the trace element content and petrogenesis of … mineralogist, volume 67, pages 223-233, 1982 the...

American Mineralogist, Volume 67, pages 223-233, 1982

The trace element content and petrogenesis of_ kimberlitesin Riley County, Kahsas, U. S. A.r

Rossnr L. Curlens, JeNIce MuI-reNex, MrcHeeL J. Drrr.ranco eNo Srpve NonoBNc

Department of Geology, Kansas State (JniversityManhattan, Kansas 66506

Abstract

Kimberlites and associated xenoliths, carbonate veins, and interstitial carbonate materialfrom Riley County, Kansas have been analyzed for certain trace elements and Fe usingneutron activation. The non-micaceous kimberlites (Bala and Randolph No. 1) are higher inthe incompatible elements (REE, Ba, Th, Hf, and ra) compared to the micaceouskimberlites (Stockdale and Leonardville.) Contamination due to the numerous crustalxenoliths and interstitial carbonate material has probably been small. The xenoliths andinterstitial carbonate would generally tend to dilute slightly the trace element concentra-tions of the kimberlites. The trace element contents (except perhaps the LREE and Th) ofthe kimberlites can be explained by slightly different degrees of incipient partial melting ofaverage upper mantle peridotite containing about 56%o olivine, 33Vo orthopyroxene, 5Voclinopyroxene, and,6Vo garnet. Significant fractional crystallization of pyroxene, olivine, orgarnet is precluded in the kimberlite melt since the compatible element contents (Co, Cr,and Sc) are quite high and similar to values expected by direct melting of peridotite.However, the LREE and Th contents of the kimberlite are somewhat higher than predictedby direct melting of peridotite possibly the result of volatile transport of these elements inH2O- and CO2-rich fluids. Alternatively, a more LREE- and Th-rich source than used inaverage peridotite in the above melting model may account for the enrichment of theseelements in the kimberlites.

None of the carbonate material (veins, xenoliths, and kimberlite groundmass) appears tobe igneous carbonate. Rather, it appears to be the result ofvaried amounts ofsecondarv orhydrothermal (H2O- and CO2-rich fluids) activity. The relative importance of theseprocesses cannot be evaluated with the present data.

Introduction

The petrology, major element contents, andstrontium isotopic compositions of the six kimber-lites of probable Late Cretaceous age in RileyCounty, Kansas have been studied extensively byBrookins and his students (e.g., Brookins, 1967,1970a, 1970b; Brookins and Naeser, l97l). Thisstudy focuses onhow certain trace elements may beused to determine the petrogenesis of these kimber-lites.

Current models for the formation of kimberlitesbased on mineralogy, major and trace elementcontents, isotopic data, and experimental petrologyindicate that kimberlites form by small degrees of

I This report was first presented at a symposium on ..Model-

ing of High-Temperature Petrologic Processes" held at IowaState University in Ames, Iowa in connection with the North-Central G. S. A. meetings April 30-May l, 1981.

m03_fi x)v82l03u--0223 $02. 00

melting of upper mantle peridotite at depth inexcess of 100 km(e.9., Wedepohl and Muramatsu,1979; Wyllie, 1979; Eggler and Wendlandt, 1979;Cullers and Graf, 1982a). This study demonstratesthat the trace element distribution of the RileyCounty, Kansas kimberlites are consistent with asimple model of their formation by incipient meltingof garnet peridotite followed by possible volatiletransport of the LREE (light rare-earth elements)and Th in COz- and H2O-rich fluids.

The kimberlites are crowded with xenoliths.mostly of crustal rocks. Several of these xenolithswere analyzed to evaluate the possible effects ofcontamination on the trace element characteristicsof the kimberlites. Samples of the carbonate materi-al found in these kimberlites were analyzed. Thismaterial does not appear to be igneous carbonate,but probably represents either secondary or hydro-thermal veins.

223

CULLERS ET AL.: TRACE ELEMENT CONTENT OF KIMBERLITES

NANDOLPH It{TRUSIOI{S

STOCKDALE IiITRUSIOI{

LEOl {ARDVILLEt l {TRU

I T r O S

i r , , s

Fig. l. The location of the kimberlites in Riley County,Kansas.

Location

The six known kimberlites are located in RileyCounty in northern and eastern Kansas, U. S. A.(Fig. 1). They are intruded into lower Permiansedimentary rocks along a synclinal-anticlinal axis.

Experimental

Samples obtained from the surface were thefreshest possible samples. In addition, cores of theBala and Leonardville kimberlites were also ana-lyzed. Few xenoliths were included in the wholerocks of analyzed kimberlites, but it was impossibleto avoid some xenolith contamination since theywere so abundant in all but our Bala core. Xenolithshave also been analyzed separately to give somefeeling as to how significantly they may contami-nate the kimberlites.

The experimental procedure for the trace elementanalyses was adopted from methods of Gordon e/al. (1968) and Jacobs et al. (1977). Typical precisionobtained by us on standard rocks is summarized inCullers et al. (1979).

Mineralogy and petrography

The mineralogy and petrology of these kimber-lites have been summarized in most detail byBrookins (1970). The following section summarizesmostly the petrography and isotopic geochemistrydone by Brookins. His petrogaphy is consistentwith the results we obtained. Brief petrographicdescriptions of our samples are included in theappendix.

The kimberlites are porphyritic with phenocrystsof olivine and smaller amounts of pyroxene, r&g-netite, ilmenite, pyrope, and/or phlogopite. Mostolivine and pyroxene phenocrysts are altered exten-sively to serpentine or calcite, and many have areaction rim of serpentine or chlorite. The pheno-crysts occur in a very fine-grained, panidiomorphicmatrix of mainly serpentine and calcite. Other mi-nor minerals such as perovskite, zircon, ilmenite,magnetite, and chromite, may also occur in thematrix.

