Chemical Geology 119 ( 1995) 79-99

.v Ott” Dcparfrmw qf Eurth Sciences, The Open University, Milton Keynes. MK7 64.4, UK

Received I6 October 1993; revision accepted I 7 June 1994

New major-, trace-element and Sr, NC! and Pb isotope data are presented for car onatites and pyroxenites from -old Jacupiranga complex in southern Erazi?. XcasGicd * ‘%r ranges from 0.7047 to 0.7055, ld3Nd/ 0.5 I25 i to 0.5 ! 264 and 206Pb/204Pb from 17.0 to 18.2. st samples have high measured Th/U ( 5-

17 ), and high 208Pb/204Pb ratios ( 37.9-40.2) relative to 206Pb/204Pb, with time-integrated T /U > 4. The pyrox- enites have higher initial 206Pb/204Pb ( 17.63- 17.67) and lower initial 87Sr/86Sr ( w 0.7047) than the carbonatites ( 17.05-l 7.47 and 0.7049-0.7054, respectively). There is a broad negative Pb-Sr isotope array between the car- bonatites and the pyroxenites which precludes simple binary mixing, because of their very different Sr/Pb ratios. The isotope differences also aypcar LU predudc ;,:ccEels in \!*h;c!, ahe carbonatitcs segregated by liquid immiscibility from magmas similar to those from which the pyroxenites crystallised, and models in which the isotope arrays are the product of crusta! contamination processes. Rather, the initial Sr, Nd and Pb isotope ratios in the Jacupiranga complex, and the negative correlation of Pb and Sr isotopes are similar to those in the Paran high-Ti basalts, and the oceanic basalts of the Walvis Ridge and Tristan da Cunha. Thus, the initial isotope ratios of the Jacupiranga rocks are considered to have been inherited from the mantle source regions associated with incipient magmatism of the Tristan da Cunha hotspot and the opening of the South Atlantic. Finally, the observed variations are used to evaluate suggestions that certain element ratios in upper-mantle rocks, such as high Ca/A! and La/Yb and low Ti/Eu, are features of carbonatite metasomatism. Mantle xepoliths altered by carbonaiite metasomatism exhibit elevated Sr contents and low Rb/Sr ratios, and most cclrbonatitzs have high U/B, low Rb/Sr and SmlNd. Thus, infiltration and/or metasomatism by carbonatitic melts is one process which may have been responsible for the negative array of U /Pb and Kb/Sr inferred for the source of certain oceanic basalts.

at carbonatites are more likely to been generated by partial melting of the mantle

Carbonatites have very low SiOz, but high in- compatible trace-element abundances and so they are markedly different from most other ig- neous rocks. Previous studies have demon-

than by limestone syntexis ( ell et al., 1982; Bell and Blenkinso 1957). Many carbsnatites oc- cur in contine al areas and so they are useful

bes of the sub-continental upper mantle. wever, on the basis of si ilar ra~~~~~~i~ kb=

-__ [MB1

“Present address: Department of Geology, LJniversity Col- lege, Belfield, Dublin, Ireland.

tope co&positions, it has been argued t bonatites were derived from an ocean island ba- salt (OIB) -type source (Basu and Tatsumoto,

CiO09-2541/95 j$O9.50 0 1995 EIsevier Science B.V. All rights reserved SSDf 0@09-2541(94)00093-N

I’. Huang et al. / Citernicai’ Geology 119 (1995) 79-99

baa

Atlantic Ocean

Mainly Precambrian rocks

m Paleozoic sedimentary rocks

17 Cretaceous basaltic and late sedimentary rocks

___I .^_.

Peridotites a Fenite nepheline Pyroxenites syenites

Carbonatites El Ijolites

Fig. 1. Simplified geological sketch maps of southern Brazil and the Jacupiranga complex, modified after Melcher ( 1966) and t Jlbrich and Gomes ( 198 I )

1980; Bell et al., 1982; Bell and Blenkinsop, 1987;

Nelson et al., 1988). The generation of carbonatites is still debated,

but they are important to any discussion of magma generation and element fractionation processes in the upper mantle. Many carbona- tites are associated with alkaline rocks and the relationship between them is relevant to under- standing carbonatite petrogenesis. Expel imental studies on carbonate and silicate systems show that carbonate melt may be segregated by liquid immiscibility from a carbonate-enriched silicate melt at certain P-T conditions (Kjarsgaad and Hamilton, 1989). Alternatively, carbonatites may be generated by small degrees of partial melting in the upper mantle (Nelson et al., 1988; Gittins, 1989). However, the primary carbona- tites are expected to have high Mg# [Mg/ (Mg+ Fe’+ ) ] and relatively low Ca# [Ca/ (Ca+Mg) ] (Eggler, 1989; Dalton and Wood, 1993), ar.d these characteristics are rare in car-

bonatites worldwide. alton and Wood ( 1993) suggested that calcio-carbonatites may be formed by interaction of primary magnesian carbonatite melts with the upper-mantle lithosphere.

The Jacupiranga carbonatite complex was in- truded into a late Precambrian mica schist and syntectonic granodiorite belt N 130 Ma ago (Fig. 1) (Amaral, 1978; Roden et al., 1985 ) and, to- gether with several nearby alkaline complexes, it comprises the Jacupiranga alkaline province (Ulbrich and Gomes, 198 1) . It occurs in an oval- shaped alkalic complex (Fig. 1, - 65 km2 ), which includes pyroxenites, peridotites, ijolites and fenite-nepheliue syenites. Two separate intru- sive stages of silicate rocks, and up to five dis- tinct carbonatite intrusions, have been identi- fied (Melcher, 1966; Gaspar and Wyllie, 1983 ) . To the west is the present outcrop of the Paran& Etendeka continental flood basalts (CFB) which erupted contemporaneously 138-l 28 Ma ago (Hawkesworth et al., 1992; Renne et al., 1992;

den et al. ( 1985 )

The Jacupiranga carbonatites an silicate rocks were collected from recently blasted outcrops to avoid weathe rock samples were crush satellite ball mills to mini and their petrography an SC ed in the Appendix.

‘or elements were measured using a A Fisons-8420 dual goniometer wavelength-dis- persive ( WD ) X-ray fluorescent trometer on fused glass discs at versity. Carbonatite samples for w’ere prepared at 50% SiOz dilution, using the normal silicate calibration. SiOz was adjusted relative to BCS368 and BCS393 carbonate stan- dards, and the data corrected forth trace elements Rb, Sr, Y, Zr, Nb, Co, Ni, Cu, Zn and Ga were determined by XRF on pressed powder pellets. Abundances of the rare-earth elements (REE) and remaining trace elements in Table 1 were obtained by instrumen- tal neutron activation analysis (INAA) , follow- ing the procedures described by Potts et al. (1985).

The nine whole-rock samples and selected mineral separates were analysed for Sr, Nd and

ave been correcte

3.1. or and trace elements

ce-element and WEE analyses le 1. Three groups can

or elements: the py- U8- I ) are character-

ised by high SiOz and Al,O,; the carbonatites (HBUO5, HBOO9-2, HBOlO and HBQII) have low SiOz ( < lo%), MgO, I&O and A@, but high CaO (35-50%); an samples (HBO04, SiOz -25%, high 2-4.5%, respectively). These extreme variations

ent compositions are reflected in the mineralogy described in the Appe._?ix. The third group of samples consisting of olivine, cal- cite and microcrystalline phlogopite, are here termed ol-phl-carbonatites. Disequilibrium tex- tures between the mineral phases were observed

82

Table 1

Y. Huang ef al. / Chemical Geology Ii 9 (199s) 79-99

Major- and trace-element abundances in the whole-rock samples from the Jacupiranga carbonatite complex

Sample HBOOJ HBU04 HBQ0.5 HBO08- I HBOOS-2 HBOO9-1 HBOQ9-2 HBOIO HBOI 1

SiOz (wt%) TiOz A1203

Fe@3

MnO MSG CaO Na,O J&G PIGS iOi S

Total

Bb (ppm) Sr Y Zr Hf Nb Ta Ba Pb Th U SC V Cr co Ni cu Zn Ga La Ce Nd Sm Eu l-b Yb LU

