Geochemical and strontium isotopic characteristics of parental ...

Contrib Mineral Petrol (1986) 94:1-11

Contributions to Mineralogy and Petrology 9 Springer-Verlag1986

Geochemical and strontium isotopic characteristics of parental Aleutian Arc magmas: evidence from the basaltic lavas of Atka James D. Myers 1, Bruce D. Marsh 2, and A. Krishna Sinha 3 1 Department of Geology and Geophysics, University of Wyoming, Laramie, WY 82071, USA 2 Department of Earth and Planetary Sciences, The Johns Hopkins University, Baltimore, MD 21218, USA 3 Department of Geological Sciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA

Abstract. Eighteen flows from a basal stratigraphic sequence on the Aleutian Island of Atka were analyzed for major elements, trace elements and initial 87Sr/S6Sr ratios. Petrographically, these lavas contain abundant plagioclase (24-45%) and lesser amounts of olivine ( < 7%), magnetite and clinopyroxene phenocrysts. Compositionally, the lavas are high-alumina ( ~ 2 0 w t % ) basalts (48-51 wt% SiO2) with low TiO2 ( < 1 % ) and MgO ( < 5 % ) . Within the section, compositional variations for all major elements are quite small. While MgO content correlates with olivine phenocryst contents, no such relationship exists between the other oxides and phenocryst content. These lavas are characterized by 8-10ppm Rb, high Sr (610-669ppm), 308-348 ppm Ba and very constant Zr (23 29 ppm) and Sc (23-29 ppm) abundances. Ni and Cr display extremely large compositional ranges, 12-118 ppm and 12-213 ppm, respectively. No correlation exists between trace element concentrations and phenocryst contents. Strontium isotopic ratios show a small but significant range (0.70314-0.70345) and are slightly elevated with respect to typical MORB. No systematic ,correlation between stratigraphic position and petrography or geochemistry is evident. REE abundances measured on six samples are LREE enriched ((La/ Yb)N = 2.20-2.81) and display similar chondrite normalized patterns. One sample has a slight positive Eu anomaly but the other lavas do not. Compared to other Aleutian basalts of similar silica content, these lavas are less LREE enriched and have lower overall abundances. The geochemical characteristics of the,se basalts suggest they represent true liquid compositions despite their highly porphyritic nature. Published phase relations indicate fractionation of a more MgO-rich magma could not have produced these lavas. The high A1203 and low MgO and compatible element abundances suggest a predominantly oceanic crustal source for parental high-alumina basalts.

Introduction

Despite considerable theoretical and petrologic discussion, the source of arc magmas is still uncertain. Four principal Offprint requests Jto :

J.D. Myers

sources have been suggested: 1. the subducted plate (4a sediment component); 2. the mantle wedge overlying the plate; 3. various combinations of these two components; and 4. a plumpudding mantle. Different geochemical data have been cited as evidence for or against one or another of these sources. Ultimately, identification of magmatic source is dependent on the presumed composition of the parental magma. Attempts to define Aleutian parental magma have generally fallen into one of two categories. One approach assumes that lavas with unusually primitive characteristics (i.e. high MgO, Ni, Cr and low SiOz and A1203) are parental (Perfit 1978; Reid and Nye 1981; Kay et al. 1982; Conrad and Kay 1984) while the other suggests the volumetrically dominant marie lava of the arc most closely approximates parental liquids (Marsh 1976, 1982). The first approach to parental magma identification explicitly equates primitive and parental thereby implying parental magmas from all tectonic settings have similar geochemical characteristics. Thus, parental magmas can only be produced by partial melting of ultramafic material and, by definition, the source of Aleutian arc magmas is fixed as the lithospheric wedge. On the other hand, the second approach suggests Aleutian parental liquids are approximated by high-alumina basalts (Marsh 1976, 1982). If these lavas are parental, the subducted oceanic crust must be the major source of Aleutian arc magmas (Brophy and Marsh in press). Clearly, determining magma source rests critically on the correct identification of parental magma composition. (These two approaches are not mutually exclusive. In the former approach, high-alumina basalt is related through differentiation and phase equilibria to the more primitive lavas.) A major objection to high-alumina basalts as parental magmas is their highly porphyritic nature (Perfit et al. 1980a; Kay et al. 1982). In particular, the high AlzO3 contents of these lavas are presumed to result from some form of crystal fractionation/accumulation. Since changes in magma composition require significant differential movement of crystals relative to the magma from which they crystallized, the mere presence of a large proportion of phenocrysts is not evidence for fractionation/accumulation. If no such movement has occurred, the crystal-magma suspension will have remained isochemical and crystallization merely transferred mass from the liquid to the solid state. Provided a representative volume is sampled, whole rock chem-

