1. Chapter 12: Layered Mafic Intrusions
Table 12.1
. Some Principal Layered Mafic Intrusions
Name
Age
Location
Area
(km 2 )
Bushveld
Precambrian
S. Africa
66,000
Dufek
Jurassic
Antarctica
50,000
Duluth
Minnesota, USA
4,700
Stillwater
Montana, USA
4,400
Muskox
NW Terr. Canada
3,500
Great Dike
Zimbabwe
3,300
Kiglapait
Labrador
560
Skaergård
Eocene
East Greenland
100
Super Laboratories for the study of fractionation
Hot
Relatively low viscosity
Slow cooling
Mafic intrusions come in all sizes
from thin dikes and sills
up to the huge 66,000 km2 x 9 km thick Bushveld intrusion of South Africa
Can occur in any tectonic environment where basaltic magma is generated
Large or particularly well-studied LMIs exposed in continents (many in flood basalt provinces)

2. The form of a typical LMI
Figure From Irvine and Smith (1967), In P. J. Wyllie (ed.), Ultramafic and Related Rocks. Wiley. New York, pp
The Muskox Intrusion

3. Layering
layer: any sheet-like cumulate unit distinguished by its compositional and/or textural features
uniform mineralogically and texturally homogeneous
Crystal accumulation and layering distinguish LMIs

4. Uniform Layering
Figure 12.3b. Uniform chromite layers alternate with plagioclase-rich layers, Bushveld Complex, S. Africa. From McBirney and Noyes (1979) J. Petrol., 20,
Uniform Layering of magnetite and plagioclase, Bushveld

5. Layering
layer: any sheet-like cumulate unit distinguished by its compositional and/or textural features
uniform mineralogically and texturally homogeneous
non-uniform vary either along or across the layering
graded = gradual variation in either
mineralogy
grain size - quite rare in gabbroic LMIs
Crystal accumulation and layering distinguish LMIs

6. Graded Layers
Figure Modal and size graded layers. From McBirney and Noyes (1979) J. Petrol., 20,
Left: Modal layering of olivine and plagioclase, Skaergaard
Right: Size layering Opx and Plag, Duke Island

7. Layering (or stratification)
Addresses the structure and fabric of sequences of multiple layers
1) Modal Layering: characterized by variation in the relative proportions of constituent minerals
may contain uniform layers, graded layers, or a combination of both

8. Layering (or stratification)
2) Phase layering: the appearance or disappearance of minerals in the crystallization sequence developed in modal layers
Phase layering transgresses modal layering
Analogy with sediment stratigraphy: layers are like sedimentary beds, layering is like bedding sequences, and phase layering is like formations
As with formations, phase layering is commonly used to delineate subdivisions in the layered sequences of differentiated LMIs

9. 3) Cryptic Layering (not obvious to the eye)
Systematic variation in the chemical composition of certain minerals with stratigraphic height in a layered sequence
Transgresses both modal and phase layering
Example: Mg/(Mg+Fe) in mafic phases and Ca/(Ca+Na) in plagioclase typically decrease upward from the floor

10. The regularity of layering
Rhythmic: layers systematically repeat
Macrorhythmic: several meters thick
Microrhythmic: only a few cm thick
Intermittent: less regular patterns
A common type consists of rhythmic graded layers punctuated by occasional uniform layers

11. Rythmic and Intermittent Layering
Figure 12.3a. Vertically tilted cm-scale rhythmic layering of plagioclase and pyroxene in the Stillwater Complex, Montana.
Left: Plag – pyroxene, Stillwater
Right: Intermittent graded and non-graded layers, Skaergaard
Figure Intermittent layering showing graded layers separated by non-graded gabbroic layers. Skaergård Intrusion, E. Greenland. From McBirney (1993) Igneous Petrology (2nd ed.), Jones and Bartlett. Boston.

