9. variation of basalt phenocryst mineralogy and … · the petrology and modal phenocryst...

9. VARIATION OF BASALT PHENOCRYST MINERALOGY AND ROCKCOMPOSITIONS IN DSDP HOLE 396B

Henry J.B. Dick and Wilfred B. Bryan, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts

ABSTRACT

Data are presented on phenocryst proportions and content for 40basalts representative of the stratigraphic section drilled at DSDP Hole396B, and olivine and spinel phenocryst microprobe analyses for allbasalts. The variations of phenocryst proportions, content, andcomposition with whole rock composition and stratigraphic position ofthe basalts are used to infer their petrogenesis. On the basis of thecolinearity of aphyric and phenocryst-free basalt compositions, itappears that crystallization of the Hole 396B basalts was controlled by alow-pressure cotectic. Phenocryst proportions vary irregularly in thebasalts, indicating that gravity separation of phenocrysts occurred due toflotation of plagioclase and sinking of olivine. The proportion of olivinewhich must have crystallized from one Hole 396B unit in order to accountfor the range in olivine phenocryst compositions in the basalts is the sameas that expected from the present plagioclase phenocryst content forcrystallization along the low-pressure cotectic. Irregular variations inmelt composition and fU2 (as indicated by olivine and spinelmicrophenocryst compositions) and the presence of normally andreversely zoned phenocrysts in the same sample, suggest periodic mixingof new magma with old. All these features and the large irregularstratigraphic variation of basalt compositions and mineralogy within apillow unit can be explained if the Hole 396B pillow lavas tappeddifferent portions of a heterogeneous magma chamber. Heterogeneity inthe chamber would arise as a consequence of the introduction of new meltand due to gravitational sorting of phenocrysts. The abrupt changes incomposition and mineralogy of the basalt and the presence of sedimentand weathering at the contacts of the various Hole 396B pillow basaltunits suggest that volcanic activity was periodic at this site, possiblytriggered or controlled by intermittent earthquake swarms and fissuringalong this portion of the old ridge axis. The differences in chemistry andmineralogy of the different units are probably the consequence of árandom combination of fractional crystallization and magma mixing orthe formation of a new magma chamber between cycles of volcanicactivity.

INTRODUCTION

Unlike dredged samples, DSDP cores provide astratigraphic section permitting the study of the localevolution of ridge basalts with time. This chapter reportsthe petrology and modal phenocryst mineralogy of 40DSDP Leg 46, Hole 396B basalt samples, and olivine andspinel compositions from 11 of these basalts. These samplesrepresent the 255-meter section of basaltic basement drilledat 22°59'N, 43°31W, 160 km east of the Mid-Atlantic Ridgein 10-m.y. old ocean crust. The samples were takenadjacent to those selected for shipboard X-ray fractionation(XRF) analysis of major and trace elements to examine thevariations in phenocryst proportions and composition inrelation to whole rock chemistry. One XRF analysis fromthe shipboard study (No. 19) has not been used here as it is

Woods Hole Contribution No. 40602.

anomalously low in MgO compared to other analyses ofrocks with similar mineralogy in the same unit.

PETROGRAPHYThe Hole 396B basalts can be subdivided into a series of

seven distinct units (Table 1) on the basis of theirpetrography, as well as on their chemistry and magneticinclinations (Site Chapter, this volume). All of these units,with the exception of Units 4 and 7, contain 3 per centplagioclase and/or olivine phenocrysts (Table 1). Units 4and 7 contain 9 to 25 per cent plagioclase and subordinateolivine phenocrysts. No clinopyroxene phenocrysts wereseen in the 40 thin sections examined. Spinel is present asan accessory phase in only 19 thin sections. Thegroundmass consists of plagioclase, clinopyroxene, glass,and occasional olivine. Opaques, presumablytitanomagnetite, were present in nearly every thin sectionas a late interstitial phase in amounts up to 2 per cent ormore. Both the grain size and abundance of

215

H. J. B. DICK, W. B. BRYAN

Sample(Interval in cm)

Unit 1

4-1, 103-1055-1, 86-885-2,51-536-1, 55-577-1, 53-557-1, 132-1348-1, 62-648-2, 60-62

10-1,51-5311-1,56-6811-2,5-712-1, 127-12913-1,45-5713-2,49-5113-2, 89-91

Unit 2

13-3,4-614-2, 17-19

F e 2 O 3

(%)

10.310.010.2

9.610.510.410.610.310.910.710.610.510.710.310.3

10.610.8

Tiθ2(%)

1.431.371.401.441.421.421.511.551.541.541.541.511.531.511.51

1.671.64

A1 2O 3

(%)

15.515.715.316.115.615.315.415.315.415.315.415.115.315.115.4

15.515.3

TABLE: lHole 396B Basalts

Cr(ppm)

355358327341357345308293319310292296323300300

297297

01

(%)

0.5< l< l< l< l< l

0.60.2

< l< l< l< l< l< l

-

<l< l

Pg(%)

0.6<l

——--1.40.41.0

< l-

< l< l< l

-

< l—

Sp

_

—-—--—trtr——tr---

—

Notes

Euhedral olivine0.5-3mm, and occasionalrounded and resorbedolivine xenocrysts;euhedral plagioclase1-2.5mm, and occasionalplagioclase xenocrystsl-6mm; spinel xenocrysts

Occasional lmm euhedralolivine and/or rounded

Unit 3

plagioclase xenocryst

15-1, 85-8715-2, 129-13115-3, 16-1915-4, 76-7915-5, 70-73

Unit 4

16-1, 83-8516-2, 40-4216-4, 20-2216-5, 96-9817-1, 132-13418-1, 117-11919-1,4-620-1, 53-5520-3, 33-3520-5,16-1821-2, 24-2622-1, 93-95

