As carbon is the element of life, silicon is the element of the earth
Silicon is an word known to the public, at least in the phrase "Silicon Valley," and is associated with computers, although the public probably does not know what a chemical element is, or how silicon is used in computers, or where it comes from, or how it behaves. The earth is constructed of silicon, and we use it every day in the form of glass and pottery. What carbon is to the living world, silicon is to the non-living. Its study is, therefore, interesting and wide-ranging, leading into many areas of technology and science. In this article, I shall strive to present the fundamentals of the most important of these areas, so that a useful understanding of silicon can be developed in those who seek it. Not only is there a great amount of information about silicon, but study brings up new questions at every turn. I am certain that somebody knows the answers to most of these questions, but I do not, and must leave them for further research. Indeed, the authorities often do not agree, figures may vary, and ignorance is glossed over in many cases. I shall do the best I can, and admit my ignorance where it appears.
In this large article, I shall first discuss the physical and chemical properties of silicon and its oxide silica. If it were not for the semiconductor applications of pure silicon, silica would be by far the more important topic. The earth is more precisely constructed of silica; except possibly for the core, the whole earth is mainly silica, with a few impurities. Then, the use of elemental silica as a semiconductor will be discussed, and how several familiar semiconductor devices made from it work. After this, the large field of silica minerals and their structures will be explained. Then, the relation of silica to life will be explored, concentrating on the diatom and the diatomite it has left, which is an important industrial mineral. Silicate minerals weather to clay, which not only forms the soils in which plants grow, but also gives us pottery, bricks and cement. Silicon also gives us glass, a substance that is probably always in our view in daily life, and which is not only useful, but also beautiful. The lore of silica is a rich garden that yields many tasty fruits. The links in the Contents will allow you to sample any subject that interests you. This article is brand new, and will be revised to eliminate errors and misprints, and to include new things that I learn.
The name silicon comes from the Latin silex, silicis, the word for "paving stone." The genitive is given to show where the "c" comes from. Flints made a very good, durable paving stone, so the word came to be associated with flints. Flints are made of silicon dioxide, so the name was acquired by this important substance, "silica." Silica is very difficult to decompose into silicon and oxygen, so was long supposed to be a simple substance, an "element," along with similar compounds, such as alumina, Al2O3, or magnesia, MgO. As sand, it has been known from time immemorial. The Greek for sand, psammites, and the Latin, arena, have not found their ways into chemical terminology as the flint has.
When Humphrey Davy decomposed soda and potash into metallic sodium and potassium by electrolysis just after 1800, powerful reducing agents became available. A reducing agent is a reagent that supplies electrons, which can turn a metal ion into the metal itself, as Fe++ + 2e- → Fe. Potassium is just about the most powerful reducing agent that exists, and can be used to reduce nearly every other element. Davy, and in Paris Gay-Lussac and Thénard, competed to decompose previously undecomposable substances and discover new elements. Davy won with boron in 1809, closely followed by the French pair. Though Davy suspected silica was a compound, Gay-Lussac and Thénard won the race for silicon in 1811. What they got was not all that impressive, and Berzelius, in 1824, finally exhibited relatively pure silicon, and is credited with the discovery.
What these workers did was dissolve silica in hydrofluoric acid (the only acid that attacks it) to get silicon fluoride: SiO2 + 4HF → SiF4 + 2H2O. The gaseous silicon fluoride was then passed through molten potassium, where the silicon was reduced: SiF4 + 4K → Si + 4KF. The potassium fluoride could then be washed away, leaving a brown powder, amorphous silicon. It is not actually amorphous, just microcrystalline. The actual reactions are not this neat, and other compounds are formed, such as fluosilicates, which means that the silicon is not very pure. It was not until 1854 that pure, crystalline silicon was produced. This is a silvery-gray, metallic-looking substance that is, nevertheless, rather brittle and low in density.
Most silicon now is made by reduction of SiO2 with C in the electric furnace. With carefully selected pure sand, the result is commercial brown silicon of 97% purity or better. This is the silicon used for semiconductors, but it must be further purified to bring impurities below the parts-per-billion level. Both sand and the brown silicon are the starting points for the synthesis of silicon compounds.
Silicon belongs to the IVA family in the periodic table, which consists of carbon, silicon, germanium, tin and lead. Its atomic number is 14, and its atomic weight is 28.086. Its isotope Si28 has an abundance of 92%. It melts at 1440°C, and boils at 2355°C. Its density is 2.36 g/cc. The outer electron configuration of each of these atoms is s2p2. Carbon and silicon are nonmetals, white tin and lead are metals, and germanium and gray tin are typical semiconductors. Of course, the important use of elemental silicon is as a semiconductor, so this matter will be discussed in its own section below. These five atoms are very different in their physical and chemical behaviors, despite being in the same column in the periodic table.
Silicon is often compared to carbon, and differences are what is significant, not similarities. Each has a valence of 4, but there the similarity ends. Carbon will share one, two or three electrons with another carbon, forming single, double and triple covalent bonds. The lengths of these bonds are typically 0.154, 0.133 and 0.120 nm. The radius of a carbon atom can be taken as 0.077, based on the single bond. Multiple bonds are also formed with other atoms, such as N, O, S and P. This leads to the great richness of carbon chemistry. Carbon will not let go of its four valence electrons, but is willing to share them.
Silicon, on the other hand, will share an electron with another silicon, showing a radius of 0.117 nm, but will not make multiple bonds. The fact that it is about 50% larger than a carbon atom makes the difference. Silicon, likewise, will not let go of its four valence electrons, and does not even like to share them very much if the alternative of sharing them with an oxygen atom is possible. Both carbon and silicon find that hybrid sp3 tetrahedral orbitals are the most stable, with the angle between bonds equal to 109° 28', the tetrahedral angle. Silicon prefers above all else to surround itself with four oxygen atoms as if in the orthosilicate ion (SiO4)----. This ion actually exists in water solution, forming the weakly ionized orthosilicic acid, H4SiO4, with ionization constant K1 = 2 x 10-10. The orthosilicate ion, which we shall call the silica tetrahedron, is shown in the diagram at the right. All you can see of it are the oxygen atoms; the silicon is safe in the middle.
Molecules of orthosilicic acid can condense at their vertices by the elimination of H2O to form Si-O-Si linkages. Each silcon atom is surrounded by four oxygens, in each of which it has a half-interest, so the composition is expressed by SiO2, the usual formula given for silica. Note carefully that there are no SiO2 molecules here, just a giant macromolecule that is the crystalline silica. If not given a lot of time, the tetrahedra will condense in a sort of mess that is not actually crystalline, but a glass. With time and luck, or if allowed to grow on a pattern, the tetrahedra will arrange themselves in the form of a crystal, perhaps the hexagonal crystal of the mineral quartz. The density of quartz is 2.65 g/cc. Silicon specializes in these large macromolecules, of which the silica tetrahedron is the building block.
The analogous compounds CO2 and SiO2 are very different. The first is a gas, or a soft solid at low temperatures, always consisting of CO2 molecules, in which the carbon and oxygen are connected by double bonds. It is soluble in water, forming carbonate ions. The second is a hard, crystalline solid not softened by heat below 1700°C, and very resistant to chemical attack. Silica is one large molecule, not an assembly of SiO2 molecules. I have not heard exactly what happens in the gaseous state.
If there is no oxygen around, the silicon atom must be satisfied by others of its own kind. The Si-Si covalent bond is 0.117 nm long, and is strong. The crystal formed is the same as in diamond, a face-centered cubic lattice. Its density is 2.36 g/cc. Large crystals are not as reactive as the brown powder of amorphous silicon (which is also fcc, but finely divided). The crystal is silvery gray, and shiny like a metal, but is lighter that you would expect from its appearance. One handbook mentions a "graphitic" silicon, but all other sources I have consulted are silent on this subject. If it were not for the electrons, pure crystalline silica would be transparent.
Silicon can be convinced to combine with halogens and hydrogen if there is no oxygen around. This combination must usually be brought about indirectly, and water must be excluded. Silicon forms silane, SiH4, which is analogous to methane, CH4. However, the two gases are quite different. Methane can be bubbled through water, but silane is immediately hydrolyzed: SiH4 + 3H2O → H2SiO3 + 3H2, since silicon has a hunger for oxygen. There are analogues to the alkanes, such as Si2H6, Si3H8, and so on, where there are Si-Si bonds. The last-mentioned compound, and all higher ones, are pyrophoric, bursting into flame even with only atmospheric oxygen available.
Treatment of brown silicon with chlorine makes SiCl4. This substance melts at -70°C, and boils at 60°C. It is tetrahedral, with bond length 0.201 nm, slightly less than expected. The contraction of about 0.016 nm in the bond length is evidence of additional stability conferred by resonance with slightly ionic structures in which electrons are moved from chlorine to chlorine. In water, SiCl4 + 3H2O → H2SiO3 + 4HCl. If SiCl4 is mixed with NH4OH, we get a dense white smoke of the metasilicic acid and ammonium chloride, that is used for military purposes and in skywriting. SiCl4 liquid is nonconducting, showing that it is not ionic.
SiF4 is curious in that its melting point is -90°C, but its "boiling point" is -95°C. What this means is that the solid sublimes to SiF4 molecules before it melts. The solid does not consist of SiF4 molecules, but of a giant molecule in silicon style. SiF4 is covalently bonded, and does not give fluoride ions in solution--instead, the molecule hydrolyzes, as we would expect, making metasilicic acid and fluosilicic acid (instead of hydrofluoric). Fluosilicic acid, H2SiF6, is a strong acid where the silicon is in the center of an octahedron of fluorines, instead of a tetrahedron of oxygens. The silicofluoride ions are stable in solution, and form salts. The silicofluoride octahedron is the only thing that can be on equal terms with the silica tetrahedron. SiF4 is evolved when silica dissolves in HF.
There are also numerous compounds with both hydrogen and halogens, such as trichlorosilane, SiHCl3. This compound is made when brown silicon is treated with anydrous HCl. It decomposes on a hot surface to elemental silicon, chlorine and hydrochloric acid. In the purification of silicon, it can be distilled to high purity, and then decomposed to silicon with impurities at the parts-per-billion level, satisfactory for many purposes.
Sand will also react directly with carbon in the electric furnace to produce silicon carbide, SiC, or carborundum: SiO2 + 3C → SiC + 2CO. This reaction proceeds because the gaseous carbon monoxide expelled. Commercial carborundum is black and impure, but still is very hard, 9.2 Mohs. When pure, it is a transparent, greenish crystal. It occurs in two modifications, α-SiC which is hexagonal in structure, and β-SiC, which is face-centered cubic, the diamond structure, and apparently the hardest form. Its electrical conductivity is 107-200 Ω-cm, so it can be used for electrodes. Its density is 3.217 g/cc, and it sublimes at 2700°C. Carborundum was discovered by E. G. Acheson in 1891, and was the first of the artificial abrasives.
