Dufek intrusion and plagioclase characteristics \bI "
Dufek intrusion and plagioclase characteristics A. B. FORD and J . L. DRINKWATER
Alaskan Geology Branch U.S. Geological Survey Menlo Park, California 94025 G. R. HIMMELBERG
Department of Geology University of Missouri Columbia, Missouri 65211
Whether feldspars float or sink in basaltic magma is a major question in interpreting results of differentiation processes in stratiform mafic complexes like the Dufek intrusion (figure 1). Textures seem to provide compelling evidence for settling of plagioclase along with other minerals at all crystallization stages (Jackson 1971). However, flotation of plagioclase seems to be indicated by some field evidence (Morse 1%9) and theoretical and experimental studies (Bottinga and Weill 1970; Campbell, Roeder, and Dixon 1978; Irvine 1978). Textures and structures are generally interpreted in terms of magmatic sedimentation and mineral accumulation on a chamber floor, to form rocks called cumulates (Wager and Brown 1%8; Jackson 1971). One criterion commonly used to infer cumulus origin is the presence of igneous lamination—the preferred orientation of tabular minerals such as plagioclase in the layering plane. Many aspects of cumulus theory are strongly questioned by McBirney and Noyes (1979), who favor in-place crystallization by chemical diffusion controlled by temperature and composition gradients. Plagioclase, an important constituent in nearly all rocks of the Dufek intrusion, characteristically shows welldeveloped lamination. Although pyroxene gabbro is the predominant rock type, plagioclase-rich leucogabbro or anorthosite layers occur in many parts of the layered sequence. Walker Anorthosite (Ford 1976), of more than 230 meters thickness, is an exceptionally thick unit of this type and is the lowest exposed part of the intrusion. Much thinner layers of anorthosite and leucogabbro (up to about 15 meters thick) occur at many of the higher levels. The thinner layers, characterized by sharp basal contacts and upward grading into overlying gabbro (figure 2), are called mineral graded (Jackson 1971). Walker Anorthosite has a sharp contact with overlying gabbro and thus is not a mineral-graded layer. Several thin layers of pyroxenite (up to about 5 meters thick in gabbro above the Walker) also show mineral grading. Channel-like trough structures are associated with both the pyroxenitic and anorthositic mineral-graded layers and are believed to be evidence of magma-current erosion and deposition. Trough structure has not been found in Walker Anorthosite. 40
A reconnaissance survey of plagioclase (Abel, Himmelberg, and Ford 1979), using samples previously analyzed for pyroxenes and iron-titanium oxides (Himmelberg and Ford 1976, 1977), shows the following features: (1) an overall upward compositional range from about 80 percent anorthite to 50 percent anorthite (An 8050), (2) the presence of a major reversal in chemical trend with height at about 1 kilometer from the top, and (3) the presence of strong re- versals near the mineral-graded pyroxenitic layers. The reversal at about 1-kilometer depth occurs near a horizon showing field evidence of a major disturbance in the chamber (Ford, Reynolds, Huie, and Boyer 1979) and is paralleled by similarly strong reversals in chemical trends of other minerals. The reversals are tentatively believed to be related to addition of a new, less differentiated batch of magma at a late differentiation stage. Plagioclase reversals near the pyroxenite layers, not associated with reversals in pyroxene trends, seem to have a different origin. Magma addition, either from an outside source or by shifting within the chamber, would be expected to be recorded in trends of all minerals.
r / •'lI%
4
\J
-
Dufek
. \ k' Explanation )ç 4—outcrop area DUFEK I MASSIF I — Inferred contact " of Intrusion
\bI
amb
FORRESTAL I RANGE )
it
0 100 kilometers (I I
Figure 1. Location of Dufek intrusion.