12. The Bushveld Complex, South Africa
The biggest:
km x 9 km
Lebowa granitics
intruded 5 Ma
afterward
Simplified geologic Map and cross section of the Bushveld complex. From The Story of Earth & Life McCarthy and Rubidge

13. Marginal Zone: the lowest unit, is a chill zone about 150 m thick
Fine-grained norites from the margin correspond to a high-alumina tholeiitic basalt

14. Stratigraphy Basal Series
Thin uniform dunite cumulates alternating with orthopyroxenite and harzburgite layers
The top defined as the Main Chromite Layer
Figure Stratigraphic sequence of layering in the Eastern Lobe of the Bushveld Complex. After Wager and Brown (1968) Layered Igneous Rocks. Freeman. San Francisco.

15. Critical Series Plagioclase forms as a cumulate phase (phase layering)
Norite, orthopyroxenite, and anorthosite layers etc
Fine-scale layering is impressively developed. The layers are quite parallel over remarkable distances
Striking rhythmic repetition of thin, homogeneous or graded layers
Figure Stratigraphic sequence of layering in the Eastern Lobe of the Bushveld Complex. After Wager and Brown (1968) Layered Igneous Rocks. Freeman. San Francisco.

16. The Merensky Reef
~ 150 m thick sequence of rhythmic units with cumulus plagioclase, orthopyroxene, olivine, and chromite
The famous Pt-Pd and sulfide-bearing sub-unit is an orthopyroxenite-olivine-chromite cumulate layer
1-5 m thick and can be followed for nearly 200 km in the western lobe and 150 km in the eastern lobe, some 200 km apart (a total of 550 km)!
It may mark the horizon at which a major fresh surge of magma was introduced into the chamber
The Pt and Pd derived from the residual liquid of the earlier stage, and the sulfides from an immiscible sulfide liquid associated with the latter stage
Figure Stratigraphic sequence of layering in the Eastern Lobe of the Bushveld Complex. After Wager and Brown (1968) Layered Igneous Rocks. Freeman. San Francisco.

17. Main Zone
the thickest zone and contains thick monotonous sequences of hypersthene gabbro, norite, and anorthosite
Cumulus olivine and chromite are absent
The layering is poorly developed compared to the lower units, but still present
Figure Stratigraphic sequence of layering in the Eastern Lobe of the Bushveld Complex. After Wager and Brown (1968) Layered Igneous Rocks. Freeman. San Francisco.

18. Upper Zone
Appearance of cumulus magnetite (Fe-rich)
Well layered: anorthosite, gabbro, and ferrodiorite
Numerous felsic rock types = late differentiates

19. Also note: Cryptic layering: systematic change in mineral compositions
Reappearance of Fe-rich olivine in the Upper Zone
Reappearance of olivine in Upper zone is explained in text
A stunning example of the problems we face in interpreting cooling and magmatic differentiation processes in what is probably the optimal natural situation: a large, slow-cooling chamber of hot, low viscosity magma
Figure Stratigraphic sequence of layering in the Eastern Lobe of the Bushveld Complex. After Wager and Brown (1968) Layered Igneous Rocks. Freeman. San Francisco.

20. Figure The Fo-Fa-SiO2 portion of the FeO-MgO-SiO2 system, after Bowen and Schairer (1935) Amer. J. Sci., 29,
The pyroxene + liquid field is bounded by the 1557oC-1305oC ternary peritectic curve on one side and the 1543oC-1178oC pyroxene-silica cotectic curve on the other
The field pinches out toward higher Fe content
Perfect fractional crystallization of a cooling liquid of bulk composition a follows the schematic path a-b-c-d-e-. Fe-poor olivine forms at point a, and the liquid evolves toward point b. Because olivine is separated as it forms, the peritectic reaction (Ol + Liq = Opx) is not possible. Rather, Fe-poor orthopyroxene forms, and the liquid progresses directly away from the pyroxene composition to point c, where tridymite forms. Fe-rich olivine reappears with pyroxene and tridymite at point d. Temperatures of invariant points are in degrees Celsius. The immiscible liquid solvus in the silica-rich field is ignored.