43.29 22.58 3.704 2.27 6.07 3.3

15.58 10.98 0.128 0.2

11.24 18.04 19.33 18.88 0.38 0.4i 0.02 3.98 0.009 1.12 0.59 17 0.06 0.54

-- - 100.34 99.3

2.3 0.66 1.16

33.64 0.25 4.34

30.78 0.18 0.26 9.04

15.96 0.26

98.83

n.d. 100 9 282 2,172 2,749

10 18 34 182 62 1,678

6.67 2.98 22.6 6 294 78 0.73 7.15 24.6

45 466 734 16 9 38 0.57 13.3 4

n.d. 0.74 0.253 52 20.9 56.2

287 131 287 65 50 11 59 45 41

150 160 n.d. 454 279 34

58 91 202 17 12 7 13.2 45 98.3 33.6 93.4 213 23.7 46.8 118 4.57 8.53 19.6 1.45 2.73 5.88 0.44 0.92 1.75 0.65 1 1.29 0.08 0.16 0.18

40 3.924 6.65

16.83 0.1 39

11.32 19.96 0.42 0.03 0.017 1.09 0

-__ 100.38

23.24 21.0 0.24 1.23 4.1 2.76

10.9 8.82 0.2 0.22

14.84 24.02 20.06 19.62

1.16 0.28 4.3 2.1 0.86 1.97

17.82 17.32 0.04 0.02

42.38 0.05 0 4.48 0.014 4.18

47.34 0.01 0.04 0.87

36.36 2.00

-- 97.85

4.3 1.32 0.86

33.0 0.31 7.3

28.56 0.2 0.36 9.7%

13.4 0.4

6.08 0.14 0.12

43.08 0.14 6.98

4528 0.08 0.06 3.07

34.36

99.92 99.4 99.43 99.79

1 75 69 4 10 6 332 2,123 2,404 5,848 2,772 5,130

10 24 18 38 33 35 208 109 20 n.d. 1,710 287

6.91 4.9 1.62 0.23 29.2 6.28 5 100 159 5 97 68 0.79 1.45 16.4 0.25 26.5 16.8

38 822 1,412 629 491 525 9 8 5 19 34 14 0.32 P 7.7 5.25 3 6.08 34.4

n.d. 2.58 1 n.d. 1.86 4.61 53.6 79.7 29.6 17 76.1 24.2

314 141 83 4 331 31 118 72 167 5 12 6 63 37 63 61 49 6

141 92 21 67 2 11 33 35 14 205 40 85 74 104 98 16 186 22 18 13 7 n.d. 9 n.d. 13.8 69.8 47.4 91.3 113 105 37.6 141 102 184 248 224 27.8 72.1 54.6 89.7 134 118

4.84 11.9 9.32 15.2 21.5 19.8 1.56 3.76 2.93 5.01 6.45 6.14 0.53 1.21 1 I .74 1.94 1.97 0.56 1.43 0.9 2.18 1.25 2.5 0.12 0.2 0.17 0.32 0.19 0.23

n.d. = not detected.

in thin sections of the ol-phl-carbonatites, in which olivines are corroded and surrounded by

The trace-element data are presented in Fig. 2,

phlogopite and calcite. Thus, the ol-phl-carbon- and they display distinctive patterns for the three

atlt :s are considered to be peridotites or dunites groups. The carbonatites have high Ba, Sr and

altered by fluid infiltration with growth of new REE contents, similar to other carbonatites

phlogopite and carmnate. worldwide (Nelson et al., 1988; Woolley et al., 199 1)) variable concentrations of ?hc high field

83

0.01’ , , , , , , , , , , , , , , , , [ , , , Pb RbBaTh K Ta Nb LaCe Sr P Nd H!’ it&n Ti Y YbLu SC

Pb RbBa Th K Tn NbLa Ce Sr P Nd Hf ZrSm Ti Y YbLu SC:

0.1

1

0 HE005 HB09-2

L HEW10

Pyroxenites

0 moo1

HBOOR-1

Ol-phi-wbonatites

o Ht3004

+a HBOOE-2 o HBQ09-1

0.01’ , , , , , , , , 1 , , , , , , , , , , #

Pb RbBaTh K TaNbLaCe Sr P NdHf ZrSm Ti Y YbLuSc

Fig. 2. Minor- and trace-element data for the Jacupiranga samples illustrated on primitive-mantle normalised diagrams. Two samples are from Nelson et al. ( 1988), and the others are from this study. Three groups are distinguished on the basis of their major-element compositions: ( 1) carbo\r&tes, which have negative Rb, K and Ti anomalies; (2) pyroxenites, which have gen- erally flat patterns with slight negative K and P anomalies; and (3) ol-phl-carbonatites, which have intermediate characteristics without the negative K, Rb and P anomalies.

( SiO2 +A1203 *2)/CaO 0.00 0.05 ‘1.10 0.15 ’ (

(SIC” ~1203 *2)/CaO

6000

20001. I i 0.00 0.05 0.10 0.15 0.20 0.25

(SiO2 +A1203 *;?.KhQ

Fig. 3. Diagrams illustrating the correlations between (%0,-t A120j*2 ) /CaO and selected minor and trace elements in the car- bonatites. The syuam arc data from this study and diwmnds are samples from Wclson et al. ( 1988). See text for discussion.

strength elements (HFSE), and relatively low Cr, Ni, Ti, Rb and K contents. The pyroxenites have flatter mantle-normalised patterns with negative anomalies at K, Rb and P. The ol-phl-carbona- tites exhibit features intermediate between those of the carbonatites and pyroxenites, but without the negative K and Rb anomalies that are pres- ent in both the carbonatites and pyroxenites. Mineral analyses indicate that apatite has the highest REE abundances (Nd > 400 ppm, Table 2). However, the observed variation in phos- phorus in the carbonatites, without much change in REE (Fig. 2), suggests that apatite does not control the whole-rock REE abundances.

phlogopite) can be used as a measure of the pro- portions of silicate and carbonate components in the carbonatites. Thus, the relationship between ( SiOz + Al&*2) /CaO and minor and trace ele- ments should reflect the distribution of these ele- ments between the silicates and the carbonates. Fig. 3 shows that ( SiOz +A1203*2) /CaO ratios correlate with selected trace elements in the car- bonatites; for example, there are positive corre- lations with K, SC, Zr, Ta and Ti, and negative correlations with Sr ( and &a, not shown).

3.2. Sr and Nd isotopes

Several trace elements, such as K, Rb and the Sr and Nd isotope data are presented in Table HFSE vary much more widely in the carbona- 2 and illustrated in Fig. 4. Present-day Sr isotope tites, than in the pyroxenites and the ol-phl-car- ratios range from 0.73467 to 0.7055 1, similar to bonatites (Fig. 2 ). The mineral assemblage in -% he those published previously by Roden et al. carbonatites is predominantly carbonate, with (1985) and Nelson et al. (1988). There is a variable amounts of silicate minerals (olivine broad correlation between Sr isotopes and Rb/ and phlogopite) and apatite. The ratio Sr within different sample groups in that, for ex- ( Si02 + A&O, * 2 ) /CaO (silicon is substituted by ample, the ol-phl-carbonatites have slightly aluminium in some mineral phases, such as higher R7Sr/86Sr (0.70538-0.7055 1 ), consistent

85

Sample

HBOQJ BOOI py

HBQOl HBQU4 cc HBOOS H wno.s CC HBOOS apt HBOfXI- I HBOO8 py HBOOB-2 HBOO9-I HBOOP-2 HBO09 cc HBCIO HBOlO cc HBOlO apt HBOlO phi HBOIl HBOdl cc HBOl I apt HBOI I phl

0.082 282 0.70467

100 2,172

9 2,i49

0.037 4,447 I 332

152 394 75 2.123 69 2.404

4 5.448

IO 2,772

0.016 3,473

6 5,130

0.04 1 3,832

0.70538 0.705 I3 0.70498 0.70494 0.70493 0.70469

0.70546 0.7055 1 0.70540 0.70534 0.70494 0.70489

0.7052 i 0.70521

3.992 19.72 0.5 I2638 0.70466 1.7 0.51253 1.2 4.998 22.89 0.512585 0.51247 0.0 8.648 45.11 0.70513 8.3

20.84 12.86 72.24

5.13 5.32

12.28

15.53 15.33 23.03 13.98 75.41

0.66 20.15

116.2 75.41

403.3 36.49 24.62 67.2 50.69 82.75

0.5

0.5 0.5 0.5

2636 0.70497 6 0.51254 1.4

2509 0.70467 1.9 0.51241 - 8.2 2492 0.51238 -4.2 2620 0.70527 10.3 Cf.51253 I.1

0.512584 0.70535 11.5 0.51248 0.2 0.5 12602 0.7054 12.1 0.51251 0.7

129.6 J7.72

4?.7.3 3.51

1P3.8

0.5 12626 0.70492 5.4 0.51253 I ’ .b 0.70489 4.9

78.96 431.4 1.52 7.06

0.5 12642 0.70520 9.4 0.51255 I.5 0.7052I 9.5

apt = apatite; cc = catcite; phl = phlogopite; py = pyroxene.

with their higher 0.08). ver, in general, the measured Sr iso- tope a /Sr ratios scatter on a 87Sr/86Sr vs. “Rb/*%r diagram (not shown ) an formation can be reasonably inferred.