ATKA ISLAND

Cape P~176

NorthCape

~

I~r:~91~5 i ~000.~'~

~.

~ _ -V'/

FIow

I~

(f

f ,A,/~"K) "~O~~egnak )l-oinr Nazan Oay

.~,~ (

)" -

4,

-

-

,

,~? Bu/dir Seg~leamisopochno i ~'%_ Kisl(t~o~ J V..a.~

f~'~OBA^

UnimaK ~t'~r~3g, ~' tP V=~,

..... V Umna~/~'~.~'~-~Ix~v ,.Girtekoi n t ./Atka ? e g u a ~ o~ o.~t~ 150mi I

%

ANOREANOFISLANDS ~O~3S•

50 o

150km ,

Fig. 1. Map showing the location of Atka as well as other important Aleutian volcanic centers. Insert map is of the northern portion of the island and shows the location of the stratigraphic section studied.

ical analysis will approximate magmatic compositions (Myers et al. 1984). Consequently, evaluating the significance of chemical analyses of porphyritic lavas requires determining if differential movement has occurred between phenocrysts a n d host magma. To make such a determination, extensive petrographic, geochemical and isotopic data are required from a large n u m b e r of basaltic lavas. Since the n u m b e r of marie lava analyses from individual Aleutian volcanic centers is very small, we have selected a stratigraphic sequence of basaltic lavas from the volcanic center of Atka for a detailed petrographic, geochemical and strontium isotopic study.

General geology Atka is a large volcanic center located in the central Aleutians (Fig. 1). Recent volcanism, which is confined to the northern part of the island, has spanned only the last few million years (Marsh 1980, 1982). Marsh (1980) has estimated the total volume of erupted material at about 200 km 3 and identified at least fourteen vents associated with central and satellite cones. Cessation of activity at central cones is signaled by the eruption of a small amount of dacitic material (Marsh 1980) and the simultaneous shutdown of the more basaltic satellite vents. The northernmost volcano, Korovin, is presently active. Although activity shifts between different volcanic edifices, overall it has been confined to a relatively small area (10-20 km in diameter) for at least several million years. Each volcanic cone consists of a thick, basaltic shield topped by a composite cone. The basal shields are composed of numerous, thin (< 1 m) flows that can often be traced for ~ 2 kin. The overall

character of the shields is best illustrated in sea-cliffs (particularly along the northern coast) as well as in the numerous stream valleys that cut the volcanic pile (Fig. 2). The stratigraphic section chosen for study forms the northern wall of such a valley (Fig. 2) and is located on the eastern side of the island (Fig. 1). At the locality sampled, the section is approximately 25 m thick and composed of numerous, thin basaltic flows that dip gently toward the coast. Eighteen flows out of a total of 22 were sampled. Each sample was taken from the center of the flow and all weathered surfaces removed. Compositionally, Atka lavas range from basalt to dacite; however, the most abundant lava is basaltic in character (Marsh 1980, 1982; Myers et al. 1985). For most major elements as well as Rb, Sr and Ba, the volcanic suite defines smooth compositional trends. With increasing silica, CaO, TiOz, MgO and Sr decrease while Na20, K20, Rb and Ba increase. Initial strontium isotopic ratios vary from 0.70320 to 0.70345, one of the smallest ranges measured in the Aleutians (Myers et al. 1985). When plotted against silica, 8VSr/86Sr ratios define a narrow horizontal band.