21. How can we explain the conspicuous development of rhythmic layering of often sharply-defined uniform or graded layers?
Experimental (and natural) systems show that a cooling magma will first crystallize one liquidus phase, joined by progressively more phases as the liquid line of descent reaches the cotectic and higher order eutectics
The repetition requires either some impressively periodic reinjection of fresh magma, or cyclic variation in one or more physical properties if it is to be produced by gravitational crystal settling alone
The pattern of cryptic layering, however, indicates a progressive differentiation that spans the full vertical height of the intrusion, precluding any model based solely on replenishment

22. The Stillwater Complex, Montana
Figure After Wager and Brown (1968) Layered Igneous Rocks. Freeman. San Francisco.

23. Stratigraphy Basal Series
a thin ( m) layer of norites and gabbros
Ultramafic Series base = first appearance of copious olivine cumulates (phase layering)
Lower Peridotite Zone
20 cycles ( m thick) of macrorhythmic layering with a distinctive sequence of lithologies
The series begins with dunite (plus chromite), followed by harzburgite and then orthopyroxenite
Upper Orthopyroxenite Zone
is a single, thick (up to 1070 m), rather monotonous layer of cumulate orthopyroxenite

24. olivine + orthopyroxene  orthopyroxene 
The crystallization sequence within each rhythmic unit (with rare exception) is:
olivine + chromite 
olivine + orthopyroxene 
orthopyroxene 
orthopyroxene + plagioclase 
orthopyroxene + plagioclase + augite
This is a common basaltic crystallization sequence, which, with the sharpness of the base of the dunite layer, suggests that each sequence is initiated by some major change in the crystallization conditions followed by a period of cooling and crystal accumulation
Once again we are faced with the problem of explaining the repetition of the cycles

25. Stratigraphy The Banded Series
Sudden cumulus plagioclase ® significant change from ultramafic rock types (phase layering again)
The most common lithologies are anorthosite, norite, gabbro, and troctolite (olivine-rich and pyroxene-poor gabbro)
Such a sudden and dramatic change suggests the introduction of a second principal magma type into the Stillwater magma chamber

26. Figure 12-9. Composite stratigraphic column of the Stillwater Complex
Figure Composite stratigraphic column of the Stillwater Complex. After McCallum et al., (1980) Amer. J. Sci., 280-A, 59-87, and Raedeke and McCallum (1984) J. Petrol., 25,

27. The Skaergård Intrusion E. Greenland
The “type locality” for LMIs. It is now perhaps the most intensely studied igneous body in the world
Figure After Stewart and DePaolo (1990) Contrib. Mineral. Petrol., 104,

28. Fine-grained chill margin
Magma intruded in a single surge (premier natural example of the crystallization of a mafic pluton in a single-stage process)
Fine-grained chill margin
Fine-grained chill margin, about a meter thick, but, like most LMI margins, it has been contaminated, and is no longer representative of the original magma

29. Stratigraphy Skaergård subdivided into three major units:
Layered Series
Upper Border Series
Marginal Border Series
Upper Border Series and the Layered Series meet at the Sandwich Horizon (most differentiated liquids)
It is generally agreed that the Layered Series crystallized from the floor upward, the Upper Border Series from the roof downward, and the Marginal Border Series from the walls inward

30. Cross section looking down dip.
Figure After After Hoover (1978) Carnegie Inst. Wash., Yearb., 77,

31. Upper Border Series: thinner, but mirrors the 2500 m Layered Series in many respects
Cooled from the top down, so the top of the Upper Border Series crystallized first
The most Mg-rich olivines and Ca- rich plagioclases occur at the top, and grade to more Fe-rich and Na- rich compositions downward
Major element trends also reverse in the Upper Border Series as compared to the LBS
The gravity-driven crystal settling model faces further problems when we attempt to explain the downward increase in differentiation if the first-formed crystals are to have settled in that direction. In this case the higher temperature crystals are near the roof
Graded layers with heavy mafics concentrated toward the top and plagioclase toward the bottom do not correspond to the gravity settling hypothesis

32. Sandwich Horizon, where the latest, most differentiated liquids crystallized
Ferrogabbros with sodic plagioclase (An30), plus Fe-rich olivine and Opx
Granophyric segregations of quartz and feldspar
F & G = immiscible liquids that evolve in the late stages of differentiation?