Initial 87Sr/86Sr ratios have been caiculated at 130 Ma, and they vary from 0.70466 to 0.70540. Several explanations are possible for the range in initial Sr isotope ratios, including post-emplace- ment alteration of Rb/Sr and/or 8”Sr/86Sr. The replacement of phlogopites by chlorites in I-BOOS might be due to post-emplacement fluid infiltra- tion. Calcites have high Sr and little Rb, and there is no evidence that they have been altered by hy- drothermal fluid. Thus, the SK isotope ratios of the calcites may provide the best estimate of ini- tial “Sr/*%r, and in general they are indistin- guishable from the initial 87Sr/86Sr ratios of their whole rocks. The range of initial *5Sr/s6Sr in the carbonatites and ol-phl-carbonatites is from 0.70489 to 0.70540, whereas the Sr isotope ra- tios in the pyroxenites are significantly lower (0.70466 and 0.70467; Fig. 4 and Table 2 ). Be-

cause the pyroxeni

a more likely explanation is t

( 1983) pointed out that the Jacupiranga com- plex underwent multiple intrusion, and it is cer- tainly possible that different i

ferent initial Sr isotope ratios. carbonatites worldwide, those from Jacupiranga have higher initial 87Sr/86Sr ratios than those from most central intrusive carbonatites ( N 0.703 ), but similar *‘Sr/?5r to the V~Aoway vein carbonatite from Australia (Nelson et al., 1988 ). Thus, the Jacupiranga rot high “Sr/*%r portion of the ENd-Esr fieId for car- bonatites from dilfereut contiuental settings (Fig. 4).

86 Y. Huang et al. I Clwm’cal Geology 1 I9 (I 995) 79-99

15 I

10

&Nd

5

Fig. 4. a. tSr diagram. 1 he Jacupiranga complex rocks plot in the high-c,, portion of the carbonatite field (Nelson et al., 198;. , Beli and Blenkinsop. 1989), and they have initial Sr and Nd isotope ratios similar to :hose of the Walvis Ridge and Tristan da Cunha basal& The j2led circles are pyroxenites, squaws are carbonatites, and ftiarlgles are ol-phl-carbonatites from this study. The diamonds are the Jacupiranga carbonatite data from Roden et al. ( 1985) and Nelson et al. (1988). The data for oceanic basalts are from Richardson et al. ( 1982), Zindler and Hart ( 1986). Ito et al. ( 1987) and references therein; and the data from the Parand basalts are from Hawkesworth et al. ( 1986) and Peate et al. ( 1992 and references therein).

Sm and Nd analyses confirm that the 14’Sm/ 144Nd ratios are systematically different in the three sample groups (pyroxenites 0.122-o. 117, ol-phl-carbonatites 0.116-O. 111 and carbona- tites 0.112-O. 107 ), consistent with the REE data in Fig. 2. Present-day Nd isotope compositions in all three groups are very similar ( ‘43Nd/ ‘44Nd=0.51264-0.5 1251). Unlike Sr, the initial Nd isotope ratios show no systematic diflerences between pyroxenites, carbonatites and ol-phl- carbonatites. The c Nd,-values range from - 1.2 to + 1.5, which are very close to the bulk-Earth composition, and the two Nd analyses on silicate rocks bracket those from the carbonatites (Fig. 4).

3.3. Pb, Uand Th data

The measured Pb, U and Th contents for the whole-rock and mineral samples are also listed

in Table 3. Whole-rock Pb contents range from 2.3-9.0 ppm, ex’cept for the pyroxenite HBOOS-I which has 0.29 ppm Pb. However, another py- roxenite HBOOl has 6.4 ppm Pb, which is higher than most carbonatites. There are no systematic differences between the three groups. However, in general, the whole rocks have higher Pb con- tents than the separated major rock-forming minerals, such as calcite, apatite, phlogopite and magnetite. This suggests that there is another phase containing significant quantities of Pb. Sulphide is a candidate, and it was observed in HBOO9-2 which has highest Pb content (9 ppm).

U abundances in the pyroxenites are much lower than those in the carbonatites (except for HBO09-2) and the ol-phl-carbonatites. How- ever, like Pb, the U concentrations in calcite and apatite are lower than those in their whole rocks. The implication is that much of the U is also car- ried by accessory minerals. Pyrochlore and bad-

Table 3

87

Sample

HBOOI 6.407 0.055 i-l,446 0.53 8.17 I7.685 15,447 3t-i.068 17.674 IS.446 38.040 Bi@J Qy

BOO4 4.506 0.703 12.24 9.77 17.4 17.222 15.394 HB004 cc 17.230 15.421 37.752 HB005 3.520 2.178 3.489 38.86 1.60 18.256 15.446 38.17C ! 7.464 15.407 37.769 HB005 cc 1.244 0.102 0.338 5.07 3.31 17.358 15.405 37.849 17.255 15.400 37.741 HB005 apt 2.247 0.473 17.16 13.62 36.27 17.644 15.443 41.003 17.367 15.429 37.816 PIBOOS- I 0.293 0.018 0.226 3.77 12.79 17.702 15.447 38.241 17.626 15.443 37.931 HB008-2 3.492 2.217 17.59 40.6 1 7.93 17.963 15.421 39.854 17.135 15.381 37.175 HB009- I 3.613 0.385 5.438 6.59 14.12 17.194 15.379 38.187 17.060 15.372 37.586 HB009-2 9.02 1 0.007 0.088 0.05 12.46 17.049 15.380 37.610 i 7.033 15.380 37.606 FIB009 cc 4.702 0.0002 0.003 17.049 15.388 37.636 17.049 ? 5.387 MB010 2.327 1.138 5.924 30.83 5.21 17.954 15.430 38.791 17.326 15.399 37.755 HBGlO cc 1.881 0.02 I 0.68 17.294 15.431 37.859 17.280 15.430 HBOlO apt 2.128 0.439 12.74 HBOlO phl 0.506 0.110 13.37 17.402 15.427 38.543 17.129 15.414 HBOII 6.315 3.46 34.4 35.39 9.94 18.191 15.448 40.320 17.470 15.413 38.049 WBOI I cc 5.104 0.003 0.041 17.231 15.415 37.766 17.230 15.419 HBOI I apt 3.036 0.368 7.77 17.435 15.43 1 40.558 17.277 15.423 HBOl J Qhl 6.112 5.120 55.08 18.417 15.432 41.456 17.295 15.377

apt = apatite; cc = calcite; phi = QhiOgOpite; py = pyroxene.

deleyite have been re ranga carbonatite complex Hussak, 1892) and they m distribution.

Th concentrations are lo pyroxeaites, but they are rath bonatites (3-26.8 ppm), except for HBOO9-2, which has CO. 1 ppm Th. Apatite data from

that it has much hipher Th contents 206Pb than its whole rock. The sam-

ples which have low P2 low Th, such as HBOOI, HBOWJ a that apatite is an important Th carrier. the samples which have high Th do not always have high P205. The lack of a positive correla- tion between P205 and Th shows that some other mineral phase(s) also contains significant amounts of Th.

