Elements covered include:

Metal extraction methods

Periodic Table

Glossary

The figures relating to the consumption of carbon are of course almost trivial when compared to the amounts of oxygen consumed in the conversion of C to CO2. The proportion of atmospheric oxygen was 23.15% by weight in mid century reduced to 22.87% today. About 280 billion tons of O2 is consumed in all processes compared to the 200 - 230 Bt returned to the atmosphere from all plants, forests, and algae.

Panting hordes of people use on average 566 - 591 kg of O2/ year with domestic animals using another 5.4 bt in total.

Respiration by the human population, each one of which is a mobile slow-burning fire, accounted for 2306.4 million tons of O2 in 1975 rising to 3378.7 million tons in 1995.

It is a paradox that oxygen is our most abundant element, the whole world has been described as a mass of oxygen ions bonded together by silicon and metals. It makes up 88.9% of the Earth and 63% of human beings by weight, yet the amount of free O2 in the atmosphere is trivial by comparison.

Projections fifty years into the future at the same rates of increasing population, declining forest areas, and increasing desertification and burning of fossil fuels all point to an inescapable total disaster, which politicians world-wide decline to acknowledge. From where I sit I can see whole mountain ranges devoid of vegetation except for the odd "Wilding" lodgepole pine or douglas fir. The Department of Conservation periodically sends out gangs to fell these "weeds".

The amount of wealth being spent on the Iraqi war, if spent on alternative fuel research and on tree planting, could have saved the human race from disaster. Ironic, ain’t it?

Potassium/Thorium

Th (along with U, Rb, Cs, Tl and Ba) are usually put together as the "K-group" elements. While Ba is divalent it can to some degree replace K+ in potassium feldspars. However K/Th does not seem to follow any logical sequence. Th can be very low in rocks of the oceanic crust (sub 0.5 ppm) but U is even lower and Th/U follows a much more logical distribution. Parental melts of basalts have a maximum of 7 ppm Th and 2 U.
K/Th is very high in mid ocean ridges, being 6800 in the SEIR, and 4810 in the EPR but drops markedly in OIB's with 2706 for Alcedo and 2920 for Sierra Negra. Some alkaline centres however stay high, 4423 for the alkali basalt, hawaiiite-mugearite Westmaneyjar off Iceland, and almost the same for the more alkaline Snaefellsnes area, (4359), but 2296 in the alkaline Vestereis seamount and 3220 for Iceland's Neo Volcanic Zone
There is not much good Th & K data from Hawaii, but Mauna Kea gives 4100 and Loihi 3959, yet Reunion Island is 2514, similar to the Galapagos, but quite different to Hawaii. In Jan Mayen we see 2810, in Macquarie 1985, 1650 in St Helena, and an average of 1068 for the Tubuai nephelinites which show suggestions of two parentages in some elements including Th-K. The potassic Sabino and Mahuvura in the African Rift Zone have 1320 and 1771.
With no data for Tristan or Gough and erratic data for Heard we cannot make any statements on sodic vs potassic OIB's. There a general lower K/Th in alkaline rocks but this is not universal.

Potassium/Rubidium

Historical Note
In the early 1960's a good deal was known of Rb as it was determined by isotope dilution for Sr87/86 age dating, and as K was also done for potassium/Argon dating there was more data on K/Rb than any other trace element though no one seemed to have looked at the data as a whole. I began work on my Ferrar Dolerites for a post-doc in ANU, Canberra in the early 1960's. The NSF had funded the project and I had major elements done commercially while I attacked the trace elements. The well-known Bruce Chappell had worked out accurate methods by XRF for Sr and Rb again for age dating purposes, so these were the first I tackled. My shift on the XRF was from midnight to 6am, winding by hand from background angle to peak to Bg. to Pk. to Bg using 45 KV and 200 sec counts and writing them down by hand. Reducing these to counts per sec net was done with a clattering Facit calculator. When the results were compared to my commercially done K's, the result was the classic shotgun scatter. Very dissapointed, I took the graph to Ross Taylor, who had once been with Louis Ahrens in Capetown. Ross peered at the graph doubtfully. "Where did these K's come from?"When I told him, he tapped the errant numbers with his glasses. "I think I would do them again. Go and see Dave Cooper!"
Dave had also developed very good K techniques by flame photometry also for the K40/Ar40 lab and soon I was digesting dozens of samples in Pt dishes with HF and nitric. The results produced a straight line of K against Rb. Ross smiled quietly, "At least,"he said, "you have achieved a remarkable coincidence of error!"
The results are still good and are shown here, the K/Rb for the Ferrar Dolerite sills average was 227, and when Janet Hergt did them for the Tasmanian Dolerites many years later, hers came to 228. This is not only a testimonial to XRF but also to the incredible homogeneity of continental Flood Basalts over distances of thousands of miles.
In Montreal we began churning out thousands of determinations for trace elements, especially Rb-Sr, by XRF and others followed suit. PhD's and post-docs came from all over the world to do their trace elements and there is still more data in existence for Rb than any other trace element.

What intrigued me was the fact that Rb, Ba and Th showed steeper trends relative to K, while Cr, Ni, showed steeper trends than Mg, Co less steep, while Zn and Mn were invariant. No one had seen element pairs with correlation coefficients of better than 0.99 before and there were bitter remarks at AGU meetings, some people saying loudly they did not believe it. But others did and now none of the work of that era would raise an eyebrow. Why did it never occur to me to put all the "residual elements" as I called them on the same diagram? Peter Baker began doing than in 1985, 20 years later and not till 1989 did "Sun Shine Soon" and Steve McDonnough make a workable system of normalising to mantle compositions though normalising is only a small extension to the chondrite normalised diagrams for the REE which had been used since pre-1960.
The K/Rb ratio is quite consistent though both elements are subject to change in any mineral alteration. Partition coefficients are much lower for Rb relative to K in plagioclase feldspar than for K/Spar, so that in basic rocks where K/Spar is absent as in MORBs, the K and Rb are mainly in plagioclase, and the K/Rb ratio is high, often over 1000. In granitic rocks, syenitic rocks, or in continental flood basalts, the ratio is much lower, often about 220. Alkali basalt series usually have K/Rb of 600-350. Excepting picritic series, K/Rb plots always diminish towards the K axis, so ratios are not constant but decrease with increasing K in fractionation. This can be marked in very depleted rocks but barely significant in high K rocks.

Production and Consumption

Though Rb is the most electronegative element after Cs, its consumption is only a few thousand Kgm pa, (Handbook of Chem. & Physics, 78th ed). The main sources are the minerals leucite, pollucite and the mica lepidolite. Rb is always found in any potassium mineral and does not form it's own. In quantity, Rb only costs $20/g.

K/Rb in Simatic Rocks

I have analysed and seen others done by ID of single MORB samples with about 0.01 ppm Rb and 0.02 max K2O with a K/Rb of over 1000 very frequently in very depleted NMORBs, but here we will only show coherent fractionated series, which are almost unkown in the very depleted NMORBs..

Parental Basalts have a range of Rb from <1 ppm in DMORB to 60ppm in extreme EMORB.

Cont. Flood Basalts

Orogenic, IAB's & Plutonic

Granite

The Oceanic rocks are reasonably predictable, the K/Rbs being over 600 for NMORBs, 400-600 for tholeiitic OIBs and mainly 3-400 for alkaline OIBs, but, rather oddly, even lower in the most enriched EMORBs of Macquarie with a range of about 330- 210 The southern Indian Ocean including the Kerguelen Plateau is however a region where potassic rocks are common, though Macquarie actually shows a negative K anomaly! (Kamenetsky et al, 2000).
The Flood basalts are as always, highly variable, one oddity being the Umatilla series in Columbia River where a high K/Rb, low Rb composition is found with the highest Ba content found in basalts, about 4500ppm The Ferrar, Tasmanian, and some of the Deccan and Parana with K/Rb of 227-245 are very much in the granitic range, a fact which has puzzled geochemists since I identified it in about 1961. Even more puzzling is the complete uniformity over thousands of miles.
Perhaps Columbia River is even more surprising as, while volumes of Grande Ronde, Roza etc basalts are enormous, and poured at in very short time periods, they are both homogenous and very different from eachother.
The Andesite series have surprisingly variable K-Rb in Proto-arc, Early arc rocks. Most andesites are similar to OIB (300-400) but the highly potassic series found on the Argentine side of the Andes, in the Roman Province (Vesuvius) and Turkey-Tibet, are very low, 95-200
Granites show a similar range, being higher in the Trondhjemites, sodic granites and in the 200-240 range for very potassic granites with abundant potash feldspar and pegmatite. Ultra-high-K andesites are all leucitic and low silica, but I do not know of plutonic equivilents.