33. Stratigraphy, Modal, and Cryptic Layering (cryptic determined for intercumulus phases)
Figure After Wager and Brown (1968) Layered Igneous Rocks. Freeman. and Naslund (1983) J. Petrol., 25,

34. Chemistry of the Skaergård
Figure After McBirney (1973) Igneous Petrology. Jones and Bartlett.
SiO2, MgO and FeO for both whole-rock analyses and estimated liquid compositions for the Layered Series
SiO2 as a differentiation index? Now looks dubious for tholeiites
Compatible (Cr and Ni) and incompatible (Rb and Zr) trace elements for the complete section
Trends are compatible with differentiation of a single surge of magma
No evidence for any cyclic variations suggestive of repeated injections of fresh magma
Complicated by the presence of layering on a finer scale

35. The Processes of Crystallization, Differentiation, and Layering in LMIs
LMIs are the simplest possible case
More complex than anticipated
Still incompletely understood after a half century of intensive study
Fractional crystallization a la Bowen and crystal settling & accumulation at the bottom of the chamber explains the major subdivisions, many textural features, and the mineral-cryptic layering in many LMIs
But it fails to explain several other features, particularly the rhythmic layering and the upside-down layering in the Skaergård

36. Rhythmic modal layering most easily explained by crystal settling interrupted by periodic large-scale convective overturn of the entire cooling unit
Reinjection of more primitive magma may explain major compositional shifts and cases of irregular cryptic variations
Rhythmic modal layering most easily explained by crystal settling interrupted by periodic large-scale convective overturn of the entire cooling unit
A sequence is deposited: settling of denser crystals beneath lighter ones, and expulsion of the late differentiated liquid
Overturn removes the late liquid, rehomogenizes the system, and the process would then be repeated
Each cycle would be more evolved due to the removal of the phases in the rhythmic unit, resulting in the phase and cryptic patterns

37. Problems with the crystal settling process.
Many minerals found at a particular horizon are not hydraulically equivalent
Size is more important than density in Stokes’ Law, but size grading is rare in most LMIs
Dense olivine in the Upper Border Series of the Skaergård
Plagioclase is in the lower layers of the Skaergård
Many of the minerals found at a particular horizon in LMIs are not hydraulically equivalent (they would not be expected to settle at the same rate via Stokes’ Law)
Size is more important than density in Stokes’ Law (eq. 11-1) since the radius term is squared, and thus size-graded layers should be more common than modal layering. Size grading is rare in most LMIs
Why is dense olivine in the Upper Border Series of the Skaergård when it should have sunk to the bottom?
Plagioclase is in the lower layers of the Skaergård, but is less dense that the liquid from which it crystallized. Why didn’t plagioclase float?

38. The Marginal Border Series shows vertical layering
Inverted cryptic variations in the Upper Border Series suggests that the early-formed minerals settled upward
The Marginal Border Series shows vertical layering
Basaltic magmas develop a high yield strength, slightly below liquidus temperatures
Cryptic variations in the Upper Border Series are inverted with the most An- rich plagioclase and Mg-rich mafics at the top, suggesting that the early- formed minerals, including heavy mafics, settled upward if they settled at all
The Marginal Border Series shows vertical layering, with mafic minerals concentrated toward the margin in graded layers
Basaltic magmas develop a high yield strength, slightly below liquidus temperatures. Crystal settling should not occur soon after crystallization began, particularly if there is some convective motion to keep the crystals suspended

39. In-Situ Processes
Nucleation and growth of minerals in a thin stagnant boundary layer along the margins of the chamber
Differential motion of crystals and liquid is still required for fractionation
Dominant motion = migration of depleted liquid from the growing crystals
Crystals settle (or float) a short distance within the boundary layer as the melt migrates away
Boundary layer interface inhibits material motion
Differential motion of crystals and liquid is still required for fractionation by any in situ process
Dominant motion = migration of depleted liquid from the growing crystals by diffusive and/or convective processes
The crystals only settle (or float) a short distance within the boundary layer as the melt migrates from the crystals in accordance with its density
The boundary layer interface inhibits material motion
Plagioclase will not easily rise out of a layer at the bottom of a chamber, and olivine will not easily sink from one at the top

40. Compositional Convection
Systems with gradients in two or more properties (chemical or thermal) with different rates of diffusion
Especially if have opposing effects on density in a vertical direction
Compositional Convection
Of particular interest for layering are systems in which there are gradients in two or more properties (chemical or thermal) with different rates of diffusion
If these gradients have opposing effects on the density of the fluid in a vertical direction, a wide range of novel and complex convective phenomena may occur, collectively known as compositional convection
Consider a container of pure water heated from below
The result is a convecting system that extends the full vertical length of the container, passing heat upward
A dumb and boring experiment!
Next try a container with an aqueous NaCl solution that has a salinity gradient in which the concentration (and density) of the solution increases toward the base

