven by surface tension,may disrupt compositional gradients established at acrystal-melt interface by diffusional processes (Anderson,1984; I(rkpatrick, l98l). It has been treated sparingly intheoretical calculations (Shaw, 1974; Lasaga, 1981;Loomis, 198 l; Anderson, 1984). In addition, bulk flow

-0.469-0.438-0.507-0.537-0.508-0.049

0.1 070.7910.071

-0.272-0 .513-0.236

0.677

0.8800.779n aeE

0.877

0.3540.1040.424

- 0 1 1 80.4950.5040 4960.752

o

ttE

e10obtroxI 1 .0CL

o

!C

8 1 0()bCo

E r.oCL

o

ttc

-e 100o

Joo

.c 10oE'

1.0

oEttc

e10obco

E r.oCL

0.1

100100

rbErCeYTCe0.1

tion to Na in the M2. Zr also shows a negative correlationwith both Ti and Al, with which it competes for the Mlsite. Again, with the assumption that charge balancing isrequired for Zr substitution into the M I site, the possiblesubstitutions are 0.5R2+ (Ml) + Si (T) - 0.5Zra+ (Ml)+ Al (T) and2Ca (M2) + R'z+ (Ml) * 2Na (M2) + Zr"*(Ml). The former is less important because it requiressubstantial t+tAl in the aegirine.

Sr, Cr, and V are all in very low concentrations in theaegirine. The basal sector shows a slight enrichment in Srrelative to the prism sectors (Table 3). This again con-trasts with trace-element observations of Shimizu (1981).Cr and V appear not to be enriched in any particularsector, but that may be a function oftheir low concentra-tions and poorer analytical precision at these concentra-tions.

In summary, the sector enrichment for major-, minor-,and trace-elements for aegirine is different from that ob-served in other pyroxenes. In contrast to calcic pyrox-enes, the study by Larsen (1981) and the present study(Table 3) indicate the following characteristics of sodicpyroxenes: (l) REE andZr arc enriched in the basal sector[001] rather than the prism sectors. (2) The REE and Zrsector enrichments generally appear to correspond withgrowth rate. The fastest growth sector is more enrichedthan the slowest growth sector. Sectors with intermediategrowth rates tend to show more overlap. (3) REE and Zrdecrease with increasing Na/(Na + Ca) from the interiorof a sector to the perimeter of a sector. (4) As in otherclinopyroxenes, Al and Ti are enriched in prism sectorsrelative to the basal sector. However, their sector enrich-ment among prism sectors in aegirine differs from thatobserved in other pyroxenes (Bence et al., 1970; Hollisterand Gancarz, l97l; Thompson, 1972 Leung, 1974;Har-kins and Hollister, 1977). Although several prism sectorenrichment schemes in augite have been documented forAl and Ti, the { 100} sector has always been observed tohave higher contents ofAl and Ti than the {010} sector.Larsen (1981) and this study indicate that the sector en-richment in aegirine for Al is {0 l0} >> { I l0} > { 100} >[001] andforTi is { l l0} > {010} > {100} = [001] (Table3 and Fig. 3).


oo

o o$$ao .o

Nd(Na+C.)

may not be an important process in trace-element parti-tioning in these pyroxenes. The aegirine crystals consid-ered here crystallized interstitially, within a frameworkof minerals in which the flow of melt would be very lim-ited. We therefore consider only the role of diffusion andsurface kinetics in our discussions.

The possible results of the interplay between diffusionand growth rate at the crystal-melt interface has beensummarized in numerous studies (Burton et al., 1953;Albarede and Bottinga, 1974;Kroger,l973; Dowty, 1976).When the rate of uptake of a major component is muchfaster than diffusion in the melt, the rate of growth iscontrolled by the rate of diffusion. When diffusion of themajor component to the interface is fast relative to therate ofincorporation, the rate ofgrowth is controlled bythe interface attachment processes. In this case, trace-element incorporation is influenced by variables such asadsorption, ion exchange, growth mechanisms, surfaceconfiguration, interface roughness, and temperature (2,AI). Can we differentiate between the importance of dif-fusion and interface kinetics in sector enrichments? If wechoose the latter as the most important, can we identifythe dominant variables?

