zircon "the key mineral"

|TYPOLOGY|

|RESEARCH|

by Serhat KÖKSAL

Middle East Technical University /TURKEY

Central Laboratory/TIMS Lab

e-mail: skoksal@metu.edu.tr

During the last decade much attention has been focused on the isotopical studies on zircons, either in igneous, metamorphic or in sedimentary rocks. Of the several isotopic systems that have been explored as a means of determining the absolute ages of granitic rocks, the U-Pb system in zircon currently enjoys a reputation as one of the most broadly applicable and reliable. Zircon is ubiquitous in granites, and is more resistant to isotopic disturbance by secondary geological processes such as metamorphism, alteration and weathering than most other minerals. Also, the U-Pb system itself has the unique feature that, because it consists of two U isotopes that decay at different rates to two Pb isotopes, any isotopic disturbance that does occur can be detected by a simple test for internal consistency between the radiogenic Pb isotopic composition and the Pb/U ratios (Williams, 1992).

As important as these factors are, it is unlikely that zircon dating would have achieved its present popularity had it not been for the major advances that have been made in U-Pb microchemistry and mass spectrometry, particularly over the last 20 years (e.g. Krogh 1973; Lancelot et al. 1975; Manhes et al. 1978). These have resulted in its being possible to make increasingly precise analyses of ever smaller zircon samples (Williams, 1992).

Much slower to develop, however, has been an understanding of the properties of the zircon U-Pb system as such. Despite the enormous amount of zircon U-Pb data now available, some very basic questions about the nature and causes of isotopic disturbance (discordance) and the preservation of older isotopic systems in younger zircon populations (inheritance) have yet to be answered (Williams, 1992).

Because zircon is resistant to physical and mechanical degradation and ages derived from detrital zircon analyses reflect the characteristics age spectrum of primary source rocks, and so detrital grains of zircons in sedimentary rocks are also important samples for age dating (Bruguier et al., 1997).

Radioactivity was discovered following experiments on the luminescence of uranyl double sulphate crystals caused by exposure to ultraviolet light. The phenomenon was noted in the walls of cathode ray tubes and this led Henri Becquerel (1896) to determine whether uranium compounds emit X-rays. In fact, both the compounds and uranium do so. Later, Marie Curie showed that thorium also emits radiation and that uranium and thorium minerals are more active than pure salts of the elements. This suggested that natural uranium ores, for instance pitchblende, might contain more powerful radiation emitters than uranium. Further research revealed two new active elements, polonium and radium.

The radioactivity of uranium and thorium was used in regard to dating minerals containing these elements, and a number of approaches developed. These included a chemical Pb-U, Th method, a Pb-alpha method, a U-He method and two which are still employed, these being the U, Th-Pb isotopic method and the common lead method. Others have been added subsequently. The abandonment of the first three resulted from the realisation that some of the assumptions upon which they are based are invalid. Thus, the chemical Pb-U, Th approach depends upon all lead in an uranium mineral being radiogenic in origin, its amount increasing as a function of time.

Uranium possesses three naturally occurring isotopes, ²³⁸U, ²³⁵U and ²³⁴U, all being radioactive. Thorium has one common radioactive isotope, ²³²Th. The half lives of the parental elements ²³⁸U, ²³⁵U and ²³²Th are longer by orders of magnitude than those of their respective daughters. The decay series fulfil the requirements for the establishment of secular equilibrium and the decay rates of the intermediate daughters are equal to those of their respective parents. If the mineral is a closed system and such a secular equilibrium exists, the rate of production of the stable daughter at the end of a particular decay series is equal to the rate of decay of the parent element. As a result, it is feasible to regard the decay of the uranium and thorium isotopes in minerals having secular equilibrium as if it took place directly to the respective Pb isotopes.

Ordinary lead has four naturally occurring isotopes (²⁰⁴Pb, ²⁰⁶Pb, ²⁰⁷Pb and ²⁰⁸Pb), but the element actually has 32 isotopes in all. The other 28 are radioactive and include some which decay by proton emission. The isotope dilution technique is applied in the dating of U- and Th bearing minerals in order to determine the concentrations of uranium, thorium and lead. The isotopic composition of lead is determined by using a suitable mass spectrometer.