Distribution of Rb in OIBs

Variation diagram for Rubidium.

Phosphate

Phosphorus is invariably recorded as the phosphate, P2O5. While P has been recorded since the very earliest days of geochemistry, it seems to seldom attract much attention, probably because early colorimetric methods were not too precise. P, like Sr however is a sensitive indicator of parental magma variations and is well worth looking at. Like Sr is reaches a peak at about 4% MgO when apatite appears on the solidus and then rapidly declines to almost vanishing point in trachytes-rhyolites.

Uses of Phosphate

P2O5 is of course one of our main fertilizers in agriculture. It is also has a heavy but declining use in detergents. World annual production of phosphate concentrate is currently about 35 million tons.

P in MORB, EMORB, OIB and Sialic rocks.
Diagrams for P invariably involving whole rocks have a lot of unexplained scatter. This could be analytical as it is usually rather imprecise by both colorimetric and XRF methods. However, basaltic andesites from the Valley of Mexico show a wide scatter between 0.2 and 0.6% both in my own data and in that of Wallace, P. et al ,1999, (C.Min.Pet 135) done 20 years later, so may be real. The scatter seems much less for MORB glasses done by electron microscope. At the peak point at about 4% MgO there are often several high points, probably cumulates which I do not believe have ever been investigated.
The most depleted NMORBs have about 0.025% rising to almost 0.7% in the most enriched Emorbs and to 0.75 in the most fractionated EPR. The range in ocean ridge basalts with some fractionation is 0.09 - 0.17 for the MAR at 37 deg N, 0.12 - 0.225 at 22N and 0.2 - 0.4 at 18N on the Cayman Rise.
In Iceland Krafla peaks in P at only 0.3%, other centres at 0.6, 0.8, to 1.0 but data is never precise.
St. Helena, Tristan and Gough all peak at about 1.0%. Some of the best data is for the Mauna Kea hawaiites with a maximum at 1.3%. Plotted against MgO, the trends describes a loop indicating a reversal in MgO as titanomagnetite and apatite both form.
In some instances, eg the Azores from which we have a good deal of unpublished data, the scatter in the basaltic rocks suddenly ceases at the peak point of 1 % and 3.5% MgO and follows a narrow trend to trachytes so the scatter in basalts-hawaiites may be real if unexplained.

Andesite series show the same scatter with a maximum of 0.2% in the Valley of Mexico, but with higher values in younger, more alkaline rocks. We see 0.2 - 0. 4 for Azufre 1, and Sangay. Nevado del Ruiz has rather less with 0.15- 0.25, Calbuco 0.2, Puyehue 0.4 - 0.6, Parinacota rather more with a variable 0.75%' Vesuvius is one of the few to show better defined trends with 0.9% at 4.5% MgO, (Belkin & Gunn, unpub data).
The general data does not warrant making definitive comparisons at this point which is disappointing.

PGE

The platinum group elements include platinum, ruthenium, rhodium, palladium, rhenium, osmium and irididium. These may occur native or as mixed alloys such as osmiridium (osmium-iridium) in placers in outwash from gabbroic to ultrabasic intrusions. More often they occur in the early-formed sulfides of basic intrusions. Long used for jewelry, the PGE are also noted for their ability to act as catalysts to a range of chemical reactions. Now widely used to reduce atmospheric pollution from car emissions by catalytic conversion, platinum is being universally sought for. As yet the distributions do not make much sense, being mainly in the ppb range. The Merensky Reef in the Bushveldt Intrusion of South Africa is still the world's biggest supply source.

Rhodium is the rarest of all metals but about 3 t /yr is produced valued at $13,000/100gm or $1000/troy oz (prices change rapidly.)

Ruthenium (At.No.44) has a mp of 2250º C and is found in Ni ores, its value being variously given as $1400/100gm and as $30/gm. It has some curious properties, one being that it increases the resistance of Ti to corrosion a hundred-fold when the Ti alloyed with 0.1% Ru. It also is used as a hardener for Pt as is iridium.

Lead

History

Pb is one of the metals that has been longest in use being easily worked and easily smelted from the main ore galena (PbS). The Romans used lead extensively in piping water to houses so that we still call a man who works on water pipes a "plumber" (Plumbum = "Lead"). Lead weights have also long been used in "Plumbobs". Lead is no longer used in water pipes as, while resistant to corrosion, enough dissolves especially in acid solutions, to cause lead poisonng.

Consumption

World use is currently about six million short tons per tyear, or 5.4 mil. metric tonnes. About 3 mt is mined directly but there is a high recovery rate of about 50%+ eg from storage batteries which is the main product. Current bulk price is $520/t.

Ores

The classic lead mine was Broken Hill, out in the desert NE of Adelaide. The ore body consisted of galena, sphalerite (ZbS) and rhodonite (MnSiO3) with many minor Mn minerals, eg, the Mn olivine, and the Mn clinopyroxene, bustamite. On the oxidised cap of the ore body was a thick layer of coronadite, the Mn-lead hydroxoxide. The galena contained about 10 oz of silver per ton which paid for recovery, the Pb, Zn, being sheer profit.
The other type is or was exemplified by the St Joe Mine in Missouri. Here galena was disseminated in reef limestone, possibly by bacterial action. The lead was present in only a few percent but by mining enormous amounts the production rivaled that of Broken Hill. Hundreds of miles of underground road and railway brought up limestone which was mined by the "Room and Pillar" method, great caverns being left with the roof supported by massive columns. No silver is present in the St Joe galena.
Some Australian figures give a production from Broken Hill since 1883 of 20 mt of lead, 15 mt of Zn and 26,000t of silver.

Lead in Oceanic Rocks

Pb appears to have a similar distribution to K being extremely low in MORB rocks and increasing quite steeply with fractionation. Only in the last decade has good Pb data been produced and there is no data for many important petrological provinces.
In the Macquarie partial melt series, Pb increases linearly with decreasing degree of melt from 0.5 - 3 ppm, with a Th/Pb of 2.25. K/Pb shows an identical slope.
The EPR has from 0.6 - 1ppm, (Regelous et al, 1999) but the N. Chilean Ridge is even lower with 0.25 - 0.5 with a U/Pb of 0.1.
In Krafla on Iceland, Pb varies from 1 - 6 ppm with fractionation and on alkaline Heard Id 0 - 7.5ppm with a Th/Pb of 1.75.
There seems to be no Pb data for Hawaii, or for most of the Atlantic Islands. Grand Comore (Class, 2000) has mainly less than 5 ppm and Th/Pb of 1.5.

Pb in Sialic Rocks

In all andesitic series, lead appears in greater amount than Th. In the Scotia Arc Pb is high but variable, but in Puyehue while Th is in the 0.5 - 8 ppm range, Pb is 2 -20 ppm, Th/Pb = 0.32. Parincota has also > 20 ppm and Vesuvius >45 with a Th/Pb of 0.6 which seems about average.
From this one might expect high Pb in Flood basalts and in fact both the Ferrar Dolerites and the Paranema (Parana) basalts have a Th/Pb of about 0.7 with up to 7 ppm leads.

The proportion of radiogenic lead included with common lead will depend on the amount of U-Th present and on the age. When time permits we will try to assemble some figures.

Lanthanides or Rare Earth Elements

The Rare Earths are usually taken to include the 15 Lanthanide elements plus Sc and Y (which are dealt with separately)
The lanthanide REE are also discussed in the REE chapter where normalised diagrams of a range of rock-types are shown also the relative ratios of the different REE.
The main sources of the REE are the minerals bastnazite, monazite, allanite and samarskite. The USA production of all REE was 30,000 tons in 1990, probably over 40,000 tons now, to a value of $32,000,000, though REE to the value of some $600,000,000 were consumed within USA. The prices range from $350/kg for Ce, to $7,500 for Lu, and $20,000/kg for Sc. (USGS Mineral Resources Program, and Commodities List).

