Using the Universal Stage to decode the cryptic cooling record of igneous rocks
Marian B. Holness
Department of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ, UK.
Decoding rock history is very much detective work: we can use the mineral assemblage, the structure of individual minerals, or the way in which its constituent mineral grains fit together to create a picture of how it got to where it is now. Having been exposed at an impressionable age to the Natural Sciences course at Cambridge University, where I studied materials science, metallurgy and physics before specialising in geology, it's the third approach that interests me most. This catholic education opened my eyes to textural features that might otherwise have passed unnoticed, and using Cambridge's vast collection of metamorphic and igneous thin sections (165,000 and still counting) gave me a lasting feeling that the way into most geological problems lies down the microscope. I count myself lucky to have been taught optical mineralogy as an undergraduate, to be teaching these courses myself to the current generation of undergraduates, and to be in a department with a long tradition of petrography: I work in the shadow of greats such as Alfred Harker (who started the enormous collection which bears his name), Cecil. E. Tilley, Alec Deer and Stuart Agrell.
My current efforts are centred on the problem of solidification. When chemically complex liquids such as magmas solidify, they generally begin by crystallising a single phase. As this phase is progressively removed from the liquid the composition of the remaining liquid changes until it become saturated in another phase that begins to crystallise along with the first. The process continues, with progressive arrivals on the liquidus (or disappearances due to peritectic reactions) until solidification is complete. The complexity of the problem arises from the possibility of relative movement of solid and liquid. Gravitational settling of solid phases to the floor of the magma reservoir, or concentration of nucleation and growth on the margins of the reservoir, creates a mushy layer. If residual liquid can be extracted from this mushy layer, either by diffusion or by gravitationally-driven compaction, this will drive the composition of the liquid remaining in the bulk reservoir towards progressively more evolved compositions. Upwards-flow of expelled liquid may also create opportunities for reaction and metasomatism as these evolved interstitial liquids encounter higher levels of the mush. The behaviour of the mushy layer therefore has enormous influence on the evolution of magmas (and hence their eruptive style), the concentration of economic deposits of ore minerals, and the timing of eruptions. Understanding this intriguing problem involves constraining the physical properties of the mush: we need to know about its mechanical response to compaction, disruption and slumping, as well as to document the progressive occlusion of porosity and destruction of permeability. My first steps in the attempt to find a solution are directed at using grain boundaries to track the cooling history of solidifying plutons.
Photomicrograph of a glass-bearing gabbroic crystalline nodule from an Icelandic basaltic lava flow, showing the randomly oriented euhedral laths of plagioclase forming a framework in which the melt-filled pores (now partly filled by clinopyroxene) are essentially planar-sided.
A glass-bearing amphibole-dominated crystalline nodule entrained in an alkali basaltic flow in Western Turkey. Note how the melt-filled pores commonly have smoothly curved walls and a generally cuspate shape.
The frequency of plag-plag-melt dihedral angles in the Icelandic gabbro nodule, measured using the Universal Stage. Note the low median and high standard deviation.
The frequency of amphibole-amphibole-melt dihedral angles in the Turkish nodule shown in (b), measured using the Universal Stage. Note the low median and low standard deviation. The spread of observed 3-D angles is most likely due to the anisotropy of interfacial energies in this system.
The variation of angle population observed in a suite of amphibole-dominated glassy nodules from Western Turkey. The samples fall on an array connecting an initial growth-dominated pore structure with a fully texturally equilibrated pore structure. Where they lie on this array is controlled by the rate at which the melt-filled mush cooled before entrainment.
In the earliest stages of solidification, crystals essentially grow unimpeded: they generally adopt shapes controlled by the relative rates of growth of different faces, forming typical euhedral forms such as the tabular habit of plagioclase or elongate amphibole prisms. As solidification proceeds, these individual crystals begin to impinge on each other. If growth is sufficiently rapid, the crystals continue to grow with planar faces, dictated by the drive towards chemical equilibrium. This creates a mush with a pore structure first described by Elliott et al. (1997) as an impingement texture (Figure 1a). The angles formed by the impingement of these planar faces, forming the corner of a melt-filled pore, vary widely and depend on the random juxtaposition of adjacent grains. However, the median value of a population of these angles will be 60°, since the sum of included angles in a triangle is 180°. If the system is held at a constant temperature, or is allowed to cool extremely slowly, it will attempt to further reduce its total internal energy by minimising the amount of energy tied up in grain boundaries and interfaces. This process of textural equilibration involves the migration of interfaces towards the position of lowest energy. Typical of an equilibrated texture is the development of characteristic angles at the pore corners due to the balancing of interfacial energies at these junctions, together with the creation of interfaces with constant mean curvature (Figure 1b). This angle, known as the dihedral angle, is generally in the range 20 - 40° for a silicate liquid-solid aggregate and is very important in controlling pore structure and permeability. A monomineralic system with negligible anisotropy of interfacial energies (i.e. a single value of dihedral angle) contains an interconnected series of channels on three-grain edges even down to vanishingly low porosities if the melt-solid-solid dihedral angle is less than 60°: this is the case for all partially molten silicate systems of geological importance (e.g. Holness, 1996). Importantly, however, no mineral is completely isotropic with respect to interfacial energies: there will be a range of dihedral angles in a texturally equilibrated mush (Figure 1d) so some three-grain edges will be dry while others contain a fluid-filled channel (Laporte & Provost, 2000). We need to know what this range is, and this is where the Universal Stage comes into its own.