Measured ~1 (238U/204Fb) -values in six of the whole-rock samples range from 10 to 4 1 (Table 2 ), although. the pyroxenites HBM.@-I ) and one carbonatite (H low p( 3-0.5 ) . Among the analysed mineral sep-

arates, calcite has

s relatively low Th an

bonatites are prese

range from 17.05 to 18.26, 207Pb/204Pb= 15.38- 15.45, and ‘osPb/204Pb= 37.61-40.32. The mea- sured Pb isotopes from the pyroxenites similar 206Pb/ b to those in the carbo but slightly elevated 207Pb/204Pb. Initia tope ratios were calculated at 130 h/la, a the exception of HBO09, the initial Pb ratios of the calcites are lower than those of rocks (Table 3 ) . This may be due recent mo ion of U. pites from 0 and HB different initial Pb isotope ratios with lower 206Pb/204Pb and/or 207Pb/2c’Pb than the cal-

88 Y. Huanget al. /Cl~ett~icalGwlogy iI9(1995) 79-99

38.5

16.8 17.0 17.2 17.4 (206pb/204pb). I

17.6 17.8

Fig. 5. a. lnitial Pb isotope ratios for the whole rocks and mineral separates from the Jacupiranga carbonatite complex. The syrnhols are as in Fig. 4, and A = apatite, C= calcite, Ph = phlogopite. The bars illustrate the degree of fractionation on repeated measurements of NBS 981. The z”‘Pb/Z04Pb ratios reported by Nelson et al. ( 1988) are slightly higher than our data at similar 2osPb/2”“Pb and *06Pb/204Pb, but these cannot be simply explained by the small differences in the measured values of NBS 981.

cites (Table 3). More measurements are needed to confirm the scale of such heterogeneities, but because most of the calcites analysed have low U/Pb ratios, their initial ratios may be more re- liable than those from the whole rocks. Nonethe- less, the initial Pb isotope ratios are consistent with three groups of samples, in that the pyrox- enites have the highest ratios with *“jPb/ 204pb - - ! 7.!Zu 17.53, he oi-phlcarbonatites tend to have the lowest initial ratios and the carbona- tites have intermediate values, except for HBOU9- 2 which is similar to HBU09-I. These initial ra- tios define a trend on a 207Pb/z04Pb-206Pb/204Pb diagram (Fig. 5a) with the pyroxenites at the upper end. All the initial 208Pb/204Pb and 206Pb/

204Pb ratios plot above the two-stage evolution line of Stacey and Kramers ( 1975) (Fig. 5a) which, in a single-stage model, requires a time- integrated Th/U ratio of > 4. Relative to other carbonatites (Nelson et al., 1988) the Jacupi- ranga carbonatite has unradiogenic Pb, but it has similar 206Pb/zo4Pb and 208Pb/204Pb to the Par- ana high-Ti and the Walvis Ridge basalts (Fig. 5b).

etrogenesis

Most carbonatites are spatially and temporally associated with mafic or alkaline complexes, and

(b,)

Walvis lhlgc

Fig. 5b. The initial Pb isotope ratios of the Jacupiranga rocks compared with those from selected oceanic basal&, the Stacey and Kramers ( 1975 ) two-stage evolution curve (S & K), and the Geochron. Szull oyfn squurcs are present-day ratios for worldwide carbonatites from Andersen and Taylor ( 1988) and Nelson et al. (198X), although some of the more radiogcnic Pb isotope ratios fall outside the range of this diagram. By contrast, the present-day Pb isotope ratios for Jacupiranga (not shown) are relatively restricted (e.g., zo6Pb/204Pb= 17.05-18.26. Table 3). Other sy&ols and data sources as for Fig. 4.

thus the relationship between the silicate and carbonatite rocks is important in any explana- tion of the petrogenesis of carbonatites. The ini- tial isotope ratios in the Jacupiranga carbonatite complex are similar to those of basalts from the Walvis Ridge, Tristan da Cunha and the Parana. Helium data on a pyroxene from the pyroxenite exhibit a mantle signature ( asu et al., 1993), and it is inferred that the Jacupiranga complex has a mantle origin as invoked for other carbon- atites worldwide (Basu and Tatsumoto, 1980; Bell et al., 1982; Nelson et al., 1988). However,

in detail the new isotope data presented here demonstrate significant isotope variations wit the Jac~p~ra~ga complex. Although the initial ‘43Nd/144Nd ratios in the pyroxenites and car- bonatites are broadly similar, their initial Sr and

atios are systematically pyroxenites have lower

(0.7047) and higher initial Pb isotope ratios (z?%j’04Pbx 17.6), whereas in the carbona- tites, initial Sr and Pb isotope ratios vary from 0.7049 to 0.7054, and 17.05 to 17.46, respec- tively, Overall, there is negative correlation be-

90 Y. Httang et al. /Chemical Geology I1 9 (1995) 79-99

(a)

17.6 -

x E 17.4 - z 2 n. 17.2 -

3

- 17.0 -

L 16.8 r 0.7045 0.7050 0.7055

(87 Sr/ *6Sr)j 0.7060

(bl 17.8

t 17.6 t

f 17.4 -

.

ii 17.2 - w a . I..ClUSl

,7,0 (O.Sl16. lb&,

. I.. CIUEl

0.5123 0.5124 OS125 0.5126

(143Nd/ IaNd). I

Fig. 6. a. (*oaPb/204Pb), vs. (87Sr/RhSr)l. Curve I illustrates an AFC model (r= 20%) for a primary silicate magma with isotope ratios similar to the pyroxenites, contaminated with lower-crustal material. The Sr and Pb contents are 900 and 5 ppm in the primary silicate magma and 200 and 6 ppm in the lower-crustal component, respectively. The bulk partition coefftcients used were 0.6 for Sr and 0.9 for Pb. Curve 2 is an AFC model for a primary carbonatite melt with Sr and Pb isotope ratios similar to those of the pyroxenite, bu? with 3000 ppm Sr, contamined with the lower-crustal component. Curve-3 lines illustrate mixing between the pyroxenite and carbonatites. The symbols are as for Fig. 4. b. (206Pb/204Pb) vs. ( ‘43Nd/144Nd), illustrating the AFC model (r= 20%). Nd = 20 ppm in the lower-crustal material, and 60 ppm in the initial silicate magma, and the Pb isotopes and abundances are as for curve I in (a).

tween initial 206Pb/204Pb and 87Sr/8hSr both in the pyroxenites and carbonatites, and within the carbonatite group (Fig. 6). Initial Pb isotope ra- tios also exhibit a linear trend between the car- bonatites and pyroxenites (Fig. Sa). These iso- tope heterogeneities in the Jacupiranga carbonatite complex may provide useful con-

straints for models of car onatite generation. However, the first step is to evaluate the effects of crustal contamination.

4 I. Crustal contamination processes

Crustal contamination is responsible for the isrbtope variations in many continental mag- matic rocks (DePaolo, 198 I), and it was consid- ered by Roden et al. (1985) to explain the Sr heterogeneity in the Sacupiranga carbonatites. With more data this hypothesis can now be re- evaluated and a number of

( 1) Because the pyroxen ST contents than the carbonatites, they should be more easily contaminated. Although there are no Sr and Pb isotope data for the wall-rocks, such mica schists and granodiorites are likely to have had high 87Sr/86Sr. Yet the pyroxenites have the lowest initial *‘Sr/*‘%r in the complex, and they therefore appear to have been least affected by any crustal contamination processes.

(2) If the carbonatites were formed from a contaminated silicate magma by liquid immis- cibility, the observed within-suite variations in- dicate that they cannot have been formed by seg- regation from a single batch of contaminated magma. Moreover, crustal roc,ks which have high Rb/Sr also tend to have elevated U/Pb, as see in most estimates of the bulk or upper continen- tal crust (e.g., Taylor and McLennan, 1985). Thus, with time, they have both high 87Sr/86Sr and 206Pb/204Pb, and so such material cannot have been responsible for the observed negative Pb-Sr correlation in the pyroxenites and carbonatites.

Alternatively, crustal contamination might have involved a component from the lower con- tinental crust, since that is regarded as a major reservoir of unradiogenic Pb. Such unradiogenic Pb may be due to U depletion, together with other large-ion lithophile elements (LILE ), during granulite-facies metamorphism (Moorbath et al., i 969; Gray and Oversby, 1972; Weaver and Tar- ney, 198 1; Rudnick and Presper, 1990; Cohen et ai., 199 1 ), or to primary low ‘J/Pb ratios in areas of significant underplating with gabbroic cumu- lates (van Calsteren et al., 1986, 1993; EIuang et

oreover, eve

relatively little change

within the carbonatites cannot be due to simple mixing between pyro ‘tes and a carbonatitic end-member, because mixing line is strongly convex-upward. T whole-rock dat.a plot on

is may indicate that they were formed with some mixture of a pyroxenite- like component, or it may be due to inadequate age corrections of the initial Pb isotope ratios as discussed above.

(4) The available 0 and C isotope composi- tions of the Jacupiranga carbonatite show that they have mantle stable isotope ratios which have not been significantly affected by assimilation of crustal material (Nelson et al., 1988). The ini- tial Sr, Nd and Pb isotope ratios of the Jacupi- ranga samples are also similar to those observed in the oceanic basalts of the Walvis Ridge and Tristan da Cunha, and in the Paranrji high-Ti ba-

ecause at pressures o

that silicate rocks i

be evaluatecl fu

partition into the silicate, w they are preferentially partitro bonate melt (Hamilton et al., 1989 ).