Characteristics of the REE

In depleted rocks, the light REE (La, Ce, Nd) are present in very low amount relative to the heavy REE Er, Tm, Yb, Lu. Both with fractionation and with evolution from tholeiitic to alkaline rocks, the ratio of LREE to HREE increases rapidly so that the ratio of, say La/Lu which may be about 10 in NMORBS may rise to about 150 in alkaline rocks and even higher in phonolitic members. Both garnet and clinopyroxene have much higher partition coefficients for the HREE, in both in the subcrustal mantle, and pyroxene in the fractionation of basalt melts where garnet does not usually occur.

REE in MORB & OIBs

If we include harzburgites in the EPR and in the SEIR and also fractionated rocks, we see a range of 0.16 - 61 ppm Ce in the tholeiites. Common little fractionated MORBs however range from 1-6 ppm Ce. Ratios such as Ce/La, La/Sm, even La/Lu tend to stay constant in fractionated series but to curves towards high La in more alkaline rocks. When a curved relationship was seen, the base of the curve was taken for the ratio given.

In the upper members of the more highly alkaline rocks, compositions become increasingly erratic. Carbonatites have extremely coarse crystals of calcite and homogenous samples are difficult to aquire. Averages of the Oka, Algerian, Turiy Peninsula and Fen carbonatites have 2000, 2500 ppm Ce with Ce/La of 1.96-2.1.
It seems that with increasing REE levels, the Ce/La decreases quite consistently from 3.1 to about 1.8 to 2 with very high correlation coefficients, almost never below 0.99. La/Sm is almost straight in any tholeiitic series becoming more curved towards the La axis with increasing alkalinity. Departures from a regular curve usually seem to indicate more than one lineage being present. La/Lu variations are extreme, ranging from 9.2 and less in depleted NMORBs up to perhaps as high as 200 in some minimal data from Tristan da Cunha. The very potassic Gough Id. remains without useable data for the REE.
Heard Id is also potassic and has Ce/La of 2.1 and reasonably good La/Lu of 191. Kimberlites of USA, Aus., and S.Af. have erratic compositions with La/Lu of 1400-1600
More data will be added as it becomes available. REE data has been accumulating for about 40 years, but the standard is so high in the more recent papers, that scatter in older data is obviously analytical.

Range of REE in Sialic rocks

The range of total REE for all active andesitic island arcs and continental margins, showing the Tongan Arc with only 25 ppm total REE up to 240 ppm for the Indonesian Arc. The Arcs are: 1 = Tonga; 2 = Izu Bonin; 3 Scotia; 4 = Marianas; 5 = Lesser Antilles; 6 = New Hebs; 7 = Fiji-Lau; 8 = Kyushu; 9 = Kuriles; 10 = Kamchatka; 11 = Aleutian; 12 = Cent. Amer; 13 = Cascades; 14 = Honshu; 15 = Andes; 16 = Mexico; 17 = Indonesia, 18 = Continental average.

REE of OIBs

	All alkaline OIBs, MgO vs Lu
	All alkaline OIBs, MgO vs Dy

The lighter REE do not show the marked increace in trachytes, commendites, pantellerites, and phonolites.

Scandium

Sc in no way emulates Y or the REE. The liking of the REE for clinopyroxenes is carried to extremes, and in crystallising basalts, with the advent of Cpx on the solidus, Sc quickly vanishes, declining with MgO and CaO. In the "basalt triangle at 7.5% MgO, Sc is at 45 ppm levels in the EPR declining to 25 at 3% MgO, in Mauna Loa it is 60 ppm, and 30 ppm in the MK HSDP drill core.
Loihi basalts have 32-38 ppm and the Vestermanneyjar Is have up to 45 ppm in ankaramites. Tristan has the most regular trend for Sc and in general it appears to follow the decline of CaO in basalt series. Gough appear to have decidedly lower levels in basalts with 15-25 ppm, but we do not know the previous ankaramite history.
In the Archaean of Quebec, there appear numerous clinopyroxenite sills. It would be interesting to know what their Sc content might be, but it is not likely to be over 80 ppm.

Selenium

Se is also recovered in electrolysis of Cu ores, 200 tons of ore yielding l lb.
Se is photo-electric and is used in rectifiers, exposure meters, photocopiers and as a steel additive.
Cost in 1989 was about $300/lb.
Both Se and Te are below detection limits in most silicate rocks.

Silicon

There seems to be a strange logic about the construction of both earth and Universe which is often overlooked. Obviously the basic home of an intelligent life form must be mainly built of highly inert materials, we would not last long if the whole earth caught fire if we dropped a march or let off a hydrogen bomb or two.
Few minerals could be more inert that the silicate series of minerals which make up the bulk of the earth. Boil any silicate you like in concentrated acid and nothing happens. Only hydrofluoric acid which forms gaseous silicon tetrafluoride when heated, has any effect, and HF is not found in nature. Pure quartz (SiO2) is even less reactive than gold, and must be one of the most stable materials known.

All ionic elements form complex compounds with grids of silicon tetrahedra and there they remain, inert and almost ungetatable. Only the few chalcophile elements such as Cu, Zn, Pb, Bi which form compounds with S (or As) are fairly easily roasted and reduced to metal.

Silicon is the substrate on which the entire silicon chip computer technology is based. The extremely large glass container and window industry is also based on silica, which alone amongst inorganic compounds is transparent to light due to it's ability, if chilled fairly quickly, to form a clear, transparent, non-crystalline glass.

Native quartz occurs in small amount in granites, in much greater amount as bands in medium-grade metamorphic schists, but due to it's resistance to abrasion, in even larger amount and often in a highly pure state in sedimentary quartzites and sandstones.

In the form of non-crystalline opal or chalcedony or agate, quartz was formed by Early Man into hand axes by peeling off slivers to form a pointed hand axe. Ian MacDougall who ran the K/Ar dating laboratory in Canberra once showed me a beautifully shaped one 500,000 years old. Acheulian man peeled chalcedonic quartz into arrow and spear points that decimated the post-glacial animal herds and probably exterminated the mammoth. Glassware is known from 4500 BC.

Optical fibres, lenses, and computer chips are a few of modern uses, quartz fume, the residue left from heating quartz is added to concrete as a filler or pozzalana resulting in compressive strengths of up to 15,000 psi.

Carbonatites are pretty well the only forms of igneous rock which do not contain large amounts of silica, in the range 30 - 80%. Our tens of thousands of analyses of MORBs show that 99% lie in the very narrow range of 59 to 51% except for the few fractionated examples, which by removal of the HFSE elements may be enriched up to 70% silica.

Alkaline rocks have less, the extreme nephelinites formed presumably under very high pressure may have as little as 30%, while the fractionated trachytes-phonolites may have between 65 and 55%.

In the island arcs, basalts with 50 - 54% SiO2 are quite rare, andesites with 58 - 62% are common, dacites of 63-69% are not very common and obsidians of > 70% quite rare. There are huge amounts of ignimbrite of 68 - 80% SiO2 but these are formed as partial melts of the continental crustal keels. Continental rhyolites may contain 80% silica.

Normal rocks may, very slowly, disintegrate and reform into clay minerals and form soils, at the rate of about 1 inch per thousand years. Any native quartz present is usually unaffected, at least in temperate zones, but silca is selectively leached in tropical soils subject to heavy, warm rainfall a process probably accelerated by acidic vegetation.

Strontium

Sr is an unsatisfactory element from the geochemists point of view. It is always present in quite adequate abundance, commonly 100-1400 ppm, it has been easily determined to great accuracy by XRF for the last 40 years and it tends to occur in markedly more abundance in alkaline rocks. However, it is neither a high field strength element nor is it a residual or LILE element. With fractionation it rises to a peak in the 6% - 2% MgO range and then tends to begin to diminish as soon as plagioclase appears on the solidus and even more rapidly with sodic and potassic feldspars. Consequently it seems to be rarely mentioned in publications in latter years.

Characteristics and source of Sr

Atomic No 38, Sr has 4 stable isotopes and no less than 16 unstable ones. One of the best known is Sr90 which is produced in atomic bomb fallout, which with a half life of 29 years is a vigorous emitter of gamma rays and is used as a light-weight energy source... there are few thorns without a rose.
Sr 87 is derived from the breakdown of Rb and is used in Sr87/86 rock fingerprinting and age dating.
The main source of Sr is the carbonate strontianite. Sr is used in TV and PC monitors also pyrotecnics because of it's brilliant characteristic colour. Price according to suppliers is $100/gm pure. China alone produces 35,000 tons a year, which seems like a lot of monitors.