Ferrosilicon, an alloy of iron and silicon, is produced if iron oxide is added to the sand and coke in the electric furnace. This alloy does not have to be especally pure, since it is thrown into molten iron to deoxidize it. The silicon combines with any oxygen present, as we might expect. Extra silicon remains and alloys with the iron. A few percent gives an excellent steel for magnetic cores, and in cast iron it aids the castability.
Sodium oxide, Na2O, reacts rapidly with water to make the strong base NaOH: Na2O + H2O → 2NaOH. NaOH dissociates completely in water to give OH- ion, and because of this is called a base. Sulphur trioxide, SO3, also reacts rapidly with water to make the strong acid H2SO4: SO3 + H2O → H2SO4. This molecule dissociates in water to give H+, and because of this is called an acid. Chemistry texts often point out that this ion is hydrated. This is no surprise: all ions in water are hydrated. If we mix the two, we get sodium sulphate and water, which is neither acidic nor basic. Therefore, we call sodium basic and sulphur acidic, representing two principles that will react with each other.
Calcium oxide, lime, CaO, gives Ca(OH)2 in water. Incidentally, this is called slaking the lime and releases much heat. Calcium hydroxide is a weak base, but we have no trouble classifying calcium as basic. Silicon oxide, SiO2, does not actually react with water, but we can imagine that it would form H2SiO3, a feeble acid, and so classify silica as acid. When the two meet in a fused state, they react to form Ca2SiO3, which has a lower melting point than either reagent. In general, acidic and basic oxides react to give a slag that is usually of relatively low melting point. This reaction is used to remove silicates in the smelting of iron. Limestone is added, which burns to lime, evolving carbon dioxide, and then the lime reacts with the silica to form a slag in which the iron is insoluble, and which can be drawn off separately. The use of the terms acid and base in this case, where there is no water to be seen, can be made reasonable by the preceding argument.
Orthosilicic acid can lose water to form metasilicic acid, H2SiO3, which we have encountered several times already. Silica is attacked by alkalis, SiO2 + 2NaOH → Na2SiO3 + H2O. If we now treat this with HCl, we get Na2SiO3 + 2Hcl → H2SiO3 + 2NaCl. The metasilicic acid is here called "water glass," since it forms a gelatinous precipitate that is insoluble in water. Water glass is actually formed from chains of silica tetrahedra in an irregular and indefinite pattern. The formula gives only the composition, not the molecule. This is, as we realize, typical of silicon. This gel can be used to seal the pores in eggs to preserve them longer, or as an adhesive. If heated, water is driven out of the abundant pore space, and the result is the familiar silica gel used as a desiccant. When it gets full of water, it can be renewed by simple heating. Sodium silicate is made commercially by fusing sodium carbonate, sand and carbon. Water glass is then produced by acidifying the product.
Silicon makes giant molecules, as we have seen. To get some perspective on atomic sizes, consider a picogram (a millionth of a microgram) of silica. This would fit in a cube 722 nm on a side, so it would be at the limit of the optical microscope if you wanted to see it, and almost as small as a colloidal particle. However, this speck would still contain 1010 SiO2 units, 10,000 million of them! Silica vapor could easily consist of clumps of SiO2 units, say hundreds of them in each fragment, but I have not heard of anyone who has investigated. The boiling point of 2230-2590°C makes measurements a little difficult. Crystalline silicon has 5 x 1022 atoms per cc.
There is another oxide of silicon, silicon monoxide, SiO. It forms cubic crystals of density 2.13 g/cc that are transparent, with index of refraction 2.0. Polycrystalline SiO is used as an optical thin film, usually to improve reflectivity. It melts somewhere above 1700°C, and boils at 1880°C. I have seen little information on its structure and technology. Apparently it can autooxidize according to 2SiO → Si + SiO2.
A silicone is a silicon-oxygen chain with hydrocarbon radicals attached to the silicons, as in methylsilicone, (CH3)3Si-O-(CH3)2Si-O-...-O-Si(CH3)3. The hydrocarbons make the molecule look like a hydrocarbon to its surroundings, while the strong silicon-oxygen chain makes it very stable at high temperatures. The viscosity of a silicone oil increases much less rapidly than that of a hydrocarbon oil when the temperature decreases, say a factor of only 70 from 100°C to -35°C, while a hydrocarbon oil's would increase by a factor of 1800. The chains can be cross-linked and polymerized by oxygen or other means to give rubbery solids that are equally inert. These remarkable compounds are strictly artificial, never found in nature.
Silicon and its compounds are not poisonous or otherwise hazardous, except for rare compounds like the silanes. Most, indeed, are almost completely insoluble. One exception is the industrial disease silicosis, which affects workers exposed to silica dust over a considerable period. The workers involved are mainly stoneworkers and miners, and the disease can be prevented by wearing respiratory filters. Some years ago, a Denver TV reporter found to her horror that children's playgrounds often contained deadly silica, and that nobody was doing anything about it.
For electronic applications, we are interested in elemental silicon as a conductor of electricity. By adding controlled amounts of impurities we can determine the sign of the charge carriers and their density. By doing this in well-defined limited areas, we can construct electronic devices that rectify, amplify, are sensitive to light, emit light, and store and move electric charge on command.
The silicon must be a single crystal with few defects, and of extreme purity, with impurities at the sub-parts-per-billion level. To produce this semiconductor silicon, we start with brown commercial silicon of about 97% purity. The silicon is treated with anhydrous hydrogen chloride to produce a chlorinated silane, which is fractionally distilled to high purity, and decomposed on a hot wire. The silicon is scraped off and cast into a polycrystalline rod. This relatively pure silicon is then suspended vertically and a molten zone is passed repeatedly from top to bottom. The heat comes from induction of a current by a radio-frequency current in a movable coil surrounding the sample. Impurities tend to prefer the molten zone to the silicon solidifying behind it. At the finish of the refining, the end of the rod containing the impurities is sawed off. This zone refining produces ultra-pure silicon. A seed crystal is placed at the lower end of the polycrystalline bar, and zone melting allows the growth of a single crystal based on the orientation of the seed crystal. In all of this processing, the silicon has never touched a crucible wall or the equivalent, and remains pure. The cylindrical crystal is then cut into thin slices, or wafers, for the fabrication process that follows.
Other materials have been used for semiconductor devices. The first transistors and diodes were made from germanium. Gallium arsenide, GaAs, an example of a compound semiconductor, has been used for specialized devices. Similar compound semiconductors must be used for optoelectronic devices, such as light-emitting-diodes, for which silicon is not suitable. Silicon, however, has remained by far the most widely used semiconductor because of its unique advantages. Among these are (1) it is elemental, so composition is not a problem; (2) it forms a tough, adhering insulating oxide, SiO2, that can be used for isolation and protection; and (3) a large variety of impurity atoms are available for controlling its conduction properties.
The charge carriers in a metal that is a good conductor are electrons, which seem to move almost as if they were free particles, but confined to the volume of the metal by a high potential barrier on the order of 8 or 10 eV. At absolute zero, when the metal ions are in their regular lattice positions, the ions have no effect on the electron motion, except perhaps for changing their effective mass. In any real metal, there are impurities and defects with which the electrons may collide, and above absolute zero the vibrating ions also hinder the electrons. These realities cause electrical resistance. The number of states which the electrons may occupy is limited, and there are usually so many electrons that all the states are filled up to an energy called the Fermi level, which is about 5 eV. This effect is called degeneracy (not a very good term, but it is the one used).
In any crystal lattice, metal or otherwise, electrons free to move about occupy states classified by their energy. The periodicity of the crystal lattice causes gaps in the allowed energies, separating them into bands. This only applies to the electrons that can roam throughout the crystal, the valence electrons. The electrons in the inner shells of the ions stay where they are and have no other influence. The band in which the valence electrons move is called the valence band. In a metal, this band is mainly empty. The electrons may be degenerate at ordinary temperatures, but they still move like free electrons.
In a silicon crystal, the valence electrons are the four outer electrons that form the covalent bonds between the silicon atoms. The possible states in the valence band are completely full. There are lots of electrons, but they can go nowhere. As soon as they would move, another electron would take their place, and the net effect would be zero. In this state, the silicon is an insulator. This is almost true in pure silicon at low temperatures. The next higher band of electron states is called the conduction band. It is, in this state, completely empty. The distance in energy between the top of the valence band and the bottom of the conduction band is, in silicon, 1.12 eV.
As the temperature rises, electrons are jostled about by the vibrating ions. At room temperature, the jostling only amounts to 0.025 eV, so it is not very likely that an electron will be knocked upstairs. This is one of the advantages of silicon, since it would require a very high temperature before many electrons were jolted into the conduction band by thermal agitation. Once this happens, we lose control of the conduction characteristics by adding impurities, and the semiconductor becomes useless. In germanium, the energy gap is only 0.67 eV, so germanium gave up at much lower temperatures than silicon does. This is an important reason why silicon has almost completely replaced germanium in semiconductors. Diamond has a band gap of 5.4 eV, so it would be useful to high temperatures, and make excellent devices. Unfortunately, it is very hard to grow perfect diamond crystals, and difficult to find impurities that will control its conduction properties.
The conductivity produced by charge carriers thermally excited across the energy gap in a semiconductor is called intrinsic. Let's suppose for a moment that we have ni electrons in the conduction band, and an equal number of places in the valence band where they used to be, called holes. If we apply an electric field, the electrons move one way, and the holes move the other way, both contributing to the current in the direction of the field. So far as semiconductors are concerned, holes act like positive charges with positive mass, very much like particles.
There is always a chance that an electron in the conduction band will encounter a hole in the valence band and fill it, or recombine. Two intrinsic charge carriers are lost in this process. The rate of recombination is proportional to the product of the concentrations of electrons n and holes p: charge pairs lost per second per unit volume = npr(T), where r(T) is some function of the temperature T. This is balanced by thermal generation of pairs at the rate g(T), which does not depend on the concentrations. In equilibrium, then, npr(T) = g(T), or np = g(T)/r(T) = f(T). In our intrinsic semiconductor, n = p = ni. f(T) does not depend on the concentration, and is a constant in a particular crystal, so we have in all cases np = ni2. For silicon at 300K, ni = 1.4 x 1010 per cc. Extreme purity of the semiconductor is necessary to achieve a figure this low.