Most of the kimberlites are crowded with alteredshale or carbonate xenoliths. In addition, the Stock-dale kimberlite has a wide range of igneous andmetamorphic xenoliths from the crust and uppermantle.

Numerous calcite veins cross-cut the kimberlites.Post-emplacement veins have 87srf6sr ratios of0.7079-0.7093 similar to the Permian limestones inthe area. These veins are clearly secondary. Otherveins have lower 875r/865r ratios of 0.7028-0.7044similar to igneous carbonate or hydrothermal veins.The groundmass carbonate contains intermediateSr isotopic compositions between the two extremesand must be a mixture of primary and secondarycarbonate.

Brookins (1970) distinguishes two groups of kim-berlites: micaceous (-15% phlogopite) and lampro-phyric (non-micaceous). The Stockdale, Leonard-ville, and Winkler kimberlites are micaceouskimberlites. The micaceous kimberlites containmore pyrope and ilmenite than the non-micaceoustypes. The non-micaceous kimberlites are the Bala,Randolph No. 1, and Randolph No. 2 kimberlites;they contain no more than traces of phlogopite.Pyrope is rare in non-micaceous types, and is foundonly in the groundmass. Ilmenite grains are smallerin the non-micaceous kimberlites than in the mica-ceous kimberlites.

Major elements

Compared to the other basic rocks like basalts,the Riley County kimberlites are higher in MgO(2W25%), CaO (13-19Vo), HzO (8-10%), and CO2(9-12%) and lower in SiO2 (21-24Vo), AlzOz(2-5Vo),K2O (0.02-0.30%) and Na2O (0.04-0.5%) (Brook-ins, 1970; Wedepohl and Muramatsu, 1979). Thesekimberlites are similar to basalts in FezOr (6.5-7.4%). Compared to the average of other kimber-lites, the Riley County kimberlites are lower in SiOzand KzO and higher in CaO and CO2 (Brookins,1970; Wedepohl and Muramatsu, 1979). The CaO

R 7 E- - fwt l {xLERil{TRUStON

B A L AINTRUS

CULLERS ET AL,: TRACE ELEMENT CONTENT OF KIMBERLITES

and CO2 contents of these kimberlites are fairlyhigh compared to other kimberlites because of thelarge amount of calcite. These kimberlites may below in the alkalies because of extensive alteration.Alteration is unaffected by depth in the intrusion;major"and trace element analyses and the petrogra-phy in cores are nearly identical to those at thesurface (Brookins, 1970; also this paper).

Results

The trace element and Fe contents of the ana-lyzed kimberlites, inclusions, and veins are given inTable 1. Compared to basalts (even alkali basalts).kimberlites are enriched in LREE, Ba, Th, Hf, andTa and depleted in Co and Sc (Wedepohl andMuramatsu, 1979).

The average trace element and Fe contents ofthese kimberlites are compared to the average of allkimberlites in Table 2. In general, Ba and Ta in theRiley County kimberlites are very much enrichedover the average of all kimberlites. Other elementsare variously enriched or depleted in these kimber-lites compared to the average. The non-micaceouskimberlites (Bala and Randolph No. l) have signifi-cantly different trace element contents compared tothe micaceous kimberlites (Stockdale and Leonard-ville). The non-micaceous kimberlites are signifi-cantly higher in REE (rare-earth elements), Ba, Th,Hf, Sc, and Ta compared to the micaceous kimber-lites and to the average of kimberlites (except Hf inthe latter). The two groups of kimberlites haveabout the same Co and Cr. The Co and Cr contentsof the kimberlites are also similar to the average ofall kimberlites.

The Leonardville and Bala kimberlites have beenanalyzed in some detail since considerable freshcore wis available. Except for Ba, there seems tobe no systematic variation in trace elements withdepth. The lowest Ba occurs in the bottom twosamples in both the Leonardville and Bala kimber-lites.

The REE contents of the Bala kimberlite, inclu-sions, and veins are shown in Figure 2. The REEcontent of the samples are as follows: kimberlite >inclusions > veins. The LREE and Th of the Balakimberlite are the highest of the kimberlites report-ed in this study. Also the carbonate-shale inclusionsin this kimberlite have the highest LREE and Thcontents compared to corresponding material in theother kimberlites. The LREE are also high but theTh contents are low in carbonate veins from the

Bala kimberlite compared to such veins in theStockdale kimberlite.

The other non-micaceous kimberlite analyzed,Randolph No. 1, is somewhat lower in Th andLREE than the Bala kimberlite, but it is higher inBa, Hf, and Ta. No carbonate veins and no inclu-sions have been analyzed in the Randolph No. Ikimberlite.

As with the other kimberlites, the REE content ofthe Leonardville kimberlite is greater than that ofthe inclusions (Fig. 3). The Leonardville kimberlitehas the highest LREE, Ba, Th, and Hf of themicaceous kimberlites (Table 2; Figs. 3 and 4). Bothmicaceous kimberlites are similar to one another inthe content of all other analyzed elements. Theelement contents in the inclusions in these mica-ceous kimberlites are lower in allanalyzed elementsthan inclusions in the Bala kimberlite.

The REE content of the Stockdale kimberlite isgreater than that of the altered shale inclusionswhich are in turn greater than that of the carbonateveins (Fig. 4). This order of REE abundance is thesame as for the Bala kimberlite.

Discussion

C arbonat e-s hale xe noliths

The carbonate-shale xenoliths in the Leonardvilleand Stockdale kimberlites have lower trace elementcontents than the corresponding kimberlite (Fig. 5).This is also true in carbonate-shale xenoliths rela-tive to kimberlite for all elements but Yb, Lu, Ba,and Hf in the Bala kimberlite. Thus, contaminationby xenoliths should act to dilute most elementsslightly in the whole rock. The values for mostelements should be slightly higher in unalteredkimberlite. Nevertheless, we avoided large xeno-liths in our sampling of Leondardville, Stockdale,and Randolph No. 1 kimberlites. There were fewvisible xenoliths in our Bala sample. Thus, webelieve that most element contents were not signfi-cantly affected by xenolith contamination in thewhole rock anlayses.