Trough structure near the pyroxenite layers suggests that current activity may be involved somehow in the origin of the layers and the associated plagioclase-trend reversals. Many complex factors are involved in movements of crystals under gravity or by currents in magma (Jackson 1971). Yield strength of magma hinders or may prevent sinking or floating of most crystals if the magma is still, but yield strength is sharply lowered if magma is disturbed by flow (McBirney and Noyes 1979). Irvine's (1978) experiments suggest that plagioclase can be carried upward in return flow of density currents and may accumulate in suspended AlsrrARcnc JOURNAL
SIMM
-
gronopityrs gab broi C cumulate
onorthosit, C
Cl) ir
-I
U I-
j .,-'--anorthositic
U 0
cp
-J
anorthosits
E -
•—. _>-; cumulate
--. — 0
(Concealed)
I-
m ognetlte *-TROUGH
CD U
anorthosit.
C-) I 4 CD
I.4
- pyroxsnjte
Ico
- pyrozenite - Walker Anor thosi ti (Concealed)
clouds of crystals near the roof of an intrusion. Differences in mineral-density and particle-size might conceivably lead to sorting during long-continued repetition of current circulation. Particles in movement may be either discrete crystals or large chains of crystals that may involve different minerals (Campbell, Roeder, and Dixon 1978). Petrographic recognition of chain structure may be difficult, but evidence for it might be present in the mineral-graded pyroxenite layers, which typically contain very coarse, poikilitic orthopyroxene crystals up to about 7 centimeters across. The crystals consist of numerous small grains of inverted pigeonite having apparent random form orientation but uniform crystallographic direction (Himmelberg and Ford 1976). The large crystals might represent chains of pigeonite later inverted to orthopyroxene. Lacking the inversion phenomenon, possible chains of plagioclase would probably be unrecognizable in anorthositic layers. Their presence, however, must be considered a possibility. Lamination is noticeably less developed in mineral-graded anorthosite layers than in adjoining gabbro, which might 1980 REvIEw
Figure 2. Stratigraphic location of major anorthositic units, with sketch showing typical features in a mineral-graded layer. Relative amounts of pyroxene (dark) and plagioclase shown by stippling.
suggest presence of chain structure. Walker Anorthosite, in contrast, is typically well laminated. Variation in plagioclase density with composition approximately parallels that reported by Campbell et al. (1978), though having slightly greater density at equal anorthite (An) content (figure 3). At an inferred magma temperature of 1,200C, density would be much lower, possibly in the range 2.65 to 2.70 grams per cubic centimeter. Density decreases along with An content generally upward through the solidification sequence, earlier formed plagioclase having greater density than later ones. Magma composition is a major factor in mineral buoyancy. Composition of the Dufek magma has not been determined, but as inferred from mineral and rock chemical trends, it must have been progressively increased in iron (Fe203 ) until a late stage. Magma variation was probably similar to that in other iron-enriched mafic layered intrusions, such as the Skaergaard intrusion of Greenland (Wager and Brown 1968). On the basis of Skaergaard data (Bottinga and Weill 1970), we can speculate that if an early Dufek magma had density of about 2.66 grams per cubic 41
2.74 -
Ui 2.72 Iz Ui 0 U) 9
2.70.• -.---,
.' . . . .. ,.- Campbell .Jgj. (1978)
Ico
0
'
2.68
50
60
I I I
70
80 ANORTH lIE CONTENT OF PLAGIOCLASE
Figure 3. Density of plagioclase in Dufek intrusion.