Petrography The lavas studied are highly porphyritic (25-50%) with phenocrysts of plagioclase, olivine, magnetite and rare clinopyroxene. In all lavas, plagioclase (24-45%) comprises well over half the total phenocryst population (Table 1). Olivine, the next most abundant phenocryst, forms less than 7 % of the lavas whereas magnetite rarely exceeds 1%. By examining a large number of Atka basalts, Marsh (1981) has determined the following crystallization sequence: plagioclase, olivine and magnetite (15% solidified), and clinopyroxene (35% crystallized).

Fig. 2. Sequence of basaltic lavas on the northeast coast of Atka used in this study (midsection of photo). Note the character of the volcanic center's basal shield

Table 1. Modes of Atka basaltic lavas Sample no.

Pl[agioclase

Olivine

Magnetite

Phenocrysts

AT-80 (top) AT-79 AT-78 AT-77 AT-76 AT-75 AT-74 AT-73 AT-72 AT-71 AT-70 a AT-69 AT-68 AT-67 AT-66 AT-65 AT-64 AT-63 (bottom)

32.1 44.1 40.9 45.1 34.6 37.2 36.6 34.1 40.0 34.9 32.4 44.4 41.1 43.1 32.5 28.8 24.3 35.6

4.2 1.9 2.0 4.8 0.9 1.9 1.7 2.8 2.1 3.7 2.7 2.0 2.4 3.5 1.7 2.0 0.8 1.3

0.8 0.9 tr 0.6 tr 0.7 0.4 0.2 0.7 0.7 0.4 1.3 0.3 0.4 0.4 0.1 0.1 0.7

37.1 46.9 42.9 50.5 35.5 39.8 38.7 37.1 42.8 39.3 35.8 47.7 43.8 47.0 34.6 30.9 25.2 37.6

a cpx-0.3 When phenocryst abundances are plotted against stratigraphic height, no simple correlation is evident (Fig. 3). The sequence can, however, be divided into seven zones separated by discontinuities in plagioclase phenocryst abundances. Within each zone, plagioclase contents are generally constant or increase slightly upward. The boundaries of these zones do not always correlate with variations in olivine and/or magnetite abundances. The lowermost zone (Zone 1) is composed of only the basal flow (AT-63). Zone 2 (AT-64 through AT-66) is the least porphyritic of the sequence ( < 3 5 % crystals) with a steady upward increase in plagioclase abundance and low olivine ( < 2 % ) and magnetite (< 0.5%) contents. Zones 2 and 3 are separated by an abrupt increase in both plagioclase and olivine abundances. At the zone boundary, plagioclase increases from 32.5 to 43.1% and olivine from 1.7 to 3.5% (Table 1). Within this zone, plagioclase abundance is nearly constant, but that of olivine decreases slightly upward (Fig. 3). A decrease in plagioclase and magnetite contents marks the transition from Zone 3 to Zone 4. While plagioclase increases upward in this zone, olivine decreases slightly. Zone 5, the largest of the se-

quence, contains four flows of nearly constant plagioclase content (34-37%), but decreasing olivine (2.8 to 0.9%) and increasing magnetite abundances (Fig. 3). An abrupt change in both plagioclase and olivine content mark the upper boundary of this zone. Going from Zones 5 to 6, plagioclase increases from 34.6 to 45.1 and olivine from 0.9 to 4.8 (Table 1). While plagioclase abundance (40.9-45.1%) is constant throughout this zone, olivine decreases substantially. The seventh zone is composed of only the uppermost flow and is characterized by higher olivine and lower plagioclase contents than the zone below.