In practice, discordant dates for uranium- and thorium-bearing minerals are obtained in many cases, reflecting the fact that a large number, perhaps indeed the majority, of them are not closed systems. The reason is that loss or gain of uranium, thorium, lead or intermediate daughters occurred subsequently to crystallization. If lead was lost, it is possible to minimize the effect on U/Pb dates by calculation of a date based upon the ²⁰⁷Pb/²⁰⁶Pb ratio which is not sensitive to this loss. Minimal sensitivity is displayed where such a lead loss took place recently and when the lost lead was of the same isotope composition as the residual lead.

Uranium and thorium are found in many minerals, of which, however, only a small number are suitable for dating using the U, Th-Pb approach. This is because only a few are adequately retentive towards them and of these the most retentive is zircon. Other useful minerals include pitchblende (or uraninite, U₃0₈), monazite, sphene and apatite. In zircon, the concentrations of U and Th average 1330 and 560 ppm respectively. When in pegmatites, zircons contain more of these elements than they do in normal igneous rocks. Uranium and thorium are found in zircon through the isomorphous replacement of Zr⁴⁺ (with an ionic radius of 0.87 Å) by U⁴⁺ (1.O5 Å) and Th⁴⁺ (1.10 Å) as well as through the presence of thorite inclusions. Such substitution is restricted by the differences in the relevant ionic radii Pb²⁺ is excluded altogether because its ionic radius is 1.32 Å and it bears a lower charge. The result is that zircon does not contain much lead at the time of its formation and has very high ratios for U/Pb and Th/Pb, making it a valuable geochronometer (Bowen, 1988). For this reason, this mineral is widely utilised in dating by the U, Th-Pb isotope method, for instance by C. J. Allégre et al. (1974).

Sometimes model dates for these minerals are discordant as consequence of non-fulfilment of some of the dating assumptions. Loss of radiogenic Pb in zircons may be due to continuous diffusion or to the influence of metamorphic episodes. It is very difficult, however, to link a discordant date with a particular geological event. The loss of radiogenic Pb is less serious and the ‘207- 206' date derived may well approach the actual age of the zircon fairly closely. Although such discordancy in dating is common, there are numerous instances of concordant dates (Bowen, 1988).

Decay of the naturally occurring isotopes of uranium to form stable, radiogenic lead provides two independent geochronometers. These are concordant in cases where dating assumptions are valid. The decay of ²³⁸U to ²⁰⁶Pb as a function of time may be expressed as follows:

²⁰⁶Pb*/²³⁸U = (²⁰⁶Pb/²⁰⁴Pb – (²⁰⁶Pb/²⁰⁴Pb)_i) / (²³⁸U/ ²⁰⁴Pb)

and the other decay series may be treated similarly. Such equations give concordant dates if a uranium-bearing mineral is analysed which satisfies the relevant assumptions. Reversing the procedure provides compatible sets of ²⁰⁶Pb*/²³⁸U and ²⁰⁷Pb*/²³⁵U ratios for specified values of t; these are the coordinates of points representing U-Pb systems with concordant dates in coordinates of ²⁰⁶Pb*/²³⁸U as ordinate and ²⁰⁷Pb*/²³⁵U as abscissa (Bowen, 1988).

Equation above and its equivalent for ²³⁵U and ²⁰⁷Pb constitute the parametric equations of a particular curve representing the loci of all concordant U-Pb systems. This curve is termed the 'concordia'. It is such that t(²⁰⁶Pb*/²³⁸U) = t(²⁰⁷Pb*/²³⁵U) and all samples which have concordant dates must plot on such a curve. Samples with an age of t₁ which lost lead at some time t₂ plot on a chord connecting t₁ and t₂ their position on which depends the quantity of lead lost. Loss may be episodic or continuous.