Development of sector zoning in pyroxenes is a meta-stable phenomenon (Dowty, 1976; Shimizu, 198 1). Yetthat does not necessarily imply whether interface kineticsor diffusion is the dominant process in sector formationor relative sector trace-element enrichments. Generally,in large magma bodies such as represented by the Ili-maussaq alkaline intrusion, crystallization proceeds atsmall undercoolings. Under these conditions, the rate-controlling growth process is the interface reactions(Kirkpatrick, 1974; Cahn, 1967). Where the ratio ofgrowth rate to diffusion coefficient is high, such as in arapidly cooled lava flow, diffusion is expected to be theimportant process. Sector-zoned pyroxenes occur in boththermal environments (Dowty, 1976). Sector-zoned ae-girine, however, has only been documented in undersat-urated, peralkaline plutonic rocks (I-arsen, l98l) and inhydrothermal veins and vugs associated with peralkalineintrusions (Ranlov and Dymek, l99l).

Morphological differences may be attributed to the op-erative process. Short-range diffusion results in the de-velopment of a cellular morphology (Elbaum, 1959),which is reflected in the crystal form (spherulites, den-dritic and skeletal crystals) and in patch-zoned regionsthat are rich in glass inclusions. Other features that mayreflect short-range diffusion processes include mineral in-clusions (i.e., apatite) that may reflect supersaturation atthe boundary layer (e.g., Bacon, 1989) and irregular zon-ing, which may reflect the extension of surfaces beyonda boundary layer into a strongly supersaturated region(Anderson, 1984). The influence of surface kinetics oncrystal growth is suggested by textural features such astabular crystals with faceted morphology (Kirkpatrick,1975), individual compositional zones that are thicker oncurvy or irrational surfaces (Hartman, 1973; Kirkpatrick,1975; Anderson, 1984), and the fact that the thickness of

0.75 0.s 06 0.F 0.s 1.o

Na4Na+Ca)

0.75 0.& 0.6 0s 0.5 t.m

Na4Na+Ca)

Fig. 3. Sector variability in major and trace elements. (A) Cevs. Na/(Na + Ca). (B) Zr vs. Na/(Na + Ca). (C) AI,O, vs. Na/(Na + Ca). (D) TiO, vs. Na/(Na + Ca).

individual compositional zones varies on different e5owthsurfaces (Dowty, 1976). A faceted morphology is evi-dence ofa stable interface, whereas the nature ofzoningthickness is dependent upon surfbce roughness, the rateof interface attachment, and surface morphology. Themorphology of the aegirine indicates surface kinetics wasa much more important growth process.

Experimental evidence (Lofgren et al., 1974) also in-dicates that sector zoning characteristics, as observed inthe aegirine, form as a result ofslower cooling rates (1.2'Clh) compared with cellular textural features (430-2.5"C/h). Plots of growth rate vs. undercooling (growth ratevs. A7", log growth rate vs. I/TAT) for experimental sec-tor-zoning studies by Kouchi et al. (1983) and surfacemicrotopographic studies of these experimentally growncrystals indicate that surface kinetic reactions and surfacenucleation mechanisms were important in sector growth(Kirkpatrick, 1975;Kouchi et al., 1983).

Observations concerning relative sector enrichmentsalso shed light on the relative importance of diffirsion vs.surface kinetics on trace-element behavior. The consis-tent lack of correlation between growth rate and incom-patible element enrichment of sectors in clinopyroxene(compare augite and aegirine) also indicates that surfacekinetics plays a more important role than diffusion insector enrichments.

Surface kinetics

The interplay among the numerous variables involvedin the reaction of the crystal-melt interface is probablyimportant in dictating differences in enrichment amongsectors of calcic clinopyroxene and between the same sec-tors in calcic clinopyroxene and aegirine. These variablesinclude magma composition, temperature, the degree ofsupercooling, the attachment kinetics of SiO., adsorption

Eo;

-

E Og6 o

too

m

O* t.o

0 5

zo

1.5

9 r oF

0 5

o 10011r{1m}o{1101

o a

'$e- t d o

Co

o

o

o"

tuNd(N!+Ca)

A. Il al|a dtbor


ctg

I

E r.o(,olt

coE

.C

.9cuJ

il S. c| C.r&Sm Ytt

o 2 4 6 8 1 0 0 1 2

zlt ' zlr "

Fig. 4. Relative enrichment of elements in the {100} sectorrelative to the basal sector ({ I I I } or [00 I ]) as a function of theCoulomb interaction (ratio of the ionic charge of the elementdivided by the ionic radius squared). (A) Ml site cations. (B)M2 site cations. Index to data: I : Carpenter (1980), 2 and 3 :this study, 4 : Kouchi et al. (1983), 5 : Hollister and Gancarz(1971), and 6 : Shimizu (1981). The experimentally grown py-roxenes from Kouchi et al. (1983) are for supercooling of l3 and45.

and desorption kinetics ofcations into surface sites, dif-ferences of surface atomic confi gurations, characteristicsofindividual sites (Ml vs. M2 sites, geometric flexibility),mechanisms of interface-controlled growth (continuous,surface nucleation, screw dislocation), and the nature ofthe ion exchange reactions occurring at the surface.