Sr in MORBs and Tholeiitic Basalt Series

In the primary parental tholeiitic melt series, Sr ranges from as low as 60 ppm (site 504), or 70-80 ppm in Kolbeinsey Ridge but rises very steeply to 700 ppm in the most enriched Macquarie primary melts. Fractionated series branch off the primary stem at very flat slopes in the most tholeiitic groups, almost certainly due to the presence of plagioclase on the solidus.

For Alcedo volcano in Galapagos and for Iceland, when fractionation passes the 3% MgO point there is a sharp decline towards rhyodacites, the maximum being 300 ppm. Iceland has a wide range in Sr contents, with the Snaefellsnes alkaline rocks peaking at 420 ppm, Westmanneyjar and Hekla being slightly less at 400. The Krafla centre has the least Sr with only 150ppm while the NVZ and the tertiary centres both east and west have a spread between 200-300, (Wood, D.A, 1977; Gunn & Siggurdsson, unpub)
The fractionated members of Puu Oo are similar to Alcedo but depleted members of ~10% MgO behave like olivine cumulates and cross the tholeiitic melt plane, heading towards olivine, ie passing from a cotectic into the olivine field..
In Hawaii, the tholeiitic members do not fractionate to the point of Sr maxima but Loihi has 400ppm at 7% MgO while the most fractionated members of Puu Oo reach 400ppm at less than 6% MgO, (Garcia et al, 2000, J.Pet. 41,(7), 967-990). Were higher fractionates present we can assume Puu Oo would peak at 500 ppm approx. The alkaline summit hawaiites of Mauna Kea have 1300 ppm ( West et al, 1988, C.M.P. 100) while the more tholeiitic Hamakua lavas have only 500, (Frey & Garcia, et al, 1991, JGR 96). Mauna Kea at 5% MgO reaches 360ppm. Many of the older centres suffer from tropical weathering and erratic Sr values for some centres, eg, Kohala, (Lanphere & Frey, 1987, C.M.P. 95) may be due to this.

Sr in Alkali Basalt Series

Perhaps better than other elements , the behaviour of Sr illustrates the difference between tholeiitic and alkaline series. Because there is no feldspar initially on the solidus, Sr increases steeply with fractionation. Not only that, the trends originate well on the Mg side of the tholeiitic plane. We have no alkaline series based on glasses only so we cannot define where the alkaline primary melts lie, but it must be around 12 - 14% MgO. At about 3% MgO the Sr trend curves over and begins to descend. The maximum reached is 1200 ppm in Heard Id, 1060 in Mt Ross, 800ppm in St Helena and Gough, at 500ppm for Ascension, 750 ppm for Westmanneyjar. The 1300 ppm already mentioned for the summit hawaiites of Mauna Kea is therefore quite high.
The oddest is Tristan da Cunha which reaches 1720 ppm but lacking trachytes, does not then decrease. The Cape Verde Islands reach 2000 ppm with an average of three unusual samples being 2700 ppm and occasional samples going as high as 4000 ppm (Gunn & Watkins, 1975,BGSA 87) while the Canary Islands are very similar (Hoernle & Schminke, 1993, J.Pet. 34). This seem to suggest Sr is higher in sodic rocks.

Sr in Orogenic Andesites

For perhaps the first time, we can say there are some similarities in element behaviour in both simatic and sialic rocks. As in alkaline series, Sr reaches a maximum at about 3% MgO and rises in abundance steadily from IAB's to the most potassic centres.
The South Sandwich and Tonga Kermadec Arc rocks have a maximum Sr of 200 ppm, 350 ppm in Deception Id., 400 ppm in Puyehue in south Chile. Mt Egmont, (NZ) has 700, Azufre 1, 600; with Lascar and Nevado del Ruiz also at 600. The Valley of Mexico is rather variable with 400 to 600 but occasional more alkaline centres reaching 1200 ppm (Gunn & Mooser,1970)
From rather few data the potassic centres on the eastern side of the Andes have 600 ppm in Cerro Tuzgle but 1200 ppm in Cerro Pocho.
The Roman province peaks at 1200 ppm with both the Campi Flegri ignimbrites and the Yellow Tuffs decreasing rapidly to almost zero. (see "Orogenic Andesites" for references.)
The "peak" for IABs is very flat and rather indeterminate but there seems to be a distinct shift in mature andesites, it occurring at 6% MgO in Puyehue, at 4.5% in Lascar, at 3% in Parinacota and at 2% in Vesuvius and associated centres.

Sr in Continental Flood Basalts

Our knowledge here is very imprecise as CFB's, while often poured out in enormous volume, are very little fractionated and different parental magmas have quite variable MgO , usually between 7 and 4% MgO. Ferrar Dolerites have a low undifferentiated 110-120 ppm Sr and not rising above 140 ppm in the type differentiated sills. Coates Land has a variable 180 ppm and the Kirwan Basalts 160-180. Tasmanian Dolerites undifferentiated have a range of 120-200; the tholeiitic Karroo approx 200; Parana, 200, while the Columbia River Basalts as usual are erratic. Prineville has an average of 275 pm, Grande Ronde 325, Picture Gorge 250, Rosa 320 and Umatilla 375. The Palisades Sill has less than 200 ppm. In general the Sr levels in CFB's seem to be much lower than for orogenic basalts.
See "Continental Flood Basalts" for more detail and references.

Sulfur

Occurrence

Sulfur in the form of SO2, H2S and minor amounts of S2, S2O, H2S2, SO, OCS, AsS, PbS, SbS, and BiS is emitted from fumaroles and from active lava flows associated with active and recently active volcanic centres world wide. The pungent smell of H2S and SO2 is familiar to all who have visited such centres. Reaction between these gases leads to the formation of native sulfur around fumaroles. (SO2 + 2H2S = 2H2O + 2S)
Much greater amounts of S however are created in the coking of coal, during oil refining, in burning natural gas and in Pb, Cu, Zn refining. Some diesel fuels, especially from Mexico are high in sulfur. The sulfur content of coal is highly variable. A relation, Peter Gunn who is a world-wide coal dealer, claims the S content can vary in coal from 0.2% to as high as 7%, being precipitated by bacterial action. About 92% of S recovered is used in sulfuric acid production, mainly for batteries. (Lead accumulators). It is also used as an agricultural fertiliser, usually as ammonium sulfate. Canola for example requires 20 - 30 lb of S per acre.
The estimated total global S emissions world-wide in 1990 were 71.5 mil. tons of which 22 mt came from China which uses soft coal extensively for household, power generation and industrial use. S becomes hydrated into sulfurous or sulfuric acid in the atmosphere and is the main cause of "Acid rain" which disastrously etches ancient classic limestone buildings and statues such as the Parthenon in Athens. The extremely high smokestacks built by INCO at Sudbury, Canada, to lift S, As and other emissions higher into the atmosphere, has caused severe acid rain problems in Norway and Sweden.
The bulk price of S is currently $30/ton.

Volcanic Emissions

Using satellite scanners and UV interferometers the GEIA (Global emissions Inventory Activity) group have made estimates of the amounts of S emitted from an average of 56 volcanoes erupting per year between 1978 and 1988, but covering the period 1965 to present. Short lived eruptions can contribute large volumes so that atmospheric S can range from 10 - 30 Tg/a. A Teragram is 10^12 gm, so if 1,000,000 gm = 1 ton, we are talking about 10 - 30 million tons, if I have not slipped a zero some place. The average is estimated at 13.4 million tons per annum. Remember man-made emission are treble this. Try calculating the height of an equilateral cone of bright yellow S formed from the total world output for a year!

S in Silicate Rocks

I once had a brilliant idea. We would measure the S content of andesitic rocks and this would indicate the centres most likely to form sulfides. The results were extremely variable and correlated more with age. Young bits of flow still had native sulfur in vesicles, old rocks had lost all except that remaining in blebs of pyrite, chalcopyrite, marcasite etc in the rock. A normal figure for S in silicate rocks is about 0.05%.
One might expect continental andesitic rocks to have a higher sulfide content, certainly their gases include more SO2. Sulfide ore bodies are found only in continental rocks to my knowledge. I have found sheets of native copper in metamorphosed MORBs but not more than trace sulfides.