Now suppose we introduce a substitutional impurity in the silicon lattice that has one more valence electron than silicon. Examples are P or As, either of which will fit pretty well. Impurities that are deliberately added to control the characteristics of the semiconductor are called dopants. P or As can be added in amounts up to 1021 per cc before coming out of solution, which is many more than we will need. Four of the valence electrons do the job of making the tetrahedral bonds to the other silicon atoms, while the fifth is left idling. It happens that it only requires about 0.4 eV to kick the fifth electron away into the conduction band, and this is easily done by thermal agitation at room temperature. This gives us a way of putting exactly the number of electrons we want into the conduction band. Accordingly, the impurity is called a donor. The diagram shows the filled valence band, the empty conduction band, and the donor levels, five of which are ionized and have given electrons to the conduction band. At the same time, from the equilibrium equation, p = ni2/n will be very small. The majority carriers, electrons in this case, will far outnumber the minority carriers, the holes in the valence band, which can be neglected in a first approximation. Silicon with a pentavalent dopant is called N-type silicon.
On the other hand, we may use a dopant that has only three valence electrons, such as B, Ga or Al. This causes a lack of one electron in one of the covalent bonds made by the dopant. It requires only about 0.4 eV to jolt an electron into the deficiency, leaving a hole in the valence band to wander around. We can add up to 5 x 1020 boron atoms in the crystal without causing a lattice disturbance, which again is sufficient for our purposes. As many holes as we want can be introduced as majority carriers, while minority carriers, the electrons, fall to a low concentration n = ni2/p. These dopants are called acceptors because they accept an electron to create a mobile hole. These levels are only just above the valence band in the figure, and have accepted four electrons. The holes are shown in the valence band. Silicon with a trivalent impurity is called P-type silicon.
There are many ways to construct semiconductor devices. I shall describe the widely-used planar process here, in its original form. It has been greatly developed and refined to produce smaller and smaller devices, but the principles are the same. The surface of the wafer can be oxidized to create a tightly-adhering layer of SiO2 by heating it in an atmosphere of oxygen. This layer is called simply oxide. Such a layer passivates or protects the silicon, and also serves as a mask for the application of dopants. Suppose we want to make a certain area of the wafer P-type by doping with boron. The oxide is coated with a photoresist, which is exposed to ultraviolet light through a photographically prepared mask. The area we want to dope is protected from the UV, while the rest is polymerized and hardened. Now the unpolymerized resist is washed off, leaving our region uncovered. The exposed oxide is etched away with dilute hydrofluoric acid, leaving the bare silicon exposed. Now the remainder of the resist can be removed chemically.
The wafer is then exposed to an atmosphere of a gas that will deposit boron atoms on the surface. Heating accelerates diffusion of the boron into the wafer. Most dopants diffuse very slowly in oxide, so the oxide layer of about 1μm thickness protects the rest of the silicon. Ga and Al diffuse well in oxide, so they cannot be used in this process. P, As or Sb can be used as dopants if an N region is desired. Finally, the oxide layer can be removed, and the process repeated to form areas with other characteristics. This use of the oxide as a diffusion mask is one of the great advantages of silicon. The wafer material we start with is called the substrate.
Interconnections can be made with Al metal. The wafer is given a thick coat of oxide, 1 or 1.5 μm thick, and then a layer of Al about 0.5 μm thick is evaporated all over. A mask is prepared, and resist exposed through it, just as for diffusion. The unexposed resist, corresponding to the areas where we do not want Al, is washed off. The unwanted Al is then dissolved in dilute acid, and all the electrical connections are made. The resistance of a conductor of length l and width w is R = 0.05(l/w) Ω, where the resistance "per square" of the Al is 0.05Ω (this is the resitivity times the thickness). Another conductor is heavily doped polycrystalling silicon, called poly.
As an example of what we can now do, consider a P-type region diffused into an N-type substrate. A side view is shown in the diagram. At a certain depth, there is a sudden change from P to N, called a PN junction, without any disruption to the crystal structure. If we make the P end positive, and the N end negative, an electric field points from P to N. It drives the holes toward the junction in the P region, and the electrons toward the junction in the N region, so the current flows in the same direction, P to N. When the holes meet the electrons at the junction of the P and N materials, they recombine, while new holes are created at the P terminal and new electrons at the N terminal (it is possible to make connections so that this takes place). The PN junction is forward-biased and carries current easily. The arrow in the diode symbol points from P-type to N-type, in the direction of easy current flow. The P-type material is the anode, the N-type the cathode.
Now reverse the polarity, making the P region negative and the N substrate positive, as shown on the right-hand side of the diagram. The electric field that results drives the holes and electrons away from the junction on each side. This creates a depletion layer with no charge carriers of either sign. Current cannot pass through this layer. As the carriers move away from the junction, they uncover the negative acceptors and the positive donors, so an electric field is created that cancels the applied field. The donors and acceptors are, of course, fixed in the lattice and cannot move. The current cannot continue, for there is no way to get new holes at the junction for the P material, and no way to get new electrons for the N material. The semiconductor now acts like an open circuit. The voltage across it is equal to the applied voltage. That is, the PN junction rectifies.
Diodes can be (and usually are) made by the planar method. We can start with an N-type wafer, then diffuse a P-type region into it. The substrate becomes the cathode, while the P region becomes the anode. A thousand of these can be made at one time on a wafer, and cut apart afterwards. Mass production is a feature of the planar process. Individual diodes, or whole microprocessors, can be made in large numbers with processes common to all, greatly reducing the expense.
When we apply forward bias and current flows, holes and electrons must recombine in large numbers near the PN junction. Holes are driven into the N region, where they become minority carriers, and electrons into the P region, where they likewise are in the minority. Recombination takes place to restore equilibrium. In silicon, the energy released by recombination is given up to the lattice. In some other materials, such as gallium phosphide, the recombination also can take place with the emission of light, at a wavelength corresponding roughly to the band gap, according to Bohr's formula hν = eV. Now we have a light-emitting diode, an LED. The wavelength emitted is given roughly by λ = hc/eV = 1240/V, if λ is in nm and V is in volt. Gallium arsenide, GaAs, has a band gap of 1.35 eV, so it will emit around 920 nm, in the infrared. Gallium phosphide, GaP, with a band gap of 2.24 eV, will give about 553 nm, a green light at the maximum spectral sensitivity of the eye.
If there is a reverse bias instead, the current through the PN junction is very small. The area from which the charge carriers have been swept, and in which there is a strong electric field, is called the depletion layer. If light passes through this region, there is a good chance that it will create an electron-hole pair, and the two charges will be quickly swept away before they can recombine again. A photocurrent results, and the PN junction is now detecting light; it is called a photodiode. The maximum wavelength that will make charge carriers is given by the same formula as for emitters. Silicon, with a band gap of 1.12 eV, will be sensitive to wavelengths shorter than 1110 nm. A GaAs emitter, at 920 nm, will be a good match for a Si detector, at 1110 nm, and this pair is used often in practice, as with TV remote controls.
A simple device easily created with planar technology is the field-effect transistor, or FET. This device controls a current by means of a voltage applied to it. Since no current is drawn by the control circuit, no power is required to operate it, and the power gain is effectively infinite. A side view is shown in the diagram, together with the circuit symbol. We start, for example, with a P-type substrate with a resistivity of about 5 Ω-cm, and diffuse two heavily-doped N regions into it. These regions are the main terminals of the device, called source and drain though they are actually alike and either can be chosen as the source. Whichever way we apply a voltage between them, no current will flow because there will be a reverse-biased junction at one electrode or the other. Conventionally, the drain has the reverse-biased junction.
A thin layer of oxide, perhaps only 0.1 μm thick, is deposited between the two electrodes, and on top of it is evaporated an Al gate electrode. If we apply a positive voltage with respect to the source to the gate, an electric field appears under it in the P-type substrate. This field drives away holes and attracts electrons to the thin layer just under the gate. At some voltage, electrons will predominate and the type of the region will be inverted from P to N. This means that the reverse-biased junction at the drain will disappear, and this will allow current to flow from drain to source (electrons from source to drain, as the names indicate). The transistor has now "turned on" and conducts. This control is exerted purely by the voltage at the gate electrode. The device is usually called a MOSFET, whether metal or polysilicon is used for the gate electrode. Microprocessors and memory chips are all MOSFET's.
Field-effect transistors were sought by workers even before there was silicon or any other solid-state device, but surface effects and impurities always frustrated their efforts, and the most they could devise were rectifiers. The galena and pyrites crystals of crystal sets are solid-state rectifiers. When how to use germanium (and later, silicon) for diodes became clear in the mid-20th century, transistor action on a different principle was discovered in 1948. The first devices used point contacts, like the crystal sets, but soon the much superior junction devices were invented. With a crystal set, you poke around with the catwhisker trying to find a crystallite that is doped just the right way to be P-type.
A junction transistor consists of two N-type regions separated by a very narrow P-type layer, called the base. A planar junction transistor is shown in the diagram. In a point-contact transistor, the N regions are metal whiskers touching the base close together, and act in the same way that we shall describe. By making the base positive with respect to one of the N-type regions, which we shall call the emitter, we can cause a large current of electrons to move from the N-type emitter into the P-type base. Meanwhile, we have made the other N-type region, the collector, positive with respect to the base, so that no current would normally flow, because the junction is reverse-biased. Emitter and collector correspond to the source and drain of an FET. The essential thing is to have a forward-biased emitter junction very close to a reverse-biased collector junction. In the collector junction there is a strong field pushing electrons into the N-type collector region. There are usually no electrons to speak of in the P-type base, so there is usually no collector current, just a small leakage current from thermally generated electrons. However, when we forward-bias the emitter-base junction, all of a sudden there are lots of electrons milling about the P-type region, and they are much more likely to fall into the collector than anything else. Some do recombine, and this requires a small current into the base to supply the holes lost this way. Most, however, just continue on into the collector. Therefore, the collector current is proportional to the small current that flows into the base, but much larger, by a factor of 100 or so, called the transistor's beta. This transistor action provides a very large power amplification, in which a small power can control a much larger power. This device is called a bipolar junction transistor, or BJT.
One final detail we should discuss is the matter of charge carrier mobility. An electric field E pushes on a charge e with force eE, accelerating it. The charge loses all this energy when it collides with a vibrating ion. It can only build up to a terminal velocity v depending on how long it moves without colliding, called the relaxation time τ. Then, on the average, v = (e/m)Eτ, proportional to E, by Newton's second law. The mobility is the constant of proportionality μ, so that v = μE, and its units are cm/sec/V/cm. Once we have the mobility, we can calculate the conductivity of a semiconductor, since the current density j = nev = neμV, where n is the charge carrier density. The conductivity is then σ = neμ, which is a measure of the carrier density. In silicon, electrons have mobility 1900 cm2/V-s, holes 500 cm2/V-s.