Episodic loss may be due to chemical weathering or a metamorphic event. If all radiogenic lead accumulated up until that time is lost, then the coordinates of the point representing the system respond in such a manner that it returns to the origin. Such a uranium-bearing mineral contains no radiogenic lead when it crystallizes, total Pb loss returns the concordia point to the position which it occupied originally.

The inference is that the U-Pb geochronometer recommences and thus all trace of the earlier history of the system disappears. On the other hand, if only some of the radiogenic lead is lost, the point moves some way long the chord, as noted above. Such systems falling on to the chord have discordant dates and, as a result, the chord is termed 'discordia'. Usually, minerals lose only a small proportion of their radiogenic Pb. Occasionally, varied fractions may be lost from zircons in the same sample of igneous rock despite the fact that they must have gone through the same conditions. This may relate to the actual sizes of the crystals concerned and their uranium concentrations as well as to radiation damage in them. Smaller grains, as well as those which have a high uranium concentration, can undergo larger lead losses than others of greater size and uranium concentration. Thus, several zircon fractions may be obtained from one rock sample and these plot as a set of points along the discordia line. Hence, it is possible to ascertain the, position of the discordia by fitting a straight line to data points which represent zircons which lost varying amounts of their radiogenic Pb. Extrapolation of discordia gives two points of intersection with concordia. The first is at the time which has passed since the original crystallization of the minerals and the second is at the time elapsed since closure of the system after an episode either of lead loss or gain of uranium.

As noted above, lead loss may alternatively be continuous. They derive from pegmatites in the White Sea area of Karelia (Belomoria) containing minerals which probably all have the same age - not that the age of zircons is always identical to the age of their host rocks, since this mineral is very refractory and therefore can survive melting of sedimentary rocks to form magma. In metamorphic rocks, zircons usually contain rounded cores representing detrital grains originating from older rocks.

Concordia diagrams allow interpretation of the geological histories of U-Pb systems as well as giving information regarding past disturbances of these systems. Thus, for instance, a parameter X on the discordia line is a measure of the degree to which the daughter-to-parent ratios of the system are affected by either lead loss or uranium gain or loss, where X = l / L. Unless the isotope composition of added lead is known, the effect of lead gain cannot be considered in the model. Implicit also is the requirement that lead loss has to take place with no discrimination against the lead isotopes on the basis of their masses. Taking this into account, X can be represented as the ratio of daughter/parent immediately after the change to daughter/parent before the change:

In a U-Pb system, if 50% of the radiogenic lead were lost without change in the quantity of U, then X would be 0.5 and l= L/2. If the uranium content doubled without change in the quantity of lead, X would again be 0.5. From this, it is clear that loss of lead in the system has the same effect as gain of uranium. In consequence, any U-Pb system lying on discordia between t_c and t_cl could have undergone loss of lead or gain of uranium (Bowen, 1988).

Several concordia models have been developed in order to interpret the discordance of U-Pb dates of uranium-bearing minerals. The first which must be mentioned is the dilatancy model first proposed in 1972 by S.S. Goldich and M. G. Mudrey, Jr. Minerals undergo radiation damage through the alpha decay of uranium, thorium and their daughters; the extent of this damage increases with increasing age, also with the uranium and thorium contents of the minerals. The phenomenon was noted by L. T. Silver and S. Deutsch in 1961 and utilized in 1963 by G. J. Wasserburg. The postulation is that the damage in question produces microcapillaries permitting entry into the crystal of water, which retained until uplift and erosion release the pressure on the mineral. The consequent dilatance of the zircons is accompanied not only by the escape of water, but also by that of dissolved radiogenic lead. In fact, such loss of radiogenic lead might be related to uplifting and eroding of crystalline basement complexes of Precambrian shields. This may be regarded as a relatively recent lead loss in geological terms and is consistent with the view that 207-206 dates frequently approximate the true ages of uranium-bearing minerals. G. R. Tilton in 1960 observed that uranium minerals from different continents appear to have lost radiogenic lead 500-600 Ma ago, despite the fact that no global metamorphic event had been recognized at that time.However, such an event is now recognized, namely the Pan-African, which produced the extensive occurrence of 500-Ma K-Ar ages in Africa and parallel revolutions in other parts of Gondwana, for instance the Brazilian Event 550 Ma ago.