Analytical procedures Using the procedure of Feigenson and Carr (1985), major and trace elements for all eighteen lavas were determined by DCP-AES at Rutgers University (Table 2). For the analyzed elements, this procedure has a precision of 2% or better. Four samples (AT-64, AT-73, AT-74, AT-76) from this sequence previously analyzed for major elements by XRF (Myers et al. 1985) agree within analytical error. Since compositional variations within this section are small and may be masked by minor interlab variations, we report the new major element analyses. Rb analyses were done by XRF at VPI using the procedure of Norrish and Chappell (1977). Six samples were also analyzed, again using the procedure of Feigenson and Carr (1985), for rare earth elements (Table 3). Fifty to ninety milligrams of sample were dissolved in H F and HNO3 for strontium isotopic analysis. Sr was separated using 3 ml exchange columns and a double elution procedure. Total Sr blanks varied between 0.2 and 0.7 ng. The accuracy of strontium isotopic analyses was monitored by running E and A standard; analytical precision was checked with replicate analyses (Table 2). Due to the extremely small isotopic range measured (0.05%), a large number of replicates were analyzed. Agreement between replicates (Table 2) is better than 0.01% or nearly one-fifth the total 87Sr/86Sr range measured. All isotopic ratios have been normalized to a 86Sr/SSSr ratio of 0.1194 and corrected to an E and A of 0.70800. In addition, strontium isotopic data were reported earlier for four lavas (AT-64, AT-73, AT-74 and AT-76) by Myers et al. (1985).

Analytical results Major elements These lavas are basaltic (48.0-51 w t % SiO2) with high A1203 ( ~ 2 0 % ) , low TiO2 ( < 1 % ) , low M g O ( < 5 % ) , a n d

4 Table 2. Major and trace element and strontium isotopic analyses of the Atka basalts AT-63

AT-64

AT-65

AT-66

AT-67

AT-68

AT-69

AT-70

AT-71

AT-72

SiO2 TiO2 A12Oa FeO T MnO MgO CaO Na20 K20 P205

48.75 0.86 20.31 8.80 0.18 3.33 9.69 3.10 0.70 0.17

49.56 0.87 20.82 8.86 0.18 3.33 10.27 3.20 0.68 0.16

49.37 0.84 20.74 8.65 0.18 3.26 10.20 3.19 0.69 0.17

50.13 0.91 20.93 9.18 0.18 3.58 10.42 3.09 0.65 0.13

49.10 0.83 20.90 8.64 0.18 3.37 9.99 3.18 0.69 0.14

50.87 0.90 20.72 9.38 0.19 3.57 10.12 3.31 0.72 0.16

49.82 0.86 20.40 9.15 0.19 3.54 9.94 3.26 0.71 0.16

49.66 0.92 19.71 9.79 0.20 4.36 10.07 3.14 0.67 0.15

49.07 0.88 20.13 9.54 0.19 4.27 9.80 3.19 0.68 0.15

50.25 0.92 20.67 9.19 0.19 3.78 10.29 3.23 0.71 0.16

Total

95.89

97.93

97.29

99.20

97.02

99.94

98.03

98.67

97.90

99.39

Rb a Ba Sr V Cr Ni Zr Sc Cu

-

324 661 245 94 55 49 24 127

8 320 664 261 93 54 49 23 123

9 310 645 267 12 12 48 24 99

9 326 655 272 59 38 49 23 128

9(9) 317 656 285 16 13 49 25 125

9 318 649 277 102 59 48 24 103

9 308 623 282 18 15 47 26 109

8 310 637 253 99 57 48 26 131

9 316 654 259 12 14 52 24 138

0.70337 • b

0.70335 •

0.70329 _+7

0.70338 _+16

0.70324 _+10

0.70331 •

0.70334 •

0.70340 _+12

0.70323 _+8

0.70335 •

0.70339 _+8

0.70334 _+8

0.70337 •

0.70345 •

325 638 264 213 118 48 23 13~

87Sr/S6Sr -

9b

0.70341 • 0.70340 _+11

AT-73

AT-74

AT-75

AT-76

AT-77

AT-78

AT-79

AT-80

avg.

