Incongruent Melting

Steven Dutch, Professor Emeritus, Natural and Applied Sciences, Universityof Wisconsin - Green Bay


Many minerals do not melt uniformly. Instead they decompose as they melt. One exampleis potassium feldspar (KAlSi3O8) which decomposes on melting toleucite (KAlSi2O6) plus liquid. Another is forsterite (Mg2SiO4),which decomposes to enstatite (MgSiO3) plus liquid.

Both forsterite and leucite are chemically incompatible with quartz. They would reactwith quartz as follows:

However, this doesn't mean that potassium feldspar (KAlSi3O8)decomposes on melting to leucite (KAlSi2O6) plus quartz liquid. Whatactually happens is that most of the potassium feldspar melts but the rest converts toleucite. The melt actually contains appreciable potassium feldspar.

This page assumes you are reasonably familiar with the behavior of simple eutectic systems.

INCONG00.gif (2024 bytes) Let's consider the system forsterite-enstatite-quartz. We know:
  • Forsterite is incompatible with quartz
  • Enstatite is compatible with both quartz and forsterite
  • Enstatite decomposes to forsterite plus liquid when it melts.
  • None of these minerals form solid solutions with each other.
Thus we can expect the phase diagram to look as at left. We expect a field where solid enstatite and forsterite occur, and another field where solid enstatite and quartz occur. When we melt pure enstatite, we expect to travel up a straight line of enstatite composition, terminating in a field where we have forsterite plus liquid. Finally, we expect sharp eutectic-like liquidus curves as opposed to solid-solution behavior.
INCONG01.gif (2194 bytes) Let's read the fields as they are now. We can identify four possibilities, labeled by A-D.

INCONG02.gif (2617 bytes) Sketching in the liquidus curves, we can infer they look pretty much as at left. Things like actual melting points and the location of the quartz-enstatite eutectic, of course, have to be determined experimentally. The one major remaining question is what happens between the Fo+Liq and the En+Liq fields?
incong03.gif (2666 bytes) Figuring out the rest works best if we start from the bottom. If we melt a mixture of enstatite plus a little quartz (at the bottom of zone B), we will get enstatite plus liquid. The solid enstatite will coexist with the liquid until we reach the conversion temperature, at which point it should convert to forsterite plus liquid. Thus we expect a boundary along the conversion temperature of enstatite as shown. Above that line there can be no enstatite.
incong04.gif (4200 bytes) We can divide this diagram into four zones as shown. The history of a melt in each is different.

Zone A

INCONGza.gif (3894 bytes) In Zone A we go from melt to melt + forsterite to solid enstatite plus forsterite. This is pretty similar to a eutectic with one twist.
  • At a, the melt begins cooling.
  • At b, the melt has reached the liquidus and forsterite begins to form.
  • At c, a bit more than half the melt has solidified.
  • At d, we have reached the melting point of enstatite. We are on the boundaries of fields containing both forsterite and enstatite. Therefore we must have both solid phases present, and enstatite begins to form. But here's the twist: when enstatite melts, some of it goes into the liquid and some converts to forsterite. As it cools, the reverse must happen. Some enstatite forms directly from the melt, but some forms at the expense of forsterite. This is quite different from what happens at a eutectic.
  • At e, enstatite is forming and the solid composition moves toward enstatite. When it reaches the original system composition, the system is completely solidified.

Note that as the liquid cools, it can actually become more silica rich than enstatite itself. Once it does, it is actually chemically incompatible with forsterite and when the melt reaches the solidification temperature of enstatite, forsterite reacts with the melt to form enstatite. This actually happens in real igneous rocks and is why Bowen's Series is called Bowen's Reaction Series. Many rocks show signs that early-formed crystals dissolved in their own magma.

We can verify that forsterite actually decreases from the diagram. Assume we begin with 100 grams of melt. When the melt and solid reach d, we can see that about 2/3 of the melt has solidified (67 grams). But when the solid entirely solidifies (equals the originalsystem composition), we see that the system is about 50 per cent each. Thus the finalsystem has 50 grams of forsterite, a loss of 17 grams. We might conjecture (correctly, itturns out) that if the system has the compostion of pure enstatite, the forsterite willdisappear entirely.