Unit 5

23-1, 27-3023-1, 72-7824-1, 98-100

Unit 6

26-1, 7-1030-1, 77-80

Unit 7

32-1, 45-4732-1, 69-71

11.011.211.011.211.2

9.29.5

10.09.29.59.4

10.58.88.79.18.68.5

10.5810.45

10.3

10.910.27

1.651.631.631.631.61

1.201.201.331.211.211.211.331.041.101.161.010.99

1.581.54

1.49

1.451.22

15.115.015.215.015.1

16.916.716.917.116.917.019.117.617.317.717.718.0

16.0116.06

15.9

15.918.0

285266263267266

323325347320335339359326347364356363

346345

358

350348

0.3

_—

-

3.22.62.13.64.24.02.42.83.01.35.44.2

10.70.9

1.10.8

0.71.1

0.4

—0.62.7

16.719.211.213.214.912.114.817.216.010.919.719.5

10.80.4

0.92.4

9.68.2

——--

trtrtrtrtrtrtrtrtrtrtrtr

trtrtr

tr

_

tr

Occasional euhedralolivine ' v lmm and anoccasional roundedplagioclase xenocryst

Euhedral olivine0.5-10mm (avg. lmm);euhedral plagioclase0.5-5mm (avg./v2mm);Plagioclase xenocrystscommon with either arounded or spongy re-sorbed appearance withan avg. size of 3-10mm

0.5-2mm euhedralolivine 3-5mm roundedand spongy appearingplagioclase xenocrysts

Euhedral olivine0.5-5mm (avg. 0.7mm);0.5-lmm rounded andragged plagioclase xeno-crysts and a few euhedral;One large anhedral spinel

Euhedral olivine <0.5mm;large euhedral plagioclaseand rounded xenocrysts

Percentage of olivine and plagioclase phenocrysts estimated by point counting a single thin section —minimum of 1000 points.

^Internal Sample numbers used in figures.

216

BASALT PHENOCRYST MINERALOGY

titanomagnetite appear to depend more on the degree ofcrystallinity than on any other factor.

Phenocryst proportions and content have at times provedto be a useful means of classifying oceanic basalts(Miyashiro et al., 1970; Hekinian et al., 1976). Thismethod is not without pitfalls, particularly wherephenocryst contents are small and modal analyses arebased on a limited area. Modal data on phenocryst contentsof basalts from Hole 396B are presented in Table 1. Theseanalyses each represent approximately a thousand pointscovering a 6-cm2 area on a single thin section. Though itwould be preferable to have counted several sections fromdifferent orientations for each basalt, this was not possibledue to the limited amount of material available. Where thetotal phenocryst content is <5 per cent, the proportionsquoted in Table 1 should be regarded as a rough estimateonly.

Plagioclase phenocrysts range from euhedralmicrophenocrysts to grains up to a centimeter across.Glomerocrysts of olivine and plagioclase are present inmany thin sections. Individual phenocrysts range from largeeuhedra to heavily corroded and rounded xenocrysts(phenocrysts clearly not in equilibrium with the liquid)sometimes with vermiform inclusions at their margins, oftenboth types occur in the same thin section. In Units 3, 4 and5, plagioclase appears to be largely xenocrystal in thepresence of euhedral olivine phenocrysts andmicrophenocrysts. In Unit 4, where plagioclase phenocrystsare most abundant, many of the xenocrystal grains containnumerous glass or matrix-filled holes and have a "spongy"appearance. As discussed by Dungan et al. (this volume),plagioclase phenocrysts may be either reversely or normallyzoned in the same thin section.

Olivine phenocrysts total 1 per cent or less in all the unitsexcept Unit 4 where they occur in amounts up to 5 and 6 percent. Euhedral grains are most common, averaging about0.75 to 1.0 mm, with rare grains as large as 10 mm.Rounded, apparently resorbed xenocrysts do occur, mostfrequently in Unit 1. Large amoeboid skeletal olivinegrains, often intergrown with plagioclase phenocrysts,occur in Unit 5; this texture is thought to indicate extremelyrapid growth (Gutmann, 1977).

Spinel occurs as a rare accessory phase in Unit 1, whereit generally is found as inclusions in olivine or plagioclasephenocrysts, or as rounded and partly resorbed xenocrystsup to 0.4 mm diameter in the groundmass. Spinel is absentin the nearly holocrystalline Units 2 and 3. In Units 4 and 5,spinel is present as small euhedra from 5 to 20 µm acrossand as larger, usually anhedral, xenocrysts up to 0.5 mmacross. Both varieties are found in most thin sections. Someof the larger grains contain numerous holes at their marginswhich are filled with vermiform glass and matrix inclusions;this texture is thought typical of rapid resorption. Spinelfrom Units 4 and 5 occurs as clusters or chains in some thinsections, in addition to isolated inclusions in phenocrystsand groundmass.

The occurrence of chromian spinel in Hole 396B basaltsappears to be favored by high Cr, low FeO*/(FeO* +MgO), and (possibly) high AI2O3 (Table 4), as found bySigurdsson (1977) for DSDP Leg 37 basalts. Many basaltswith Cr contents above 300 ppm, however, do not havespinel, suggesting that Cr content alone does not control thepresence of spinel.

OLIVINE CHEMISTRY

Analyses of 37 olivine phenocrysts and microphenocrystsfrom eight Hole 396B basalts are given in Table 2. All theanalyses were done with the Massachusetts Institute ofTechnology electron microprobe. Natural standards wereused for all elements: olivine (#P-140) for Mg, Fe, and Si;diopside for Ca; and manganese ilmenite for Mn and Ti.Estimates of precision are given in Table 2 (and Table 3)based on repetitive analyses of a single standard grainmade throughout the course of analysis. This estimate doesnot allow for errors due to variations in the thickness ofcarbon coatings between samples and standards.

Olivine ranges in composition from Foes. 0 to Foβ9.5, witha range of composition in each thin section examined. Asseen in Figure 1, there is a rough correlation between theapparent grain size and forsterite content, with the smallestolivine crystals having the highest iron contents.