Using an average for the FEE in t

92 Y. Huang et al. / Chemical Geology I1 9 (199.5) 79-99

ranga carbonatite and available partition coeffr- cient data, the REE pro& in an inferred paren- tal magma can be calculated (Fig. 7 j . Model d is the REE profile calculated using data from ilton et al. ( 1989), and model 2 is that obtained using the data of Wendlandt and Harrison ( 1979). The LREE contents in both the calcu- lated parent magmas and the carbonatite melts are similar, but similar HREE abundances are only obtained using the partition coefficients of Hamilton et al. ( 1989 j (model I, Fig. 7 j. Fig. 7 also includes data from syenite and carbonatite rocks from the Juqula complex situated w 50 km north of Jacupiranga (Fig. 1). This complex is approximately coeval ( 1 XL 145 Ma) with that at Jacupiranga and it has similar Sr isotope ra- tios (*‘Sr/*%r=0.7052-0.7056) (Beccaluva et al., 1992). It was argued by Beccaluva et al. ( 1992) that the Juquia carbonatites were de- rived by liquid immiscibility from the syenites, but this can only be reconciled with the REE data if the partition coefficients of Wendlandt and

0.1 1 I I I I I \ r LO CC Nd Sm Eu Yh Lu

Fig. 7. REE mantle-normalised diagram, illustrating poten- tial silicate parent magmas for the Jacupiranga carbonatite magmas, calculated on the basis of available experiment data for REE distribution between carbonatite and silicate melts. The pyroxenite magma (crosses) is that calculated to have been in equilibrium with the pyroxenites. Open sqltares rcp- resent the average of the Jacupiranga carbonatites, and open diamonds are the average of Juquia carbonatites; the shaded field is for the Juquia syenites. The REE profiles for silicate magma in equilibtium with the Jacupiranga carbonatiu b *cre calculated using the partition coefficient data from Wen- dlan It and Harrison ( 1979) (model 2, Jill& frianglp.r) and for 6 kbar from Hamilton et al. ( 1989) (model I, filled cir- */es). See text for discussion.

Harrison ( 1979) are use Hamilton et al. ( 1989 ). more experimental data are required on partition coefficients.

The Jacupiranga pyroxenites consist la clinopyroxene, as seen in the similar Nd contents in separated clinopyroxenes and the bulk-rock pyroxenites (Table 2 ). The REE pro- tile of a magma in equilibrium with the pyrox- enites can therefore be calcu!ated using REE par- tition coefficients for clinopyroxene Dunn, 1993) and that in eq~~~ibr~~ jacupiranga pyroxenites has relatively P igin LREE and high LREE/HREE ratios (Fig. 7). Thus, it is different from both the calculated magmas in equilibrium with the carbonatites, and from the Jacupiranga carbonatites them- selves (Fig. 7). Since the LREE are incompati- ble (D =ZK 1 j, these differences cannot easily be attributed to fractional crysvallisation processes. If the carbonatites were derived from the same magma from which the pyroxenites crystallised, but at a later stage than the pyroxenites, they should have steeper REE profiles and higher LREE contents than those observed. Although early experimental work icated that LREE might be preferentially re ved in CO2 vapour (e.g., Wendllandt and Harrison, 1979), Paterson ( 1993) argued that REE variations in carbona- tites from New Zealand conflict with these ex- perimental data. One explanrztion might be that carbonatite crystallised earlier than the pyrox- enites but that is not consistent with the field ob- servations (Caspar and Wyllie, 1983 ). Since the Jacupiranga pyroxenites and carbonatites cllso have different initial Sr and Pb isotope ratios, it is more likely that they crystallised from differ- ent parental magman,.

4.3. Partial melting

Sr and Nd isotope differences between car- bonatites and alkaline silicate rocks in some complexes have been cited as evidence that their associated silicate rocks may rat be directly ge- ne”lically related to the carbonatites (Nelson et al., !988; Bell and Peterson, 1991). Nelson et al. ( 1988) invoked very small degrees of melting

93

Ti K Rb

Ol 0.63 0.0005 0.0005 0.0003 0.0006 0.009 0.0 LO9 0.0005 0.000 I 0.06 0.000 1 0.000 1 opx 0.24 0.01 0.01 0.01 0.01 3.12 0.163 00067 0.007 0.3 0.000 1 0000 1 cpx 0.095 0.02 0.03 0.06 0.1 0.43 C.435 0.3055 0.067 0.5 0.00 I 0.0005 pbt 0.01 0.0 0.05 0.03 0.03 0.05 0.0471 0.2806 2 0.3 20 20 gar 0.005 0.0001 0.005 0.03 0.09 12 14 1.7 0.0006 1.7 0.000 I 0.000 1 spine1 0.02 0.03 0.03 0 035 0.04 0.1 0.09 0.2 0.0001 0.15 0.000 I 0.000 1

cpx=clinopyroxcne; garzgarnet; ol=olivine: opx=orthopyroxece; phi= phlogopite.

0.0, I lib IJa ti Id CL- Sr Nd 23 %I Ti l’h Lu

Fig. 8. Mantle-qormaliszd trace-clement diagram, iHustra:- ing the batch melting model for the generation of the carbon- atite magmas. The residual m,-ntle is assumed to be pcrido- title with I% phlngopite and 0.5% garnet. and with trace- element abundances similar to the spine1 pcridotitc xcnohths studied by McDonough ( 1990). PAal melting calculations were undertaken using both the avcragc (/illed .r~bols) and the mean (open s~mboh) values for the spine1 peridotites. The shadedjicld is the data for the Jacupiranga carbonatites (Nelson et al., 1988 and this study).

( < 1%) from eclogitic sources to generate pri- mary carbonatite melt with extreme LREE abundances. Experimental studies in mantle peridotite-C02-HZ0 systems have revealed a P- T window (Falloon and Green, 1989; Meen et al., 1989) in which partial melting of depleted lherzolites can generate primary carbonatite melts with low alkali contents (Dalton and Wood, 1993 ) . Our calculations show (Fig. 8 ) that the minor- and trace-element patterns of the Jacupiranga carbonatites can be reproduced by small degrees (0.5% ) of partial melting of man- tle peridotite in the presence of phlogopite ( 1%) and garnet (0.5%) using the partition coeffr-

ses on trace-element patterns are mot clear,

patterns of calcio-carbonatites and dolomite- carbonatites, even though the 1

to primary compositions than

4.4. Fractional crystallisation

Somt elements, including K, Ta, Nb, f, Zr and Ti, exhibit considerable variation in the Ja- cupiranga carbonatites, as seen in other carbon-

94 Y. Huang CI a/. / Chemical Geology 119 (I 99s) 79-99

atites worldwide. The partial melting calcula- tions outlined above indicate that these variations cannot be due to different degrees of partial melting, because the highi;y incompatible elements, such as Ba and La, vary less than Nb, Ti and Zr, which vary by over two orders of ma+ nitude (Fig. 2). The variations in elements, such as P, Nb (Ta) and Zr (Hf ), which are not con- centrated in minerals like olivine, phlogopite or magnetite, may rei”lect the presence of accessory mineral phases (Nelson et al., 1988). Apatite, pyrochlore and baddeleyite have been observed in the Jacupiranga carbonatites (Hussak, 1892; Melcher, 1966), and other phases which may cause significant variations in such elements in- clude perovskite, monazite and sphene (Nelson et al., 1988). The samples with high (SiOz+A1203* 2 )/CaO have higher Ti and Zr, and lower Sr and Ra (Fig. 3). Since the major observed mineral phases are phlogopite, olivine, magnetite and carbonate, the ratio of ( SiOz + A&O3 * 2 ) /@a0 represents the propor- tion of silicate and carbonate mineral phases. If it is inferred that the silicates crystallised before most of carbonates, the observed correlations between Ti, Zr, K, and Nb and ( SiOz + A1203 * 2 ) /CaO might be explained by accumulation of the silicates, with these ele- ments preferentially distributed into either sili- cate or associated accessory mineral phases. The negative correlation between Sr (as well as Ba, not shown ) and ( SiOz + A&O3 * 2 ) /CaO indi- cates that Sr was compatible into the carbonate phases, consistent with its known geochemical characteristics.

antle sources of the Jacupiranga complex

The Jacupiranga carbonatites have esn = O- 12 and eN& = - 1.2 to 1.5 (Fig. 4), and their Pb iso- tope ratios straddle the Geochron (Fig. 6a and b). Relative to other carbonatites worldwide (Fig. 43, the Jacupiranga carbonatite complex has higher Sr and lower Nd initial isotope ratios. Thus, their source regions were less depleted than those for many carbonatites and for Sr, Nd and

Pb isotopes they were si bulk Earth.