SiOz TiO2 A1203 FeO T MnO MgO CaO Na20 K20 P205

49.08 0.89 20.36 8.81 0.18 3.39 9.93 3.25 0.71 0.16

49.70 0.87 20.84 8.86 0.18 3.39 9.94 3.24 0.72 0.17

49.62 0.87 21.03 8.94 0.18 3.43 10.01 3.15 0.64 0.18

50.42 0.92 20.59 9.17 0.18 3.49 10.27 3.27 0.70 0.13

50.41 0.89 20.50 9.18 0.19 3.59 10.29 3.19 0.71 0.15

50.09 0.87 20.69 8.86 0.18 3.66 10.46 3.20 0.71 0.15

49.99 0.87 20.87 8.64 0.17 3.44 10.25 3.30 0.75 0.15

49.74 0.98 19.38 10.15 0.19 4.56 10.06 3.22 0.74 0.15

49.76 0.89 20.53 9.10 o.18 3.63 10.11 3.21 0.70 0.16

Total

96.75

97.91

98.05

99.15

99.10

98.87

98.43

99.18

Rb" Ba Sr V Cr Ni Zr Sc Cu

9b 317 645 268 90 54 50 23 92

JO b 316 658 255 63 38 50 24 137

9 348 669 262 98 59 51 24 127

8b 321 661 277 13 15 50 25 114

9(9) 310 640 273 34 26 49 23 96

9(9) 296 653 263 128 72 46 24 95

10 312 665 271 39 33 49 25 120

303 610 333 157 87 46 29 145

0.70331 ~ 10 u

0.70321 _+5

0.70319 • b

0.70328 _+ 10

0.70319 •

0.70314 _+10

0.70335 _+7 b

0.70318 •

0.70320

0.70330 •

878r/86Sr 0.70331 _+6 b

0.70337 _+7 8

_+7 8

-

9 317 649 270 74 46 49 24 119

(0.56) (0.04) (0.44) (0.41) (O.Ol) (0.38) (o.21) (0.06) (0.03) (0.01)

(11) (15) (19) (56) (29) (2) (1) (16)

0.70331 _+9

0.70323 •

a X R F analysis at VPI b Myers et al. 1985 relatively high K 2 0 (Table 2). A s a whole, the sequence is m a r k e d by a relatively small degree o f c o m p o s i t i o n a l variability. F o r example, SiO2 varies by o n l y 4 % and A1203 by 8 % . M a x i m u m a n d m i n i m u m values for the less a b u n -

d a n t oxides (i.e. TiO2, F e O v, M g O , C a O , N a 2 0 , K 2 0 ) differ by less t h a n 1.5 w t % . W h i l e the slightly l o w totals o f several analyses m a y suggest alteration, all the a n a l y z e d samples are very fresh in thin section a n d great care was

TOp

!

SO-

Q

9 7e

P

9

i

__

9

7'

9

(,

9

i'

q~

9 76

9

9 9

74

q

9

Q.

.

.

9

9

.

.

.

I

9

.

.I

70

9 9

q,

I

68

9

9

9

9

q 64~-

I

9

~)

615

I

9

3

i! .

9

20

"

/

I, b

9

__

-2BOTTOM

5

Q

30

,r

50

O

2

4

I.

0.5

olivine

Nag

9

I.O

.

1.5

.

.

20

30

9

40

50

~ phenocryst

magnetite

TOP 8O-

T

9

Fig. 3. Phenocryst modes (volume %) plotted against stratigraphic position. Numbers on left refer to sample numbers. Section is approximately 25 m thick. Horizontal dashed lines delineate zones discussed in text

9

7

i 78-

i

76-

__ (

g 749 .o

7z

9 '

o

~

9

9

t

5

-

9

9 e

70r-

I 6~

9

9

,

9

9

,4 I

i

9

-4

4

Fig. 4. Major elements as a

9

9

-4

,

9

9

-4

3

I

I

9

-t I

2

9 9

9

i I

66-

! 9

9

64BOTTOM

50

0.8

Si02

0.9 1.0 YiO 2

]9

20

2~

8

A~O 3

9 I0 FeOT

4 MgO

5

9

I0 CaO

Table 3. Rare earth element abundances (ppm) AT-63 La 6.2 Ce 15.5 Nd 10.5 Sm 2.74 Eu 1.04 Gd 2.99 Dy 3.20 Er 1.99 Yb 1.88 Y 18.5

AT-67

AT-70

AT-73

AT-76

AT-80

5.4 14.2 8.1 2.54 0.99 2.74 2.67 1.83 1.57 1,6.4

6.5 16.8 9.9 3.06 1.07 3.34 3.04 1.79 1.70 18.8

6.4 15.7 10.8 2.78 1.06 3.17 3.15 2.07 1.82 18.7

5.0 12.9 8.2 2.18 1.01 2.71 2.66 1.58 1.48 15.5

6.4 15.8 11.4 2.82 1.05 3.24 3.04 1.90 1.52 18.2

In

4 NazO

function of stratigraphic position. All oxides except TiOz and K20 are plotted at the same horizontal scale. Zone boundaries are from Fig. 3. For each oxide, error bars are shown except when they are smaller than the plotted symbol. See text for discussion