centimeter and later iron-enriched magma had density of about 2.69 grams per cubic centimeter, plagioclase was slightly denser than magma at an early stage and became less dense at some later stage. Though factors such as water content of the magma cannot presently be evaluated, such possible variation in buoyancy might result in differences in sorting during operation of Irvine's (1978) mechanism in different crystallization stages. Distribution of anorthosites and their different characteristics at different heights possibly reflect in part the variation in plagioclase buoyancy. Plagioclase composition is about An7580 in Walker Anorthosite and An" in the mineral-graded layers. Walker Anorthosite may represent an early stage at which plagioclase density was greater than magma density, with the others showing a later stage with opposite density relation. If so, an approximately 1,500meter-thick gabbroic sequence, containing almost no anorthosite, above the Walker possibly represents the stage of density-trend crossover, when plagioclase and magma densities were nearly equivalent. According to Jackson (1971, p. 139): ". . . most coexisting cumulus minerals in layered intrusions were not deposited by magmatic currents, but are, for the most part, unsorted crystallization products that reflect exactly what the magma was crystallizing at a particular period of time." In this view, the well-laminated, trough-free, and nonmineralgraded Walker Anorthosite apparently accumulated without significant sorting at a time of greatly predominant plagioclase crystalization. Association of current activity, as inferred from trough structure, and mineral grading with the poorly laminated, higher anorthositic layers indicates a different mechanism in their formation. In the context of Irvine's (1978) experiments, we theorize that buoyant plagioclase was transported upward and sorted in crystal clouds (perhaps involving chain formation in the upper part of the chamber) and then carried down by currents and deposited
42
across the floor. However, it is more difficult to envision how this mechanism operates in the concentration of pigeonite crystals to form pyroxenite layers having a form similar to that of mineral-graded anorthosite layers. Though an explanation does not exist for pigeonite concentration, the activity of currents may help explain plagioclase reversals at these layers. Calcic plagioclase (though not in abundance) may have been carried upward to reside near the roof for some time while more sodic plagioclase crystallized from fractionating magma near the floor. Later deposition by currents on the floor would result in chemical-trend reversal. Plagioclase brought to the floor might be expected to float free again but could be prevented by attachment to the substrate, by deposition of overlying material, or by properties of the magma itself. Some crystals might continue to recirculate (Irvine 1978). Our explanation of plagioclase features in the intrusion is highly speculative. Work in progress focuses on such critical aspects as exact location of trend reversals and determination of whether reversals are also present at the mineral-graded anorthositic layers and of whether mixtures of plagioclase of different generations are present at reversal horizons, as might be expected. These studies are part of work supported by National Science Foundation grant DPP 77-22765 to the Geologic Division, U.S. Geological Survey. References Abel, K. D., Himmelberg, C. R., and Ford, A. B. 1979. Petrologic studies of Dufek intrusion: Plagioclase variation. Antarctic Journal of the U.S., 14(5), 6-8. Bottinga, Y., and Weill, D. F. 1970. Densities of liquid silicate systems calculated from partial molar volumes of oxide components. American Journal of Science, 269, 169-182. Campbell, I. H., Roeder, P. L., and Dixon, J. M. 1978. Plagioclase buoyancy in basaltic liquids as determined with a centrifuge furnace. Contributions to Mineralogy and Petrology, 67, 369-377. Ford, A. B. 1976. Stratigraphy of the layered gabbroic Dufek intrusion, Antarctica. U.S. Geological Survey Bulletin, 1405-D. Ford, A. B., Reynolds, R. L., Huie, C., and Boyer, S. J . 1979. Geological field investigation of Dufek intrusion. Antarctic Journal of the U.S., 14(5), 9-11. Himmelberg, C. R., and Ford, A. B. 1976. Pyroxenes of the Dufek intrusion, Antarctica. Journal of Petrology, 17, 219-243. Himmelberg, G. R., and Ford, A. B. 1977. Iron-titanium oxides of the Dufek intrusion, Antarctica. American Mineralogist, 62, 623-633. Irvine, T. N. 1978. Density current structure and magmatic sedimentation. Carnegie Institution of Washington Yearbook, 77, 717-725. Jackson, E. D. 1971. The origin of ultrámaflc rocks by cumulus processes. Fortschritte der Mineralogie, 48, 128-174. McBirney, A. R., and Noyes, R. M. 1979. Crystallization and layering of the Skaergaard intrusion. Journal of Petrology, 20,487-554. Morse, S. A. 1969. Layered intrusions and anorthosite genesis. In Y. W. Isachsen (Ed.), Origin of anothosite and related rocks (New York State Museum and Science Service Memoir, Vol. 18). Wager, L. R., and Brown, C. M. 1968. Layered Igneous Rocks. Edinburgh: Oliver and Boyd.