The low Pb, IOW ENd,

of the Jacupiranga car lar to those of the ?ara Walvis Ridge and Tristan salts. Such compositions are different MORB, and other OIB, such as Hawaii, St. ena and the Canaries (Figs. 4 and 5 ). they have been termed the Dupai anomal 1984) and while it is still debated whether thd Dupal anomaly is a deep-seated or a shallow phenomenon, it appears to e present in basalts from both oceanic (Wa!vis idge and Tristan da Cunha) and continental areas (Parana high-Ti ) (Hawkesworth et al., 1986). The Jacupiranga complex exhibits similar isotope ratios, consis- tent with the presence of a regional mantle iso- tope signature. Moreover, the negative correla- tion between Sr and Pb isotope compositions in the carbonatite complex is also similar to those observed in the basalts of the Parand (high-Ti ), the Walvis Ridge and Tristan da Cunha.

#/SF fractionation by ca~b~~atit~

The negative ‘06Pb/‘04Pb-s7Sr/s6Sr correla- tion in the Jacupiranga carbonatite complex is likely to have been inherited from their mantle source. Kwon et al. ( 1989) also documented a negative correlrrt,ion between initial Pb and Sr ratios in a variety of carbonatites of different ages, except for a group at 2.7 Ga, and they em- phasised that such trends are observed in certain oceanic basalts, such as those from the Walvis Ridge. If U is more incompatible than Pb during partial melting in the upper mantle (Sun and McDonough, 1989; McKenzie and O’Nions, 199 1 ), both Rb/Sr and U/Pb should be low in depleted mantle, high in enriched mantle and with time there should be positive correlations between Pb and Sr isotopes. The implication is that other processes or components were in- volved, and a number of authors have invoked the subduction of altered ocean crust to explain such isotopic variations in mantie-derived rocks

art agents (Green and gerty, 1989; Jones, 1989; Several geochemical featu Al and La/I% and low Ti have also been re-

1993). Although the compositions of primary carbon-

atite melts remain controversial, a striking fea- ture of carbonatites is that they tend to have Rb/Sr, and variable, but often high U/Pb. perimental studies (Dalton and Wood, 1993) have shown that primary melts generated by par- tial melting of depleted lherzolite can have low alkaline contents. Most of the carbonatites stud- ied here and by Nelson et al. ( 1988 ) have Sr contents in the range of 2000-6000 ppm and Rb/ Sr< 0.003. Thus, the empiacement of a few per cent of carbonatite melt wou change the Sr budget and significantly reduce /Sr ratio in the host mantle. Mantle xenoliths which are consid- ered to have undergone carbonatite metasoma- tism have low Rb/Sr, and positive correlations between Sr contents and (La/Yb)n or Ca/AP,

3

II calcite

i Carbonalites I

.01-i -- .O&ll

1 I 0 .oOl .Ol .l 1 10

87Rb/ *"Sr

Fig. 9. fl (238U/204Pb) vs. “Rb/‘%r illustrating the mea- sured ratios from carbonatites, melilitites and Group-II kim- ber ites (Fraser and Hawkesworth, 1992; Rogers et al., 1992 ). Most of the carbonatites have high p and low “Rb/‘“Sr. Some carbonatites with very high i! rnx~ ~dlert large amounts of fractional crystallisation. The Do/DPb=O. 1 line represents the maximum p gtueidtcd by SPiCli partial melting of primitive mantle material using the partition coefficients summariseo in McKenzie andO’Nions ( 1991 ). BSE=bulk silicate Earth.

96 Y. Huang cl al. / Chemical Geology 119 (I 995) 79- 99

( 1992 ) argued that a CO,-rich environment has an important role in keeping phlogopite stable during small degrees of melting in the generation of melilitites. Such small-degree silicate melts have high U contents and U/Pb, but with 10~ Rb/Sr because of the presence of residual phlo- gopite (Fig. 9). Thus, mantle metasomatism by either carbonatitic melt or a melilitite-like sili- cate melt will fractionate U/Pb and Rb/Sr, and with time will result in low Sr and high Pb iso- tope ratios, in addition to low Sm/Nd and hence low EN,,. At present, the effects of carbonatite emplacement cannot be distinguished from those of melilitite on the basis of radiogenic isotopes alone, primarily because 1.0~ Rb/Sr ratios can be due to the high Sr contents of carbonatites and/ or to low Rb retained in residual potassic phases.

7. Conclusions

( 1) The initial Sr, Nd and Pb isotope ratios in the Jacupiranga carbonatite complex are similar to those of basalts from the Parand (high-Ti), the Walvis Ridge and Tristan da &&a. They ap- pear to have been derived from similar mantle source regions, and not to have been affected sig- nificantly by crustal contamination processes. They also exhibit a broad negative correlation between Pb and Sr isotopes similar to that in cer- tain oceanic basalts.

(2) The initial Sr and Pb isotope ratios in the Jacupiranga carbonatites are different from those in the associated pyroxenites. These differences are considered to be primary features related to their mantle source regions.

(3) Model calculations indicate that liquids in equilibrium with the carbonatites and the pyrox- enites had different REE profiles, and that these differences cannot readily be attributed to frac- tional crystallisation or segregation of immisci- ble liquids. Thus, both the isotope and trace-ele- ment data indicate that the Jacupiranga carbonatites were not derived from the same magma as the pyroxenites. Instead, it is argued that the Jacupiranga complex was generated in multiple episodes of small-degree melting.

(4) Both experimental and mant!e xenohth

studies have demonstrated that carbonatite are important agents of mantle mctasomatism, and some distinctive geochemical s have been recognised (Haggerty, 1989; et al., 1989; Yaxley et al., 199 1 ick et al., 1993). Carbonatites are chaldc by very high Sr contents, high and variable U/Pb, but Sr. Thus, infiltration and/or metasom carbonatite melts may significantly reduce Rbj Sr and increase U/Pb in the upper mantle. This is consistent with the low contents observed in the ilied by carbonatite melts. With time this will sult in negative correlations between Sr and isotopes, and thus carbonatite metasomatism is one process which may have been responsible for the negative fractionation of U/Pb and Rb/Sr inferred for the source of certain oceanic basalts.

Y.-M.H. acknowledges funding from the partment of Earth Sciences, The Open Univer- sity. We thank Nick Rogers for many discussions on these and related igneous ks, and Keith Bell, Catherine Chauvel and ke Roden for their helpful and detailed reviews that resulted in significant improvements to the manuscript. Isotope Research at the Open University is funded by the NERC, and sample collection was with Professor M.S.M. Mantovani, with finan- cial assistance from FAPESP, CNPq and FINEP. The final manuscript was prepared by Janet Dryden.

hy and mineralogy

HBOOf is a pyroxenite with cuhedral granular texture. The pyroxencs are high-Ca diopside with Wo over 50%, indica- tive of crystallisation in a high-CaC@ environment. There are a few small flakes of micas in fractures between the py- roxenes, and these are considered to be secondary. Analyses of magnetite show high TiOl ( 10-l 1%) and AljO3 _ 3.5%. IIt is inferred to be cumulitic based on texture and mineral compositions.

HB004 is a brown-and-white banded sample with a total SiOz content of -, 20%. The white iayers are mainly composed of

scattered olivine, phlogopi and magnetite. The brown laye ‘s consist of olivine, phlogo te and calcite. Oliv- ines have been corroded by phPogopite icrocrystals a& Cal- cite. suggesting that phlogopite was formed by reaction be- tween fluids and olivine. The phlogopites nave .&O, ( i@- 7.5%) and BaQ ( c 0. I %) and Fe0 ( > 4%).

is a carbonatite with Si@< lO%, consisting mainly te, dolomite, npatite, phlogopite, oiivine and magnc-

tite. The carbonate minerals are predominantly calcite with a few dolomite grains (2 in 20 analyses), occurring as xeno- morphic crystals. Twinning is common. The calcites contain .-2Oh MgO and 0.68~0.8% SrQ and 0.08-0.2% BaO, whereas

the dolomites contain less SrO (0.2) and almost no BaO. Phlogopite is euhedral ( > 0.5 mm) and appears to be pri- mary. It has higher Al$& ( 12-15%) and BaO (0.2- 1.3%). and lower Fe0 (2-3%), and hence it is chemically different from those in HB004. Olivine has Fo 95-96. which is much higher than that in olivines (Fo 88-92) in basic and/or ul- trabasic silicate rocks, but similar to olivines (Fo 96 ) in mcta- morphic limestone. Magnetite occurs scattered throughout the sample and the analyses show TiO, 0.5-5%, and low Aiz03 of m 1.5%.

HB008 was split up into t s: one (HBOO&I) is similar to HBOOl and the other ( 2) is similar to HBOO4.