Shimizu (1981) concluded that in cases of rapidSiO;'zattachment, trace-element substitution into the Mlsite in augite is based on both the configuration of thegrowth surfaces (the number of available Ml sites) andthe ability of the cation to be retained in the structuralsite. The retention of a cation in a structural site is afunction of the cation bond strength. The importance ofthe latter was based on the enrichment of the { 100} prismsector relative to the { I I I } basal sector in augite. Ifsurfaceconfiguration were the only variable (four times as manyMl sites are available on the {100} growth surface as onthe { I I I } growh surface) the enrichment factor betweenthe {100} sector and { I I 1} sector should consistently beequal to 4. However, Shimizu observed enrichment fac-tors of l-2 arrd a linear relationship between the enrich-ment factor and Coulomb interaction between cation andthe M I site (Z/ 12 , wherc Z is ionic charge and r is ionicradius). This suggests that although the { 100} growth sur-face had four times as many available Ml sites as did thebasal { I I I } growth surface, the { 100} surface had a lowerdegree oftrace-element adsorption or a higher degree oftrace-element desorption (two to four times) than that ofthe {1 l l } sur face.

Figure 4 shows the relationship between preferentialenrichments of elements in the {100} prism sector rela-tive to the basal sector and the strength of Coulomb in-

teraction for cations that occupy the Ml site (Fig. 4A)and those that occupy the M2 sites (Fig. 4B). Data arefrom the aegirine of this study, augite (Shimizu, l98l;Hollister and Gancarz, l97l), sodic augite (Carpenter,1980), and laboratory-grown calcic clinopyroxene (Ko-uchi er al., 1983). The study by Shimizu (1981) and thisstudy are the only ones that report trace-element data.

The behavior of cations that occupy the Ml site gen-erally shows an increase in the preferential enrichment ofthe prism sector relative to the basal sector with strongerCoulomb interaction. This is best illustrated with the datafrom Shimizu (1981). This behavior appears to confirmconclusions reached by Shimizu (1981) and Dowty (1976)that those cations that are most firmly bound to the siteon the growth surface and therefore have a lower degreeof desorption are those that have a high charge and asmall ionic radius. Major- and minor-element data fromHollister and Gancan (197 l), Carpenter ( I 980), and Ko-uchi et al. (1983) show a similar relationship. However,unlike the data from Shimizu (1981), the data in the otherstudies show relationships that are not strictly linear, par-ticularly those elements with the lowest Z/r'z (Fig. 4).Comparisons of the {100} prism sector with the basalsector in the aegirine ofthis study shows even a furtherdeviation from the observations made by Shimizu (1981)in augite. In particular, the {100} prism sector is depletedin Zr relative to the basal sector. These initial observa-tions imply that coulombic interactions are not the onlyfactor controlling sector enrichments.

If trace-element substitution were strictly a function ofCoulomb interaction, a positive correlation similar to thatobserved by Shimizu for the Ml site cations (enrichmentvs. Z/r2) also would be observed for the M2 site cations.Observations from aegirine (this study) and augite indi-cate a preferential depletion in the {100} sector relativeto the basal sector with increasing Coulomb interaction(Fig. aB). The initial interpretation would be that thosecations most firmly held in the M2 site are those with theleast Coulomb interaction (Fig. aB). This completely con-tradicts what we would expect and strongly suggests othersurface kinetic processes are involved. These processesmust account for (l) the lower degree of retention of par-ticular trace elements in the { I 00 } prism sector comparedwith the basal sector in aegirine, (2) the negative corre-lation between Coulomb interaction and relative sectorenrichments of M2 site cations, and (3) the irregular en-richment behavior of Zr in the aegirine relative to augite.