ANTARcTIC JOURNAL
Alaskan Geology Branch U.S. Geological Survey Menlo Park, California 94025 G. R. HIMMELBERG
Department of Geology University of Missouri Columbia, Missouri 65211
Whether feldspars float or sink in basaltic magma is a major question in interpreting results of differentiation processes in stratiform mafic complexes like the Dufek intrusion (figure 1). Textures seem to provide compelling evidence for settling of plagioclase along with other minerals at all crystallization stages (Jackson 1971). However, flotation of plagioclase seems to be indicated by some field evidence (Morse 1%9) and theoretical and experimental studies (Bottinga and Weill 1970; Campbell, Roeder, and Dixon 1978; Irvine 1978). Textures and structures are generally interpreted in terms of magmatic sedimentation and mineral accumulation on a chamber floor, to form rocks called cumulates (Wager and Brown 1%8; Jackson 1971). One criterion commonly used to infer cumulus origin is the presence of igneous lamination—the preferred orientation of tabular minerals such as plagioclase in the layering plane. Many aspects of cumulus theory are strongly questioned by McBirney and Noyes (1979), who favor in-place crystallization by chemical diffusion controlled by temperature and composition gradients. Plagioclase, an important constituent in nearly all rocks of the Dufek intrusion, characteristically shows welldeveloped lamination. Although pyroxene gabbro is the predominant rock type, plagioclase-rich leucogabbro or anorthosite layers occur in many parts of the layered sequence. Walker Anorthosite (Ford 1976), of more than 230 meters thickness, is an exceptionally thick unit of this type and is the lowest exposed part of the intrusion. Much thinner layers of anorthosite and leucogabbro (up to about 15 meters thick) occur at many of the higher levels. The thinner layers, characterized by sharp basal contacts and upward grading into overlying gabbro (figure 2), are called mineral graded (Jackson 1971). Walker Anorthosite has a sharp contact with overlying gabbro and thus is not a mineral-graded layer. Several thin layers of pyroxenite (up to about 5 meters thick in gabbro above the Walker) also show mineral grading. Channel-like trough structures are associated with both the pyroxenitic and anorthositic mineral-graded layers and are believed to be evidence of magma-current erosion and deposition. Trough structure has not been found in Walker Anorthosite. 40
A reconnaissance survey of plagioclase (Abel, Himmelberg, and Ford 1979), using samples previously analyzed for pyroxenes and iron-titanium oxides (Himmelberg and Ford 1976, 1977), shows the following features: (1) an overall upward compositional range from about 80 percent anorthite to 50 percent anorthite (An 8050), (2) the presence of a major reversal in chemical trend with height at about 1 kilometer from the top, and (3) the presence of strong re- versals near the mineral-graded pyroxenitic layers. The reversal at about 1-kilometer depth occurs near a horizon showing field evidence of a major disturbance in the chamber (Ford, Reynolds, Huie, and Boyer 1979) and is paralleled by similarly strong reversals in chemical trends of other minerals. The reversals are tentatively believed to be related to addition of a new, less differentiated batch of magma at a late differentiation stage. Plagioclase reversals near the pyroxenite layers, not associated with reversals in pyroxene trends, seem to have a different origin. Magma addition, either from an outside source or by shifting within the chamber, would be expected to be recorded in trends of all minerals.
r / •'lI%
4
\J
-
Dufek
. \ k' Explanation )ç 4—outcrop area DUFEK I MASSIF I — Inferred contact " of Intrusion
\bI
amb
FORRESTAL I RANGE )
it
0 100 kilometers (I I
Figure 1. Location of Dufek intrusion.