Melting models

Most models of kimberlite petrogenesis focus ontheir formation by a very small percentage of melt-ing of upper mantle peridotite at depths in excess of100 km. Trace elements can be used to model thisprocess if a reasonable mineralogy of the source,trace element content of the source, and percentmelting can be assumed, and if it is known how the

po

!5

I

oc

o+

> o! !5 l9 @

o oN T

d oE >

O Ec ot !op o

,E OO E

o 63

6 !

9 d

O . FP O

E o

o oa

o =

6 0!P g

@

E r uO . 99 6 6

O FU Ou +E G O

! 6o =

\ oJ E

6 E! o

9 Q 6O N

o

o 6 0p E r6 ! PE O

c 9P r O6 Qo p !

oo ! E

E ! =

o e9 EE O O

cO . F

r t r t su €dH @ L

a oo o

F F

t t l

d N O

o o o +6 < l 6 0o o o oo o o oN N

o @ @ No @

o H d N ' s =1 . .

N O r @ O N

d o d 6 t + 6j ; a i ; ' d d

s

3 : RR l edc o d +d . j d a i o o i

N d

@ N @ dN N d N

o o o oRs qq{o o

[E HHH

N O O O t s O O @N N O N N N N N

o o + @ + @ @ N o e a o dN d d d i H o a - ; . a a

- ol l

I o o o o o o e N o oN @ @ o N N o N s r ( q e l

H O O O O O O O O OI r t l

x3 qqqo o

3s qqq. : . :

d d d

c o 6@ @ e a \o i o i N € @

@ N O O d S@ N N @ N N

< t @ @ @ @o t s o d @ 6

O N T D O N N

N @ + O O O

N d o o o od d d d

o o o o o o 6d o o o N I o od N o o o I

N o * No o N o o t s o i o + N o No o N d N o o o H H d r o

t .o o o o o o o o o o o o o

o o o o o o o H N @ d o od 4 r d @ d o o o v i o r Nd o o o d d N o I oH d d d d d d


o @r d @r o o

N @ Od d o

o o +

I N Hr o d

o o

o oH O

o o

o o

oO + o @ t N oc j i

" ; " ; t - ; " j

@ o N N^ ; ; g i l dd

N N O d d 0 6o I d d d d N o

. 1 . . . . . .o o o o o o o

o @ o + H o

o o N o o o

@ o o o @ o

o o o o o o o o o Q O oo @ N O O d n + r r o r 6 6€ @ N d o a o N o d i o 66 + + + N N O O I N N

o o o o o o oO H N 6 N O Oo N @ i @ o s@ @ 6 @ @ € < t

o @ o r( o o @ od o o d

d @ c o

a o o o

o 6 0 < i

r o r@ d o

o N o o

o @ Ha r q N

o o or d @

N O @ @

o o o o

I N F @ d 6 O @

o @ @ o @ r €d

+ N O O O + Nt rI d 6 + N N N d

N d H d H d d

N b I r @ o oo N I r + o o

o H d

o o N o + o qN N N N N N N

O NO N

r . @t o

N N O

o o o

O N

d c i *

6o o N + @ o o o @

o o o o o o o o +

o d o o + o o d

o o 6 0 0 N @ N oN N d N d d H

€ @ O N E N NN N N N N N N

o o o o o o o

@ o + d o o +6 r d o N N oo o o o o o o

o + @ o N n @O Q N O N H d

@ @ N O O O OO + N N O N d

d d d d H H H

o o N + @ b @i o o N q o o

d H o d o H o

o @ oN O H O @ C O N

o c + o + o +

o o o $ @ + NN @ @ @ @ 6 @

+ o o o @ | |N N N N N I I

N N O O O Ot s @ @ t s @ t s

@ @ + n 6 0N N d i d d

r @ @

@ @ | r + l oI t d r o

sH o o o

l .

R R ? 9 - r o+ oo o

O N O O N I Ni i V V N I N

o o 6 0 N t @N O O + O r N

r @N + N + O

N H d i

N H d O T N Qo @ 6 0 0 0 @b + d d o ( p t s

No o

o N o 6 N O@ @ o r o N N

. 1 .

o @ 6o o I t @ b I

| | . . lo o o

S N N @I o o r 4 0 @| . . t .

o o o o

o + N o o o +6 6 0 0 + o od d o o N a o

o o oo o l ! L! L O O Oo o 9 u op o

P P O O Oo o o o oo o F 9 +

o o oN @ N H t sN Q d N N

@ @

+ r c o + o d oo + o 6 c o o o

N N

+ o a o N 6 +

n + b a s + €

o o t oo o t s

N . o o o o @ o \ e i i b g g 3 - gO @ d O O O O d @ @ i

o d + o@ + d o

s f @ N d o O €+ < i + + + + +

+ d o @ N o oO o d O d O Oi l H d d N N

N d c o do o o o

N $ O No o o H

Nl l l

' l l lo

$ N N No d o d

o 6 t s @N O < l O

i o o o

o @ d N

t s r 6 N

6 N N F o o c o o @

N9 s : 933 R e+eo ot ! _o o - '9 0 ^

i o o o € N + N+ 6 O N O O O +6 b 6 0 + +

o o6 6

O N N + @ + O O d e NN O O O O O O O N N @ O6 0 6 0 0 0 0 N N N

N + 6 N @ d O Oo @ o @ o o H oO O N N N N N N

3R 8ssN N

i d @ + o F oo o 6 H o o oO O N O N N O

c o @ d o o oN O @ O N 9d H i d H H

o o

T T

d N

o @

o @

@ @

+ o +< f @ d H o

@ c o d d o

6 0 0 0 d o b

@ @ t s @ @ N O

o

o

L

(J

O

o

o

oo

o

oo

a!i

c)