The second model is based on chemical weathering, which affects almost all samples of rocks and minerals collected from surface outcrops. It was discussed by T.W. Stern et al. (1966). Zircons were removed from residual clay formed by the chemical weathering of a gneiss near Redwood Falls in Minnesota, USA. Their dates demonstrated considerable discordancy. In fact, the weathered samples may have lost as much as 85% of their radiogenic lead, assuming that the uranium content is unchanged. The dates lie on straight line chords commencing from discordia and running to the origin.

The third model is a continuous diffusion one, radiogenic lead having diffused from crystals at a rate governed by a diffusion coefficient, D, an effective radius, a, and the concentration gradient. The crystals are assumed to be spheres having the effective radius and there is presumed to be practically no diffusion of uranium and intermediates (or at least if there is, it is assumed negligible). It is also assumed that the radiogenic lead diffusion follows Fick's law. The following expression refers to the change of lead concentration with time for any radial volume element in the sphere:

Solutions of the diffusion equation generate curves on the concordia diagram which are the loci points representing U-Pb systems of specific ages which underwent continuous Pb loss governed by the parameter D/a². The curve is practically a straight line in the case of systems with D/a² < 50 X 10^-12 a^-1; for values exceeding this, the curve deviates from linearity and approaches the origin.

U-Pb concordias are paralled by others involving ²³²Th in association with either the decay of ²³⁸U or ²³⁵U. Actually the concordia curve of ²⁰⁸Pb/²³²Th and ²⁰⁶Pb/²³⁸U ratios is straight because of the similar half lives of the two parent elements. Therefore, intersections of discordia with it are erroneous. On the other hand, concordia based upon the decay of ²³²Th to ²⁰⁸Pb and ²³⁵U to ²⁰⁷Pb are similar in shape to conventional U-Pb concordia (²⁰⁸Pb/²³²Th being plotted along the ordinate and ²⁰⁷Pb/²³⁵U along the abscissa). If no fractionation of either parents or daughters takes place, the response in U-Th-Pb concordias to loss of parents or daughters or gain of parents is analogous to that of U-Pb concordias. However, where the alteration of a U-Th-Pb system entails a preferential gain or loss of parents or disproportionate losses of ²⁰⁸Pb compared with ²⁰⁷Pb or ²⁰⁶Pb, a different response occurs. Uranium and thorium have differing geochemical properties so that changes in the ratios between these elements are probable in cases of episodic alteration or continuous gain or loss of parents. It is also possible that ²⁰⁸Pb, ²⁰⁷Pb and ²⁰⁶Pb are lost at slightly different rates. Such fractionation induces discordant U-Th-Pb systems to lie on discordia which do not pass through the origin. Such lower intersections of resultant discordias with U-Th-Pb concordias are not significant in terms of geological time, merely recording changes which took place in the Th/U ratios or lead isotope ratios through the factors alluded to above (Bowen, 1988).

U-Pb and Th-Pb isochrons have been employed to date samples of granite in the USA, the USSR and Australia. U-Pb isochrons are not so optimal, mainly because chemical weathering causes loss of much uranium from exposed rocks. Since thorium and lead are retained preferentially, their isochrons are feasible. Where uranium loss has not taken place, closed U-Pb systems lie on the concordia curve and, if lead is lost or uranium gained, movement occurs towards the origin along a straight line discordia; if U is lost, movement takes place along a straight line in a direction away from the origin. Thus, rocks which have lost varying amounts of uranium through chemical weathering should lie above concordia and form a straight line passing through the origin (if the loss is fairly recent). The age of such rocks is indicated by the point of intersection on the concordia curve (Bowen, 1988).