Trough structure near the pyroxenite layers suggests that current activity may be involved somehow in the origin of the layers and the associated plagioclase-trend reversals. Many complex factors are involved in movements of crystals under gravity or by currents in magma (Jackson 1971). Yield strength of magma hinders or may prevent sinking or floating of most crystals if the magma is still, but yield strength is sharply lowered if magma is disturbed by flow (McBirney and Noyes 1979). Irvine's (1978) experiments suggest that plagioclase can be carried upward in return flow of density currents and may accumulate in suspended AlsrrARcnc JOURNAL
SIMM
-
gronopityrs gab broi C cumulate
onorthosit, C
Cl) ir
-I
U I-
j .,-'--anorthositic
U 0
cp
-J
anorthosits
E -
•—. _>-; cumulate
--. — 0
(Concealed)
I-
m ognetlte *-TROUGH
CD U
anorthosit.
C-) I 4 CD
I.4
- pyroxsnjte
Ico
- pyrozenite - Walker Anor thosi ti (Concealed)
clouds of crystals near the roof of an intrusion. Differences in mineral-density and particle-size might conceivably lead to sorting during long-continued repetition of current circulation. Particles in movement may be either discrete crystals or large chains of crystals that may involve different minerals (Campbell, Roeder, and Dixon 1978). Petrographic recognition of chain structure may be difficult, but evidence for it might be present in the mineral-graded pyroxenite layers, which typically contain very coarse, poikilitic orthopyroxene crystals up to about 7 centimeters across. The crystals consist of numerous small grains of inverted pigeonite having apparent random form orientation but uniform crystallographic direction (Himmelberg and Ford 1976). The large crystals might represent chains of pigeonite later inverted to orthopyroxene. Lacking the inversion phenomenon, possible chains of plagioclase would probably be unrecognizable in anorthositic layers. Their presence, however, must be considered a possibility. Lamination is noticeably less developed in mineral-graded anorthosite layers than in adjoining gabbro, which might 1980 REvIEw
Figure 2. Stratigraphic location of major anorthositic units, with sketch showing typical features in a mineral-graded layer. Relative amounts of pyroxene (dark) and plagioclase shown by stippling.
suggest presence of chain structure. Walker Anorthosite, in contrast, is typically well laminated. Variation in plagioclase density with composition approximately parallels that reported by Campbell et al. (1978), though having slightly greater density at equal anorthite (An) content (figure 3). At an inferred magma temperature of 1,200C, density would be much lower, possibly in the range 2.65 to 2.70 grams per cubic centimeter. Density decreases along with An content generally upward through the solidification sequence, earlier formed plagioclase having greater density than later ones. Magma composition is a major factor in mineral buoyancy. Composition of the Dufek magma has not been determined, but as inferred from mineral and rock chemical trends, it must have been progressively increased in iron (Fe203 ) until a late stage. Magma variation was probably similar to that in other iron-enriched mafic layered intrusions, such as the Skaergaard intrusion of Greenland (Wager and Brown 1968). On the basis of Skaergaard data (Bottinga and Weill 1970), we can speculate that if an early Dufek magma had density of about 2.66 grams per cubic 41
2.74 -
Ui 2.72 Iz Ui 0 U) 9
2.70.• -.---,
.' . . . .. ,.- Campbell .Jgj. (1978)
Ico
0
'
2.68
50
60
I I I
70
80 ANORTH lIE CONTENT OF PLAGIOCLASE
Figure 3. Density of plagioclase in Dufek intrusion.