The P-type substrate for an FET mentioned above was said to have a resistivity of 5 Ω-cm. Using σ = neμ, we find that this corresponds to a doping density of 6.25 x 1016 cm-3, which is rather lightly doped. There are 5 x 1022 silicon atoms per cc, so only one in a million is replaced by a boron. A hundred times this is a rather high doping level.
Oxygen and silicon alone constitute 74% of the mass of the earth's crust. If the composition of the whole earth were known, they would probably be even more predominant. Approximate percentages are O 46%, Si 28%, Al 8%, Fe 5%, Ca 4%, Na 3%, K 3%, and Mg 2%, for a total of 99%. All other elements make up the remaining 1%. It is hot within the earth, but not molten in the usual sense. The contents of the mantle grind around in slow convection, while the foam is scraped off the top to make the crust. The crust is only 33 km thick on the average, and is the only part we know directly. The lighter metals, Al, Na and K, are concentrated at the top, in rocks called the "sial," while the heavier metals, Fe, Mg, Ca are concentrated at the bottom, in the "fema." Silicon predominates at every level. The lighter rocks have been crushed down into mountain roots by continental drift, where they may form molten blobs called magma where a lot of rock chemistry occurs. Erosion brings these areas to the surface as batholiths of granitic rocks, surrounded by the tortured writhing of gneisses, showing us what once happened deep in the crust.
All these rocks are mainly silicates, which form perhaps 80% of all minerals. The simple minerals, many of which are metallic ores, like galena and pyrites, are rare and have been concentrated in local deposits. By far the majority of minerals have crystallized from liquid and solid solutions of mixed silicates, so what we observe is the result of a process that produces infinite variety, controlled by a few basic principles. I will try to explain the principles, with a few examples of how they give rise to the observed rocks, without getting lost in the confusing variations that make igneous mineralogy so complex and puzzling.
An igneous rock is one that has crystallized from a liquid magma, or has been formed by diffusion and recrystallization below the melting point. The mantle of the earth (the thick layer between the crust and the core that makes up most of the volume of the earth) is not composed of melted rock, as was once assumed, but rather magmas are created locally by the grindings of plate tectonics, which also provides active fluids that dissolve and modify rocks. Old magmas are exposed by erosion as batholiths here and there, and their rocks have intruded surrounding rocks, dissolving and wedging into every crack. Rocks that show evidence of later modification by heat, pressure and the action of these active fluids are separately classified as metamorphic. Igneous and metamorphic rocks are quite distinct in structure from those that are composed of material weathered from pre-existing rocks, and deposited in the ocean or on dry land, where they have become cemented, hardened and consolidated into sedimentary rocks like sandstone, shale and limestone. Sandstone and shale are mostly silica, but the limestone represents the carbon dioxide that was once present in the atmosphere.
The basic structural element of silicates is the silica tetrahedron, formed by a central Si atom surrounded by four O atoms, each with an unpaired electron. This unit can be written SiO4----. It can form four bonds, covalent or electrovalent, with other atoms. This orthosilicate ion is stable and can actually exist, as in orthosilicic acid, H4SiO4. However, it normally forms covalent bonds with an adjacent Si atom, so that the Si atoms are connected by an oxygen bridge: -Si-O-Si-. This is a strong bond. This can be considered as formed from condensation of two -OH groups attached to Si atoms: -Si-OH + HO-Si- → -Si-O-Si- + H2O. If every Si atom has four such bonds, then the structure can be represented by (SiO2)n. Normally, we write just SiO2, but this molecule does not actually exist, at least in the solid, so the formula shows only the ratio of Si to O, which is 1 to 2.
If our melt contained only silicon and oxygen, then obviously silica would be the result on cooling. Comparing silicon and carbon, you might expect that the solid would have a diamond structure (as indeed crystalline elementary silicon does) but silica does not behave like this. There are more stable forms of more complex structures. The familiar quartz crystals are of a structure called α-quartz, which forms hexagonal crystals. It is a very complex structure with over 100 atoms in the unit cell, in which the silica tetrahedrons form a kind of spiral. It is so difficult for the atoms to arrange themselves in this structure that molten quartz solidifies as a glass, called fused quartz. Using special crystal-growing techniques, α-quartz crystals can now be grown in the laboratory, and have replaced the natural crystals, which were growing rare in the required sizes. Nevertheless, it is not an easy thing. Natural crystals have required 100,000 years or so to grow.
Quartz is also found in crystal forms that are stable at higher temperatures. Cristobalite is cubic, and tridymite is hexagonal. Both have structures similar to those of ice, a curious thing because quartz was once considered petrified ice! Keatite, coesite and shishkovite are more recently discovered crystal forms, that do not occur in the normal course of events. The last two are found in meteor impact craters, where they were produced by the high pressures and temperatures. The crystal structure of quartz is a very complex matter. Quartz is birefringent and optically active, properties used in the handling of polarized light. It is piezoelectric, meaning that distortion of the crystal produces electrical polarization, used in the crystal resonators that can control the frequency of electrical oscillations. The "quartz" legend on the face of a wristwatch means that there is such a crystal inside. It is also pyroelectric, meaning that heating produces electrical polarization, for which a use is yet to be discovered. All these topics are worthy of exploration, but would require an article of their own devoted to the optical and electrical properties of quartz. Crystalline quartz is often colored by impurities, making a variety of attractive gems, such as purple amethyst (Fe+++), yellow citrine (iron hydrates), pink rose quartz (Mn++), and blue quartz (needlelike colloidal particles of TiO2). Microcrystalline quartz also occurs in many attractive forms, such as variegated agate, red jasper, chalcedony, layered black-and-white onyx, red-and-white sardonyx, green chrysoprase, green with red drops heliotrope or "bloodstone," and orange-red carnelian, a variety of chalcedony including limonite and hematite, which are iron oxides.
Opal is a curious form of silica that includes some water. Its formula is usually written SiO2·nH2O. It consists of small, regular spheres of cristobalite, cubic silica, about 300 nm in diameter, cemented by hydrated silica gel. The fact that the size of the spheres is comparable to the wavelength of light, and that they are arranged regularly, creates interference that causes an attractive play of color. Mother-of-pearl, or nacre, secreted by oysters and abalone, has a similar iridescence and color, but is calcium carbonate, layers of thin aragonite films. However, a grain of silica is often the irritant at the center of a pearl. Precious opal is white or black and has a good play of iridescent colors, but common opal may be less showy and lack iridescence. The best comes from Australia. Fire opal from Querétaro is red, but not iridescent. Siliceous sinter deposited at hot springs, geyserite, petrified wood, and diatomite, which will be discussed in the next section, are composed of opaline silica. Opal may deteriorate on exposure to air due to loss of water.
The natural magma, however, also contains aluminium Al and other cations, such as postassium K. If an extra electron is given to the Al atom, it can form four tetrahedral bonds just as silicon can, and we can make alumina tetrahedra (AlO2)- that are very similar to the silica tetrahedra, except that they have a (-) charge. These alumina tetrahedra can make covalent bonds with silica tetrahedra, or with other alumina tetrahedra. Anything we can build with silica tetrahedra can be built with a mixture of silica and alumina tetrahedra, provided only that we neutralize the negative charge by adding the same amount of positive charge. This can be done quite simply by the unit (KAlO2) which substitutes exactly for the unit (SiO2). A phase with the constitution (KAlO2)(SiO2)3 commonly crystallizes out. The formula may be written in the less evident form KAlSi3O8. This mineral is orthoclase, a potassium feldspar commonly found in granites.
Granite is formed from a melt (just our name for the original substance; it does not have to have been actually molten) rich in quartz, which at high temperatures is a homogeneous solid solution. When it cools, the components are no longer miscible, and it separates out into distinct phases. These may be quartz, and the potassium feldspar we have just mentioned. A granite is mainly feldspar, perhaps pink in color, with large masses of quartz embedded in it. Granites are a plutonic or abyssal rock, meaning that they have crystallized in a leisurely manner deep in the earth, so that the individual crystals can grow large enough to be easily recognized by the eye. The granite is then phanerocrystalline. If it crystallized in intrusive sills and dikes, that is in hypabyssal conditions, the crystals are of microscopic size, and we have a rhyolite. If it had partially crystallized before being intruded, it would be a granite porphyry, of large crystals in a microcrystalline groundmass, or a granophyre if the groundmass also had visible crystals. If it flowed out over the ground, and cooled quickly, under extrusive conditions, it would be a black glass, obsidian. If it were blown into the air as a foam, it would form pumice. Geologists are very interested in the appearance of the rock and its crystal size, greatly multiplying the descriptive names that are used for the same mixture of minerals. Mineralogists like to have one rock, one name for ease of labelling their collections, and look askance at the collective terms that are so useful to us.
If the magma also contained Ca and Na, the units (CaAl2O4) and (NaAlO2) might also appear in the solid phases. The radius of the Ca++ ion is 0.099 nm, and the radius of the Na+ ion is 0.095 nm, while the K
The magma probably also has some iron and magnesium in it as well. These ions crystallize in a phase that is usually biotite mica, small black, brown or dark green scaly plates. "Mica" is Latin for crumb, referring to these flaky bits. The minerals we have so far described have been tectosilicates, 3-dimensional structures like geodesic domes, with the K, Na and Ca of the feldspars inside. The silicon:oxygen ratio in tectosilicates is 1:2, as in silica, and all oxygens are shared. The net formal charge is 0, so aluminium tetrahedra must be added to include cations. Magnesium and iron ions would rattle around in these spaces, so a new tighter structure is favored. Imagine the silica tetrahedra as sitting on their bases in a plane, and rearrange them in your minds as forming hexagonal rings, the vertices having condensed together. The whole plane can be tessellated in this way, and the composition is (Si2O5)--. If magnesium is used to get neutrality, then the unit is (MgSi2O5), and the layers are electrically neutral. This is very similar to the soft mineral talc, which adds a Mg(OH)2 for each two of the above units. The mineral comes apart in thin sheets, which is typical of talc. Silicates based on sheets of tetrahedra are called phyllosilicates, from the Greek "phyllon," "leaf."
The phyllosilicate net is shown at the right, and should be considered as extending in all directions in the plane. All the silica tetrahedra are shown with one orientation, but the unbonded oxygens could point to either side, and could be random or ordered. The black dots show the locations of the silicon atoms. In a ring, there are 6 silicons, 12 oxygens belonging exclusively to the ring, and 6 shared oxygens. The silicon:oxygen ratio is then 2:5 for this structure. It is also possible for some of the silica tetrahedra to be replaced by aluminium tetrahedra. For all silica tetrahedra, the net charge of a ring is -2, which must be balanced by cations in the crystal structure. Each aluminium tetrahedron adds an additional -1 to the formal charge. There is electron resonance to give additional stabilization, so the negative charges are not associated with any particular unbonded oxygens.