SPINEL CHEMISTRY

Analyses of 28 spinels from 11 Hole 396B basalts aregiven in Table 3 and are plotted in Figures 2, 3, 4, and 5.NiO, CoO, ZnO, and V2O3 were not determined, althoughthey are likely to be present in small quantities. Fe2 +/Fe3 +

was estimated by calculation from total iron based on theassumption of perfect stoichiometry of the spinels(R + +R2+ + + Q4).

Basalts from Hole 396B have chromian spinelsintermediate in composition between chromite (Fe2Cr2θ4)and spinel (MgAUOV). As with most terrestrial chromianspinels, they have low Fe2θ3 contents of 4.6 to 9.2 weightper cent. The Cr/(Cr+Al) ratios are similar to those foundfor most other spinels from Mid-Atlantic Ridge basalts, buthave decidedly higher iron content (Figure 3). This mayrefelct either a lower temperature of crystallization or agreater FeO content in the melt.

The most striking feature of the analyses is the largerange in compositions found for individual spinels in asingle thin section (Figure 4). The variation is erratic and atleast two patterns are apparent. The most prominent is alarge progressive increase of Cr/(Cr+Al) with a smallvariation of Mg/(Mg+Fe2 +) e.g., Samples 396B-16-2,40-42 cm and 16-5, 96-98 cm (cited as Samples 2 and 4) inFigure 4. The second trend is increasing Fe 2 + and Fe 3 +

with variable Cr/(Cr+Al). This trend is seen most clearlyin zoned groundmass spinels in basalts from Unit 1, e.g.,Samples 396B-8-2, 60-62 cm; 10-1, 51-53 cm; and 12-1,127-129 cm in Figure 4.

Irvine (personal communication, 1977) has found twotrends for spinels experimentally crystallized at oneatmosphere from a basaltic melt: (1) decreasingCr/(Cr+Al) with co-precipitating olivine, and (2)increasing Cr/(Cr+Al) with co-precipitating plagioclaseand olivine. The partitioning of Mg and Fe 2 + betweenolivine and spinel is temperature-dependent andcrystallization at lower temperatures (or higher ironcontents) shifts spinel to a more iron-rich composition(Irvine, 1967). Accordingly, the spinel zoning patternsobserved in the three Unit 1 basalts are consistent withsimple progressive crystallization of spinel with olivine orwith plagioclase and olivine, and a consequent decrease intemperature and Mg/(Mg+ Fe2 +) in the melt.

217


Sample

Grain

I eOMgOSiO2

CaONiO

Total

I oGrainsize (mm)Pointscounted

Shape

Sample

Grain

I eOMgOSiO2

CaONiO

Total

l oGrainsize (mm)Pointscounted

Shape

Sample

Grain

l'eθMgOSiO2

CaONiO

Total

FoGrainsize (mm)Pointscounted

Shape

1

13.2846.4838.87

0.320.16

99.1

86.2

0.15

4

sb

1

13.1446.8038.82

0.300.14

99.2

86.4

1.2

7

eu

1

11.2048.3839.14

0.320.20

99.2

88.5

0.5

4

eu

16-1, 86-83 cm

2

12.7847.2538.82

0.300.16

99.3

86.8

0.5

5

eu

2

12.0147.5738.98

0.320 . 1 3

99.0

87.6

0.7

6

eu

3

13.1146.8038.85

0.320.16

99.2

86.5

0.25

7

eu

4

13.2247.1838.46

0.300.15

99.3

85.6

0.30

4

eu

5

11.6148.0439.34

0.300.16

99.4

88.1

1.2

9

eu

19-1, 7-10 cm

3

12.5747.1638.87

0.320.13

99.1

87.0

5

eu

20-3, 35-39 cm

2

10.8248.5539.47

0.310.20

99.3

88.9

0.5

4

3

12.6847.1339.32

0.340.18

99.7

86.9

0.05

4

gdm

4

12.9847.2938.82

0.320.14

99.6

86.7

0.1

5

eu

4

12.0947.6838.36

0.310.19

98.6

87.6

0.3

2

eu

5

13.0447.2339.16

0.340.19

100.0

86.6 -

0.2

5

cu

1

12.2947.1239.43

0.290.18

99.3

87.2

0.5

6

eu

TABLE 2Hole 396B Olivine Analyses, Unit 4

16-2, 37-40 cm

1

11.0347.8539.33

0.310.22

98.7

88.5

2.0

3

eu

6

11.8748.1139.31

0.310.14

99.7

87.8

2.0

4

eu

2

13.1146.6339.01

0.320.16

99.2

86.4

0.1

4

sb

7

13.6146.8238.33

0.300.16

99.2

86.0

0.2

4

eu

3

13.5946.2038.38

0.330.18

98.7

85.9

0.1

3

1

13.2746.4539.04

0.280.17

99.2

86.2

0.22

8

eu

21-2, 24-27 cm

2

11.0847.9839.70

0.250.19

99.2

88.6

1.0

5

an

3

12.2746.9939.57

0.290.17

99.3

87.3

0.1

4

sb

4

10.1848.7739.71

0.280.19

99.1

89.5

3.0

4

sb

l

10.3249.0639.66

0.310.21

99.6

89.5

2.5

3

en

20-1

2

12.3947.0739.43

0.300.16

99.3

87.1

0.4

3

an

5

11.6647.3639.71

0.270.14

99.1

8 7.9

2.0

sb

16-5,95-9

2

12.7746.9339.07

0.330.18

99.3

86.8

0.4

5

eu

3

13.4446.4638.86

0.310.18

99.3

86.0

0.4

5

eu

,94-97 cm

3

12.1147.1439.56

0.280.18

99.3

87.4

0.5

7

eu

4

12.3546.9439.49

0.270.17

99.2

87.2

1.8

8

sb

4

14.3345.4938.59

0.330.16

98.9

85.0

0.05

3

an

5

11.0747.9539.81

0.300.20

99.3

85.6

1.8

6

1

12.1546.7439.45

0.320.17

98.8

87.3

0.35

6

an

22-1,

2

10.7047.9739.81

0.300.17

99.0

88.9

2.0

6

sb

95-99 cm

3

11.7647.1339.33

0.320.17

98.7

87.7

0.07

4

eu

4

11.5947.3239.61

0.280.20

99.0

87.9

0.9

8

sb

Replicate Analyses (15),Standard Olivine Grain (p. 140)

x lσ Quoted Analyses

7.26 0.07 7.2352.02 0.53 51.6340.14 0.41 40.86

- - -0.30 0.02

99.73 0.42

92.7 0.01 92.7

Notes: eu = euhedral, sb = subhedral, an = anhedral.