centimeter and later iron-enriched magma had density of about 2.69 grams per cubic centimeter, plagioclase was slightly denser than magma at an early stage and became less dense at some later stage. Though factors such as water content of the magma cannot presently be evaluated, such possible variation in buoyancy might result in differences in sorting during operation of Irvine's (1978) mechanism in different crystallization stages. Distribution of anorthosites and their different characteristics at different heights possibly reflect in part the variation in plagioclase buoyancy. Plagioclase composition is about An7580 in Walker Anorthosite and An" in the mineral-graded layers. Walker Anorthosite may represent an early stage at which plagioclase density was greater than magma density, with the others showing a later stage with opposite density relation. If so, an approximately 1,500meter-thick gabbroic sequence, containing almost no anorthosite, above the Walker possibly represents the stage of density-trend crossover, when plagioclase and magma densities were nearly equivalent. According to Jackson (1971, p. 139): ". . . most coexisting cumulus minerals in layered intrusions were not deposited by magmatic currents, but are, for the most part, unsorted crystallization products that reflect exactly what the magma was crystallizing at a particular period of time." In this view, the well-laminated, trough-free, and nonmineralgraded Walker Anorthosite apparently accumulated without significant sorting at a time of greatly predominant plagioclase crystalization. Association of current activity, as inferred from trough structure, and mineral grading with the poorly laminated, higher anorthositic layers indicates a different mechanism in their formation. In the context of Irvine's (1978) experiments, we theorize that buoyant plagioclase was transported upward and sorted in crystal clouds (perhaps involving chain formation in the upper part of the chamber) and then carried down by currents and deposited
42
across the floor. However, it is more difficult to envision how this mechanism operates in the concentration of pigeonite crystals to form pyroxenite layers having a form similar to that of mineral-graded anorthosite layers. Though an explanation does not exist for pigeonite concentration, the activity of currents may help explain plagioclase reversals at these layers. Calcic plagioclase (though not in abundance) may have been carried upward to reside near the roof for some time while more sodic plagioclase crystallized from fractionating magma near the floor. Later deposition by currents on the floor would result in chemical-trend reversal. Plagioclase brought to the floor might be expected to float free again but could be prevented by attachment to the substrate, by deposition of overlying material, or by properties of the magma itself. Some crystals might continue to recirculate (Irvine 1978). Our explanation of plagioclase features in the intrusion is highly speculative. Work in progress focuses on such critical aspects as exact location of trend reversals and determination of whether reversals are also present at the mineral-graded anorthositic layers and of whether mixtures of plagioclase of different generations are present at reversal horizons, as might be expected. These studies are part of work supported by National Science Foundation grant DPP 77-22765 to the Geologic Division, U.S. Geological Survey. References Abel, K. D., Himmelberg, C. R., and Ford, A. B. 1979. Petrologic studies of Dufek intrusion: Plagioclase variation. Antarctic Journal of the U.S., 14(5), 6-8. Bottinga, Y., and Weill, D. F. 1970. Densities of liquid silicate systems calculated from partial molar volumes of oxide components. American Journal of Science, 269, 169-182. Campbell, I. H., Roeder, P. L., and Dixon, J. M. 1978. Plagioclase buoyancy in basaltic liquids as determined with a centrifuge furnace. Contributions to Mineralogy and Petrology, 67, 369-377. Ford, A. B. 1976. Stratigraphy of the layered gabbroic Dufek intrusion, Antarctica. U.S. Geological Survey Bulletin, 1405-D. Ford, A. B., Reynolds, R. L., Huie, C., and Boyer, S. J . 1979. Geological field investigation of Dufek intrusion. Antarctic Journal of the U.S., 14(5), 9-11. Himmelberg, C. R., and Ford, A. B. 1976. Pyroxenes of the Dufek intrusion, Antarctica. Journal of Petrology, 17, 219-243. Himmelberg, G. R., and Ford, A. B. 1977. Iron-titanium oxides of the Dufek intrusion, Antarctica. American Mineralogist, 62, 623-633. Irvine, T. N. 1978. Density current structure and magmatic sedimentation. Carnegie Institution of Washington Yearbook, 77, 717-725. Jackson, E. D. 1971. The origin of ultrámaflc rocks by cumulus processes. Fortschritte der Mineralogie, 48, 128-174. McBirney, A. R., and Noyes, R. M. 1979. Crystallization and layering of the Skaergaard intrusion. Journal of Petrology, 20,487-554. Morse, S. A. 1969. Layered intrusions and anorthosite genesis. In Y. W. Isachsen (Ed.), Origin of anothosite and related rocks (New York State Museum and Science Service Memoir, Vol. 18). Wager, L. R., and Brown, C. M. 1968. Layered Igneous Rocks. Edinburgh: Oliver and Boyd.
ANTARcTIC JOURNAL