Superimposed on factors such as the availability of sitesand Coulomb interaction are the characteristics of a sitethat dictate its ability to accommodate a cation and dif-ferences in growth rate and growth mechanisms amongsectors and surface reactions (i.e., coupled substitutions).Figure 5 illustrates the relationship among prism-basalsector enrichments, ionic radii, and cation valence. Thisdiagram is similar to an Onuma diagram (McKay, 1989;Onuma et al., 1968) and reflects shifts in the optimumionic radius a potential site can accommodate. In addi-tion, the differences in cation site characteristics between


augite and aegirine appear to dictate the relative enrich-ments. The efect of the shift in the optimal radius isparticularly noticeable in the Ml site. In aegirine, wherethe Ml site is dominated by Fe3* (mean site radius :62.5 pm) rather than Fer* (hedenbergite mean site radius: 73.1 pm), large Ml cations such as Zr are further offsetfrom the optimum ionic radius for the Ml site. This ac-counts for the deviation of Zr in aegirine from linearityrelative to augite, as observed by Shimizu (1981). It alsosuggests that the differences between hedenbergite andacmite in the ideal charge characteristic of the Ml site(Table l) are less important. On the basis of only the idealcharge, the higher ideal charge of the Ml site in acmite(3.02) compared with hedenbergite (2.30) suggests Zrshould be better accommodated in acmite.

The relative enrichments of all cations in the M2 site(Figs. 4 and 5) can be attributed to both optimal radiusand ideal charge. Nearly identical M2 site radii for hed-enbergite and acmite (Table l) result in similar slopes inenrichment factors, whereas differences in optimal charge(hedenbergite : 1.79, acmite : l 16) result in a higherenrichment factor for highly charged cations in augite.The negative correlation between Na/(Na * Ca) and manyincompatible elements should not be interpreted as in-dicating that charge-balance reactions such as 2Ca (M2)* Na (M2) + REE (M2) are not imporrant. Rarher, rhiscorrelation indicates the importance of the acmite sub-stitution for the optimum radius of the Ml site and theideal charge of the M2 site. In general, as a cation deviatesfrom the optimal ionic radius and ideal charge of a site,it becomes increasingly incompatible.

In the pyroxenes from this study and other studies, theobservation is that the sequence of relative sector enrich-ments does not strictly correlate to growth rate. The rel-ative enrichments should be related to the number ofrelevant sites available on the growth surface and the me-chanics ofadsorption and desorption on that growth sur-face. As illustrated in this study, what is important to thelatter is the actual nature of the sites, cation deviationfrom the optimal site size and ideal charge, and growthrate. The effect ofthis interrelationship on sector enrich-ment is illustrated in Figure 6.

In addition to relative growth rate, the growth mech-anism should also be important in the incorporation oftrace cations into a growth surface. In silicates such aspyroxene, lateral growth (surface-nucleation and screwdislocation mechanisms) rather than continuous growthappears to be the most common. Growth rate vs. super-cooling relationships and the surface microtopography oflaboratory-grown sector-zoned pyroxenes (Kouchi et al.,1983) suggest that both surface nucleation and screw dis-location mechanisms may be functioning. As illustratedby Hollister and Gancarz (1971), growth surface config-uration may be dependent on growth mechanism. As thedegree of supercooling will affect both crystal morphologyand growth mechanism, the degree of supercooling mayaffect the configuration of the growth surface. This willhave implications for the relative sector enrichments

0.6 0.8 1.0 1.2lonic Radius

Fig. 5. Relative enrichment of elements in the {100} sectorrelative to the basal sector as a function ofthe ionic radius. Solidlines are augite data from Shimizu (1981). Dashed lines are ae-girine data from this study.

(Kouchi et al., 1983). For example, in experiment withlarge degrees of supercooling (A?" : 45), clinopyroxenecrystallized with the prism sector actually was depletedin Al and Ti relative to the basal sector. Other studies,by Reeder (1992) and Paquette and Reeder (1990a,1990b), also illustrated the importance of growth mech-anisms in trace-element behavior. Although supercoolingmay play an important role in producing relative sectorenrichments in some cases, it is not likely to have beenthe controlling factor for the aegirine in this study becausethere is no evidence that a large degree of supercoolingoccurred in the llimaussaq magma.