Thorium-Uranium

These two radioactive elements are present in all crustal rocks, with a range for Th of 1.5 to 100ppm and a Th/U ratio varying from 2.5 in the most basic MORBs to about 5 in highly fractionated potassic phonolites.
Data is scarce, for some of the most important volcanic islands there is none at all. The standard of investigation ranges from 55 determinations of U-Th from Macquarie Id, (Kamanetsky et al, 2000) with a correlation coefficient of 0.99, to no data at all or data with a CC of zero. Selection has been made on correlation coefficients of > 0.98, but as low as 0.75 has been accepted when no other data is available but marked "?".

Consumption

We currently consume 72,000 tons of U3O8 yellow cake uranium oxide pa, mainly in power stations, the major nations producing 30% to 50% of their electricity from atomic power. Australia has 27% of the world's known reserves and exports 10,000 tons annually. Current price is about $10US/lb. U3O8 is quite soluble and precipitates in porous sandstones in desert country. It is not conceivable that silicate rock with 1-10 ppm should be used as a primary ore at this time.
Th is mainly recovered from monazite, also a beach-sand by-product of ilmenite mining. Though thorite (ThO2) has the highest melting point of any substance (3,300ºC) and Th has been used in atomic power production the current demand is low, 13 tons/pa for USA.

U/Th in Simatic Crust

Considering the importance of U-Th in maintaining heatflow, and, it is claimed, volcanism, in the Earth, the amount of accurate U, Th data available is minimal. A great deal of basic work was done by gamma emission spectrometry in the "Nuclear Arms Race" era but the accuracy of determination is no longer acceptable. Amounts present in MORB rocks is so low that only the advent of ICPMS techniques have allowed good determinations to be made.

The data is not too precise in both Site 504 and for the Galapagos Rift at these low levels but for the EPR the regression line has a correlation coefficient of 1.0 while 25 random samples from the MAR also shows a coefficient of 1.0 as has the data for the NMORBs of the N. Chilean Ridge so we may take the best approximation for Th/U in an N-MORB series to be 2.5 ranging up to 5.0 in commendites. It is a great pity U data is lacking for Pantellaria, also for Heard, Tristan da Cunha and Gough. There is very detailed data for Macquarie Id which is a series of EMORB basalts of a range of composition but little fractionation, and in 55 samples the Th/U stays close to 3.65 throughout.

Th/U x = EPR (Regelous et al, 1999) + = Kilauea and Mauna Loa (Jochum, 1995,AGU) * = Macquarie Id. Partial melt Series (Kamenetsky et al, 2000)

Note that the fractionated Hawaiian and EPR series only diverge very slightly from the partial melt plane.ie, a very low degree partial melt will not differ much from a higher degree of melt. This explains the small range in Th/U ratio found.
The Macquarie Parental melts are constant at a Th/U of 3.5, while equally good dat for the Romanche Fracture Zone, which appear to be also close to aprental melts but less potassic, have Th/U 0f 3. KMost MORB however seem to close to 4. Pb/Th ratios also differ.

Th/U in Arc Andesites

While in general similar to OIB's and MORB's, the ratio Th/U tends to be much more variable in sialic (andesitic) rocks. In 2000 Andean samples for example of which perhaps 3-400 have been determined for Th-U, there is a concentration at a rather high level of about 4. This is reflected in a rather high Pb208/204 level as well. However, the range is large with samples ranging from 1.5 to 10. This might indicate some contamination of primary andesite with oceanic sediments, as the Sr/Rb isotopes are equally variable, as are many other elements.

In the more primitive oceanic arcs, Th/U is much lower being commonly in the range 1 - 1.5 with a marked negative Th anomaly being seen in the fingerprint normalised diagrams. Thus proto-arc centres such as Kao, Tafahi, Late, Curtis, Macauley plus the Leg 135 Tongan Back Arc samples all show a negative Th anomaly, but not the more enriched islands of Raoul and Esperance. This also seen in proto-arc IBM, Scotia Arc, Merelava in the New Hebrides, and in Statia - St Kitts in the Lesser Antilles. Pb/ Th may be quite high, in the region of 10.

In mature arcs we find a dominant level of 4 in the Andes, 3.3 in Mexico, 2.7 in the Cascades, 2.4 in the Aleutians, 2.3 in Kamchatka, about 2 in Honshu, 1 - 3 in Tonga, 3.3 in the Scotia arc but 4.2 in Indonesia.(Sunda Arc). All however show considerable variability.

Kluyuchevskoy Volcano in Kamchatka, while a massive cone, is of general Early Arc composition and has a very low level of U-Th with a Th/U little greater than 1 (Kersting & Arculus, 1995, EPSL 136; Dorendorf et al, 2000, EPSL 175)

Tellurium

Te (At No. 52) is associated with sulfur but may form separate telluride minerals especially in sylvanite and calaverite. It is recovered in the anode mud during the electrolysis of copper sulfides, 500 tons of ore yielding about 1 lb of Te. Te has a total of 30 isotopes ranging from atomic weight 108 - 137. The ore in the famous Cripple Creek mine in Colorado consisted of gold telluride in fluorite.
Te is used in Fe, Cu, and stainless alloys, in semiconductors and ceramics.

Thallium

Tl (At. No. 81) About 15 tons of Tl are produced yearly, from flue dust from smelting Zn, Cu and Pb, with a value of $600/lb. It is used in high temp superconductors, in alloys and in glass. It is poisonous and an estimated 1000 tons Tl released into the air from metal refining and cement manufacture in VietNam is reputedly causing severe health problems. (Note: This sounds unreasonably high, we will try to check the source.)

Titanium

Ti Uses and Consumption

Titanium is mainly produced from ilmenite (FeTiO3) in beach sands, the concentrate costing only $87/ton in Australia. The oxide rutile (TiO2) is rather more expensive at $563 per ton. World consumption is now 4,100,000 tons of ilmenite concentrate per year and 370,000 tons of rutile, about a quarter of which total is mined in Australia. Ti metal is as strong as steel and half the weight and is increasingly used in aircraft and the space industry. The latest laptop also has a titanium cabinet. Being non-magnetic titanium is used in building atomic submarines, but 90% of all production goes into the paint industry as a pigment, unlike lead it is non-toxic. It is the only metal known to burn in nitrogen!
World reserves are some 1 billion tons.

Titania Fractionation Pattern

Ti climbs very steeply relative to MgO during early fractionation, in some cases more steeply than the primary melt series. It reaches a maximum at about 4.25% MgO when titanomagnetite forms. This seems to be temperature controlled rather than compositionally as the Ti may peak at 2 - 3 - 4 - 5% but almost never higher, while iron may reach 10 - 16% at the same point. Closely related rocks such as in the EPR may show several sub-parallel paths peaking at different levels, it may be PO2 plays a part in regulating the amount of FeO present.

Ti in NMORB - EMORB

The most depleted, highest degree melts may have as low as 0.5% TiO2, eg in Kolbeinsey Ridge. The range in the Macquarie primary melts is only 1 - 2%. The glass data of Melson and O'hearn show that fractionated groups along the MAR may have 1-2% at 22N, and 0.6 - 3.5% in the South Atlantic. The EPR data of Regelous et al, 1999, show a peak of 3.75% as does Volcan Alcedo in the Galapagos.

Ti in OIBs

Iceland is variable with Krafla peaking at only 2.5% while the Austerhorn and East Iceland have up to 4%. The Hekla - Katla group show variable data according to the author, but in the range 4 - 4.7% with the alkaline Westmanneyjar at slightly over 3%
The Atlantic islands reach a high in Tristan da Cunha with 4.25% and Gough and Ascension peaking at 3.5%. The Cape Verde island ankaramites-limburgites-phonolites are variable but may peak at >5% while Heard Island, also variable lies in the range 3-6%.
There seems to be little difference between tholeiitic and alkaline islands, but mafic melilotite mela-nephelinites such as the Bermudites or the East Greenland Nunataks may have in excess of 6% TiO2.

Ti in Orogenic Andesites

There seems to be no compositional pattern here. About 1% is the common maximum found in centres as different as the South Sandwich IABs, in Sangay, Calbuco and Mt Vesuvius. Planchon Petaroa and Azufre 1 are a little higher at 1.2% while Deception Island and Puyehue have 1.75%. There seems to be considerable shift in the MgO point at which the TiMt IN point occurs, more work needs to be done on this as it is not fully understood.
Commonly the andesite series have about 1/3 the TiO2 of the Oceanic rocks, but there can be overlap. We shall add histograms as time permits.