The spinel composition trends are less simple in the Unit4 basalts. The crystallization history of the Unit 4 magmaapparently was complex, as noted in a subsequentdiscussion, and it is possible that several generations ofspinels may have been mixed together in the magmachamber to produce composite trends. Although the Unit 4spinels occur with large amounts of co-precipitatedplagioclase and olivine, the most prominent trend is a largeincrease in Cr/(Cr+Al) with progressive crystallization,opposite to the findings of Irvine (personal communication).Large aluminous spinels, in fact, often have spongy marginsapparently due to resorption. A similar trend from early,partially resorbed high-alumina to late euhedralhigh-chrome spinels has been found in some primitiveMg-Ca basalt glasses from the FAMOUS region which haveolivine but no plagioclase (Dick, 1976). Further, theaphyric Unit 1 basalts contain no high-alumina spinels,even though they may have undergone little plagioclasecrystallization.

High-alumina spinels were first reported in Mid-AtlanticRidge basalts by Bryan (1972) who suggested that their

alumina contents indicate relatively high crystallizationpressures. Frey et al. (1974), Sigurdsson and Schilling(1976), Dick (1976), and Sigurdsson (1977) have attributedhigh alumina contents of such spinels to crystallization athigh pressures, noting their similarity to spinels in manymantle xenoliths. Experimental evidence for such a pressureeffect was found by Green et al (1972) for spinelscrystallized from lunar basalts. It is possible, therefore, thatthe high-Al spinels in Leg 46 basalts are xenocrystscrystallized at higher pressures.

Spinel composition fields for different occurrences areshown in Figure 3. The majority of the Leg 46 basalts andabyssal tholeiites have spinels with Cr/(Cr+Al) ratiosbetween 0.42 and 0.55. The similarity of the field for a fewabyssal peridotites to that for the majority of spinels fromabyssal tholeiites suggests that many abyssal peridotites arecumulates precipitated from the basalts. Alpine-typeperidotites, of which the Josephine Peridotite (Figure 3) isrepresentative, generally have a large range of Cr/(Cr+Al)ratios, more restricted Mg/(Mg+Fe2+) ratios than abyssaltholeiite spinels, and a restricted composition in a single

218


TABLE 3Chrome Spinel Analyses From DSDP Hole 396B Basalts

Sample

Grain

C r 2 O 3

MnOFeOF e 2 O 3MgOA12O3

Tiθ2Total

Pointscounted

Sample

Grain

C r 2 O 3


TiO2

Total

Pointscounted

Sample

Grain

C r 2 O 3


TiO2

Total

Pointscounted

1

34.500.11

12.596.34

15.6229.22

0.4898.9

5

16-1,1

1

37.990.13

14.004.65

14.2827.24

0.20

98.5

5

22-1, 95-99 cm

2

35.310.11

12.235.70

15.7629.00

0.4598.6

4

S6-83 cm

2

35.940.13

14.977.69

13.9426.23

0.55

99.5

4

19-1, 7-10 cm

][

35.870.17

14.6,

14,

,33,24,19

27.450.36

98.

5

6

3

36.290.12

13.045.70

15.1327.80

0.4598.5

5

1

34.170.17

12.766.60

15.6929.90

0.46

99.8

3

20-1

1

37.380.15

14.507.15

13.6924.18

0.6397.7

2

4

35.080.11

13.355.79

14.9528.77

0.3598.4

4

8-2, 61-64 cm

1 core

36.300.14

11.265.86

16.7229.43

0.38100.1

4

16-2, 37-40 cm

2

26.520.16

11.006.13

16.8235.96

0.27

96.9

3

3

38.450.22

15.149.15

13.5222.00

0.96

99.4

4

, 94-97 cm

2

35.490.14

12.355.22

15.6429.37

0.3298.5

6

3

35.180.10

13.125.78

15.2928.95

0.48

98.9

5

1 rim

34.090.15

12.616.62

15.7329.87

0.4399.5

5

4

35.530.20

12.396.92

15.9328.49

0.57

100.0

10-1, 50-52 cm

1

35.220.19

14.208.81

14.4826.11

0.6999.7

3

2

40.540.19

14.668.67

13.8521.13

0.8899.9

3

12-1, 129-130 cm

1 core

38.110.17

10.615.46

16.8828.20

0.3299.75

5

16-5, 95-97 cm

1

36.680.16

13.426.60

14.7526.45

0.57

98.6

4

20-3, 35-39 cm

1

34.730.11

13.275.82

15.1029.02

0.50

98.6

4

2

35.920.12

12.814.97

15.3829.03

0.35

98.6

5

2

31.510.16

14.448.17

14.6730.21

0.54

99.7

1

avg.

44.68

22.09

12.4119.55

0.42

99.15

3

37.930.17

13.977.48

14.1123.49

0.79

97.9

3

1 rim

38.470.18

11.865.53

15.9427.23

0.3499.55

4

21-2,

1 2

38.68 37.470.16 0.18

13.06 13.557.25 7.58

15.03 14.7524.73 25.10

0.59 0.67

99.5 99.3

4 3

Standard 52 NL 11

lσ

0.28

0.14

0.120.260.03

0.47

24-27 cm

3

38.130.18

12.847.19

15.1225.21

0.56

99.2

4

Quoted Analysis

44.55

22.12

12.2119.400.46

98.74

4

37.940.17

13.156.96

15.0725.87

0.51

99.7

4

TABLE 4Chemical Character of Spinel Bearing

and Spinel Free Hole 396B Basalts

Basalt CrFeO*/

(FeO*+MgO)

Average

AL.O-

Spinel 337±21D 52.5±2.6 16.9+1.1 15.6±0.4Spinel free 311±33 55.0±1.2 15.4±0.3 15.4±0.3

^Excluding phenocryst-rich basalts.blσ.

thin section. These differences undoubtedly reflect there-equilibration of spinel to low temperatures in a slowlycooled mantle peridotite, while spinels in abyssal tholeiitesreflect a range of conditions during crystallization andrapid cooling. It is interesting to note that in addition tomore aluminous chromite ores (not shown in Figure 3),alpine-type peridotites have Cr-rich chromite ores entirelyoutside the composition field of abyssal basalt andperidotite accessory spinels.