In summary, the relative sector enrichments in pyrox-ene seem to correlate better with ionic radius (Fig. 5) thanZ/ r'1 (Fig. 4). Three observations, in particular, illustratethis point: (l) the depletion of Zr (Ml site, high Z/r2) irtthe {100} sector relative to the [001] sector in aegirine,(2) the systematic enrichment of M2 site cations in { 100}relative to [001] with decreasing Z/r2 in both sodic andcalcic pyroxenes, (3) the lower enrichment factor gener-ally observable in aegirine 'trc. 4rrgite in most M2 sitecations. These observations may be interpreted in twofundamentally different ways.

l. Crystal chemical or geometric factors dominate for

Ml Sitefs3+ ilg fs2+

M2 SiteCa Na

2

oo

oo

o l(tCUlrc,oE-co

EI

0

O{ augite1' -O aegirine

oNa

""'11e

, j5t""YbtT sm

3+

Sc^ t Io * r l

t ll l

L r+v r

tr . z t

9II a.

c.

b.

d.

e.


oo

oo

o616I!

CoEE(,

cuI

Ml SiteFeo. llg Fez.

M1 SiteFe3+ ilg Fe2.

dlr*doo d Il dlr

+

'P

0.6 0.8 1.0lonic Radius

0.6 0.8 1.0lonic Radius

Fig. 6. Diagrammatic summary illustrating the influence ofvarious factors on the relative enrichment of elements in the{100} sector relative to the basal sector of clinopyroxene as afunction of ionic radius. (A) Influence of the acmite substitution(Fe3+ + Na+ - Fe2+ + Ca,*) on relative sector enrichments.(B) Illustration of the additive effect of a variety of factors onrelative sector enrichments. (a) Pattern if only the number of M Isites available on the growth surface were important. (b) Patternif factor a and Ml site characteristics (optimum radius) wereimportant. (c) Pattern if factors a, b, and Coulomb interaction(Z/r2)were important. (d) In addition to factors a, b, and c, thispattern reflects differences between sectors in rate of growth,SiO;'zattachment, and cation adsorption and desorption. (e) Sameas d but with a greater difference in growth rate between sectors.(f) Same as e but with a compositional shift influencing the Mlsite characteristics (i.e., a shift ofthe optimum radius).

all elements in both aegirine and augite. In this interpre-tation, the hypothesis proposed by Nakamura (1973),Dowty (1976), and Shimizu (1981) is not valid, and thecorrelation between enrichment factors and Z/12 for Mlsite cations evident in the data presented by Shimizu(1981) is fortuitous. In addition, the M2 site cations fromthis augite show the opposite relationship. In this case,site factors dominate for all elements in the aegirine. Theacmite component (NaFeSirOu) in each sector controlsthe abundance of the other elements through the influ-ence of optimal radius for the Ml site and optimal chargefor the M2 site. Although optimal radius for the M2 sitedictates the slope of enrichment factors in Figure 5, thedifferences in the M2 site cation enrichments betweenaugite and aegirine is mostly a function of the differencein ideal charge characteristics ofthe site. Each sector mayhave different partition coemcients because of differentsite size and ideal charge provided by different Na andFe3+ contents. The difference in Na and Fe3+ content be-

tween sectors might be influenced by differences in elas-ticity of partially formed M1 and M2 sites in distinctcrystallographic directions.

2. Site characteristics dominate for most minor andtrace elements in aegirine because differences in structur-al components [Na/(Na + Ca) and Fe3+/(Fe2+ + Fe3* )]among sectors make for large differences in site factors(optimal size and ideal charge). However, in contrast tol, adsorption and desorption kinetics still controls Na/(Na + Ca) and Fe3+/(Fe2+ + Fe3+ ) sector enrichments inaegirine and thus is ultimately responsible for producingthe trace-and minor-element sector zoning. Moreover,because augite has smaller differences in major structuralcomponents between sectors (compare characteristics ofMl and M2 sites between diopside and hedenbergite inTable l), geometric factors might be less important inaugite, and thus adsorption and desorption kinetics mightbe the dominant mechanism producing minor- and trace-element differences among sectors. In that case, the factthat a plot of sector enrichments in augite vs. ionic radiusresembles an Onuma diagram is largely coincidence or isto be expected for any element lying along the Z/r2 cor-relation line. It is, therefore, without genetic significance.

AcrNowlrocMENTS

The authors wish to thank the Institute for the Study of Mineral De-posits, the Institute ofMeteoritics, and the Geological Survey ofGreen-land for their support in this study. The analytical results on GGU ma-terial are published with the permission of the Geological Survey ofGreenland. We also appreciate Nobu Shimizu's guidance during SIMSanalysis at WHOI and his insight concerning the dynamics of surfacekinetics on trace-element behavior We also acknowledge the assistanceof Mike Spilde in performing electron microprobe analyses on these sam-ples. As always, J.J. Papike provided a reservoir of information concem-ing the crystal chemistry of pyroxene and unlimited enthusiasm. Hugh E.O'Brien and Gordon McKay provided insightful reviews that improvedthe overall quality of the manuscript. This research was supported byNSF grant RII-89-02066 (to C.K.S.).