Measuring the characteristics of a population of angles between grain boundaries is something that has been done by the metallurgical community for some time. Since they are dealing with optically opaque materials they developed ways of interpreting the population of angles measured on a polished section, i.e. a random cut through a 3-D material. Early work by geologists followed suit, with many published studies documenting the variations of the median of a population of angles in geological systems, all measured on a conventional microscope stage. The median of such a population is very close to the true median of the population (Riegger & Van Vlack, 1960) but we lose valuable information by confining ourselves to this simple measure: what we really need is an idea of the true range of angles. While it is possible to construct numerical models which can be used to predict what a complex population might look like in a random 2-D cut, there is nothing like direct observation to constrain it properly and, of course, direct observation doesn't rely on model-dependent assumptions. As pointed out by Ron Vernon (1997) this can be neatly and elegantly achieved for rocks if you use a Universal Stage: any three-grain junction can be rotated so that the true angle between grain boundaries can be measured, and the true spread of the population can be observed. There is currently no other way of doing this - the Universal Stage is the only instrument capable of collecting this information. The Universal Stage also has the great advantage of letting us measure all angles in a section, even those cut so obliquely that they can't be properly seen on a conventional stage - this is particularly important in a coarse-grained sample in which the number of suitable junctions may be small.
There is more useful information encoded in the angles at pore corners in melt-bearing rocks. The median and standard deviation provides a measure of how close the crystal mush has approached textural equilibrium. A mush which is actively solidifying (e.g. Figure 1a) will have a population of angles with a median of 60° and a high standard deviation (Figure 1c), while one which is experiencing a period of constant temperature, and has therefore stopped crystallising, has the opportunity to reach textural equilibrium (Figure 1b): this will result in an angle population with a lower median and a lower standard deviation (Figure 1d). A suite of glass-bearing amphibole-dominated crystal nodules erupted by recent alkali basalt eruptions in Western Turkey shows this process frozen in time (Figure 1e). While some of the nodules have dihedral angle populations indicative of continuing crystallisation interrupted only by entrainment and eruption, other nodules provide evidence that they remained at high constant temperatures for a while, permitting variable extents of approach to textural equilibrium at the pore corners (Holness et al., 2005).
Photomicrographs of gabbroic rocks in various stages of textural equilibration.
Glass-bearing gabbroic nodule entrained in a basaltic lava flow from Iceland in which the pore spaces between plagioclase grains are almost entirely filled with clinopyroxene. The arrowed junction shows the pyroxene inheriting the original impingement angle between the two plagioclase grains. This is the starting point for fully solidified gabbros. Scale bar is 200 μm long.
Fully solidified gabbro from the Skaergaard Intrusion, East Greenland, in which the plag-plag-cpx junctions (arrowed) are beginning to rotate towards solid-state textural equilibrium. The original angles have been increased to something of the order of 80-90°. Scale bar is 200 μm long.
A fully texturally equilibrated metabasic granulite, with granoblastic texture and ~120° dihedral angles (arrowed). This is the point at which the previous two textures would arrive if they remained sufficiently hot for sufficiently long.
If we now consider a continuously crystallising system we need to think about the nucleation and growth of a second solid phase in the melt-filled pores. These pore-filling, or interstitial, phases commonly inherit the shape of the melt-filled pore, including the angles at the pore corners (Figure 2a). Irrespective of whether these pores had reached textural equilibrium or whether they still had the growth-controlled impingement texture, the median angle of the population inherited by the interstitial solid is much lower than that expected for solid-solid textural equilibrium (Figure 2c). These solid-state angles are of the order of 120°, since grain boundary energies are very similar regardless of the solid phases involved (Vernon, 1968). If the now-solid igneous rock is cooled slowly enough, it will attempt to reduce its internal energy by rotating grain boundaries into these higher angles (Figure 2b), and the extent to which it can do this is a function of its time-integrated thermal history (Holness et al., 2005). The angle population therefore provides an easily accessible measure of how long the rocks stayed hot. Since the closure temperature for this solid-solid dihedral angle change is rather high (it is accomplished by significant diffusive mass transport via grain boundaries) we can use the angle population as a thermal probe for processes occurring close to the magma-mush interface.