Our magma, however, must take care of its aluminium and iron somehow, but is low in magnesium, so the micas form. The micas include alumina tetrahedra in the construction of the sheets, which have a negative overall charge. They are held together by positive ions between the negative sheets. The micas still come apart in flakes, splitting apart along the weaker interfaces between sheets, but the sheets are harder and more coherent than those of talc. Now we finally have a complete granite: mostly pink orthoclase, with some plagioclase, free silica in transparent crystals, and dark flecks of biotite or clear scales of muscovite. Rocks of this general composition are called felsic, from feldspar and silica, their principal components. Felsic rocks contain more than 65% silica.
Silica tetrahedra can also be put together in chains. Such silicates are called inosilicates. A single chain, with silica tetrahedra connected at one point, has the composition (SiO3)--, with a silicon:oxygen ratio of 1:3. Such a chain is shown at the left. There are plenty of opportunities for cross-linking to make a solid crystal, but the chains still tend to cleave easily along their length. These minerals are called pyroxenes, of which a simple example is enstatite, (MgSiO3). This formula looks like just magnesium silicate, a simple salt, but it is not--it is a crystal with this unit. Enstatite is a common mineral, cleaving easily in the direction of the chains, and sometimes found in fibrous masses. It is hard, heavy and green or olive-colored.
Corresponding side links of the chain can be condensed, forming one double chain, as shown in the figure at the right. Minerals with this structure are amphiboles, also inosilicates, and the unit is (Si4O11)-6, with silicon:oxygen ratio 4:11. In tremolite, the negative charge -12 of two chain units is neutralized by CaMg5. If the Ca++ is replaced by Fe++, the result is actinolite. In either case there is an extra Ca(OH)2 in the molecule. Tremolite and actinolite are members of a series, like the plagioclase feldspars. Actinolite is green and fibrous, giving us nephrite jade. Jadeite, the other form of jade, is a pyroxene in which NaAl neutralizes two chain units.
Tremolite is often fibrous, and was the material originally called asbestos. In Greek "asbestos" means "unquenchable, irrepressible" as an adjective, and "unslaked lime" as a noun, which seem exactly opposite to the characteristics of the mineral fiber. It was a wonderful substance, that could be picked apart into a fiber of which a cloth could be woven that could not be ignited or damaged by acids. It was later found that the metamorphic rock serpentine (see below) also has a similar fibrous phase called chrysotile that made a better asbestos. Unfortunately, chrysotile dust causes a virulent form of cancer, a fact discovered late in the 20th century, after asbestos, a very useful substance, had been widely used. Tremolite is not dangerous.
Orthosilicates are an important type of silicates. The formulas look like salts of orthosilicic acid, Mg2SiO4, but the crystals are not, of course. They consist of silica tetrahedra, (SiO4)----, in which the tetrahedra are joined not by condensation at the corners, but by cations that hold the oxygens of two silicons in each hand. If the unit is (MgSiO4), the crystal is olivine, which may be the most common mineral of all on earth, if the earth's mantle is indeed made of it. Fe++ can easily substitute for Mg++, since their radii are 0.083-0.075 nm and 0.078-0.065 nm, respectively. Again we have a series like that of plagioclase. The magnesium end is forsterite, the iron end fayalite. Olivine is the generic name for the whole series of minerals, and its crystal lattice is orthorhombic. A specimen of transparent forsterite with the characteristic color is shown at the left (image ©Amethyst Galleries). The green is due to ferrous iron.
The forsterite end gives us the beautiful gem peridot, of deep olive-green color. Stones with a yellow-green color are called chrysotile, not to be confused with its anagram chrysolite, which is asbestos. The hardness is nearly Mohs 7, equal to quartz. Optically, it is a positive biaxial crystal, with indices of refraction 1.6359, 1.6507, 1.6688, greater than that of beryl (1.58) or quartz (1.5442ω, 1.5533ε), about that of ruby and sapphire (1.77), but much less than that of zircon (1.92) or diamond (2.42). The orthorhombic crystals are distinctly pleochroic. The angle between optic axes is 2V = 85° 16'. The Geological Museum in London once had on display a perfect, faceted peridot of 136 carats, the most beautiful gem I have ever seen. Most gem peridots are at most 30 or 40 carats. The pleochroism was evident, and gave a subtle variation of color. Gem peridot has been found predominantly in St. John's Island in the Red Sea, in a deposit of dunite, a mysterious rock from the deeps that is mainly olivine. It is named after the Dun Hills in Australia, I am told. I have not been able to find the Dun Hills in my atlases. St. John's Island may now be Shadwân Island at the north end of the Red Sea. Another source says it is the island of Zebirget off the port of Berenice in Egypt (equally ex-atlas). Other sources are said to be Burma, Queensland, Brazil and the Navajo country of New Mexico and Arizona. Much modern peridot is said to be reworkings of earlier stones. Artificial forsterite has been made as a high-temperature refractory material. It is made by fusing olivine rock with magnesia, to bring the magnesium content up to that of forsterite.
Olivine has been detected in interstellar dust by means of an infrared absorption near 9.7 μm. This was done by Krätschmer in 1979. This helps the case for olivine as a primary constituent of planets.
The minerals we have just discussed crystallize from a melt that is poor in aluminium and alkali metals (otherwise feldspars would form instead of pyroxenes), but richer in magnesium and iron. Instead of granite, we get gabbro, containing plagioclase and pyroxene. Its hypabyssal equivalent is basalt, instead of rhyolite. The feldspars give light-colored rocks, the pyroxenes dark-colored rocks. The classification of rocks by silica content is explained in the table at the right. The percentages are calculated for the oxides; for example, silica is calculated as SiO2. Forsterite is considered as 2MgO·SiO2, so its percentage of silica is 43%. Olivine is an ultramafic rock. Albite, a plagioclase feldspar, is 69% silica, and so is felsic. The classification mainly refers to rocks that are an assembly of minerals, rather than to the minerals themselves.
Olivine and dunite are found in characteristic assemblies of rocks called ophiolite suites. In Greek, "ophis" means snake, and this is an allusion to the typical occurrence of serpentine, a metamorphic rock that is produced from olivine, pyroxene and similar mafic and ultramafic rocks by heat and pressure, in the assembly. Serpentine occurs as chrysotile, antigorite and lizardite, and is commonly greenish like olivine. We have already talked about its fibrous form, chrysotile. These rocks are found at spreading centers on the edges of continental plates, where basic magma from the mantle is brought to the surface. The mid-Atlantic ridge is one such place, Cyprus is another, and the Red Sea overlies another line of spreading that has separated Africa and Arabia, and continues down the Rift Valley of Africa. The occurrence of lizardite on the Lizard Peninsula in Cornwall, and dunite in Australia, may be relics of earlier spreading centers. Before plate tectonics, the origin of these rocks was a subject of great perplexity.
Silicate minerals show clearly that what is important about an atom are only its size and charge, and that they are assembled into compounds any way that gives the least free energy. What results depends on the constitution of the melt. There is no calcium analog to forsterite. Calcium, at 0.099 nm, seems to be too much larger than magnesium at 0.065 nm. Zinc, on the other hand, has a radius of 0.074 nm, which is close enough, and there is an analog to forsterite, which is the fluorescent mineral willemite. We have series of minerals in which one ion replaces another, Ca-Na and Mg-Fe for example, making minerals that are carefully distinguished by the mineralogist, but are really the same thing to a good approximation. Zn fits in with Mg and Fe, but it is too rare to make much of an impact. Oxygen and fluorine negative ions are the same size, but differ in charge. F often substitutes for OH, notably in micas. SiF4 can't make macromolecules, so Si always prefers the oxygen.
Life burgeoned in the sea, but this world is still largely unknown to the general public, except for the large sea life they plunder and eat. The occupants of the sea are the plankton that float passively at the surface, the nekton that swim in its volume, and the benthos that lie on the bottom. Plants exist only among the plankton, since they are dependent on sunlight to drive photosynthesis, constituting the phytoplankton, the "plant-plankton." They are the source for all the food in the ocean, which covers 70% of the area of the planet. All other life eats them, smaller things that eat them, and whatever drops to the bottom. Phytoplankton is largely ignored in ecological arguments, though it must play a major role not only in biology but also in atmospheric and oceanic science. It is well out of sight, and hence out of mind.
Plants contain chlorophyll, contained in organelles inherited from earlier, simpler organisms who could not survive in the new oxygen-rich atmosphere created by the forms of life that released it to the atmosphere as a waste product, and poisoned themselves. The chlorophyll manufactured complex compounds of carbon from which energy could be derived by oxidation. Animals ate what they could find of these substances. Movement and sense organs were stimulated by the search for food. Much of this life was single-celled or acellular, microscopic and soft, not preserved in the fossil record.
Cellular plants usually made use of strong, rigid cell walls constructed from cellulose to give them the ability to grow large and complex. Animals used chitinous or cartilaginous substances for the same purpose, as framework to grow on, or as protective enclosures. Then animals found they could survive more comfortably by eating smaller animals that grazed on phytoplankton, and spare themselves the trouble of filtering huge amounts of water. The response to predation was the favoring of armored coverings, and none proved better than calcium carbonate, which could be extracted from the sea water (the richness of the atmosphere in carbon dioxide led to thick deposits of calcareous sediments, and sea water was full of carbonates).
This occurred at the beginning of the Cambrian era, and suddenly the fossil record is full of calcareous shells, which were easily preserved in conditions favorable to calcareous sediments. It should be remembered that the presence of a fossil is evidence of the existence of an organism, but the absence of a fossil is by no means evidence of the absence of an organism.
The foraminifera, of which globigerina is an example, were acellular animals that chose a carbonate skeleton to grow in and around. Their fossils are found beginning in the Ordovician. However, another protozoan that chose a siliceous skeleton, the radiolarians, is found from the Cambrian on. These fossils, or tests, look like spiky balls full of holes, from which the pseudopodia of the animal projected. The porifera, sponges, were multicelled animals with undifferentiated cells and no sensory organs. Some chose to grow on siliceous spicules, which also left a fossil record beginning in the Cambrian. It is easy to call these animals "primitive" as if they were just an inferior stage of development, but it should be remembered that they still exist today, which shows that they are well-adapted to their environment, and very successful.