RprunrNcns crrnoAlbarede, F., and Bottinga,Y. (1972\ Kinetic disequilibrium in trace el-

ement partitioning between phenocrysts and host lava. Geochimica etCosmochimica Act^. 26. | 4l- | 56.

Anderson, A.T. (l 984) Probable relations between plagioclase zoning andmagma dynamics, Fuego Volcano, Guatemala. American Mineralogist,69,660-676.

Bacon, C.R. (1989) Crystailization ofaccessory phases in magmas by localsaturation adjacent to phenocrysts. Geochimica et Cosmochimica Acta,53 ,1055 -1066 .

Bence, A.E., and Albee, A.L. (1968) Empirical correction factors for theelectron microanalysis of silicates and oxides. Journal of Geology, 76,382-403.

Bence, A.E., and Papike, J.J. (1971) A martini-glass clinopyroxene fromthe Moon. Earth and Planetary Science ktters, 10,245-251.

-(1972) Pyroxenes as recorders of lunar basalt petrogenesis: Chem-ical trends due to crystalJiquid interactions. Proceedings oflunar andPlanetary Science, 3, 431-469.

Bence, A.E., Cameron, K., and Papike, J.J. (1970) Sector zoning in calcicclinopyroxenes (abs.). Eos, 5 l, 435.

Burton, J.A., Prim, R.C., and Slichter, W.P. (1953) The distribution ofsolute in crystals grown from the melt. Journal of Chemistry and Phys-ics, 21, 1987-1996.

Cahn, J.W. (1967) On the morphological stability of growing crystals. InH.S. Peiser, Ed., Crystal gowth, 856 p. Pergamon, Oxford, England.


C-ameron, M., and Papike, J J. (1981) Structural and chemical variationsin pyroxene American Mineralogist, 66, l-50.

Cameron, M., Sueno, S., Prewitt, C.T., and Papike, J.J. (1973) High-temperature crystal chemistry of acmite, diopside, hedenbergite, jade-ite, spodumene, and ureyite. Arnerican Mineralogist, 58, 594-619.

Caporuscio, F.A., and Smyth, J.R. (1990) Trace element crystal chemisrryofmantle eclogites. Contributions to Mineralogy and Petrology, 105,550-56 l .

Carpenter, M.A. (1980) Composition and cation order variations in asector-zoned blueschist plroxene. American Mineralogist, 65, 313-320.

Clark, J.R., Appleman, D.E., and Papike, J.J. (1969) Crystal-chemicalcharacterization ofclinopyroxene based on eight new structure refine-ments. Mineralogical Society of America Special Paper, 2, 3l-50.

Czamanske, G K., and Atkin, S.A. (1985) Metasomatism, ritanian acmite,and alkali amphiboles in lithic-wacke inclusions within the Coyote Peakdiatreme, Humboldt County, California. American Mineralogist, 70,499-5r6.

Deer, W.A., Howie, R.A., and Zussman, J. (1978) Rock-forming miner-als, vol. 2A, Single-chain silicates, 668 p. Wiley, New York.

Downes, M.J. (1974) Sector and oscillatory zoning in calcic augites fromMt. Etna, Sicily. Contributions to Mineralogy and Petrology, 47,187-196.

Dowty, E. (1976) Crystal structure and crystal growth II Sector zoningin minerafs. American Mineralogist, 61, 460-469.

-(1977) The imponance of adsorption in igneous partirioning oftrace elements. Geochimica et Cosmochimica Acta. 41. 1643-1646.

Elbaum, C. (1959) Substructures in crystals grown from the melt. In B.Chalmers, and R. King, Eds., Progress in metal physics, vol. 8,438 p.Pergamon, London.

Ferguson, A.K. (1973) On hour-glass sector zoning in clinopyroxene.Mineraf Magazine, 38, 321-325.

Ferguson, J. (1964) Geology of the Ilimaussaq alkaline intrusion, SouthGreenland. Description of map and structure. Meddr Grsland, 172, l-82.

-(1970) The significance of the kakortokire in the evolution of theIlimaussaq Intrusion, South Greenland. Meddr Groland, 186, 193.

Harkins, E., and Hollister, L.S. (1977) Sector zoning of clinopyroxenefrom a weakly metamorphosed diabase. American Mineralogist, 62,39U394.