As shown in Figure 5, the Leg 46 basalt spinels also havehigher Fe 3 + contents than accessory spinel from alpine-typeperidotites. The Fe 3 +/Fe 2 + ratio of spinel is a function of f‰which, in turn, is a function of temperature. Accordingly,the lower Fe 3 + content of accessory spinels from alpine-typeperidotites may reflect their equilibration to lowertemperatures and a low relatively uniform fθ2. Thecomposition of spinel from the relatively massive podiformchromitite ores found within alpine-type peridotites,however, probably reflects more closely the conditions oftheir initial crystallization (Dick, 1977). These ores arethought to be residues precipitated from magmas thatcoalesced from and rose out of the mantle (e.g., Dick,1977). The large range of Fe 3 +/Fe 2 + found for spinels inbasalts from Hole 396B, particularly in the same thinsection, indicates a large variation if fθ2 during the coarseof crystallization. Figure 6 suggests that fθ2 varied fromthat at which podiform chromities crystallized in the mantleto much higher values, perhaps due to interaction withcrustal water or to progressive fractional crystallization ofthe melt.

219


oδ

.18

.16

.14

100

- 12"H.

<0,4 mm •0.4-0.8 mm0.8-1.2 mm #1.2-1.6 mm •> 1.6 mm •&

.50 .54 .56 .58

(FβOVMgO) R o c k

.60

Figure 1. (FeO*/MgO) basalt versus (FeO/MgO) olivine forHole 396B basalts from Unit 4. Numbers are internalsample numbers given in Table 1. Symbols indicatephenocryst size. The dotted lines represent equilibriumolivine compositions if all iron is Fe^+. Departures fromthis line of microphenocryst compositions must largelybe due to variable amounts of Fe3+ in the melt at thetime of crystallization.

Figure 2. Plot(RmR2i~M'O4) from Hole 396B basalts.

spinels

COMPOSITION AND PHENOCRYSTPROPORTIONS

The range of variation in basalt compositions representedby the Hole 396B basalts is small compared to the totalvariation of abyssal tholeiites, or to basalts recovered in theFAMOUS region of the Mid-Atlantic Ridge. So-called"primitive" MgCa and Ti-poor basalts (Melson et al.,1976) and relatively fractionated FeTi basalts (Melson etal., 1976), representing the known compositional extremes,are entirely absent at Site 396.

The normative compositions of 39 basalts from Hole396B, calculated from the shipboard analyses, are plottedin Figure 6. Tie lines connect these points to the point alongthe plagioclase-olivine join representative of their modal

80

60

CrxlOO

Cr+AI

40

20

- M.A.R.basalts

high-Crpodiformchromites

layered 'intrustions /

M.A.R. peridotites

100 80 60 40 20

Mgx100

2+Mg+Fe

Figure 3. Plot of R** versus Principal R+++ componentsfor spinels from Hole 396B basalts. Composition fieldsare given for Mid-Atlantic Ridge basalts, (a) Mid-Atlantic Ridge, Sigurdsson Schilling, 1976; Co) FAMOUSbasalts (Dick, 1976), layered intrusions (Irvine, 1967),high-Cr podiform chromite ores (high-Al ores notshown) and alpine harzburgite spinels from the Jose-phine Peridotite (Dick, 1977), and Mid-Atlantic Ridgeperidotites (Aumento and Loubat, 1971; Dick, unpub-lished data) are given for comparison.

phenocryst proportions (volume % ) . The total modalphenocryst content of each sample is indicated along thejoin. Note that basalts which contain predominantly olivinephenocrysts are almost aphyric (bulk composition liquidcompositions), while plagioclase is dominant in most of thephenocryst-enriched basalts. A similar relation wasreported by Bryan and Moore (1977) for FAMOUS basalts.

With the exception of Units 4 and 7 (which contain largeamounts of phenocrysts) and the slowly cooledcomparatively coarse-grained Unit 3 cooling unit, the Hole396B basalts all lie quite close to a single line. The best-fit

220


55

50

CrxlOOCr+AI

45

40

35

11

75

Hole396Bbasalt spinels

65Mgx100/Mg+Fe

60

Figure 4. Blow-up of Figure 3 showing range in spinelCompositions in individual thin sections. Hyphenatednumbers are sample numbers, others are internal samplenumbers for Unit 4. Arrows indicate last crystallizedspinel (crystallization trend) where petrographic criteriaare available. Points for 8-1, 10-1, and 12-1 representseparate analyses of core and rim from a single grain.

line to these points (Figure 7) lies in a position close to thatof the experimentally determined cotectic in the ternarysystem Fo-Di-An (Osborn and Tait, 1952) and is similar tothe liquidus boundaries inferred for abyssal tholeiites byMiyashiro et al. (1970), and for FAMOUS basalt glasses(Figure 7[A]).

The Unit 4 (and Unit 7) plagioclase-olivine basalts lie onthe plagioclase side of the inferred cotectic boundary. InFigure 7b, the proportions of phenocrysts have beenrecalculated to weight per cent from Figure 6 and used toproject the basalts to their phenocryst-free compositions inthe ternary. With one exception,2 the projected

2Section 396B-16-1 contains exceptionally coarse phenocrysts for whichthe modal analysis may not be representative of the small portion analyzedby X-ray fluorescence.

compositions all lie along the cotectic inferred from theremaining relatively aphyric basalts from Hole 396B. AsMiyashiro et al (1970) suggested, this indicates thatcrystallization and fractionation of the basalts wascontrolled by the cotectic at shallow levels. Crystallizationmay also have occurred at greater depths, as suggested byO'Hara (1968), or plagioclase since fractionation of olivinefrom an ascending magma would drive it toward the lowpressure cotectic.