Ti in Continental Flood Basalts

These are remarkably variable, ranging form 0.6% in Ferrar Dolerites (and overlapping with Orogenic Andesites) to 3% in undifferentiated massive flow series such as the Grande Ronde.

Vanadium

V behaves like Ti, building up in the early stages of fractionation in silicate rocks to the 250 - 500 ppm levels and then precipitating in titanomagnetite in which it may have a content of about 2-4%. There is however, proportionately less variation than in the Ti in high and low Ti series, as seen in Flood basalts for example.

Uses of V

V is used in alloys, especially of rust resistant steels. Time was when no spanner would have sold if it did not have "Chrome Vanadium" stamped on it. The alloy, while less resistant than say Type 504 Ni-Cr, is somewhat harder. V is also used in ceramics and V-Ga alloy is used in superconducting magnets.

World production

This was some 35,000 tons in 1998, half coming from South Africa, the price being only $4.00/lb. Reserves are estimated at 63 mt. V is found, not only in TiMt but in phosphates, uraniferous sandstones, bauxite, coal, crude oil and tar sands.

V in Ridge basalts and OIBs

The Macquarie primary melt curve shows a slight negative slope at an average of 200 ppm from which we can guess that there will not be a great difference between tholeiitic and alkaline rocks. However with fractionation, V in tholeiitic rocks rises steeply, to a maximum of 400 ppm for the EPR, to then declines sharply at the TiMt point.

Alkaline rocks, below the TiMt point have a flat trend extending towards ankaramites, so we might guess that Cpx includes several hundred ppm V. Marc Norman (1997) has shown in a laser ablation study of clinopyroxenes from 1955 Kilauea, that the V range is from 200 to 400 ppm. However, the Ol+Cpx+Pl of tholeiites also includes about 35% cpx, so why is the trend for tholeiites so steep compared to the flat ankaramite trend? An olivine trend only ought to be steep. Interestingly, Puu Oo shows a flattish trend at 280-290 ppm in the magnesian endmembers, so just what is the V content of olivines? Sure enough, Rhodes et al, (1995) show a V trend in the picrites of Mauna Loa to decline steeply from 300 ppm in undifferentiated basalt to about zero in pure olivine. Puu Oo has about 40 ppm higher V and appears to have a flatter trend towards higher Mg but the range in MgO is not great enough to be positive. Rocks like the Ferrar Dolerites seem to suggest there is a fair amount of V in orthopyroxene-pigeonite also, but so far no data has been found.
In general the vanadium will peak at a higher point in those rocks with high Ti, so that ferrobasalts usually have the highest V. High Ti ferrobasalts in the Flood Basalts, eg Paranema and Esmeralda in the Parana Basalts peak at 500 ppm, as do the high Ti Coates Land dolerites of Antarctica. The Karoo, being mainly low Ti, show a marked concentration at 250 ppm but cover a range of 150-500, (Erlank et al, Karoo File). The Palisades Sill, a low Ti type, peaks at 250 ppm and in fact replicates exactly the Mauna Loa picrites.
Calc-alkaline series of low iron contnent continental to high K regionsshow a lower V content. Puyehue peaks at 200-300 ppm as do the South Sandwich Islands. Sangay has a maximum of 200 ppm, Parincota 230, Planchon Petaroa 230 and Vesuvius 280 ppm. However the "tholeiitic" Island arc basalts which may have ferro-basalts with 18 - 26% FeOT (eg, Izu-Bonin-Marianas Arc) show a close correlation with iron to a maximum or 600 ppm V.

Water

Of all comodities, water is likely to be the one soonest and most urgently in short supply. A consultantcy group, Lenntech, have written us a URL explaining the situation, and detailing alternatives.

Information about water quantities from a water purification company.

Yttrium

Yb is usually taken to be geochemically associated with the lanthanides, it being element 39 and being in the same column in the periodic table as Lu, (el.No. 71). As with the REE the main source of Y is in monazite beach sands and carbonatites. The current cost of Y is $75/oz.

The distribution of Y against MgO for all alkaline oceanic island basalts. Alkali basalts at 5-6% MgO, ankaramites to right, trachytes, commendites, phonolites and pantellerites to left.

Distribution of Y in MORBs and OIBs

The distribution of Y is to say the least, unexpected. All basalts of whatever composition appear to have close to 25ppm of Y at 7.5% MgO but the fractionation paths differ widely being steep and linear in tholeiites but flat in alkali basalts but sweeping up very steeply at sub 0.5% MgO compositions.

Historical Note
In 1970 we were working on the Deep Drill project with cores recovered by the 'Glomar Challenger' on Leg 37. At Site 332b we had a series of undifferentiated MORBs, two series of anorthosites and two sets of picrites. We had just begun analysing for Y by XRF and a MgO/Y diagram clearly showed that the Y content of calcic feldspar and olivine was vanishingly low, and high for clinopyroxene. The resulting distribution of points looked like a stick drawing of a running bird, and was easily the most interesting diagram produced. We showed at a preliminary AGU meeting and as a result, all the other investigators in the project also analysed Y for the first time. However, until the EPR data was produced by Regelous et al, not a great deal more was learnt, picrites and anorthosites being rare in the oceanic environment.

|__| Mg/Y for Site 332b.
Being an HFSE element, the behaviour of Y is the not the same as the LILE. However it does increase with fractionation even if very slowly at first in the alkaline series.
The fractionation path for the EPR can be computer simulated assuming the usual tholeiitic crystal cumulate of about 50% plagioclase, 35% Cpx and 15% Ol. It is also necessary to assume partition coefficients for plag and Ol of close to zero and of about 0.3-0.5, (Norman, M., 2000) for Cpx . The bulk crystal cumulate has an MgO content of 8.5% MgO and has perhaps 10-12 ppm Y. A basalt of the usual MORB 7.5% MgO will fractionate very steeply though not as fast as the LILE, eg in the EPR we see Y increase by a factor of 5 compared to 7, 7, & 8 for Cs, Nb and Ba. The ratio Yb/Y appears to be quite constant at 10 in ORB rocks.
Alkaline rocks do not accumulate plagioclase but instead varying amounts of Ol+Cpx. The ratio was 50-50 in the Crozet ankaramites and 60-40 on Possession Id. in the same group. (Gunn et al, 1971, J.Pet.32). Gough Id. (le Roex, 198 ) appears to have a higher Ol to Cpx ratio. At sub 1.0% MgO levels, the alkaline rocks increase in Y very steeply, obviously Cpx has ceased to form and plag + sanidine have very low partition coeficients for Y. Does the sudden increase in Y (and Nb) coincide with the collapse of Ba in trachy-phonolites? We do not know yet.
The very final products may have a relatively high Y content as it reaches 200ppm in Pantellaria and 300 ppm in the African Naivasha commendites. Tholeiites OUGHT to have the highest Y content, but their low alkali content means that only very rarely will a higher fractionate be produced. Suppose we look at the rhyolites of Iceland? When we do so, the trend is less steep than the EPR and the high-Nb "rhyolites" do sweep up at the end point, but only to 160ppm levels, (Wood, 1977, J.Pet.19).

Y in Sialic Rocks

Early arc rocks such as the Scotia or Tonga-Kermadec arc have much flatter trends than do MORBs. Y rises relative to SiO2 rather steeply from 7-8 ppm in IABs to 30 in basaltic andesites and then flattens with rhyodacites having only 35. Deception is similar, this being about 1/4 the amount seen in the EPR. This rather puts paid to the theory that arc andesites = MORB plus a small amount of sediment, unless a very different kind of fractionation pertains.
Puyehue is rather different with a maxumum of 60 ppm, while Sangay is very low with a slightly negative slope against silica with only 15ppm. Higher fractionates of the EPR have about 7 times as much Y as Sangay.

The very potassic Vesuvius has a flat distribution with 25-30ppm Y but the more fractionated Campi Flegri ignimbrites sweep up steeply, to a maximum of 50ppm but the scatter is bad.
Y has much the irregularity in andesites as seen in Zr.

Zinc

Zn has a number of very different interests, without it as an anti-corrosive coating on steel cladding whole sections of the construction industry would fail and without sacrificial zincs on ships hulls a steel ship would not last two years and life would be extremely difficult. Zn deficiency can kill off whole forests and it is fundamental to brass metallurgy.