F

o

tr

CULLERS ET AL.: TRACE ELEMENT CONTENT OF KIMBERLITES 227

Table 2' Average Trace element and Fe contents of Riley County kimberlites compared to the average contents of kimberlites('Wedepohl and MuramatsV 1979; values in ppm except where indicated)

non-mi caceous mr caceousAverage in,

Ki mberl i tes'Average Bala

Ki mberl i te(7 samples)

AverageRando lph No. I

K imber l i te

AverageLeonardv i I Ie

Kimberl i te

AverageStockda l eKimberl i te

2 samol 6 samol or 3 samol

89

r426 . 7L . 7 50 .500. 820 . t 2

2650t22 . 6

20

79

1 0 . 5

t . >

1150

0 . 0 9

La

Ce

Sm

Eu

Tb

Yb

Lu

Ba

Th

Hf

Ta

Co

Sc

Fer0 , (%)

Cr

Nar0(%)

150200

I J

? n

1 . 0

1 . 2

0 . 1 6

r00016

9

10.2

I 1 1 0

0 . 2 7

302

516

1 8 . 3

+ . 5 b

I . 0 7

1 , .24

0 . 2 6

57 40

43. 6

4 . 8

24

65

1 9

9 . 5

I 100

0 . 1 2

231

3811 q ?

, 1 7

0. 90

u . t 5

6760

27

5 . 9

58

2 I

1 ,0 .7

r0200. 07

1 5 t

243

1 0 . 9

2 . 9

0 . 7 8

0. 85

0 . 1 6

4160

2 l

18

o v

1 3

7 . 4

1040

0 . 1 1

trace elements partition between the source andmelt (e.g., Shaw, 1970). The problems involvedwith these assumptions will not be elaborat-ed onhere in detail as they have been discussed in greatdetail in many other papers (e.g., Haskin, l9g2;Cullers and Graf, 1982a and b).

First, a D. C. (distribution coefficient) needs to bedefined. A D.C. is defined as the equilibrium con-centration of a trace element in a mineral relative toits concentration in a melt. The D. C. values used inthis study for mineral-basic melt pairs are summa-rized in the appendix along with the correspondingreferences. We have used D. C.'s which are at thelow end of all measured ranges for these mineral-melt pairs. This results in an upper limit for thetrace element contents in the melt. One problem isthat the listed D. C.'s are usually obtained at lowpressure, and we are not certain if the D.C.'schange at the high pressure needed to form kimber-lites. Also these are the D. C. for minerals relativeto basaltic melts instead of kimberlitic melt.

Upper mantle peridotite with 56Vo olivine, 33Voorthopyroxene, 5Vo clinopyroxene, and 6Vo garnetwith a trace element content of average peridotite(Wedepohl and Muramatsu, 1979) is assumed toundergo a small percentage of equilibrium melting.

F T A l t l c o F E i t r a E i l t r E !

i

\

\

\\

(nt_dF

E.oz.oIO

UJJo_

a

tooo

500

200

too

50

Lo C. Sm Eu Tb yb Lu

RARE-EARTH ATOMIC NUMBERFig. 2. The REE content of the Bala kimberlite, inclusions

(xenoliths), and veins normalized to chondrites (values ofHaskinet al., 1968) (Range of kimbertte, cross-hatch lines; kimberlitewith carbonate vein, solid square and dashed line; carbonate-shale inclusion, open squaxe and solid line; carbonate veins,circles and dash-dot or long dash-short dash line).


200

t o

al!

EozoI(J

Lulrl&.

@UJ

tEt_u(n

)<ulu.lE

roo

Lo Ce Sm Eu Tb Yb Lu

RARE - EARTH ATOMIC NUMBERFig. 3. The REE content of the Leonardville kimberlite and

carbonate-shale inclusions normalized to chondrites (range ofkimberlite, cross-hatch lines; carbonate-shale inclusion, opencircles).

This model is consistent with models proposed bymany others for the formation of basic and ultraba-sic rocks enriched in the incompatible elements(e.g., Gast, 1968; Kay and Gast, 1973; Cullers andMedaris. 1977: Mitchell and Brunfelt, 1974). Theincompatible elements with low D. C.'s for perido-tite relative to kimberlite melt will concentrate inthe melt (LREE, Ba, Th, Hf, and Ta). The compati-ble elements with high D.C.'s for peridotite relativeto kimberlite melt will concentrate in the peridotite(Co, Cr). Sc has a D. C. close to one for this systemso it should partition equally between peridotite andkimberlite melt.

The predicted trace element contents of the meltassuming 0.5% eqrilibrium melting of average pe-ridotite are compared to the Bala (example of a non-micaceous kimberlite) and Leonardville (mica-ceous) kimberlites ir.. Figure 6. Note the traceelement contents in this figure are normalized to theaverage peridotite of Wedepohl and Muramatsu soenrichment or depletion of an element in the meltrelative to the source rock can easily be noted. Webelieve this model is supported by the data sincethere is good agreement for most elements betweenthe predictions and observations. In addition, thenon-micaceous kimberlites have higher contents ofthe incompatible elements than the micaceous kim-

berlites, and all kimberlites have similar concentra-tions of the compatible elements. These differencesand similarities are precisely what is to be expectedif small differences in partial melting at small de-grees of melting are producing the variations intrace element content of the melt. The incompatibleelements are expected to decrease with increasedmelting (especially in the range of 0.5Vo to severalpercent melting) so presumably the non-micaceouskimberlites may represent a smaller degree of melt-ing than the micaceous kimberlites. The compatibleelements are insensitive to the degree of melting inthis range of melting and should have about thesame concentration in the melt as long as the sourceis homogeneous in these trace element contents.