0,60,8 K20

taken to remove any weathered surfaces before sample preparation. The low F e O / ( F e O + F e 2 0 3 ) ratios (average o f 0.615 for 16 samples) o f other A t k a basaltic andesites ( M a r s h 1986, unpublished data) suggest these low totals p r o b a b l y result from reporting all iron as F e e . Recalculating the analyses assuming a similar ferric-ferrous ratio yields totals much closer to 100. As noted by M a r s h (1982), these lower F e O / ( F e O + F % Q ) ratios p r o b a b l y reflect a slightly m o r e oxidized state than n o r m a l l y assumed for mafic magmas. A l t h o u g h m a j o r oxides do vary slightly, there is no simple correlation with stratigraphic position (Fig. 4). In the lower two thirds of the section, SiO2 varies from flow to

,oo

. . . . . . .

,oo r

;

rc:o:,

, -

oc.oo=: ....

I0

] ""

9 o

_ ....... -~cc=ool

9

,"

20

OOoo

o

j

&

i f

o

" o

o

"

"

0--''-- ~ r

~-----0--

_.--. . . . . . .

('C C = 0 . 0 5

~"

~'eO

o

O.

-- - - - o=

o'~ o

o.o ~ oo o

Oo 0

o

CC= -0.01

1

P

60

I

o

I

CC=O.Oo)

o o "*

--

. . . . . . . . ~

-~-g 0

OUo -

...o

o

. . . .

z

2o

9

OoO

9

I

.

9 ~L

le I

I

80

/ 4 /

2 _ ~ _ _ _._ _~. _ -_~_:. __-~_ _._-., % "'q ,

C=0.21 . . . . . .

I

I00

E

It)

I

~ ~ -

I

I

,

i

,

/ 0

0

7

0

48

o

0

I0

210

3

I0

I

40

0

CC = O. 85 50

volume % pl0gioclase

0

O0

Fig. 5. SiO2, A1203, CaO and Sr contents versus plagioclase abun-

dance. No significant correlation exists between plagioclase abundance and composition. Symbols: 9 - this study; o - Marsh, unpublished data; dashed line - linear regression for basaltic sequence; solid line linear regression for entire data set. Each line is labelled with its correlation coefficient (cc)

1 o

flow ( ~ 1 wt%) but remains fairly constant in the upper third. Starting from the base of the section A1203 increases gradually to 21 wt%, passes through a minimum near the center of the section then increases again to a constant, high value (>20.5 wt%) until the top of the section. In the uppermost flow, A1203 abruptly decreases to 19.4%. Total iron and MgO behave in a similar fashion throughout the sequence. In the lower third of the section, both oxides vary slightly from flow to flow. After reaching maximums a little below the center of the section, FeO T and MgO decrease slightly and remain constant until the top of the section. In the second flow from the top, both oxides decrease slightly. A significant increase in FeO T and MgO abundance occurs between this and the topmost flow (Fig. 4). The remaining oxides (i.e. TiO2, CaO, Na20, K20) are quite constant throughout the sequence. While the major element variations are apparently real (i.e. outside experimental error), they are subtle. Compositional breaks that mark some oxides in the middle and at the top of the section do not extend to other elements. Clearly, the entire section is mafic in composition and the variations in individual oxides are minor. The degree of correlation between the plagioclase zones and compositional breaks is quite variable (Fig. 4). The best agreement between these zones and composition occurs between Zones 3 and 4, and 6 and 7. Except for A1203, the compositional breaks at these boundaries are not, however, in the abundances of the major oxides of plagioclase (i.e. S i O / a n d CaO) but in MgO and FeO T contents. This lack of correlation between composition and plagioclase abundance is also apparent when SiO2, A1203 and CaO are plotted against plagioclase percentage (Fig. 5). Extremely poor correlations (r