Despite the colinearity of the Hole 396B aphyric andphenocryst-free phyric basalt compositions, virtually nonehave phenocryst proportions expected for crystallizationalong the inferred cotectic [Pg/(Pg+01) 0.64]. Rather,the basalts are rich in either olivine or plagioclasephenocrysts. This could reflect low phenocryst proportionsand non-representative thin sections; however, this isclearly not the case for Units 4 and 7. This suggestsseparation of crystals during crystallization, most likely byflotation of plagioclase and sinking of olivine.

The excess plagioclase in basalts from Unit 4 may be dueto loss of co-precipitating olivine or to accumulation ofplagioclase. Titanium, however, decreases systematicallywith increasing phenocryst content in Unit 4 basalts. This isthe reverse of the trend expected for fractionalcrystallization, where Tiθ2 normally increases. Such apattern might emerge, however, in a suite of magmaserupted from different parts of the same magma chamberwhere some magmas are Tiθ2-rich as a result of loss ofphenocrysts and others are Tiθ2-poor where Tiθ2-poorphenocrysts have accumulated due to gravitational sortingof crystals.

Many of the differences between the compositions ofbasalts from Unit 4 and those from other units might beaccounted for by the excess plagioclase in Unit 4 basalts.Specifically, the high AI2O3 contents in Unit 4 basalts mightresult from plagioclase accumulation. A least-squaresregression line through all the points in Figure 8, however,yields a Tiθ2 content of -0.62 weight per cent at 100 percent phenocrysts. This unlikely situation indicates that thereare real differences in the liquid compositions of thedifferent Hole 396B basalt units, differences due to factorsother than phenocryst accumulation. These differences mayreflect primary differences in magma compositions:fractionation of a parent liquid to different liquids; or acomplex combination of random mixing, fractionation, anderuption of magmas from a chamber(s) over a period oftime.

VARIATION OF ROCK ANDOLIVINE COMPOSITIONS

Roeder and Emslie (1970) found that the FeO/MgO ratioof olivine in equilibrium with a liquid is independent oftemperature and is a function of the FeO/MgO ratio of themelt.Using Roeder and Emslie's partition coefficient, Irvineand Findlay (1972) calculated the change in meltcomposition with percent olivine crystallization. UsingIrvine and Findlay's (1972) diagram in reverse, one canestimate that for an average ocean tholeiite, the per centolivine crystallization represented by the average range ofolivine compositions in basalts from Unit 4 is about 8 percent. This contrasts with the average of 3.1 per cent olivine

221


15

10

Fe3+x100

Fe3++Cr+AI

100

Hole 396B spinel

ALPINE PERIDOTITEpodiform oresaccessory spinels

90 80 70

Mgx100

2+

60 50

Mg+Fe

Figure 5. Plot of R3+ versus R2+ component ratios for spinels from Hole 396B basalts. Lines connect analysesfrom the same thin section. Sample numbers are the same as in Table 1. Fields for accessory spinel in harzburgiteand podiform chromitite ores are for the Josephine Peridotite (Dick, 1977).

PYROXENE

Unit

45 55 65OLIVINE

Figure 6. Corner of Px-Pg-01 ternary plotted with normative compositions (volume %) of Hole 396B basalts. Lines connectthese points to the Pg-01 join to indicate proportions ofphenocrysts in the basalts and the modal phenocryst contents aregiven on the joins. Arrows point to 100% olivine.

222


Pyroxene

Pyroxene

A.

Plagioclase

Plagioclase 5 10 15 20 25 30^ – Olivine

proportion of phenocrysts

Figure 7. (A) Pg-01-Px ternary with lines of organic correlation calculated for Leg 46aphyric basalts and FAMOUS basalt glasses, and line drawn by Miyashiro (1970)between plagioclase and olivine tholeiites. (B) Blow-up of the plagioclase corner of thePg-01-Px normative ternary with all data recalculated to weight percent instead ofvolume percent. Dotted line represents the inferred cotectic boundary based on therelatively aphyric unit one, five and six basalts (same as in A) indicated by crosses.Coarsely crystalline Unit 3 basalts and phenocryst-rich basalts were excluded incalculating the line. Solid stars indicate Unit 4 basalt compositions. Open stars indicateprojected Unit 4 phenocryst-free compositions.

phenocrysts found in these basalts. Thus, presuming initialequilibrium crystallization, one cannot account for therange of olivine compositions in Unit 4 basalts by theirpresent olivine phenocryst content. The basalts must haveeither fractionated-out olivine or incorporated additionalolivine by mixing of magmas and their phenocrysts, or bydisaggregation of older rocks.

Using the proportion of olivine indicated by the range ofolivine phenocryst compositions and the averageplagioclase phenocryst content of the Unit 4 basalt yields a

Pg/(Pg+01) ratio of 0.65 for the proportions of phenocryststhat crystallized from the Unit 4 magma. This is virtually thesame proportion as that expected for shallow-levelcrystallization along the inferred cotectic in Figure 7(0.64). This could indicate that the Unit 4 basalt magmacrystallized olivine and plagioclase at a shallow level,fractionating olivine. It also suggests that the average Unit4 basalt contains a representative proportion of theplagioclase phenocrysts crystallized from the Unit 4 magmaprior to eruption.

223


12 16

Phenocrysts

Figure 8. Plot of T1O2 contents of Hole 396B basaltsversus their phenocryst proportions.