Zn Production

Zn is produced usually along with Pb and Ag being mined as sphalerite (ZnS) and galena,(PbS). The galena usually (eg at Broken Hill) contains a few oz of silver (about 10) to the ton. Zn is the fourth most important metal, total world production in year ending Oct 2001 being 8,177,000 tons. This is very hard to break down to a by country basis as the big metal companies are international and have a finger in every pie, eg Con-Zinc Rio Tinto are supposed to own Mt Isa but many other companies have a share in Con-Zinc. Con Zinc in turn owns a chunk of Port Hedland and many other mining ventures. Canadian miners Cominco have a major share in the very large "Red Dog" Zn mine in Alaska.
Zn and Pb are chalcophile elements, ie they preferentially bond with S++ but the origin of massive sulfide deposits is not easy to explain. Old andesitic volcanoes ooze sulphur from fumaroles for centuries and it is not exactly surprising to find massive pyrite-chalcopyrite in old rock, but where do the enormous amounts of Pb-Zn come from? In Broken Hill the ore body goes (or went, as it is largely mined out) down over a mile and must have been a mile or more in diameter. To drop down in the cage a few thousand feet, walk along a drive to the face, shine your headlamp over the roof and see it sparkle as though from a million diamonds (galena crystals may have cleavage faces an inch across) is an experience long remembered. Mining geologists talk of "Mineralising solutions" which is not helpful. Presumably streaming sulfur plasma gas driven off from down in mantle, at one time reached high up into the crust scavenging the Pb-Zn as it came.
Some silicate minerals can take in a small amount of Zn in ionic form into their lattice, eg the Kilauean magmas have about 100ppm and olivines take in about 86ppm, ie the Zn(ol)/Zn(liq) is about 0.9. Was this silcate Zn (and Pb) the origin of the massive suphides? Or do the sulfides occur as immiscible droplets even in the mantle? I think at this point we can only guess.

Zn in MORBs and OIBs

Again many key islands and centres have no Zn data. The minimum in unfractionated NMORBs appears to be 60 ppm rising with fractionation to 200. EMORBs may have slightly more, perhaps 80 but such data as is are not good.
The Hawaiian picrites of Mauna Loa and Mauna Kea begin at about 120 ppm and decline slightly with increasing olivine as I found for Kilauea. The Mauna Kea summit Hawaiites elevate slightly to about 150 ppm.
In Iceland the Hekla-Katla volcanics and those of East Iceland (Wood, D.A. 1977) Zn rise to a little less than 180 ppm at low MgO levels and then in rhyolites the Zn falls off to about 80ppm.
In the alkaline islands the trends are quite different. Zn remains constant at 80-100 ppm levels but rising to 2-300 in trachytes-phonolites with a maximum in the African Naivasha commendites of 400 ppm but a little less in Kerguelen phonolites.

Zn in Andesites

This can only be termed "low and erratic". In Deception Id and Puyehue Zn increases somewhat with fractionation from 65 -110 and 90 -120. In Planchon Petaroa, Zn is fairly constant at 80-90 ppm but in Parinacota it decreases from 150ppm at 6% MgO to under 45 ppm in dacites. A similar decline is seen in the Valley Of Mexico where basaltic andesites of 75-90 ppm decline to ~40 ppm in rhyolite.

Zirconium

As fingerprint diagrams in these pages show, the HFSE elements, including Zr, Ti, Eu, Gd, Y, Yb, Lu while showing less variation than extreme LILE elements such as Cs, Rb, Ba, Th, U, Nb, K, can nevertheless still vary a good deal. This is usually between about x0.8 and x10 EMORB in common rocks, occasionally up to x 30. How much of this is due to differences in parental magma type and how much to fractionation we shall see.

Consumption & Sources

If asked to name the rock with the highest content of the mineral zircon (ZrSiO4) most people of geological background would probably name granite or rhyolite, probably because those little euhedral zircons are so prominent under the microscope in granites. As we will see, this is not the case, as commendites-trachytes-phonolites contain much more. World production of zircon as a source of ZrO2 is now almost a million tons, almost all recovered from beach-sands as a by product in the recovery of ilmenite-rutile for titanium. Zircon is a tough little mineral and resists abrasion much better than feldspar does. 57% of zircon concentrate is produced in Australia, 45% in W.A. The main usage is in ceramic glazes and tiles, and as a refractory. It is also used as a radiation screen in TV sets.
Zr (atomic no. 40) occurs in association with an isomorph, hafnium, (atomic no. 72) which is present always present at about 2% (see below). The 1999 prices for Australian zircon was $A514/tonne.

Zr in the Simatic (Oceanic) crust

Zr is one of the more interesting elements to the geochemist. It is present in amounts ranging from <40 to 2500 ppm in common igneous rocks and varies widely with fractionation by factors of up to 6-7 from basic to felsic rocks; in the primary melt series it may be as low as 40 ppm in depleted NMORB of high degree melt, but up to 160ppm in the smallest degree of melt rocks of Macquarie (Kamenetsky et al, 2000), a relatively small range which is important in understanding it's distribution. Zr increases somewhat as we progress from tholeiitic to alkali basalt to basanite but as we do not have glass data for the latter it is impossible to identify exactly. About 150 ppm for larger degree melt alkali basalts and up to about 250 ppm in basanites is a best guess. As it is unlikely a "high degree melt" can exist for basanites the comparisons are not simple. The variation in the range tholeiite - basanite is much less than we see in Nb however but it is there.
When we look at two diagrams, one of MgO/Zr and the other of Zr/Nb they are remarkably different. Why? Understanding the reasons is essential to any understanding of the chemistry of basalts.
Macquarie Parental melts have from a minimum of 40 ppm in NMORB to 162 in extreme EMORB while Ta has a minimum of 1.5. Magnesian DMORBs might have slightly less than 40 ppm

	The distribution of Zr against MgO for all alkaline oceanic island basalts. Alkali basalts at 5-6% MgO, ankaramites to right, trachytes, commendites, phonolites and pantellerites to left.
	MgO/Zr diagram for the EPR(x), Alcedo(*), (Galapagos), Piton Fournaise(+), St. Helena(o), Gough(sq) and from Pantellaria(tr). Note that Gough has the highest Zr among the basaltic rocks, but whether this is true for all potassic rocks relative to sodic is not yet known. The EPR and other N-MORBs are the lowest.
	Zr vs Nb. MORB and OIB.

In the MgO/Zr diagram we see a wide range of rocks from tholeiites to basanites following almost similar paths with a logarithmic increase as MgO decreases. Why?
The reason is that MgO does not differ in amount much between tholeiite - alkali basalt - basanite. It does however diminish sharply as olivine and clinopyroxene form. In other words, MgO is an index of fractionation effects, we also could also use the Fe/Mg ratio, or the Na/Ca ratio but not the K, Ba, Rb, Cs, Th etc amounts which go up with fractionation but also increase between tholeiite - basanite and also from depleted NMORB to enriched EMORB. In other words they are not a pure fractionation effect index but the change in MgO amount is.
Zr changes little between tholeiite and basanite but shows quite pronounced fractionation effects. So both MgO and Zr are showing mainly fractionation effects. So why do they not form a straight line of negative slope? Because some Zr is going into the clinopyroxene and not until olivines and pyroxenes have stopped forming is a large increase in Zr seen as partition coefficients must be low for albitic and potassic feldspar and for nepheline. Norman (2000) gives a partition coefficient Zr(Cpx/Zr(melt)) of only 0.1 but it appears be higher in ankaramites, more like 0.5.
Why do tholeiites-basanites show the same trend when they are so different for say Zr/Nb? Because the partial melt trend as shown by the Macquarie rocks lies exactly parallel to the fractionation trend in the range 10% to 4% MgO and the partial melts only have a range of 40 -160 ppm anyway. Notice those points for the EPR (X's) sticking out at the bottom and the basanites of Gough Id at the top, (squares)). So there is not much difference in the Zr content of rocks of different degree of melt or of different alkalinity nor is there much change with fractionation until we get down to rocks with under 4% MgO. Get it? If not, read again.
In the Zr/Nb diagram we are plotting against Nb which increases very steeply as we go from tholeiite to basanite and which also increases steeply from NMORB to EMORB and also with fractionation. So while Zr is an index of fractionation, Nb both fractionates (note the Nb for basanites-trachytes of St Helena) and is a good indicator of alkalinity and of increase in LILE generally and Nb also increases steeply with decreasing degree of melt, almost vertically in the diagram. Imagine a cluster of cumulate minerals located near the origin of the diagram and you will see the trends all radiating from it.
The result is we can plot Zr/MgO for the EPR, the SE Indian Ocean Rise, from Alcedo and Sierra Negro volcanoes on the Galapagos and from the Austurhorn basalt - high Nb-Zr rhyolite of Iceland (Furman, et al 1992, J.Pet) and from the Piton Fournaise on Reunion (Albarede, 1992) and find they lie on exactly the same trend though the end-point in fractionation may differ widely. Pu'u O'o series almost coincides with all other seafloor and island tholeiites for Zr but is a little more enriched but much less so than for Gough Id.
The Austurhorn shows an unusual feature in that Zr rises to about 1000 ppm in rhyodacites and then collapses in rhyolites to only 1-200 ppm. Is this due to formation of Zircon? How can such tiny crystals be segregated from a viscous rhyolitic melt? There are one or two erratic points with up to 1150 ppm. Are these cumulates?! If so why does Zr not separate out at 2500 ppm levels in trachytes? More problems to solve!