41. One gradient (in this case rtemp) is destabilizing (although the total density gradient is stable)
The diffusivity of the destabilizing component (heat) is faster than the diffusivity of the salt
Figure After Turner and Campbell (1986) Earth-Sci. Rev., 23,

42. Double-diffusive convection situation A series of convecting layers
Figure After Turner and Campbell (1986) Earth-Sci. Rev., 23,
The result: A classic double-diffusive convection situation
Opposed gradients with different diffusivities -> series of convecting layers rather than a single large overturning convection cell
Heat  convection only in a thin bottom layer, which homogenizes both the salinity and thermal density gradients
Heat diffuses faster than the salt, so it is transferred across the diffusive interfaces between the convecting layers and causes a thermal density gradient to develop in the next layer upward
This creates convective overturn in that layer which homogenizes the salinity and thermal gradients within it and overcomes the thermal density instability while transferring heat upward, where it will diffuse through the interface to the next layer etc.
-> a series of layers, each with a different composition and temperature, and the gradual upward transfer of heat from the bottom of the container to the top by convection within layers and diffusion across the interfaces between them
The layers operate in the same fashion as the two-layer mantle model

43. Dense crystals held in suspension by agitation
Density currents
Cooler, heavy-element-enriched, and/or crystal-laden liquid descends and moves across the floor of a magma chamber
Dense crystals held in suspension by agitation
Light crystals like plagioclase also trapped and carried downward
Another model- Density currents
cooler, heavy-element-enriched, and/or crystal-laden liquid may descend (perhaps along the cool walls) and move across the floor of a magma chamber and deposit layers
Irvine (1980) proposed gravity-driven currents in several LMIs, supported by such current features as scour-and-fill channels and cross-bedding
Dense crystals are held in suspension by the agitation of the flow
Light crystals like plagioclase can also be trapped and carried downward in the currents

44. Figure 12. 15a. Cross-bedding in cumulate layers. Duke Island, Alaska
Figure 12.15a. Cross-bedding in cumulate layers. Duke Island, Alaska. Note also the layering caused by different size and proportion of olivine and pyroxene. From McBirney (1993) Igneous Petrology. Jones and Bartlett
Figure 12.15b. Cross-bedding in cumulate layers. Skaergård Intrusion, E. Greenland. Layering caused by different proportions of mafics and plagioclase. From McBirney and Noyes (1979) J. Petrol., 20,

45. Neil Irving’s Vortex model
Figure After Irvine et al. (1998) Geol. Soc. Amer. Bull., 110,
Cross section of a vortex cell in a density surge current along the boundary layer between floor cumulates (shaded) and magma. Streamlines and arrows portray the instantaneous flow relations relative to points V, R, and S. The floor is actually stationary, of course, and points V, R, and S move as shown with the arrow at V. Flow within the surge moves faster than the vortex (as in a tractor tread), so that the flow material separates from the floor at S and reattaches at R. Forward motion of the vortex results in deposition between it and the chamber floor.
Black flow lines and arrows indicate motion relative to the cell

46. Figure 12-17. After Irvine et al. (1998) Geol. Soc. Amer. Bull
Density currents initiate at cool roof and descend along walls to cross floor
Create layering along whole path- roof, walls, floor
Periodic slumps and drops of accumulated material (autoliths) -> scour/fill and craters

47. Figure Cold plumes descending from a cooled upper boundary layer in a tank of silicone oil. Photo courtesy Claude Jaupart.

48. Figure Schematic illustration of the density variation in tholeiitic and calc-alkaline magma series (after Sparks et al., 1984) Phil. Trans. R. Soc. Lond., A310,

49. Figure Schematic illustration of a model for the development of a cyclic unit in the Ultramafic Zone of the Stillwater Complex by influx of hot primitive magma into cooler, more evolved magma. From Raedeke and McCallum (1984) J. Petrol., 25,