It is not possible to rule out disaggregation of older rockas the source of some of the Unit 4 phenocrysts.Disaggregation of an older rock, however, should producean excess proportion of phenocrysts in the melt as aconsequence of the heat loss required for disaggregation(Anderson, 1976). Projecting the average Unit 4 basaltback along the cotectic a distance corresponding tocrystallization of 14.6 weight per cent plagioclase and 8weight per cent olivine yields an initial liquid scarcely moremafic than the most mafic Hole 396B basalt and well withinthe range of Mid-Atlantic Ridge basalt compositions. Thus,the phenocryst content of the Unit 4 basalts can beexplained by simple crystallization from a relatively maficparent. The bulk of the Unit 4 phenocrysts could beexplained by disaggregation of older rock only if thephenocryst content and composition of the disaggregatedrock were similar to those of the intruding melt. Due todisproportionation of phenocrysts during crystallization, itwould be fortuitous if disaggregated wall rocks containedthe same phenocryst proportions as that crystallizing fromthe intruding magma. The absence of excess phenocrystsand the lack of xenoliths in the Unit 4 basalts, suggest thatsignificant amounts of disaggregation are unlikely to haveoccurred.

Plotted in Figure 1 is the line representing equilibriumolivine and melt compositions based on Roeder andEmslie's partition coefficient. It is reasonable to assumethat small euhedral microphenocrysts (<0.4 mm) nowfound in the basalts were in equilibrium with the melt at thetime of eruption. As expected, the olivine microphenocrystsplot below the line in Figure 1, obviously reflectingFe 2 + <Fe total and a significant, highly variable Fβ2θ3content in the melt at the time of eruption. Subtracting theFeO and MgO content of the phenocrysts from the wholerock analysis only increases the amount of Fe2θ3 in the meltsuggested by the microphenocryst compositions.

It is reasonable to assume that the Fe3+/Fe2+ratioof meltwas largely a function of ‰ Variation of fθ2 in the meltmight be produced by progressive fractional crystallizationor interaction between the melt and water trapped in thecrust. Both of these, however, should act to progressivelyincrease fθ2 in the melt. Alternatively, recent experimentalwork on fluid inclusions in xenocrystal olivine from oceanicbasalts (Delaney et «!., 1977) suggests that the fθ2 of

"primary" mantle derived magmas is very low. The large"irregular variation in Fe2(V indicated by the olivinemicrophenocrysts from Unit 4 plotted in Figure 1 maysuggest intermittent feeding of the magma chamber by lessfractionated magmas.

SUMMARY AND CONCLUSIONS

The Hole 396B stratigraphic sequence contains a numberof distinctly different pillow basalt units. Their boundariesare marked by abrupt changes in mineralogy and chemistry,and often coincide with weathering and sedimentdeposition, and a change in magnetic inclination. Given thepredominance of pillow lavas, the Hole 396B suite islargely surface flows in stratigraphic sequence of theireruption. It appears, then, that they represent periodicvolcanism with intervening eruptive hiatuses marked bysedimentation and weathering.

There is no systematic variation in chemistry andmineralogy through the sequence of units that can beexplained by simple progressive fractional crystallization ina magma chamber of a single batch of parent liquid. It canbe seen in Figure 6, in fact, that each of the different unitshas basalts with a range of compositions overlapping thoseof the other units. Relatively mafic compositions often lielate in the Hole 396B sequence, and intra-unit variationsmay be irregular or the reverse of those expected for simplefraction ation.

It is evident, from the variation of whole rockcompositions and phenocryst mineralogy, thatcrystallization of the pillow basalts was controlled atshallow levels by a cotectic and that gravitational sorting ofcrystals in the liquid occurred prior to eruption. In the caseof Unit 4, we can infer that olivine and plagioclasecrystallization was extensive (about 22 weight per cent)along this low-pressure cotectic. The negative correlationof phenocryst proportions and whole rock Tiθ2 contentssuggests that accumulation of plagioclase and loss ofolivine, or the reverse, occurred in the different Unit 4basalts. All of this can be explained by derivation of thesurface flows from different portions of a shallow magmachamber, including the irregular intra-unit stratigraphicvariations of phenocryst proportions and compositions andwhole rock chemistry. It is a common observation fromvolcanic stratigraphy that some magma chambers appear tohave emptied from bottom to top, some from top to bottom,and others randomly.

The large variation of phenocryst compositions, and theirdifferent glass inclusions in a single pillow, can beexplained by mixing early and late-formed olivine andplagioclase. It is well-known that crystals accumulated atthe bottom or top of a chamber undergo little growth fromthe interstitial liquid due to a low ratio of liquid to crystals.In this manner, early formed crystals can be preserved.Turbulence accompanying eruption may mix these withlater-formed crystals, both in the eruptive unit and in thechamber itself. This phenomenon should give rise tocomplex co-existing growth patterns such as are observed inmany of the Leg 46 thin sections.

The presence of both normal and reverse zoning anddifferent glass inclusion compositions in some Hole 396Bbasalts could be explained as the result of incorporation of

224


crystals from disaggregated wall rocks. The proportion ofolivine thought to have crystallized from many of thesebasalts and the observed plagioclase phenocryst content,however, is that expected for crystallization along the lowpressure cotectic from a reasonable parent composition.The remaining Hole 396B basalts have low phenocrystcontents which would not be expected if they haddisaggregated appreciable amounts of wall rocks.

Mixing of magmas and their phenocrysts may alsoexplain different glass inclusion and normal and reversezoning of crystals in many Hole 396B basalts (Dugan et al.,this volume). As discussed in this paper, it is also a meansof explaining irregular variations in melt composition, ‰(as indicated by olivine microphenocryst and spinelcompositions), and the presence of possible high-pressureAl-rich spinel xenocrysts.

Magma mixing, however, is a term which can encompassa multitude of sins. One might explain many of the featuresobserved in the Hole 396B basalts as the product ofintermittent feeding of a magma chamber with freshmantle-derived melt during the eruptive cycle with acomplete homogenization of the two. This is not likely to bethe case. We view the intermittent volcanic activity inIceland and Hawaii as analogous to the situation at Hole396B. In these regions, a given volcanic episode, such asmight produce one of the Hole 396B pillow basalt units, isrelatively short; eruptions generally occur over a period ofdays to a year. The introduction of a single batch of newmelt into a magma chamber is likely to give rise to liquidheterogeneities in the chamber which last the duration ofthe eruptive episode. Thus, flows from different parts of thechamber and turbulence accompanying eruption could giverise to all the chemical and mineralogic variations seen inindividual Hole 396B pillow basalt units.