Hartman, P. (1973) Structure and morphology. In P. Hartman, Ed., Crys-tal gowth: An introduction, p.367-402. North-Holland, Amsterdam.

Hollister, L.S. (1970) Origin, mechanism, and consequences of compo-sitional sector-zoning in staurolite. American Mineralogist, 55, 742-766.

Hollister, L.S., and Gancarz, A J. (1971) Compositional sector-zoning inclinoplroxene from the Narce area, Italy. American Mineralogist, 56,9s9-979.

Irving, A.J., and Frey, F.A. (1984) Trace element abundances in mega-crysts and their host basalts: Constraints on partition coefrcients andmegacryst genesis. Geochimica et Cosmochimica Acta, 48, 12Ol-l22l.

Kirkpatrick, R.J. (1974) The kinetics of crystal growth in the systemCaMgSi,Ou-CaAl,SiOu. American Journal of Science, 273, 215-242.

-(1975) Crystal growth from the melt: A review. American Min-eralogist, 60, 798-814.

-(1981) Kinetics of crystallization of igneous rocks. In Mineralog-ical Society of America Reviews in Mineralogy, 8, 398 p.

Kouchi, A., Suqawara, Y., Kashima, K., and Sunaqawa, I. (1983) Iabo-ratory gowth ofsector zoned clinopyroxenes in the system CaMgSirO6-CaTiAl,O". Contributions to Mineralogy and petrology, 8], 177-184.

Krdger, F.A. (1973) The chemistry of imperfect crystals (2nd edition),vol. l, 313 p. North-Holland, Amsterdam.

Larsen, L.M. (1976) Clinopyroxene and coexisting mafic minerals fromthe alkaline Ilimaussaq intrusion, South Greenland. Journal ofpetrol-ogy, 17,258-290.

-(1977) Aenigmatites from the Ilimaussaq intrusion, South Green-land: Chemistry and petrological implications. Lithos, 10, 257-210.

- (l 979) Distribution ofREE and other trace elements between phe-nocrysts and peralkaline undersaturated magmas, exemplified by rocksfrom the Gardar igneous province, South Greenland. Lithos, 12, 303-3 1 5 .

-(1981) Sector zoned aegirine from the Ilimaussaq alkaline intru-sion, South Greenland. Contributions to Mineralogy and Petrology, 76,285-291.

Lasaga, A.C. (1981) Implications of a concentration-dependent gowthrate on the boundary layer crystal-melt model. Earth and PlanetaryScience lrtters, 56, 429-434 -

kung, J.S. (1974) Sector-zoned titanaugites: Morphology, crystal chem-istry, and gowth. American Mineralogist, 59,127-138.

Iofgren, G.E., Donaldson, C.H., Williams, R.J., Mullins, O., and Ussel-man, T.M. (1974) Experimentally reproduced textures and mineralchemistry ofApollo 15 quartz normative basalts Proceedings ofthe5th Lunar Science Conference, 549-567.

Loomis, T.P. (1981) An investigation ofdisequilibriurn growth processesof plagioclase in the system anorthite-albite-water by methods of nu-merical simulation. Contributions to Mineralogy and Petrology, 81,219-229.

Mahood, G.A., and Stimac, J.A. (1990) Trace-element partitioning inpantellerites and trachytes. Geochimica et Cosmochimica Acta, 54,2257-2276.

McKay, G. (1989) Partitioning of rare earth elements between major sil-icate minerals and basaltic melts. In Mineralogical Society of AmericaReviews in Mineralogy, 21, 45-77.

Nakamura, Y. (1973) Origin of sector-zoning ofigneous clinopyroxenes.American Mineralogist, 58, 986-990.

Nakamura, Y., and Coombs, D.S. (1973) Clinopyroxenes in the Tawhi-roko tholeiitic dolerite at Moeraki, northeastern Otago, New Znaland.Contributions to Mineralogy and Petrology, 42,213-228.

Nielsen, T.F.D. (1979) The occurrence and formation of Ti aegirines inperalkaline syenites. Contributions to Mineralogy and Petrology, 69,236-244.

Onuma, N., Higuchi, H., Wakita, H., and Nagasawa, H. (1968) Traceelement partition between two pyroxenes and the host lava. Earth andPlanetary Science ktters, 5,47-51.