Even though predictions and observations formost elements agree well in the above model, thepredicted values for the LREE and Th are too lowcompared to the observed values in any of thekimberlites. We do not believe this difference be-tween observed and predicted values for theseelements negates the model considering the manyassumptions involved. For example, inappropriateD. C.'s may have been used for these elements dueto differences in composition or pressure between

akJFE.oz.oIo

ul-J(L

a

Lo Ce Sm Eu Tb \t Lu

RARE- EARTH ATOM IC NUMBERFig. 4. The REE content of the Stockdale kimberlite'

carbonate-shale inclusions, and carbonate veins (range ofkimberlite, cross-hatch lines; carbonate-shale inclusions, opencircle: carbonate veins. closed circle).


the peridotite-kimberlite melt system at high pres-sure and the mineral-basaltic melt system at lowpressure in which the D. C.'s were measured.Indeed, Th D. C.'s are not even well established atlow pressure. Also the LREE and Th contents ofthe peridotite source may have been higher thanthat used in this model. Nixon et at. (l9gl), forexample, report mantle xenoliths that are muchmore enriched in the incompatible elements (espe-cially the LREE) used in the above model. Meltingof these enriched xenoliths could still generate therange of trace element contents of the Riley Countykimberlites using D. C. values that are not as low asthose used in the above model. Finally, there maybe other processes, as discussed below, affectingthe REE and Th contents in kimberlites.

High pressure fractional crystallization of eclo-gite has been used to explain increased LREEcontents in kimberlite melts over that predicted inpartial melting models (e.g., Mitchell and Brunfelt.1974). Large amounts of fractional crystallization ofgarnet and clinopyroxene from kimberlite couldproduce the higher REE and Th contents of kimber-lite over those predicted in the melting model. Thismodel, however, would result in increased amounts

lrJFtrooE.l!o_

IJFJtt_dm

E

500

roo

EL EM ENTSFig. 6. The element content of Bala (solid circle) and

konardville (open circle) kimberlites are normalized to averageperidotite. AIso plotted is the predicted element content of a meltproduced by 0.5Vo equilibrium melting of peridotite witholivine/orthopyroxene/clinopyroxene/garnet = 561331516.

of the incompatible elements (Ba, Th, Ta, Hf) anddramatic decreases in the compatible elements (Co,Cr, Sc, and probably the heavy REE) over thoseobserved in the kimberlites. The decrease expectedfor the compatible elements is the main reason thefractional crystallization model must be rejected.

Crustal contamination has also been suggested asnot having a significant effect on the trace elementcontents in this study. Minimal contamination bycrustal rocks is also supported by the very lowSiO2, Na2O, Al2O3, and strontium isotopic contentsof the kimberlites (Brookins, 1970).

Kimberlites contain abundant volatiles. and vola-tile transport of the LREE and Th in CO2- and H2O-rich fluids may be a viable mechanism to explain thehigh LREE and Th contents in kimberlites (e.g.,Fesq er al., 1974). No experimental work has beendone on Th, but recent experimental studies sup-port the mobility of the REE in CO2- and H2O-rich

I

o.5o.3

Lo Eu Tb l-u Th To ScCe Sm Yb Bo Hf Co Cr

t!F

JEUJ=:<u ) lz.oa o.5fJ()z'- o.zlrJ

I

I(r)Iro

o() Lo Sm Tb Lu Th To ScCe Eu Yb Bo Hf Co Cr

ELEMENTFig. 5. The element content of carbonate-shale inclusions

relative to their corresponding kimberlite (Bala, open circles;Leonardville, solid circles; Stockdale, solid square).

230 CULLERS ET AL.: TRACE ELEMENT CONTENT OF KIMBERLITES

fluids if the mass of these fluids is large enough(Mysen, 1979; Wendlandt and Harrison, 1979). Forexample, the LREE become soluble in H2O-richfluids relative to silicate solids and melts at mantlepressure. The LREE cbuld concentrate in a largemass of H2O in contact with a moving kimberlitemelt. At lower pressure, the LREE are less solublein the H2O-rich fluid relative to silicate melt (Cull-ers €t al.,1973:' Mysen, 1979) so the LREE mightconcentrate back into the kimberlite melts as theyrise.

Thus, we believe the trace element content ofthese kimberlites can be explained by a very smallpercentage of melting of garnet peridotite with atrace element content similar to Wedepohl andMuramatsu's average peridotite followed by possi-ble volatile transport of the LREE and Th. Alter-nately, a peridotite source more enriched in theLREE and Th than that used above could partiallymelt and produce the range of trace element con-tents of the Riley County kimberlites. Crustal con-tamination or fractional crystallization is not be-lieved to be a major process in affecting these traceelement contents during the formation of the RileyCounty kimberlites.

Formation of the carbonate veins and dispersedcarbonate

Brookins (1970) distinguishes post-emplacement(secondary) veins with high Sr isotopic ratios and

low Sr cbntents similar to the limestone countryrock and pre-emplacement veins (primary) with lowSr isotopic ratios and high Sr contents more similarto igneous carbonate (i.e., carbonate formed fromcarbonate melts) in the Riley County kimberlites.The carbonate in the groundmass of the kimberlitehas intermediate values between the two extremesand this carbonate appears to be a mixture ofigneous and sedimentary carbonates.

