Differentiation is the process by which different ingredients separate from an originally homogenous mixture. An example is the separation of whole milk into cream and nonfat milk. In the early part of the twentieth century, N. L. Bowen conducted a series of laboratory experiments demonstrating that differentiation is a plausible way for silicic and mafic rocks to form from a single parent magma.

Bowen's Reaction Series is the sequence in which minerals crystallize from a cooling magma, as demonstrated by Bowen's laboratory experiments. In simplest terms, Bowen's reaction series shows that those minerals with the highest melting temperatures crystallize from the cooling magma before those with lower melting points. However, the concept is a bit more complicated than that.

Crystallization begins along two branches, the discontinuous branch and the continuous branch. In the discontinuous branch, one mineral changes to another at discrete temperatures during cooling and solidification of the magma. Changes in the continuous branch occur gradationally through a range in temperatures and affect only the one mineral, plagioclase. Although crystallization takes place simultaneously along both branches, we must explain each separately.

Discontinuous Branch
All minerals in the discontinuous branch are ferromagnesian. In this branch, as the completely liquid magma slowly cools, it reaches the temperature at which olivine begins to crystallize from the magma. Olivine is a mineral with an exceptionally high proportion (2:1) of iron and magnesium to silicon - its formula is (Fe, Mg)₂SiO₄. The liquid left after olivine has crystallized is relatively depleted in iron and magnesium and relatively enriched in silicon (because only one part of Si is used for two parts of Fe and Mg).

As the melt cools further, the melting temperature for the next mineral of the series is reached, and pyroxene begins to crystallize at olivine's expense. The previously formed olivine now reacts with the remaining melt, and the original crystal structure of olivine rearranges into that of pyroxene (from isolated silicon-oxygen tetrahedrons to single chains of tetrahedrons). The crystal structure of pyroxene, with a formula of (Mg, Fe)SiO₃, accommodates a higher amount of silicon relative to the iron and magnesium - a ratio of 1 to 1. After all of the olivine has reacted with the melt to form pyroxene, the temperature of the magma can decrease and pyroxene will crystallize directly from the melt.

However, if the original melt was basaltic, all of the liquid would likely be used up before all of the olivine has reacted with the melt. In this case the rock formed would have only olivine and pyroxene as its ferromagnesian minerals, which (along with plagioclase that crystallized simultaneously in the continuous branch) would be a basalt. If, on the other hand, the original melt were more silicic or if early formed ferromagnesian minerals were removed from the melt, there would still be melt left after pyroxene crystallized and the next mineral (amphibole) could crystallize.

Assuming there is melt remaining when the crystallization temperature for amphibole is reached, pyroxene reacts with that melt. Its crystal structure rearranges into amphibole's double chains of silicon-oxygen tetrahedrons. More of the silicon leaves the melt (along with aluminum, calcium and minor amounts of sodium) and is incorporated into the newly developing amphibole crystals.

If melt is left after amphibole has formed, on further cooling amphibole reacts with the melt to produce biotite (which is a sheet silicate). Biotite is the last of the ferromagnesian minerals to crystallize. Any magma remaining after biotite has finished crystallizing contains very little iron or magnesium.

Continuous Branch
Plagioclase feldspar is the only mineral in the continuous branch. Silicon and aluminum, which are part of all feldspars, combine with calcium and sodium to form plagioclase. Calcium-rich plagioclase will crystallize first and, upon slow cooling, increasingly more sodic plagioclase will crystallize. If a basaltic melt, which is enriched in calcium relative to sodium, is cooled very slowly, a very calcium-rich plagioclase will crystallize first. With progressive cooling, the plagioclase crystals react with the melt and grow larger. The growing plagioclase crystals react with the melt and grow larger. The growing plagioclase crystals will have an increasingly higher amount of sodium relative to calcium. Crystallization will stop when the plagioclase crystals have the same calcium-to-sodium ratio as did the original magma. In the case of a basaltic magma this will be at a fairly high temperature (approximately the temperature at which pyroxene crystallizes in the discontinuous branch). If there is a lower ratio of calcium to sodium (or if calcium-rich plagioclase is removed from the melt), plagioclase will continue to crystallize through lower temperatures.

Any magma left after the crystallization is completed along the two branches is richer in silicon than the original magma and also contains abundant potassium and aluminum. The potassium and aluminum combine with silicon to form potassium feldspar (If the water pressure is high, muscovite may also form at this stage). Excess SiO₂ crystallizes as quartz.

Normally a newly erupted cooling basalt lava progresses only a short distance down the reaction series before all the magma is consumed by growing crystals. Olivine develops, but only part of it reacts with the melt to form pyroxene before all the magma is solidified. Simultaneously, calcium-rich plagioclase grows and becomes increasingly sodic; but its growth ceases when all liquid is consumed. The rock becomes a completely solid aggregate of calcium-rich plagioclase, pyroxene, and olivine - in other words, what one expects to find in a basalt.

However, Bowen used this experimentally determined reaction series to support his hypothesis that all magmas (mafic, intermediate, and silicic) derive from a single parent (mafic) magma by differentiation. The early-developing minerals are separated from the remaining magma. These minerals collectively result in a rock that is more mafic than the original magma. The remaining magma is deficient in iron, magnesium, and calcium; therefore, upon cooling, it solidifies into a silicic or intermediate rock.

Crystal Settling
Only if the original basaltic magma cools slowly, and the earliest-formed minerals physically separate from the magma, can the minerals on the lower part of the reaction series crystallize. Crystal settling is the downward movement of minerals that are denser (heavier) than the magma from which they crystallized. What is pictured happening is that as the olivine crystallizes from the magma, the crystals settle to the bottom of the magma chamber. Calcium-rich plagioclase also separates as it forms. The remaining magma is, therefore, depleted in calcium, iron, and magnesium. Because these minerals were economical in using the relatively abundant silica, the remaining magma becomes richer in silica as well as in sodium and potassium. If enough mafic ingredients are removed in this manner, the remaining residue of magma eventually solidifies into a granite.

Undoubtedly this method of differentiation does take place in nature, though probably not to the extent that Bowen envisioned. The lowermost portions of some large sills are composed predominantly of olivine, whereas the upper levels are considerably less mafic. Even in large sills, however, differentiation has rarely progressed far enough to produce any granite within the sill.

If we assume that the mafic minerals settle ever deeper in large magma bodies, there is still a problem in trying to explain the origin of granite by Bowen's theory. Calculations show that to produce a given volume of granite, about ten times as much mafic rock first has to form and settle out. If this is true, we would expect to find far more mafic plutonic rock than granite in the continental crust.

This is not to say that Bowen's work is discredited. Quite the opposite. His work has led to other theories on the behavior of magmas. Moreover, differentiation does occur and can explain relatively minor compositional variations within intrusive bodies, even if it does not satisfactorily explain the origin of large granite bodies.

Ore Deposits Due to Crystal Settling
Crystal settling accounts for important ore deposits that are mined for chromium and platinum. Most of the world's chromium and platinum come from a huge sill in South Africa. The sill, the famous Bushveldt Complex, is 8 kilometers thick and 500 kilometers long. Layers of chromite (a chromium-bearing mineral) up to 2 meters thick are found and mined, at the base of the sill. Layers containing platinum overlie the chromite-rich layers.

For additional information, read Chapter 3 of your textbook.