This provides us with an astonishingly sensitive new tool: a nice example of serendipity in science. One of the most powerful ways of using dihedral angles is to track changes in the liquidus assemblage. When I first started this work I concentrated on the plagioclase-rich cumulates of the Rum Layered Intrusion in which augite fills in gaps between the earlier-crystallising plagioclase grains. I discovered that the pyroxene-plagioclase-plagioclase angles in all the gabbroic layers (in which plagioclase, augite and olivine are liquidus phases) had medians hovering around 90°, while those in the troctolitic layers (in which only plagioclase and olivine are on the liquidus) were in the region of 80°. I had absolutely no idea why this might be until I looked at the Layered Series of the Skaergaard. Unlike Rum, which was an open chamber with a consequent random jumble of cumulate layers, Skaergaard crystallised as a closed system, with the liquid tracking down the phase diagram. The Layered Series is thus a set of progressively more fractionated layers and it was immediately apparent that the arrival of cumulus augite near the base of the chamber made the median pyroxene-plagioclase-plagioclase angle jump from about 80° to 100° (Holness et al., 2007a, b). The reason for this is absolutely fundamental and is to do with the change in slope of the liquidus at the arrival of a new phase. This change in slope means there is a step-change in the fractional contribution of the latent heat of crystallisation to the total enthalpy budget, and so the pluton takes longer to cool down: the cumulate have more time to approach textural equilibrium.
Materials scientists have known about this effect for a long time - many undergraduate courses include an experiment tracking the temperature of a crystallising binary system in a beaker on the benchtop - but this is the first time we have found a way of decoding fully crystalline rocks to detect the same thing. Suddenly there are a whole lot of interesting questions we can ask: at the moment I am busy using this new information to constrain the thickness of the mushy layer within the Skaergaard intrusion. It looks as though there was an unstable highly porous mushy layer on the walls, which got progressively thicker with time, reaching a maximum of about 200m. The continual erosion of the wall mush created detrital fans on the floor where the mush may have been several hundreds of metres thick: but far from the walls the floor mush was a few metres thick - crystallisation essentially occurred at a hardground.
Although it is possible to get an approximate idea of textural change using a conventional stage, the Universal Stage makes it possible to extract extraordinarily detailed information about cooling histories. Measuring angles is not difficult - it certainly doesn't exploit the full potential of the U-stage and the technique is very easy to learn - and perhaps the biggest obstacle is actually getting your hands on a functioning U-stage. I have derived immense satisfaction over the last few years in giving these beautiful and sophisticated instruments another lease of life, obtaining hitherto inaccessible information on the cooling history of igneous rocks.
Elliott, M.T., Cheadle, M.J. & Jerram, D.A. (1997) On the identification of textural equilibrium in rocks using dihedral angle measurements. Geology, 25, 355-358.
Holness, M.B. (2006) Melt-solid dihedral angles of common minerals in natural rocks. Journal of Petrology, 47, 791-800.
Holness, M.B, Cheadle, M.J., & McKenzie, D. (2005) On the use of changes in dihedral angle to decode late-stage textural evolution in cumulates. Journal of Petrology, 46, 1565-1583.
Holness, M.B., Nielsen, T.F.D. and Tegner, C. (2007a) Textural maturity of cumulates: a record of chamber filling, liquidus assemblage, cooling rate and large-scale convection in mafic layered intrusions. Journal of Petrology, 48, 141-157.
Holness, M.B., Tegner, C., Nielsen, T.F.D., Stripp, G., and Morse, S.A. (2007b) A textural record of solidification and cooling in the Skaergaard Intrusion, East Greenland. Journal of Petrology, 48, 2359-2377.
Laporte D., & Provost, A. (2000) Equilibrium geometry of a fluid phase in a polycrystalline aggregate with anisotropic surface energies: Dry grain boundaries. Journal of Geophysical Research, 105, 25937-25953.
Riegger, O.K. & Van Vlack, L.H.W. (1960) Dihedral angle measurement. Transactions of the Metallurgical Society of the AIME, 218, 933-935.
Vernon, R.H. (1997) Comment: On the identification of textural disequilibrium in rocks using dihedral angle measurements. By Elliott, M.T. & Cheadle, M.J. Geology, 25, 1055.
Vernon, R.H. (1968) Microstructures of high-grade metamorphic rocks at Broken Hill, Australia. Journal of Petrology, 9, 1-22.