The data for the trace elements in the carbonatematerial in this study may be considered prelimi-nary since we plan to do more detailed work on thismaterial. The data for the carbonate fractions arecompared to average primary carbonatites, replace-ment carbonatites, and limestones in Table 3. Aver-age carbonatite is very much more enriched in allelements compared to average limestone' Yet thereis quite a difference in the degree of enrichment ofthese elements in the carbonate fractions of thesekimberlites. None of the carbonate fractions appearto be all igneous carbonate or sedimentary carbon-ate. The Bala veins are very enriched in the LREEand Ba contents similar to carbonatite, but they aresignificantly depleted in heavy REE, Th, Co, Sc,and Cr contents compared to carbonatite. TheStockdale veins have lower LREE and Ba contentsand have higher Co contents than the Bala veins.Other elements are similar in veins of the twokitnberlites. This suggests these veins are not igne-ous carbonate. They could, for example, be second-

Table 3. Trace element contents of carbonate fractions in veins, inclusions, and groundmass of Riley County kimberlites compared to

average limestones and carbonatites (ppm)

L O 5 C L rI ATh

Bala vei ns 56 <0.0037

S tockda le ve ins 4 -15 .8 0 .01 -0 .08

Frcrn carbonate-I 64-778 0.1-0.75sha le i nc l us i ons

F rom Leona rdv i l l e 166 0 ,19kimber l i te- groundmass

carbonat i te-pr imary2 600 0.5

Carbonat i te-2 480 2.5repl acement

Linestone 2 7 0.2

< 0 . 0 4 < 0 . 1

0 . 3 7 - 0 . 6 6 0 . 2 1 - 0 . 4 8

-0 ! 0

5 t . d

1 ,500

130

0 . 3

< 0 . 7 0 . 2 3 1 1

0 .07 - t 2 .6 3 .4 -22 .5 0 .38 -3 .1 6 -68

1 5 ,900

300-4400

- 0-570

2,570

1 5,000

5,300

1 0 7 . 7

0 . 2 - 2 . 5

1 , 2

0 . 0?

32-79 -0-0.4 29-125

87 -0 630

77 10 48

33 25 470

0 . 1 1 1 1

I - These are inferred values based upon mass balance calculat ions knowing the f ract ion of carbonate re lat ive to non-carbonatefract ion and knowinq the concentr i t ion of theelements before and af te i d issolut ion in 0,3M acet ic acid. The quant i ty oft raceelements lCach-ea f rom the k imber l i te could not be checked direct ly , but the t race element contents of a var iety oflayer s i l icates did not change before and af ter using th ' is technique'

values given are f rom many sources but the main ones are Ri is ler and Lange (1972), Armbrustmacher (1979) ' cul lers2 - Average

and Graf ( 1 % 1 ) .


ary veins in which the LREE and Ba were moremobil than the other elements so the LREE and Babecame enriched in the carbonate veins over aver-age limestone concentrations. Alternatively, theseveins might be hydrothermal, but not enough isknown about the behavior of these elements inhydrothermal veins to evaluate this possibility(Cullers and Graf, 1982b). Determination of the Srisotopic compositions of these veins should distin-guish between these two possibilities.

Even the carbonate material from the carbonate-rich shale inclusions is not clearly sedimentary. TheBa, Th, Hf, Ta, and Sc contents of the carbonate inthese inclusions are similar to limestones, but theLREE, Co, and Cr contents are quite high com-pared to those of limestones.

Finally, the carbonate material from the ground_mass of the kimberlite is also neither clearly igneousor clearly sedimentary, but it must result from acomplex mixture of processes. The REE, Ba, Co,and Cr contents are similar to carbonatite, but theTh and Sc contents are similar to average lime-stone. This is consistent with Brookins' (1970)previously discussed interpretation that the Sr con_tents and Sr isotopic compositions of this materialresults from mixed primary and secondary process-es. Also if the carbonate material from the ground-mass is not clearly igneous, it should be consideredas a contaminant for the trace elements in thekimberlite. The Co content of the carbonategroundmass is about the same as that of the kimber-lite although all other elements are variously deplet-ed in the groundmass relative to kimberlite. Thus,the groundmass carbonate like the carbonate-shalexenoliths should tend to dilute slightly the kimber-lite in most trace elements.

Note added in proof:The Randolph No. 2 Kimberlite has recently been analyzed,

and it has a similar trace element content to the other non-micaceous kimberlites. Thus, it presumably formed in a similarfashion as the non-micaceous kimberlites. Samples at Winklercrater have not been analyzed since they are highly altered.

Acknowledgements

We thank the Kansas Geological Survey and most especiallyPieter Berendson for the support of most of this research. TheBureau ofGeneral Research, Kansas State University, providedadditional support. We thank Tom Grifrn for help with theneutron activation, Bob Barnett for drafting the figures, andRen6e Hambleton for typing the manuscript. We also thank theNuclear Engineering Department at Kansas State University forirradiating the samples and the use oftheir counting equipment.Douglas Smith and Bill Mansker provided useful reviews, andIan Duncan and Michael Holdaway are thanked for their editori_al corrections.

ReferencesArmbrustmacher, T. J. (1979) Replacement and primary mag-

matic carbonatites from the Wet Mountains area, Fremont andCuster Counties, Colorado. Economic Geology, 74, 888-901.

Baker, B. H., Goles, G. G., Leeman, W. P., and Lindstrom, M.(1977) Geochemistry and petrogenesis of a basalt-benmoreite-trachyte suite from the southern part of the Gregory Rift,Kenya. Contributions to Mineralogy and Petrology, 64, 303-332.

Brookins, D. G. (1967) The strontium geochemistry of carbon-ates in kimberlites and limestones from Riley County, Kansas.Earth and Planetary Science Letters, 2,235-240.

Brookins, D. G. (1970a) Kimberlite at Winkler Crater, Kansas.Geological Society of American Bulletin, 81, 541-546.

Brookins, D. G. (1970a) The kimberlites of Riley County,Kansas. Kansas Geological Survey Bulletin, 20f., t-32.

Brookins, D. G. and Naeser, C: S. (1971) Age of emplacement ofRiley County, Kansas kimberlites and possible minimum agefor the Dakota Sandstone. Geological Society of AmericaBulletin, E2, 1723-1726.