HB009 has a dark layer sandwiched between carbonatile, and it was also split into two subsamples. HB009-I consists of olivine, microcrystalline phlogopite and calcite, with SiOZ z 20%. It is similar to HBU04. HBOO9-2 is mainly com- posed of calcite, with some apatite and a few sulphides.

HBOfO is a carbonatite simila; to HB005. The olivines have Fo 95-96, and serpentinisation is seen along the fractures in the olivine crystals. Apatite is distributed heterogeneously as mineral aggregates. It appears in two distinct habits; ( I ) as small isolated oval grair?s within the carbonatitic minerals, with diameters of 0. I-1 mm (these apatite cry!,tals usually are slightly rounded and corroded); and (2) as idiomorphic prisms I-5 mm in diameler in aggregates. Electron probe analyses indicate that the ripatites are fluctrapatite with 5-6% F.

HBOII is a carbonatite similar to HBOIO, but with fewer sil- icate phases (olivine and phlogopite) and magnetite.

kmaral, G., 1978. Potassium-argon studies on the Jacupi- ranga alkaline district, State of SHo Paulo, Brazil. Proc. 1st Int. Symp. on Carbonatites, Depafimento National da Produ@o Mineral, Brasilia, pp. 297-302.

Andersen, T. and Taylor, P.N., 1988. Pb isotope geochemis- try of the Fen carbonatite complex, S-E Norway: Age and petrogenetic implications. Geochim. Cosmochim. Acta, 52: 209-2 ! 5.

Baker, M.B. and Wyllie, P.J., 1990. Liquid immiscibility in a nephelinite-carbonatite system at 25 kbar and imp!ica-

tion for carbonatite origin. Nature (Eonaon ), 446: i68-

1 ?O.

BaSU. AR. iI& Ti3tWmOt0, M., I9SO. Nd, isotopes in selected mantle derived rocks and minerals and their implications for mantle evolution. Contrib. Mi era!. Petrol., 75: 43- 54.

asu, A.R., Renne, P. ., Mertz, D. and Poreda, alit igneous complexes of the Deccan and ation for the origin ofcontinenta! flood ba

Am. Geophys. Union Fall Meet., Eos (Trans. Am. Geo-

~orbide~~i, L., Ruberti, E., Siena, F. and Traversa, G., 1992. Fractional crystallisation and liquid immiscibility processes in the alkaline-carbonatite complex of Juquia (Sao Paula, Brazil). J. Petrol., 33: 1371-1404.

ell, K. and Blenkinsop, J., 1987, Nd and Sr isotopic com- positions of east Africa carbonatites: Implications for mantle heterogeneity. Geology, 15: 99- 1 OP.

Bell, K. and Bienkinsop, J., 1989. Neodymium and st tium isotope geochemistry of carbonatites. In: K. (Editor), Carbonatite: Genesis and Evolution. Unwm

yman, London, pp. 278-299. Bell, K. and Peterson, T., 199 I. Nd and Sr isotope systemat-

its of Shombole volcano, East Africa, and the links be- tween nephelinites, phonolites and carbonatites. Geol- ogy, 19: 582-585.

Bell, K., Blenkinsop, J., Cole, T.J.S. and Menagh, D.P., 1982. Evidence from Sr isotopes for long-lived heterogeneities

r mantle. Nature ( London ), 298: 25 l-253. e, W.R.. Ellis, D.J., Green, D. hikov, I.D., 1983. Pyroxene

tionsin the upper mantle. Earth Planet. Sci. Lelt., 62.63- 74.

Chase, C.G., 198 I. Ocean island Pb: two stage histories and mantie evolution. Earth Planet. Sci. Lett., 52: 277-284.

Chauvel, C., ofmann, A.M. and Vidal, P., 1992. HIMU- EM: The French Polynesian connection. Earth Planet. Sci. Lett., 110: 99-l 19.

Cohen, AS., O’Nions, R.K. and O’Hara, M.J., I99 1. Chro- nology and mechanism of depletion in Lewisian granu- lites. Contrib. Mineral. Petrol., 106: 142-l 53.

Dalton, J.A. and Wood, B.J., 1993. The compositions of pri- mary carbonate melts and their evolution through wall- rock reaction in the mantle. Earth Planet. Sci. Lett.. I 19: 51 l-525.

Dawson, J.B., L984. Contrasting type of upper mantle meta- somatism. In: J. Komprobst (Editor), Kimberlites, IF: The Mantle and Crust-Mantle Relationship. Elsevier, Am- sterdam, pp. 289-294.

DePaolo, D.J., 1981. Trace element and isotopic effects Of combined wallrock assimiiation and fractional CvSla%S-

ation. Earth Planet. Sci. Lett., 53: 189-202. Eg@er, D.H., 1987. Solubility of major and trace elements in

mantle metasomatic fluids experimental constrains. In: MA. Menzies and C.J. Hawkesworth (Editors), Mantle

98 Y, Huang et ul. / Chemical Gedvgy I I9 (1 Y95) 79- 99

Metasomatism. Academic Press, London, PP. 2 1-4 1. Eder, D.H., 1989. Carbonatites, primary melts and mantle

dynamics. In: K. Bell (Editor), Carbcnatite: Genesis and Evolution. Unwin Hyman, London, pp. 561-579.

Falloon, T.J. and Green, D.H., 1989. The solidus of carbon- ated, fertile peridotite. Earth Pianet. Sci. Lett., 94: 364- 370.

Fraser, K.J. and Hawkesworth, C.J., 1992. The petrogenesis of grout, 2 ultrapotassic kimberlites from Finsch Mine, South Africa. Lithos, 28: 327-345.

Fujimaki, H., Tatsumoto, M. and Aoki, K., 1984. Partition coefftcients of Hf, Zr, and REE between phenocrysts and groundmass. J. Geophys. Res., 89: B662-B672.

Gaspar, J.C. and Wyllie, P.J., 1983. Magnetite in the carbon- atites from the Jacupiranga complex, Brazil. Am. Min- eral., 68: 195-213.

Haggerty, SE., 1989. Mantle metasomes and the kinship be- tween carbonatites and kimberlites. In: K. Bell (Editor), Carbonatite: Genesis and Evolution. Unwin Ilyman, London, pp. 546-560.

Gittins, J.. 1989. The origin and evolution of carbonatite magmas. In: K. Bell (Editor), Carbonatite: Genesis and Evolution. Unwin Hyman, London, pp. 580-600.

Gray, C.M. and Oversby, V.M., 1972. The behaviour of lead isotopes during granulite facies metamorphism. Geo- chim. Cosmochim. Acta, 36: 939-952,.

Green, D.H. and Wallace, M.E., 1988, Mantle metasoma- tism by ephemeral carbonatite melt. Nature (London ), 336: 459-462.

Hamilton, D.L., Bedson, P. and Esson, J., 1989. The behav- i3cr of trace elelments in the evolution ofcarbJnatites. In: K. Bell (Editor), Carbonatite: Genesis ard Evolution. Unwin Hyman, London, pp. 405-427.

Hart, S.R., 1984. A large-scale isotope anomaly in the South- em Hemisphere mantle. Nature (London ), 309: 753-757.

Hart, S.R. and Dunn, T.. 1993. Experimental cpx/melt par- titioning of 24 trace elements. Contrib. Mineral. Petrol., 113: 1-8.

Hat% B., 1983. Mantle peridotites and prccesses - the kim- berlite sample. In: C.J. Hawkesworth and M. Norry (Ed- itors), Continental Basatts and Mantlt: Xenoliths. Shiva, Nantwich, pp. 46-9 I.

Hauri, E.H., Shim& N., Dicu, J.J. ant\ Hart, S.R., 1993. Evidence for hotspot-related carbonatite metasomatism in the oceanic upper mantle. Nature (London), 356: 22!- 227.

Hawkesworth, C.J., Mantovani, M.S.M., Taylor, P.N. and Palacz, Z., 1986. Evidence from the Parana of somh Bra- zil for a continental contribution to Dupal basalts. Nature (London), 322: 356-353.

Hawkesworth, C.J., Kempton, P.D., Rogers, N.W., Ellam, R.M. and van Calsteren, P.W., 1990. Continental mantle lithosphere and shallow level enrichment processes in the Earth’s mantle. Earth Planet. Sci. Lett., 96: 256-268.

Hawkesworth, C.J., Gallagher, K., Kelley, S., Mantovani, M., Peat% D.W., Regelous, M. and Rogers, N.W., 1992. Par- ana magmatism and the opening of the South Atlantic. In:

B.C. Storey, T. Alabaster and R.J. Pankhurst (Editors), Magmatism and the Causes of Continental Break-up. Geol. Sot. London, Spec. Pub]., 68: 22 I-240.