Now we come to our main subject, the diatoms. Diatoms are microscopic plants, algae, that belong to the phytoplankton, plants that float wherever the sea takes them. In the Jurassic, they appear for the first time in the fossil record, since they then developed tests, or shells, or valves, or more precisely frustules (Latin, "little morsels"), of opaline silica. The "opaline" means that it contains some water in its structure. They probably had always been around, since they are essential to life in the sea. There are other algae that help diatoms with photosynthesis, but diatoms are the primary sea plant, living in astronomical numbers in oceans and fresh water. Every drop of sea water (and many drops of fresh water) will contain diatoms. They prefer higher latitudes, upwelling currents and continental shelves, and float at depths less than 70 metres, where they have sunlight and nutrients. The land of the earth produces about 40 million tons of carbohydrates from photosynthesis, but the sea produces 80 to 120 million tons, mainly due to diatoms. Beneath a square metre of surface of the Gulf of Maine, for example, there are from 7 to 8 billion diatoms. One diatom can have 100 million descendants in a month. They are astonishingly prolific and efficient, making the sea as productive as an equal area of tropical rain forest. They are one of the dominant factors in the CO2 economy of the world, though seldom regarded.
Diatoms are the foundation of the food chain in the sea. They are eaten by copepods and other small zooplankton. The small crustacean Calanus grazes on diatoms, herring eats the crustaceans, and people eat the herring. Problems with the diatoms, calanus, or herring can interrupt the supply of fish. Herring was made extinct in the Baltic by overfishing long ago. This probably encouraged the crustaceans, who ate all the diatoms, and the whole chain collapsed. There is still no herring in the Baltic. Any economy of the sea must take the diatoms into account.
Although our main interest here is in silica-using diatoms, as far as photosynthesis in the sea is concerned we should not neglect the minute cyanobacterium Prochlorococcus, discovered as late as 1988, which probably accounts for half of the marine photosynthesis. I have no idea how these tiny cells (500-700 nm diameter) are eaten, since they would be difficult even to filter unless clumped, but they are no doubt the foundation of an important food chain. These cyanobacteria, one of the earliest forms of life, are also the most successful, it would appear.
The frustules of a diatom are in two parts, like a pillbox and its lid, that fit into one another, as shown in the diagram below. They are not hinged like the shell of a clam. This allows the plants to multiply by cellular division. Half of the protoplasm sticks to one half of the frustule, half to the other, and they come apart. Each half grows a new part of a frustule that inserts into the other, which forms a lid. The cells get a little smaller each time. Sexual reproduction creates spores that grow into a full-sized diatom to maintain the average size of the plant. This two-part frustule is the reason for the name "diatom" (Greek, "cut in two"). Although the protoplasm contains green chlorophyll, coloring agents in the shell give the plant a beautiful golden color, so they are the family of the chrysophyta, "golden plants."
The shapes of the frustules are either pennate, with bilateral symmetry, or most commonly centric, with radial symmetry, as illustrated in the figure. The pennate forms may even be motile to a degree, but all the centric forms are sessile. There is no reason for them to want to move, since sunlight and water are all they need. Their decoration can be elaborate. The centric forms usually have a central pore and radial lines, while the pennate forms have a medial line and decorated edges. 5- and 10-fold symmetry is often seen. These attractive forms can be examined under a microscope. It is estimated that there are 6000 to 10000 species of diatoms living at the present time, so they are eminently successful in what they do.
When the diatoms are finished with the frustules, they sink to the bottom, where they form diatomaceous ooze, if they are not dissolved. Diatoms like sunlight, salinity appropriate to their species, cool water, phosphate and nitrate nutrients, and, of course, silica. Sea water on the average contains from 0.02 to 4.0 ppm of silicon by weight, in water of standard 19% chlorinity. Diatoms are happier near the upper limit, where they can thrive. When frustules sink to the bottom, they take their silica with them, so it must constantly be supplied to the oceans. Where silica is plentiful, and the other requirements are met, diatoms will thrive, and make thick deposits of their shells. This rock is called diatomite, or "fossil flour," which has many uses, and is highly valued.
Diatoms can also live in fresh-water lakes. These lakes normally have very little silica, but vulcanism can supply volcanic ash and silica-laden hot waters, which run into the lakes and encourage diatoms. Something of this nature happened in the U. S. West, and especially California, in the Tertiary, which produced huge deposits of diatomite both in lakes, and in the ocean border. An arm of the sea reached in near Lompoc, just north of the Santa Inez Mountains, and with the nearby volcanic action, a thousand metres and more of diatomite was deposited in the Upper Miocene and Lower Pliocene, the largest deposit in the world.
Diatomite is 85-92% SiO2, most of the rest being clay. When pure, it is brilliant white, but also can be yellowish or buff. Less pure rock is called diatomaceous earth or Kieselguhr. It has 75% porosity, more or less filled with water that can be driven out by heating. It can be quarried in blocks, that can be used directly for heat insulation. Its voids give it a high thermal resistance. Each cubic inch contains 40 million frustules or more, all of which are nicely sharp and all surface and pore space. The nature of the frustules is important in the application of the diatomite. It is a gentle abrasive, so it is used in silver polishes. In paint, it gives a matte finish as each frustule roughens the surface. It has excellent absorptivity for liquids. It was used as an absorbent for nitroglycerine, rendering it insensitive to shock, in Alfred Nobel's dynamite.
Its most important industrial use is probably as a filtering agent. A suspension in water is first passed through a screen, depositing a layer of diatomite on the screen. Then the suspension to be filtered moves through the prepared layer, leaving behind all but the finest matter. One source says that it even filters colloidal solutions, which pass through ordinary filter papers. This filtration is suitable for the "hard jobs" in clarifying liquids.
This is a suitable place to talk about the other siliceous rocks, most of which are biogenetic. The term organogenetic sometimes found is clumsy, looking more like "tool-made" than "made by life." The Cretaceous Chalk of England is a famous rock, mostly white calcium carbonate from foraminiferal tests. However, it also contains silica, now in the form of the familiar dark flint nodules, which typically occur in bands. The thick beds were originally deposited as a siliceous chalk, the silica from opaline sponge spicules, and the chalk from foraminifera. Percolating waters replaced the opal in the spicules with calcium carbonate, and redeposited the silica as much less soluble silica concretions, called flints. This separation into two purer constituents is not rare in sediments. They are roughly brick-sized elongated and rounded stones that remain covering the ground after the chalk weathers and erodes. They have been used like bricks, indeed, for walls and buildings, and even churches, since they are very durable.
Flint is deserving of special study, especially in the ways it has been used by man. Flint was prized by neolithic man for its conchoidal fracture, by which sharp, hard edges could be produced for tools and weapons. It was the earliest mined substance, since good flints for making tools are not common. The flint mines in the chalk of Norfolk, in England, are famous. Flint-mining technology was later adapted for the mining of coal and metals. Flint was used for edged tools and weapons of all kinds, and great skill was involved in its shaping, called "flint-knapping." Egypt, a country poor in metals, used flint ploughs, showing how meaningless the term "neolithic" is as except as a rough cultural indicator. Flint was the origin of technology.
Flint and steel are well-known as fire-starters. When struck together, fat sparks are produced (probably from the steel, heated from impact and then burning in air) that are caught in tinder. The glowing bit of tinder is put on dry grass and wood shavings, and coaxed into flame by gentle breaths. This was a much easier way to start a fire than any available alternative. The flintlock musket was invented around 1630, in which a spring-loaded flint hammer strikes steel to send sparks into the firing pan, where there is fine gunpowder, the priming charge, that ignites the granular gunpowder in the chamber. Flintlocks were still used in the early 19th century. The United States Army cooled its heels in Piqua, Ohio waiting for flints in one phase of the War of 1812, on its way to Lake Erie. The "flints" in cigarette lighters are not flints, but an alloy of 70% mischmetall (50% cerium with 50% rare earth metals, mainly lanthanum), and 30% iron, patented by Auer von Welsbach in 1903, that gives off sparks copiously when struck by steel. Flint and steel are used in the "sparkers" used to ignite a gas torch with one hand.
Flint is a kind of chert, though chert is bedded, rather than occurring in nodules. Flints may be blackened on the outside, but when broken show a white center that is almost pure silica. Chert is also a biogenetic rock associated with limestones, but occasionally occurs alone in thick beds. Like flint, it is fine-grained or microcrystalline quartz or chalcedony. There has been a great geological controversy over the origins of flint and chert. Some hold that it is syngenetic, or created at the same time as the rock in which it is found, while others maintain that it is epigenetic, or formed later. It is probable that some flints began as silica gel blobs formed in the alkaline carbonate deposits, and so are syngenetic, but most flints appear to have formed later, as described above. Petrified wood and similar bodies where silica has replaced the original material, are clearly epigenetic. The mechanism of this faithful metasomatic replacement is unknown. As in so many other cases, both origins are probably involved under different circumstances. Bedded chert probably began as radiolarian or diatomaceous ooze, but time and metamorphism have erased the original fossils. It is difficult for the silica concentration to rise to the point where silica would precipitated out of solution. No such case is known, except for the siliceous sinter that forms at hot springs.
Radiolarite is like diatomite, except that the known examples are from the palaeozoic, and are hard and indurated, not soft and powdery like the Tertiary diatomites. Radiolarite has been used for whetstones and grinding stones.
When a felsic rock weathers, water leaches out the potassium and sodium of the feldspars, which crumble into a soft phyllosilicate called clay, containing mainly aluminium, silicon and oxygen. The flakiness of clay minerals are the reason for the fissility of shales. Where for some reason they have not become aligned in layers, the result is nonfissile mudstone. The crystals of silica are little affected by weathering, and remain as loose sand. Mixed with organic matter, humus, the result is soil in which plants can grow. Long weathering in humid tropical or semitropical areas may even leach out the silica, leaving only ferric and aluminium hydroxides, which are even more insoluble. This poor, reddish soil devoid of most nutrients is called laterite from its resemblance to brick clay (which, of course, it is not). Bauxite, the ore of aluminum, is a lateritic deposit consisting mainly of aluminium hydroxide. When basaltic rock weathers, the equivalent of clay is greenish chlorite, also a flaky mineral.
As our example of a clay, let's take the valuable kaolin, a pure form of clay that forms a slippery white mud, named after its original source in China. It is composed of six-membered rings of silica tetrahedrons, rendered electrically neutral by Al+++ and OH- ions strategically arranged. Its formula may be written Al2Si2O5(OH)4, or Al(OH)3·HAlSi2O6, or even Al4Si4O10(OH)8, which shows the Si to O ratio of 4 to 10 typical of the phyllosilicates. A "hydrated aluminium silicate" is a good description. It is, at any rate, a layered silicate like the micas. Clay usually is a complex mixture of related compounds with numerous impurities, but kaolin is a good representative of the family. Kaolin is called "pottery clay" because it is plastic, but it is mostly used for paper sizing, and only a minor fraction is used for pottery. Large amounts come from the weathered granite of Cornwall in south western England, where the clay separated in purification makes tall, conical white hills.