Zone B

incongzb.gif (4161 bytes) This zone has the most complex history of all. We go from melt to forsterite plus melt to enstatite plus melt to solid enstatite plus quartz. Therefore we must form forsterite, then lose it completely.
  • At a, the melt begins cooling.
  • At b, the melt has reached the liquidus and forsterite begins to form.
  • At c, about a third of the melt has solidified.
  • At d, we have reached the melting point of enstatite. We are on the boundaries of fields containing both forsterite and enstatite. Therefore we must have both solid phases present, and enstatite begins to form, some directly from the melt, some at the expense of forsterite.
  • At e, enstatite is forming and the solid composition moves toward enstatite.
  • At f, the solid composition is equal to enstatite and forsterite must be completely gone.

But that's not the end of the story. The solid has not reached the initial systemcomposition yet and there must still be some melt present (about 20% judging from thediagram). Between f and the original system composition (open square d), both adjacentfields contain liquid. Therefore the system cannot completely solidify by movingdirectly right from f. The only way the system can solidify completely is to movedown to a field where everything is solid: the enstatite + quartz field.

The only solid that can form now is enstatite. The solid composition moves down thevertical enstatite line, and the melt moves down the liquidus to the eutectic, wherequartz begins to form. The history is like that of Zone C, shown below.

In reality this is a quite narrow zone out in the middle of the diagram, thus thissequence of evolution is uncommon. Also it lies far from both forsterite and quartz incomposition, so the amounts of quartz and forsterite that can form are fairly small. Itmay happen that the olivine doesn't react completely with the melt before being rimmed byenstatite. When that happens the forsterite is isolated from the melt and may persist whenthe rock solidifies. Thus on rare occasions we do find rocks with both olivine and quartz,but they are not in direct physical contact.

In extreme cases of magmatic segregation, we could start with a very mafic magma (ZoneA), and form a great deal of forsterite which sinks to the bottom of the magma chamber.That would have the effect of resetting the system composition equal to that of the melt,because the segregated olivine is effectively removed from the system. The system couldnow be in zone B. Thus we could easily (and do) end up with a layered igneous complex witholivine-bearing rocks at the base and quartz-bearing rocks at the top.

Zone C

incongzc.gif (4238 bytes) Zone C is simple; it is just a binary eutectic.
  • At a, the melt begins cooling.
  • At b, the melt has reached the liquidus and enstatite begins to form.
  • At c, about a quarter of the melt has solidified.
  • At d, we have reached the eutectic. Quartz begins to form along with enstatite.
  • At e, enough quartz has formed to push the solid composition well to the right. When the solid composition reaches the original system composition, the system is completely solidified.

Although part of the solid composition path lies in Zone B, that has no effect on the evolution of the melt.

Zone D

INCONGzd.gif (3988 bytes)  Zone D is simple; it is also just a binary eutectic.
  • At a, the melt begins cooling.
  • At b, the melt has reached the liquidus and quartz begins to form.
  • At c, about a third of the melt has solidified.
  • At d, we have reached the eutectic. Enstatite begins to form along with quartz.
  • At e, enough enstatite has formed to push the solid composition to the left. When the solid composition reaches the original system composition, the system is completely solidified.

A Familiar (But Paradoxically Little Known) Example

This is a familiar system but few people really know what goes on. It's the system water - NaCl

On the left side, we see the familiar fact that salt lowers the melting point of water, so that a solution of 23% salt melts at -21 C. This is why you put salt in ice to make homemade ice cream and why you do not put salt in ice for first aid. It's the familiar example of a eutectic.

However, what crystallizes out of a saturated salt water solution at low temperatures isn't salt. It's a hydrate of NaCl called hydrohalite and is a recognized mineral in places like the lakes in the dry valleys of Antarctica, where conditions are suitable for it to form. Hydrohalite is 61% NaCl and 39% water.

When we heat hydrohalite ("heat" being a relative term because we only get to near zero C) it doesn't melt. Instead, it breaks down to brine plus halite. The water "sweats" out of the hydrohalite, or the solid hydrohalite turns to a mush of saturated brine plus NaCl. The process here is just what happens with enstatite melting to liquid plus olivine, just at a much lower temperature.

If the thought of salt melting at 0 C bothers you, consider this. Despite the name, hydrohalite has two molecular units of water for every one of NaCl. It has a low density (1.61) and a pseudohexagonal crystal structure (actually monoclinic). Rather than regard it as a hydrate of sodium chloride, it is just as accurate to regard it as ice with incorporated NaCl (although actually its structure is distinct from both.) Once we regard it as salty ice, the mystery of it breaking down at 0 C ceases to be a mystery.


Return to Mineralogy-Petrology Index
Return to Thin-Section Index
Return to Crystals and Light Index
Return to Crystal Structures Index
Return to Mineral Identification Tables
Return to Professor Dutch's Home Page

Created 15 October 2009, Last Update

Not an official UW-Green Bay Site