Cox, K. G., Bell, J. D., and Pankhurst, R. J. (1979) Theinterpretation of igneous rocks. George Allen and Unwin,Ltd., London, 450 p.

Cullers R. L. and Graf, J. (l9E2a) Rare earth elements in igneousrocks of the continental crust: predominately basic and ultra-basic rocks. In P. Henderson, Ed., Rare-earth Geochemistry,in press, Elsevier.

Cullers, R. L. and Graf, J. (1982b) Rare earth elements inigneous rocks of the continental crust: intermediate and silicicrocks; ore petrogenesis. In P. Henderson, Ed., Rare-earthGeochemistry, in press, Elsevter.

Cullers, R. L. and Medaris, L. G. (1977) Rare earth elements incarbonatite and co-genetic alkaline rocks: Examples fromSeabrook Lake and Callander Bay, Ontario. Contributions toMineralogy and Petrology, 65, 143-153.

Cullers, R. L., Medaris, L. G., and Haskin, L. A. (1973)Experimental studies ofthe distribution ofrare earths as traceelements among silicate minerals and liquids and water. Geo-chimica Cosmochimica Acta. 37. 1499-1512.

Cullers, R. L., Chaudhuri, S., Kilbane, N., and Koch, R. J.(1979) The rare earth contents in different size fractions andwhole rocks from Pennsylvanian and Permian platform sedi-ments from the mid-continent of the U. S. A. GeochimicaCosmochimica Acta, 43. 1285-1302.

Duke, J. M. (1976) Distribution of the period four transitionelements among olivine, calcic clinopyroxene, and mafic sili-cate liquid: experimental results. Journal gf petrology, 17,499-52t.

Eggler, D. H. and Wendlandt, R. F. (1979) Experimental studieson the relationship between kimberlite magmas and partialmelting of peridotite. In F. R. Boyd and H. O. A. Meyer, Eds.,Kimberlites, Diatremes, and Diamonds: Their Geology, pe-trology and Geochemistry, 330-338.

Fesq, H. W., Kable, E. J. D., and Gurney, J. J. (1974) Aspects ofthe geochemistry of kimberlites from the Premier mine, andother selected South African occurrences with particular refer-ence to the rare earth elements: Physics and Chemistry bftheEaxth.9. 6E7J07.

Frey, F. A., Bryan, W. B., and Thompson, G. (1974) Atlanticocean floor: Geochemistry and petrology ofbasalts from legs 2and 3 of the Deep-sea Drilling Project. Journal of GeophysicalResearch, 79, 5507 -5527 .

232

Gast, P. W. (l%8) Trace-element fractionation and the origin oftholeiitic and alkaline magma types. Geochimica et Cosmochi-mica Acta. 32. 1057-1086.

Gordon, G. E., Randle, K., Goles, G. G., Corliss, J. B., Beeson,M. H., Oxley, S. S., and Gram, O. E. (1968) Instrumentalactivation analysis of standard rocks with high resolution X-ray detectors. Geochimica et Cosmochimica Acta, 32' 369-396.

Grifrn, W. L. and Murthy, V. R. (l%9) Distribution of K' Rb,Sr. and Ba in some minerals relevant to basalt genesis.

Geochimica et Cosmochimica Acta, 33, l3E9-1414.Gunn, B. M. (1971) Trace element partitioning during olivine

fractionation of Hawaiian basalt. Chemical Geology, 8' l-13.Hart, S. R. and Brooks, C. (1974) Clinopyroxene-matrix parti-

tioning of K, Rb, Cs, Sr, and Ba. Geochimica et Cosmochi-mica Acta, 38, 1799-1806.

Haskin, L. A. (19E2) Petrogenetic modelling of the rare-earths.In P. Henderson, Ed., Rare-earth Geochemistry, in press'

Elsevier, Amsterdam.Haskin, L. A. and Korotev, R. (1977) Test of a model for trace

element partition during closed-system solidification of a sili-cate liquid. Geochimica et Cosmochimica Acta, 41,921-939.

Haskin, L. A., Frey, F. A., Schmitt, R. A., and Smith' R. H.(1966) Meteioritic, solar, and terrestrial rare-earth distribu-tions. Physics and Chemistry of the Earth, 7, 167-321.

Henderson, P. and Dale, I. M. (1969-70) The partitioning ofselected transition element ions between olivine and ground-

mass of oceanic basalts. Chemical Geology, 5,267-271.Irving, T. and Frey, F. (1977) Distribution of trace elements

between garnet megacrysts and host volcanic liquids of kim-berlitic to rhyolitic coinposition. Geochimica et CosmochimicaActa. 42.771-:788.

Jacobs, J. W., Korotev, R. L., Blanchard, D. P., and Haskin, L.A. (197'?-) A well tested procedure for instrumental neutronactivation analysis of silicate rocks and minerals. Journal ofRadioanalytical Chemistry, 40, 321.

Kay, R. W. and Gast, P. W. (1973) The rare-earth content andorigin of alkali-rich basalts. Journal of Geology, 81, 653-682.

Leeman, W. P. and Scheidegger (1977) Olivine-liquid distribu-tion coefficients and a test for crystal-liquid equilibrium. Earthand Planetary Science Letters, 35,247:257.

Mitchell, R. H. and Brunfelt, A. O. (1974) Rare earth elementgeochemistry of kimberlite. Physics and Chemistry of theEarth, 9, 671-686.

Muir, E. D., Tilley, C. E. and Scoon, J. H. (1964) Basalts fromthe northern part of the rift zone of the mid-Atlantic Ridge.Journal of Petrology, 5, 409-434.

Mysen, B. O. (1979) Trace element partitioning between garnetperidotite minerals and water-rich vapor: Experimental datafrom 5 to 30 kb. American Mineralogist, U,274-287.