An exotic clay is bentonite, found in the Upper Cretaceous Benton Formation near Denver, and also in northeastern Wyoming and other places. It is formed from the weathering of volcanic ash (whose major ingredient is feldspar), and occurs in irregular lenses in formations near the Cretaceous-Tertiary boundary in this region. Its major constituent is montmorillonite, a common clay mineral similar to kaolinite, in a colloidal form. Montmorillonite is a non-plastic clay with a maximum particle size of 1 μm. Its distinction is that it has a great appetite for water, of which it is starved during its long dry burial in impermeable shales. When water can get to it, it will swell faster than a thirsty horse in a pond. It can adsorb 10 to 30 times its volume, and swell accordingly, the water coating the phyllosilicate sheets. A large amount thrown into water makes a gummy mass that is good for plugging leaks, since it is nearly impermeable. It was used for this purpose during the construction of the Grand Coulee Dam in 1937, when a water leak threatened to undermine some important towers. Bentonite was forced into the sand under pressure, reducing the leak from 30,000 gpm to 250 gpm almost immediately. This makes it useful to plug leaks while drilling oil wells, but a rather unsatisfactory foundation for a house or runway. However, undaunted brave people keep building houses and airports on bentonite. Denver International Airport is built on bentonite in a tornado alley, but so far, so good.
Montmorillonite is also know as fuller's earth. It was used in the process known as fulling, which was the cleaning of wool from fats, performed in a slurry of montmorillionite by pounding. Now that good detergents are available, other methods are used. Cats like to bathe in montmorillonite if they can find a suitable clay to roll in. It cleans much more effectively than soap and water, and is much easier on the cat.
It was discovered that one could shape useful vessels from clay mud, and heat them in a furnace, or kiln, by the gases from a fire, and that they would harden into a porous stone and could be used for holding liquids and for cooking (if only by dropping hot rocks in them). The impurities in the clay form compounds, like slags, that melt at lower temperatures than the clay itself, and cement together the particles of the object. Kaolin is too pure to do this by itself. Pottery clay is carefully mixed from clay--the "right" clay--and sand, and perhaps other ingredients, to give the best product. Pottery that has been baked in this way is called "biscuit." It is usually red or buff colored from the iron impurities, which are oxidized in the heat of the kiln.
To give the objects a durable, nonporous surface, they must be glazed with some glassy substance melted so that it will flow over the surface and fill the porosity. This glaze must have a melting point lower than that of the slags binding the biscuit, so that the objects will not soften and sag when they are fired. Decoration may be applied to the biscuit object, and then the glaze, in the form of a liquid slip, is applied to the surface. The object is fired again, to melt the glaze. The heat often changes the colors of the decoration as well as the glaze, as substances oxidize. Glazes are made from salt, NaCl, and metal oxides, such as those of tin or lead. Chemical changes occur when the object is fired. The salt becomes something else, so the glaze does not dissolve in water. The glaze can be clear, showing the decoration or the color of the biscuit beneath, or may be colored. Borax with metal oxides produces good colored glazes.
A mix of white orthoclase, kaolin and quartz fuses at 1100-1300°C to a white, vitrified, translucent material called porcelain, long a Chinese trade secret. By using phosphates from burnt bones in a pottery clay, a similar material called bone china can be manufactured. The secret of making porcelain was finally decoded in Meissen, near Dresden, in the 18th century. Good porcelain is also made in Golden, Colorado using a local clay. Besides its use in fine tableware and ornaments, porcelain is useful in laboratory ware and as electrical insulators. Unlike glass, porcelain does not adsorb moisture that would spoil its insulating properties.
The clay may have a humbler destiny, as construction brick. Brick was originally made from clayey earth and a binder such as straw, mixed with just the right amount of water and pressed into molds. When the brick dries in the sun, the clay becomes hard and coherent. This makes a very satisfactory construction material, familiar as adobe in the Southwest. Such bricks were generally used in all parts of the ancient Mediterranean world until the first century BC. Egyptian palaces were made of them, as well as peasant's homes. In time, all melt into shapeless heaps of earth, since adobe is not preserved in the long haul. Homes in Greece and Rome were also of adobe, even substantial buildings. Only temples were in the expensive and hard-to-work, but very durable, stone.
Curiously, right now in Denver homes are built of plywood covered with tarpaper, which is inferior even to adobe, and are externally decorated with veneers that make them appear to be substantial buildings. They will rot and collapse even faster than the Egyptian palaces. They must be considered only temporary buildings.
A more durable material, rivalling stone in durability, can be made by baking the bricks in a kiln, like pottery. Baked brick was known in Babylonia, and used for limited purposes, but its widespread use in Europe was introduced by the Romans. The proper clay must be chosen, and it is mixed with water to make a stiff mud that is molded into the bricks. The Romans preferred a brick that was a relatively thin square, while we use a brick that is about 2"x4"x8". The bricks are fired, or burnt, in a kiln at a higher temperature than pottery, so they become strong, and usually red from oxidation of iron. Bricks for external use must be vitrified on the surface so they will not absorb water which will shatter the brick when it freezes.
Tiles are just bricks, but made in special shapes adapted for different duties. There are wall tiles for decoration, floor tiles for paving, and roof tiles for roofing. Ordinary bricks were also used for paving, even for streets, a service they performed very well and attractively. Corsicana, Texas, was called "The Bricks" on account of its paved streets. Bricks from Coffeyville, Kansas paved station platforms on the Santa Fe. Tall chimneys of brick kilns marked the outcrop of excellent brick clay in an arc from Oxfordshire to Huntingdonshire north of London. Brick is a warm, pleasant and human building material. Ignatz the mouse heaved brickbats at his admirer, Krazy Kat. A brickbat, incidentally, is half a brick, fitting neatly into the hand and of use in labor disputes and political demonstrations.
Bricks are laid using mortar, which is slaked lime, Ca(OH)2, mixed with sand. Lime, CaO, is made by roasting, or "burning," limestone. A thin layer of mortar is spread to seat the bricks. The bricks absorb water from the mortar, and the rest of the water evaporates. The calcium hydroxide, some converted into calcium carbonate by the carbon dioxide in the air, crystallizes into a tangle of crystals that bind the bricks firmly by preventing relative sliding motion. Brick walls depend on weight for stability. They would not do on the space station. Mortar will not set in large amounts, like cement, only in thin sheets.
Denver was plagued by fires in its early, jerry-built history, but the abundance of good brick clay in the vicinity allowed Denver to rebuild and grow in brick. Many houses even had tile roofs. After the second world war, building codes were cut and shaped to maximize the profits of the builders, and this meant cheap wooden houses, as mentioned above. Unless the brick houses are torn down for space to build impressive million-dollar shacks, they will long outlast them. Much Roman brick survived its first use, and was reclaimed for use in later buildings. Some of Denver's recent wooden palaces were covered by an impermeable stucco-like sheathing. Water got behind it, and now the wood will rot out in less than a decade unless replaced.
The Romans also invented a substance like mortar, but which would set in large volumes and even under water, and made an inexpensive substitute for stone. This material, Portland cement, is made by roasting finely crushed limestone and clay in long, rotating cylindrical kilns. The limestone and clay are considered as bringing the constituents CaO, Al2O3, and SiO2. These ingredients are mixed as a wet slurry in exactly the right amounts to produce the desired substances, which are: (1) Ca3Al2O6, called Ca3A; (2) Ca3SiO5, called C3S; and (3) Ca2SiO4, called C2S. The slurry is introduced at the top of the kiln, it soon is dried by the hot gases, and and tumbles downward until it fuses into clinker in the heat. The clinker from the kiln is ground to a very fine powder, since the reactions occur only on the surfaces of the particles, and everything inside is wasted.
It is called Portland cement because a Yorkshire bricklayer named Aspdin roasted limy road dust with river mud to make a powder that, when mixed with water and sand, would harden to a reasonable imitation of the excellent and expensive Portland stone from Southern England. He received a patent in 1824 for this discovery. Smeaton had used a less perfect cement at the Eddystone Rock lighthouse in 1756 by burning a clayey limestone. Cement does not set by drying. The water combines chemically with the cement.
When cement powder is mixed with water, reactions start immediately. Compound C3A hydrolyzes to give calcium and aluminum hydroxides. Crystallization of these hydroxides gives the initial strength of the cement, up to about 24 hours from pouring. Meanwhile, the hydroxides are working on compound C3S to form complex calcium aluminum silicates that increase the strength of the cement for up to about 7 days. C2S works the slowest, giving the increase in strength from 7 days to 28 days, when the cement is considered to be fully set. Changing the relative amounts of the three ingredients controls the setting time and other characteristics of the cement. A little gypsum, CaSO4 may be added to retard the initial set and give sufficient time to place the cement. Heat is given off when cement sets, and this must be allowed for so that the cement does not crack. A typical cement has an analysis of 60-70 CaO, 20-24 SiO2, 3-8 Al2O3.
Cement is not usually used by itself, but in the cheaper but equally effective form of concrete, in which it is mixed with aggregate. Aggregate consists of coarse stones or gravel, and a finer material, sand. It is best for the sizes of the aggregate to be graded so that the voids are filled as well as possible. Concrete can be mixed on-site, but advantage is taken of the fact that if freshly-mixed cement is kept agitated, it will not set until the agitation stops. The crystals that would give the initial hardness are continually disturbed, and new material is continually supplied from the surfaces of the grains of cement. The distinctive trucks with the rotating tanks are evidence of this. If you stopped the rotation with the tanks full of concrete, you would have a tank full of solid concrete in a short time, but as long as it rotates, you can pour it.
An excellent construction material known as cinder block is made from cement mixed with rough cinders, and poured into forms. What is used now that coal fires are not as common, I do not know. Cinder was the slaggy material that resulted from reactions between sandy and limy material in the coal ash in a hot fire. The best was in small, sharp pieces that would interlock. It had many excellent characteristics, and was useful product that could be had for the effort of carrying it away. It made good footpaths, and much athletics took place on the "cinder track." Cinder block is much cheaper than brick, and lighter, and much more durable than wood. Why it is not more widely used is a mystery to me. It can be tarted up just as attractively as plywood and tarpaper can.
I had thought of titling this section "Silica and Sand" but sand is usually silica, so it seemed redundant. Glasses are substances that are not crystalline, but tangled webs of long molecules. Silica loves to make long molecules, so it is used in glass for this purpose. A glass is a hard solid at lower temperatures, then softens at higher temperatures, usually above red heat, when it becomes soft and can easily be worked. At still higher temperatures, at white heat, it liquefies, and can be cast or a blob put on the end of a hollow tube so it can be blown into hollow shapes. Glass, like pottery, has been an outlet for artistic skills, and is of great beauty in addition to utility.