Paquette, J., and Reeder, R.J. (1990a) New types ofcompositional zoningin calcite: Insights into crystal-gowth mechanisms. Geology, 18,1244-1247

-(1990b) Surface symmetry constraints on compositional zoningand cation ordering during crystal growth. Geological Society ofAmer-ica Abstracts with Programs, 22, 4216.

Ranlsv, J., and Dymek, R.F. (1991) Compositional zoning in hydrother-mal aegirine from fenites in the Proterozoic Gardar Province, SouthGreenland. European Journal ofMineralogy, 3, 837-853.

Reeder, R. (1992) Intrasectoral zoning in dolomite: Further evidence forsurface structural control on element incorporation during crystal growth.V.M. Goldschmidt Conference Program and Abstracts, A88.

Robinson, P (1980) The composition space ofterrestrial pyroxenes-rn-ternal and external limits. In Mineralogical Society of America Reviewsin Mineralogy, 7, 419-494.

Shaw, H.R. (1974) Difftrsion of HrO in granitic liquids: Part l, Experi-mental data; Part II, Mass transfer in magma chambers. In A.W. Hof-mann, B.J. Giletti, H.S. Yoder, Jr., and R.A. Yund, Eds., Geochemicaltransport and kinetics. Carnegie Institution of Washington, Publication634,353 p.

Shearer, C.K., Papike, J.J., Simon, S.B., Galbreath, K.G., and Shimizu,N. (1989) An ion microprobe study of the intra-crystalline behawior ofREE and selected trace elements in pyroxene from mare basalts withdifferent cooling and crystallization histories. Geochimica et Cosmo-chimica Acta. 53. l04l-1054.

Shearer, C.K., Papike, J.J., and Spilde, M.N. (1991) Pyroxene/melt traceelement behavior: A study of pyroxenes from the Valley of Ten Thou-sand Smokes, Alaska. Geophysical Research Iftters, 18, 1557-1560.

Shimizu, N. (198 l) Trace element incorporation into growing a"gite phe-nocryst. Nature, 289, 57 5-577.

Shimizu, N., and Hart, S.R- (198 l) Isotope fractionation in secondary ionmass spectrometry. Journal ofApplied Physiology, 53, 1301-I31 l.

Sm1th, J.R. (1992) Crystal structure controls on element partitioning ingeologic systems. V.M. Goldschmidt Conference Program and Ab-stracts, A-105.

Smlth, J.R., and Bish, D.L. (1988) Crystal structures and cation sites inthe rock-forming minerals, 386 p. Allen Unwin, London.


Strong, D.F. (1969) Formation ofthe hour-glass structure in augite. Min-eralogy, 37, 472-479.

Thompson, R.N. (1972) Oscillatory and sector zoning in augite from aVesuvian lava. Carnegie Institution of Washinglon Year Book, 7 | , 463-470.

Ussing, N.V. (1912) Geology ofthe country around Julianehaab, Green-land. Meddr Gronland, 38,376 p.

MrNuscrom rucErvED Ocronen 21,1992M,llr-rscnryr ACCESTED Novpnsen 15. 1993

ERRATUM

Plagioclase-melt equilibria in hydrous systems, by T. B. Housh and J. F. Luhr (v. 76, p. 477-492). Some of the plagioclase compositions shown in Figure 2 were incorrectly plotted. A cor-rected version of the figure is given below.

1400

1300

1200

1 100

rb1 000

900

800

700

600

40 50 60 70

An (mol%)80 100

thls studv Rutherford Helz

P(kbar ) i 1 2 2 .5 4 1323

s y m b o l : O a A L l I

Fig. 2. Albite-anorthite binary diagram showing the l-barliquidus-solidus loop of Bowen (1913) and the 5-kbar HrO-sarurated loop ofYoder et al. (1957). Also shown is the collapse ofthe loop when projected from diopside saturation in the systemdiopside-albite-anorthite (Wyllie, I 963; Morse, I 980). Symbolsshow plagioclase and projected glass compositions from thisstudy, Helz (1976), and Rutherlord et al. (1985). Multicompo-nent glass compositions are plotted as l0O[normative anl(an *ab)]. Half-solid symbols indicate the eight samples whose pla-gioclase or glass compositions deviate from the general trendsand are deleted from the modeling. Values near open squaresindicate experimental pressures in kilobars for data from Ruth-erford et al. (1985). Sol id and dash-dotted l ines labeled 1,2,2.5,4, and 5 kbar indicate mean loci of projected isobaric melt andplagioclase compositions, respectively.

90

sector-zoned aegirine from the ilimaussaq alkaline intrusion, south

Documents