The data suggest, then, the presence of a magmachamber(s) at some depth beneath a ridge axis. Thischamber appears to have been fed intermittently by newmagma either from a deeper chamber or directly from themantle. Crystallization within the chamber, sometimesextensive, was controlled by the cotectic with gravitationalsorting of precipitated olivine and plagioclase. In additionto variation in phenocryst proportions, the chamber couldalso have been zoned or heterogeneous with respect to meltcomposition. Disaggregation of wall rocks was probably aminor process at best. The abrupt changes in chemistry andmineralogy and the presence of sediment and weathering atthe upper contacts of the Hole 396B basalt units indicatesthat eruption from the chamber(s) was periodic. A possiblecontrol may have been intermittent earthquake and riftingepisodes along the ridge axis. The differences between themajor pillow basalt units are the likely consequence ofrandom combinations of fractional crystallization andintroduction of new magma, or of the formation of a newmagma chamber between eruptive episodes.

ACKNOWLEDGMENTS

We would like to thank the officers and crew of the R/VGLOMAR CHALLENGER. In addition, we would like toacknowledge helpful reviews of the manuscript by Dr. GeoffreyThompson of the Woods Hole Oceanographic Institution and Dr.James Kirkpatrick of the Deep Sea Drilling Project. This work was

supported by the National Science Foundation (OCE 76-21967).We would also like to thank Angela Sousa and Mary Berry fortyping and editing several versions of the manuscript.

REFERENCES

Anderson, A.T., 1976. Magma mixing: petrological process andvolcanological tool,/. Vole. Geoth. Res., v. 1, p. 3-33.

Aumento, F. and Loubat, H., 1971. The Mid-Atlantic Ridge near45°N. Serpentinized ultramafic intrusions, Canadian J. EarthSci, v .8 ,p . 631-663.

Bryan, W.B. and Moore, J.G., 1977. Compositional variations ofyoung basalts in the mid-Atlantic ridge rift valley near36°49'N, Geol. Soc. Am. Bull, v. 88, p. 556-570.

Delaney, J.R., Muenow, D., Ganguly, J., Royce, D., 1977.Anhydrous glass-vapor inclusions from phenocrysts in oceanictholeiitic pillow basalts, Trans. Am. Geophys. Union, v. 58, p.96.

Dick, H.J.B., 1976. Spinel in fracture zone " B " and medianvalley basalts, FAMOUS area, mid-Atlantic ridge, Trans. Am.Geophys. Union, v. 57, p. 341.

, 1977. Partial melting in the Josephine Peridotite I, theeffect on mineral composition and its consequence forgeobarometry and geothermometry, Am. J. Sci., v. 277, p.801-832.

Frey, F.A., Bryan, W.B., and Thompson, G., 1974. Atlanticocean floor: geochemistry and petrology of basalts from Legs 2and 3 of the Deep Sea Drilling Project, J. Geophys. Res., v.79, p. 5507-5527.

Green, D.H., Ringwood, A.E., Ware, N.G., and Hibberson,W.O., 1972. Experimental petrology and petrogenesis ofApollos 14 basalts, Proc. Third Lunar Sci. Conf., Geochim.Cosmochim. Acta, Suppl. 3,v. l , p . 197-206.

Gutmann, J.T., 1977. Textures and genesis of phenocrysts andmegacrysts in basaltic lavas from Pinacate volcanic field, Am.J. Sci., v. 277, p. 833-861.

Hekinian, R., Moore, J. G., and Bryan, W. B., 1976. Volcanicrocks and processes of the mid-Atlantic ridge rift valley near36°49'N, Contrib. Mineralo. Petrol., v. 58, p. 83-110.

Irvine, T.N., 1967. The Duke Island ultramafic complex,southeastern Alaska. In Willie, P.J. (Ed.), Ultramafic andRelated Rocks: New York (John Willie and Sons), p. 84-97.

Irvine, T.N. and Findlay, T.C., 1972. Alpine peridotite withparticular reference to the Bay of Islands Complex. In TheAncient Ocean lithosphere: Ottawa, Canada (Dept. Energy,Mines and Resources, Earth Physics Branch), v. 43, p.97-126.

Melson, W.G., Valuer, T.L., Wright, T.L., Byerly, G., andNelen, J., 1976. Chemical diversity of abyssal volcanic glasserupted along the Pacific, Atlantic and Indian Ocean sea-floorspreading centers. In Geophysics of the Pacific Ocean Basin:Geophys. Mono. 19, Am. Geophys. Union, p. 351-367.

Miyashiro, A., Shido, F., and Ewing, M., 1970. Crystallizationand differentiation in abyssal tholeiites and gabbros frommid-oceanic ridges, Earth Planet Sci. Lett., v. 7, p. 361-365.

O'Hara, M. J., 1968. Are ocean floor basalts primary magmas?,Nature, v. 220, p. 683-686.

Osborn, E.F. and Tait, D.B., 1952. The systemdiopside-forsterite-anorthite,/4m. J. Sci. Bowen vol., p. 413.

Roeder, P.L. and Emslie, R.F., 1970. Olivine-liquid equilibrium,Contrib. Mineral. Petrol., v. 29, 275-289.

Sigurdsson, H., 1977. Spinels in Leg 37 basalts and peridotites:Phase chemistry and zoning. In Aumento, F., Melson, W.G., etal., Initial Reports of the Deep Sea Drilling Project, v. 37,Washington (U.S. Government Printing Office), p. 775-794.

Sigurdsson, H. and Schilling, J.G., 1976. Spinels in mid-Atlanticridge basalts: Chemistry and occurrence, Earth Planet. Sci.Lett., v. 29, p. 7-20.

225

9. variation of basalt phenocryst mineralogy and … · the petrology and modal phenocryst...

Documents