We think of glass as transparent, but this is a secondary characteristic. A glass must be quite pure to be transparent, or even translucent. One might wonder why glass can be so perfectly transparent, when it is a disorderly random tangle of silica chains. It is not hard to believe that an ordered crystal, like quartz or salt, could be transparent if the lattice spacings are smaller than a wavelength, and this can be shown mathematically. It is also true that a random structure can also be transparent, if the randomness is uniform on the average over distances of a wavelength or so. Gases, indeed, are an example of this, as is water. As Rayleigh showed, the blue of the sky is scattering from the fluctuations in density, not from the molecules as individuals. The disorder in glass is over distances less than a wavelength (about 500 nm), and so pure glass is transparent. Disorder over larger distances, or the inclusion of larger particles, makes a glass first translucent and milky, and finally opaque.
I have a beautiful glass bottle, an unguentarium, that is two thousand years old. It is 90 mm long by 15 mm in diameter, and weighs 6 g. It looks like a test tube with a somewhat bulbous end, and the glass is colorless. Clear glass was developed by the Romans, and was rather common. It was even used for windows, as we now use glass. Afterwards, houses just had holes in the wall in Western Europe, where the wind certainly did blow through, or at best oiled paper. Most Roman glass that is seen in museums is of common, crude objects that would have been cheap at the time and were discarded in trash, and so are easy to find. If you go to Colchester, for example, you will see in the Roman Museum there some fine Roman glass, that is clear, brilliant, symmetrical and very attractive, as good as that made today. Today, "pressed glass" made from cast glass is very beautiful, comparable to the much more expensive cut glass, but available for pennies.
One glass is made from quartz alone. Quartz (pure sand is used, not the large crystals) melts at 1800°C, and when it cools cannot figure out how to get back into the hexagonal quartz structure. This is a homogeneous and isotropic substance known as fused quartz or quartz glass. Fused quartz softens at 1665°C and anneals at 1140°C. These are very high temperatures, so fused quartz is difficult to work. Its density is 2.2 g/cc, its hardness Mohs 4.9 (not nearly as hard as quartz). Its bulk modulus is 5.3 x 106 psi, shear modulus 4/5 x 106 psi, Young's modulus 10.4 x 106 psi, and Poisson's ratio 0.16. In tension its strength is at most 7 ksi, but in compression it is very strong, 160 ksi. Its dielectric constant is 3.75, its index of refraction 1.4585. Its specific heat is 0.18 cal/g-K, its heat conductivity 0.0033 cal/cm-s-K. A very useful property is that its coefficient of linear expansion is only 5.5 x 10-7 per °C. Borosilicate glass (Pyrex) has a coefficient of 33, and ordinary glass 90. Fused quartz will, therefore, withstand thermal stress very well. One indispensable property of quartz is that it is transparent to the ultraviolet, so it can be used for envelopes of ultraviolet lamps for sterilizing, exposing photoresist, erasing programmable read-only memories, exciting fluorescence, and other such duties.
Ordinary glass is called crown glass, or soda-lime glass, or simply soft glass. It is made by fusing sand, sodium carbonate and lime. An average composition by oxides is SiO2 73, Na2O 15, CaO 7, MgO 4, Al2O3 1. It softens at below red heat, and so is easily worked in the Bunsen flame, as in making chemical apparatus. It even melts in the flame, so it can be flame-polished. It is called crown glass not because it is used in crowns, but from the early method of manufacture, which came down from Roman times. A large globe of hot, soft glass is blown, and then the closed end is cut off with glass scissors. In this state it looks like a crown. It is then rotated rapidly to flatten it, and cooled as a round plate. Window panes are cut from it. The "bull's eye" piece where the blowpipe is broken off is less desirable as a window, but makes a good decoration and is even simulated these days. Elizabethan window panes were small and set in lead mullions to make a large window. Large, flat panes were not available until the 19th century.
The blowpipe was introduced by the Romans, as well as most of the associated techiques for the hand manufacture of glass objects. Blown glass is not only easy to make, but gives far wider scope to artistic expression than earlier methods, such as casting. A blowpipe is an iron pipe about four feet long, with a mouthpiece at one end. The "gaffer" collects a "gather" from the pot of molten glass in the furnace, and then shapes it by blowing and rotating, and by the use of iron tools. The work is reheated at the furnace to keep it plastic. The Greco-Romans were known for cement and building techniques, but an equally great contribution was the technology of glass.
In borosilicate glasses, boron trioxide, B2O3 replaces all the soda and lime. This reduces the coefficient of thermal expansion by a factor of 3, so the glass will resist thermal stress, and also increases its resistance to chemical attack. In most parameters, it resembles fused silica, except that it softens at 700°C, a red heat, making it much easier to manage than fused silica, which requires a thousand degrees more.
Flint glasses were developed originally for decorative purposes, to have indices of refraction greater than the usual 1.5 for crown and borosilicate glasses. The name refers to the original use of flints to supply the silica, but this is not an essential matter, and flints were used in making other kinds of glass as well. In flint glass, lead oxide, PbO replaces the lime, and potassium carbonate K2CO3 some of the soda. A high-lead glass might be SiO2 70, PbO 12, Na2O 10, K2O 6, CaO 2. The problem is keeping the lead in solution. An index of refraction of over 1.7 can be attained.
Flint glass has less dispersion relative to its index of refraction than crown glass. Dispersion is the change in refractive index with wavelength. For all glasses, the index increases at shorter wavelengths. In lenses, this causes chromatic aberration, which was very troublesome in early telescopes with simple lenses. If a strong positive lens of crown glass is followed by a weak negative lens of flint glass, the net power may be positive, while the chromatic differences are equal and opposite and cancel each other. Such a lens is called an achromat. Optical glasses have been intensively developed and improved. For an introduction to this subject, see Jenkins and White.
Colored glass is not only very attractive, but has uses as signal lenses as well. Clear glass can be colored by transition metal ions in solution, which are well-known chromophores, or by colloidal suspensions. The colloid particles must be smaller than a wavelength of light, or they would make the glass cloudy. Transition metal ions generally come from dissolved oxides. The principal metals used are Ni, Co, Fe, Mn and Cr. The colors produced depend on the oxidation state of the metal ion, and on the composition of the glass. Cobalt gives a deep blue, which is much prized. Fe++ gives a green color, but Fe+++ makes yellow. Adding FeO to glass gives a green color, and makes the glass strongly IR-absorbing. Such glass can be used as "heat filters" in slide projectors. Iron is a common impurity in sand, so inexpensive "bottle glasses" may be greenish. The glass insulators used on American telegraph and power lines were of greenish glass. Glass, incidentally, does not make a good electrical insulator because of its propensity of picking up a surface film of water that is conductive. There are several ways to make a red glass. Colloidal suspensions of selenium or gold may be used. Cuprous oxide reacts to form Cu++ ion and colloidal Cu, producing red by two mechanisms. Another source says that the copper must be in the Cu+ state to give red; cupric copper gives the usual green. Selenium works better in potash glass than in soda glass. A little selenium and cadmium sulphides added to molten glass give a good result.
Colloidal gold in glass gives a deep red color. This ruby glass was first made by Kunkel in 1679, when he added Purple of Cassius (a gold precipitate on gelatinous stannic hydroxide, a sensitive analytical test for gold) to molten glass. It can also be made by adding AuCl3 to the glass; the heat is enough to reduce the gold. The same thing happens in aqueous solutions. As little as 0.1% of gold gives a good color. A little stannous oxide will prevent the coagulation of the gold particles. If the particles become larger, the color turns to blue.
In colored glass, the color seen by transmission is the complement of the color that is absorbed by the ion or scattered by the colloid. If all the shorter wavelengths are absorbed, then the color will be red. If the green is absorbed, the result is purple. The energy absorbed by a transition ion is lost to the lattice by emission of phonons or other non-radiative process. Gustav Mie became interested in the explanation of the red color of colloidal suspensions of gold in 1908, and this led to his famous theory of scattering by small particles. Gold particles are hydrophobic (so they are not hydrated). The sol formed by particles 0.04 μm in diameter has a peak absorption at 530 nm, and is red by transmitted light. There is a little yellow-green scattering, but mainly an absorption in the green and shorter wavelengths, with good transparency in the red. A diameter of 0.14 μm gives the blue gold sol in which scattering is large in the red. The optical properties of the gold are important in the effect, not just the size of the particles. The effect does not depend greatly on the transparent medium, whether water or glass.
Opaque or translucent glass that is used for decoration may be colored by other means as well. Borax is well known for reacting with metal oxides in glass to produce good colors. Glasses can be made opaque and white by adding oxides that give suspensions of materials of a different index of refraction. SnO2, TiO2, CaF2, As2O3, Ca3(PO4)2 and cryolite are typical additions. Opaque glass applied to a metal or other substrate by fusion is known as enamel, and is an important artistic medium. As was mentioned above, pottery glaze is a glass. The mixture applied to the substrate to be fused into enamel is called a frit. This enamel should not be confused with the paint of the same name, which imitates it with an organic film.
Glass, pottery and gems show that there is much beauty in silicon, as well as the everyday utility of semiconductors, stone and cement. The study of silicon leads into many areas of pure science, including the structure and dynamics of the earth, life in the seas and on land, semiconductors, colloids and other areas that we have touched on in this article.
M. Prinz, et al., editors, Simon and Schuster's Guide to Rocks and Minerals (New York: Simon & Schuster, 1978). Excellent colored illustrations.
C. S. Hurlbut, Jr., Dana's Manual of Mineralogy, 16th ed. (New York: John Wiley & Sons, 1952).
R. L. Bates, Geology of the Industrial Rocks and Minerals (New York: Dover, 1969). pp. 360-370. Includes an excellent short account of diatomite.
A. Bar-Lev, Semiconductors and Electronic Devices, 2nd ed. (Englewood Cliffs, NJ: Prentice-Hall, 1984). Any good textbook for the undergraduate course in physical electronics gives a satisfactory introduction to semiconductors, mainly silicon. Beware of authors who do not understand the subject very well.
F. A. Jenkins and H. E. White, Fundamentals of Optics, 4th ed. (New York: McGraw-Hill, 1976). Optical glass is discussed on pp. 176-182.
R. B. Leighou, Chemistry of Engineering Materials, 4th ed. (New York: McGraw-Hill, 1942). Chapters XI-XVI.
H. C. van de Hulst, Light Scattering by Small Particles (New York: Dover, 1981). Gold sols are discussed on pp. 397-400.
C. P. Idyll, Abyss (New York: Crowell, 1976). p. 73f (diatoms).
Excellent photographs of diatoms can be seen at Diatoms.
Picture of olivine kindly furnished by Amethyst Galleries, Inc.. This is an excellent website and specimens can be purchased online. This is probably the best mineral website, and the company should be supported for making it available.
Composed by J. B. Calvert
Created 30 November 2002
Last revised 15 September 2007