Herz, N., t 977. Timing of spreading in the South A?lantic: information from Brazi an a:kalic rocks. Geol. Sac. Am. Bull., 88: 101-l 12.

Hofmann, A.W. and White, W.M., 1982. Mantle plumes ftom ancient oceanic crust. Earth Planet. Sci. Lett., 57: 421- 436.

Huang, Y.-M., van Calsteren, P.W. and Hawkrsworth, C.J., 1994. Implication for the evolution of the lithosphere in scuthem Africa from isowpe systematics in granultie xenoliths. Abstr. 8th Int. Conf. Geochronol. Cosmo- chronol. Isot. Geor.( ICOG-B), Berkeley. Ca1.f.. June 5- 11, 1994, p.145.

Hussak, E., 1892. Uber Brazilit, ein neues Tantal (Niob) Mineral von der Eisenmine Jacupiranga. Neues Jahrb. Mineral. Geol. Pallontol., 2: 141-159.

Ito, E., White, W.M. and Gopet, C., 1987. The 0, Sr, Nd and Pb isotope geochemistry of MORB. Chem. Geot., 62: I57- 176.

Jones, A.P., 1989. Upper-mantle enrichment by kimberlitic or carbonatitic magmatism. In: K. Bell (Editor), Carbon- atite: Genesis and Evolution. Unwin Hyman, London, pp. 448-463.

Kjarsgaad, B.A. and Hamilton, D.L., 1989. The genesis of carbonatites by immiscibility. In: K. Bell (Editor), Car- bonatite: Genesis and Evolution. Unwin Hyman, Lon- don, pp. 388-404.

Kwon, S.-T., Tittan. G.R. and Griinenfelder, M.H., 1989. Lead isotope rcaltionships in carbonatites and alkalic complcxcs: An overview. In: K. Bell (Editor), Carbona- titc: Genesis and Evolution. Wnwin Hyman, London, pp. 360-387.

McDonough, W.F., t !)90. Constraints on the composition of the continental hthospheric mantle. Earth Pta.net. Sci. Lett., 101: I-18.

McKenzie, D. and O’Nions, R.K., 199 1. Partial melt distri- butions from inversion of rare earth element concentra- tions. J. Petrol., 32: IOZt-i@:.

Meen, J.K., Ayers, J.C. and Fregeau, E.J., 1989. A model of mantle metasomatism by carbonated alkaline melts: Trace-element and isotopic compositions of mantle source regions of carbonatite and other continent igneous rocks. In: K. Bet1 (Editor), Carbonatite: Genesis and Evolution. Unwin Hyman, London, pp. 464-499.

Melcher, G.C., 1966. The carbonatites of Jacupiranga, S?io Paulo, Brazil. In: OF. Tuttle and J. Gittins (Editors), Carbonatites. Wiley, New York, N.Y.. pp. 169-l 81.

Mcnzics, M.A.. 1983. Mantle uitramafic xenoliths in alkaline magmas: evidence for mantle heterogeneity modified by magmatic activity. In: C.J. Hawkesworth and M. Norry (Editors), Continental Basalts and Mantle Xenoliths. Shiva, Nantwich, pp. 92-l 10.

Moorbath, S.H., Wclke, H. and Gale, N.H., 1969. The signif- icance of lead isotope studies in ancient high-grade meta- morphic basement complexes as exemplified by the Lew-

isian reeks ofnorthwest Scotland. Earth

.R., Chivas. A.R., Chappell, Geochemical and isoto C SysPcmatics in car-

e unusual examples from Waast, New Zealand. Eos (Trans. Am. Geophys. Union), 74: 340 (abstract ).

Peate. D.W., Yawkesworth, C.J. and antovani, M.S.M., 1992. Chemical stratigraphy of the Parana lavds (South America ): classification of magma types and their spatial distribution. Bull. Volcanol., 55: 119- 139.

Potts. P.J., Thorpe, Q.W., Isaacs, MC. and Wright, D.W., 1985. High-precision instrumental neutron-activation analysis of geological samples employing simultaneous counting with both planar and coaxial detectors. Chem. Geol.. 48: 145-155.

Renne, P.R., Emesto, M., Ptevot, M. and Perrin, volcanism, rifting of Go Cretaceous boundary. Science. 258: 975-979.

Richardson, S.N., Erlank, A.J., Duncan, AR. and Reid, D.L., 1982. Correlated Nd, Sr and Pb isotope variation in Wal- vis Ridge basalts and implications for the evolution oftheir mantle source. Earth Planet. Sci. Lett., 59: 327-342.

Roden, M.F., Murthy, V.R. and Gaspar, J.C., 1985. Sr and Nd composition of the Jacupiranga carbonatite. J. Geol., 93: 2 12-220.

Rogers, N.W., l-lawkesworth, C.J. and Palacz, Z..4_., ! 992. Fhlogopite in the generation of olivine-melilitites from Namaqualand, South Africa and implications for element fractionation processes in the upper mantle. Lithos, 28: 341-365.

Rudnick, R.L. and Prespcr, T., 1990. Geochemistry of intrr- mediate- to high-pressure granulites. In D. Vielzeuf and Ph. Vidal (Editors), Granulites and Crustal Evolution. Kluwer, Dordrecht, pp. 523-550.

Rudnick, R.L., McDonough, W.F. and Chappell, B. W.. 1993. Carbonatite metasomatism in the northern Tanzanian mantle: petrographic and geochemical characteristics. Earth Planet. Sci. Lett., 14: 463-475.

Ssacey, J.S. and Kramers, J.D., 1975. Approximation of ter- restrial lead isotope evolution by a two-stage model. Earth Planet. Sci. Lett., 26: 207-22 I.

Sun, S.-S. and McDonough, W.F., 1989. Chemical and iso-

topic systematics of oceanic basahs: irnp!icarions for manale composition and processes. In: A. Saunders and

Norry (Editors). GA. Sec. London

agmatism in the ocean Basins. Publ., 42: 313-345.

Sweeney, R.J.. Green, and Si:, SPH.. 1992. Trace and minor element par ing between garti~ knd amphi- bole and carbonatitic melt. Earth Planet. Sci. Lett., 1 13: l-14.

Taylor, S.R. and MC , 1985. The Continental Crust: Its Compo olution. Blackwell, Lon- don, 3 I:! pp.

I-U US, M., Kelley, S., Nawkesworth, C. and 1994. WIagmatism and continental break- Atlantic: high precision 40Ar-“9Ar geo-

chronologyy. Earth Plans:. Sci. Lett.. 121: 333-348. UIbrich, E!.G.J Comes, C.B., I98 1. Alkaline rocks from

continental h-Sci. Rev., 17: 13 I- 154. van Calsteren, arris. N.B.W.. Hawkesworth, C.J.,

enzies, MA. and Rogers, N.W., 1986. Xenoliths from uthern Africa: a perspective on the lower crust - The

nature of the lower continental crust. Geol. Sot. Eom.!on. Spec. Pubi., 24: 35 l-362.

van Calsteren, P.W., keswor in, L..$ ., I993. Pb isotope tal lithosphere, southern Africa: a xenolith perspective. Terra Nova, l(5): 33.

Watson, E.B., Ben Qthman, D.. Luck, J.- and Hofmann. A.W., 1986. Partitioning of U, Pb, Cs, , Hf, We and 0s between chromian diopsidic pyroxene and haplobasaltic liquid. ihem. Geol., 62: I9 l-208.

Weaver. B.L. and Tarncy, J., 1981. Lcwisian gneiss geochem- isicy and Archaean crusial development models. Earth Planet. Sci. Lett., 55: I7 I-l 80.

Wendlandt, R.F. and Harrison, W.J., 1979. Rare earth parti- tioning between immiscible carhonatc and silicate liquids and CO, vapor: Results and implications for the forma- tion of light rare earth-enriched rocks. Contrib. ineral Petrol., 69: 409-419.

Woolley, A.R., Barr, MW., Jones, G.C., Wall, F. and Wil- liams, C.T.. 1491. Extrusils carbonatitcs frc.m the Uyay- nah area, United Arab Emirates. J. ?etrol., 32: I i ? 3-1167.

Yaxley, GM., Crawford, W.J. and Green, D.F,., 1991. bi- dence for carbonatite metasomatism in spine1 peridotite xenoliths from Western Victoria, Australia. Earth Planet. Sci. Lett., 107: 305-3 17.

Zindler, A. aud Mart, S., 1586. Chemical geodynamics. Annu. Rev. Earlih Planet. Sci., 14: 493-57 I.