Nixon, P. H., Rogers, N. W., Gibson, I. L., and Grey, A. (1981)

Depleted and fertile mantle xenoliths from southern Africankimberlites, In F. A. Donath, Ed., Annual Reviews of Earthand Planetary Science, p. 285-309. Annual Reviews, Pirl,oAlto. California, U.S.A.

Onuma, H., Higuichi, H. H., Wakita, H., and Nagasawa, H.(1968) Trace element partition between two pyroxenes andhost lava. Earth and Planetary Science Letters, 5,4'7-51.

Paster, T. P., Schauwecker, D. S. and Haskin, L. A' (1974) Thebehavior of some txace elements during solidification of the

Skaergaard layered series. Geochimica et CosmochimicaActa,38, 1549-1577.


Philpotts, J. A. and Schnetzler, C. C. (1970) Phenocryst-matrixpartition coefficients for K, Rb, Sr, and Ba, with applicationsto anorthosite and basalt genesis. Geochimica et Cosmochi-

mica Acta, 34,307-322.R<isler, H. J. and Lange, H. (1972) Geochemical Tables. Else-

vier, Amsterdam, New York, 468 P.Schreiber, H. D. and Haskin, L. A. (1976) Chromium in basalts:

Experimental determination of redox states and partittioning

among synthetic silicate phases. Proceeding of the 7th Lunar

Science Conference, Geochimica et Cosmochimica Acta Sup-plement, 7, 122l-1259.

Shaw, D. M. (1970) Trace element fractionation during anatexis.

Geochimica et Cosmochimica Acta, 34,237-243.Wedepohl, K. H. and Muramatsu, Y. (1979) The chemical

composition of kimberlites compared with the average compo-

sition of three basaltic magma types. In F. R. Boyd and H. O.

A. Meyer, Eds., Kimberlites, Diatremes, and Diamonds:Their Geology, Petrology and Geochemistry, 300-312.

Manuscript received, JulY 30' 19EI;accepted for publication, October 21, 1981.

Appendix 1. Brief petrographic descriptions of therocks analyzed in this studY

Bala Kimberlite.This porphyritic rock contains upto 30 percent phenocrysts (up to 2 mm) of olivineand pyroxene (altered to serpentine, calcite, andmagnetite) and a few phenocrysts of pyrope. Thegroundmass contains fine-grained serpentine andlesser calcite, opaque oxides, and hematite. Theveins are composed of nearly all calcite. There arefew xenoliths in the rocks analyzed.Randotph No. 1. Kimberlite. This is a porphyriticrock containing 25 percent phenocrysts (up to 2mm) of predominately olivine and pyroxene alteredto serpentine, calcite, hematite, and black oxides.'The phenocrysts occur in a very fine-grainedgroundmass of lizardite, calcite, and iron oxyhy-droxides. There are few xenoliths in the rocksanalyzed.Leonardville Kimberlite. This porphyritic rock con-tains up to 30 percent phenocrysts (up to 5 mm) ofolivine and pyroxene (altered to calcite, serpentine,or magnetite) and a few phenocrysts of pyrope,phlogopite, and magnetite. The fine-grained matrixis composed of serpentine of varied grain size withlesser phlogopite, opaque oxides, calcite' and gar-

net: There are up to 10 percent xenoliths of shalewith finely disseminated calcite'Stockdale Kimberlite. This porphyritic rock con-tains about 20 percent phenocrysts (up to 2 mm) ofolivine (altered to serpentine), pyroxene (altered toserpentine), and phlogopite in a fine-grained matrix

of lizardite with minor calcite and black oxides.About 20 percent of the rock is composed of alteredxenoliths (dunite, eclogite, phyllite, shales) of var-ied size and shape. The rock is brecciated and

riddled with calcite veinlets. The veins analyzed arecomposed primarily of calcite although the samplethat has 4.0 ppm La has about 30 percent serpentinein addition to calcite.


Appendix 2. D.C. used in the modeling during the melting of peridotite to produce kimberlite melt.l

La Ce Sm Eu yb Lu Ba T h C o \ e l n

0 l i v i ne 0.001 0 .001 0 .002 0 .002 0 .002

0rthopyroxene 0.002 0.003 0.010 0.013 0.05

Cl inopyroxene 0 .10 O.7Z 0 .26 0 .30 0 .Zg

Garnet 0 .02 0 .02 O.ZZ 0 .32 4 .0

0 .002

0 . 0 6

0. 28

4 . 5

0. 001

0.001

0. 001

0.002

0 . 1 0 . 0 1 *

0 .03 0 .01*

0 .07 0 .013

0.03

3 0 . 3 2

2 7 . 2 1 . 8

L J f ,

0 . 3 1 0 2

I - Many sources were surveyed to obtain these D.c. 's , and the lowest values avai lable tended to be used for the model ing.The re fe rences I i s t ed be low had a t l eas t one va lu i c l ose i o t ne - "a i r e i - o i - t he .above D .c . ' s : pas te r e t a | . , 1974 ;Ph i l po t t s and Schne tz te r , J ! lQ ; i r ey e t a1 . , 1974 ; nenae r j on ana 'O i i i , iSOs_zO; Mu i r e r a l . , 1964 ; Gunn , I gz l ;onuma et aI . ' 1968; Duke, 1976; Leeman and Scheidegger, 1977; Hasi in ina-korotev, 1977; Schreiber and Haskin, 1976iHa r t and B rooks , 1974 ; G r i f f i n and Mur thy , 1969 ; B i l e r ' e i " l : ,

i t t i ; i i v i ng ana i " i v , i g l i i co i - l t - i r . , r s zs .

* - These va lues a re no t es tab l i shed . The D .c . ' s shou ld be cons ide rab l y l ess t han one .

the trace element content and petrogenesis of … mineralogist, volume 67, pages 223-233, 1982 the...

Documents