Zr in OIB's other than tholeiites

Trachytes and phonolites of the oceanic islands, being richer in alkalis show more extended fractionation effects than do tholeiites of the ocean basins and ridges and big increases in Zr may be seen when the MgO content is below 0.5%. Trachytes from Kerguelen reach >1200ppm (Weiss & Frey et al, 1993, EPSL 18); while Heard Island also in the southern Indian Ocean (Barling, et al, 1994, J.Pet.35) has 600 ppm at 4% MgO and reaches 1600 in trachytes with the minimum in ankaramites being ~150ppm.
Ascension Id in the South Atlantic (Kar, 1998) attains up to 1800 ppm, and Tristan and Gough >1800. The sodic St Helena (Chaffey, et al, 1989, Geol.Soc.Lond) reaches only 1400 ppm but the difference may be due to accident of collection. The phonolitic Mt Ridley in West Antarctica attains 2400 ppm, the Naivasha commendites of East Africa 2000, the Mopanui Phonolites of Dunedin, N.Z. 1600, and the Kapsiki phonolites of the Cameroon, west Africa, 2000, (Ngounouno et al, 2000, JVGR 102)
The island of Pantellaria, south of Sicily in the Mediterranean has a high percentage of low alumina "Pantellerites" with between 2000 and 2500 ppm.
While there are small differences between tholeiitic and alkaline rocks, and large differences in the degree of fractionation attained, the main Zr trends are the same for all simatic rocks. For basalts at about 7% MgO the potassic basanites of Gough Id seem to have the highest Zr and NMORBS the lowest, but whether all potassic rocks are the same is not known.

Zr in Sialic Crust

The very depleted proto-arc basalts of Candlemas in the South Sandwich Id and those of the Tonga-Kermadec Arc and those of the IBM Arc (see Orogenic Andesite pages for references) have a common, quite flat trend rising at about 30 ppm at 6-7% MgO with a maximum in rhyodacites of only 100 ppm. This contrasts with Deception Id also in the South Sandwich group where the range is 60 - 550 ppm. Unfortunately there is a problem as of four sets of determinations for Zr from Deception Island, the interlaboratory differences span some 30%. Deception appears to be the most Zr enriched of all andesite series and no equivilent has been found elsewhere, as yet. However, all andesites series appear to have less than half the Zr found in EPR series.
In the Colombian, Peruvian and N.Chilean Andes, including Sangay and the nearby Nevado del Ruiz, Galeras and Parincota, all have fairly high K, Ba and Rb, but the Zr has a flat distribution relative to silica or MgO, the Zr(60) being about 120 for Sangay and 200 for Parincota as well as Mt Egmont and the rhyolites-ignimbrites of the Taupo Volcanic Zone in New Zealand where we see a range at <0.5% MgO of only 100-200 ppm Zr. Thaty is, the more potassic andesites with K(60) of 2-4 overlap with the IABs.
By contrast, in south central Chile the less potassic centres there show a much steeper SiO2/Zr or MgO/Zr trends. PuPuyehue for example has rhyolites with up to 400ppm, (Gerlach & Frey, 1988, J.Pet.29). There is no sweep-up as seen in trachytes, the trend is quite linear but stems from the basic end at about 6-7% MgO, not at 10-11 as seen in simatic rocks, the basalts only having 50ppm. The data dates from 1988 but the data of Professor Frey and associates is depressingly reliable so we must accept the fact that Puyehue is in some fashion different to the general run of calc-alkaline volcanoes. If we plot the neighboring volcanoes on the same diagram, though some have a limited range of composition, it seems that Azufre, Llaima, Villarica, Copahue, Cerro Antuco and Planchon Petaroa, all follow the Puyehue trend as they do for Y, Tb and other HSFE, though sometimes different on a Fe-Mg-Alk ternary diagram for example.
Pinon Hachado, a centre a little north appears to have two series one with HFSE less and one x10 more than EMORB, (Munoz & Stern, 1989). Some Aleutian Arc centres are also close to Puyehue.
Most andesite series have a limited range from basaltic andesite to andesite, rarely dacite, and any associated rhyolite generally occurs as ignimbrite. Obsidian domes analysed for Zr are not common. The general andesite trend elsewhere is quite flat, eg in the Trans-Mexican Volcanic zone, between 150 & 225 ppm, (Wallace, 1999), and at 200+ in Parinacota in N. Chile.
Not only the amounts present but the slopes when plotted against silica change. It is definite that the very depleted IAB, s also have the lowest Zr, while Early Arc and Active arc rocks have more, but there is no direct correlation with K content or any other parameter. This may be that the slope of elements as shown in the fingerprint diagrams hinges close to Zr. However we have no theory at present why HFSE including Zr should be so high in Deception Id.

|__|	Average orogenic Andesite distribution for Zr.
	Zr/MgO for Candlemas Id, (which is common to all proto-arc rocks, and the unusually steep trend for Puyehue in South-Central Chile. Most andesite series have a flat distribution at 100-300 ppm, often declining in rhyolites. + = Puyehue rhyolites; * - Candlemas and other arcs.

North Chilean ignimbrites actually show Zr sloping down to ~100ppm with increasing silica but ignimbrites are always subject to the suspicion of mechanical sorting. The Topopah Tuff of western USA also shows Zr declining to almost zero at very low MgO levels. We cannot tell is this is a fractionation effect or a mechanical effect.
Not much Zr data is available on granite pegmatites. A survey of 20 granites shows many to be often less than 1.5 times E-type MORB.
We might suspect that highly fractionated syenite pegmatites might contain even more Zr than Pantellerites and this seem to be the case, as Ivigtut granophyres (related to the alkaline Garda syenites) have up to 4000 ppm Zr, (Goodenough, Upton and Ellam, 2000, Lithos 51).

Continental Flood Basalts

These follow much the same pattern as the more tholeiitic of the simatic basalts. Basalts of 6-7% MgO in the Ferrar Dolerites, Antarctica, have 80 ppm but one series in the Coates Land dolerites, (Brewer, et al, 1992), the so-called high Ti series, have elevated HFSE including Zr with about 150 ppm. Tasmania dolerites are similar to the FD (Hergt, pers comm) and unfortunately in the Parana rocks where they are labelled high and low Ti series, the latest data includes no major elements.
The Karoo range from tholeiites to highly alkaline rocks so the Zr in the more basic rocks is about 100-125 up to 500 ppm in alkaline members, but the more alkaline ones sweep up with fractionation and declining MgO to 1250 ppm.
The Columbia River Flood Basalts vary widely in their HFSE, but show little fractionation with Prineville having 140 ppm at 4% MgO and Umatilla about 520 at 3% MgO. (Hooper, WSU Data File)
Deccan "High Ti" Traps, have 2.5% TiO2 at 7.5% MgO and about 170 ppm Zr. (Melluso et al, 1995, J.Pet.36). The rather alkaline Snake River flood basalts have in their closing stages, extremely high Zr at about 2400 ppm

Conclusions

The winner appears to be the island of Pantellaria. If zirconium is desperately needed try the Pantellaria beaches or some late-stage CFB's.. The Garda area has been mined out for cryolite but we may see a rejuvenation with Zr prospecting when the beaches run out but separation from syenitic pegmatites is not easy. However any highly fractionated trachyte or phonolite should be a likely source with on average five times as much as in rhyolites! It would be interesting indeed to find a rhyodacite derived from the Umatilla CRB's but perhaps it would be in the same range as the late-stage Snake River rocks with 2400, 2500 ppm.