Common lead occurs in many minerals such as galena, PbS, and cerussite, PbCO₃, in ore-forming bodies. It is also present as a trace element in many minerals of low uranium and thorium content as well as in rock-forming minerals such as feldspars (in potassium feldspar, Pb²⁺ replaces K⁺). Common lead contains isotopes of masses 204, 206, 207 and 208, of which the first is not the decay product of a radioactive decay series, the others of course originating with ²³⁸U,²³⁵U and ²³⁸Th respectively. The content in these three isotopes of lead may be related to that of ²⁰⁴Pb and the atom number ratio may be described thus:

Every lead contains a measurable quantity of ²⁰⁴Pb and may be represented in a three-dimensional diagram of the variables a , b and g . There are large variations in the isotopic composition of common lead of different origins. These were difficult to understand in view of earlier observations that common lead seemed to have a constant atomic weight, from which a constant isotopic composition might be inferred. However, it was realised that this apparent constancy was fortuitous, reflecting the fact that the increase in the ²⁰⁶Pb/²⁰⁴Pb ratios are frequently accompanied by a comparable increase in the ²⁰⁸Pb/²⁰⁴Pb ratios.

Attempts have been made to construct quantitative models for the isotopic evolution of lead in the Earth from which the age of the planet and also that of common lead minerals could be determined. It is based upon the following set of assumptions. Firstly, the Earth is supposed to have been originally both fluid and homogeneous at which time uranium, thorium and lead were distributed uniformly and the isotopic composition of this primeval lead was the same everywhere. Later, the Earth became rigid and small regional differences in the U/Pb ratio arose. In any given area, the ratio altered solely as a consequence of radioactive decay from uranium to lead. Finally, at the time of formation of a common lead mineral, the lead was separated from uranium and thorium and its isotope composition has remained constant ever since.

The model permitted the comprehension of differences in a , b and g for various types of common lead and also allowed the calculation of the time which has elapsed since the formation of the crust of the Earth as well as the time of ore formation of common lead occurrences, at least in some cases. Most of the lead contained in those parts of the crust which are accessible occurs in igneous and sedimentary rocks with a content of a few tenths to, say, 50 ppm of lead. Only a very small proportion of the lead in the crust is contained in minerals or ore formations containing more than 0.1% of lead (this is also true of uranium and thorium). Consequently, the incorporation of lead into a lead mineral is a very rare event so that the probability of it occurring twice in the history of a given sample of lead is negligible.

In the source rock from which a given sample of common lead has been formed by mineralisation or ore formation processes, lead has been associated with a given quantity of uranium and thorium. This may be described by the chemical milieu index, m , the atom number of ²³⁸U over ²⁰⁴Pb comprising the milieu from, which the sample of common lead is derived. As the ²³⁸U is subject to radioactive decay with time, the content of it may be extrapolated to the present as if the original mother rock were still in existence. The chemical milieu index may be defined as, m = (²³⁸U)_today/(²⁰⁴Pb). In a similar way, it is possible to define another chemical milieu of the mother rock of a given sample of common lead by the ratio of the number of atoms of ²³²Th over the atom number of ²³⁸U, both extrapolated to the present, in the mother rock from which the sample of common lead has been separated:

A considerable impetus to these investigations was given by the measurement of the isotopic composition of minute quantities of lead contained in iron meteorites, particularly in the iron as well as in the triolet phase of Canon Diablo. Values of a and b obtained were much lower than those in any other sample of terrestrial lead or lead contained in stony meteorites. Meteorites constitute fragments of larger parental bodies, which formed very early in the history of the solar system. The parents probably originated through partial melting and chemical differentiation, thereafter solidifying - a process in which troilite (FeS) formed. The troilite phase, while containing lead, is practically uranium- and thorium-free. Hence, the isotopic composition of the lead is regarded as almost constant since crystallization. Such lead is the least radiogenic available and probably approaches nearest to primeval lead in isotopic composition.

The Holmes model assumes for the parent matter a constant value for the chemical milieu index m from the time w to p. The measured a , b and g for a given sample permit the model age p, the chemical milieu index m and the value x to be calculated. There seems to be a rather uniform cluster of m -values of the parent rock in many samples of common lead.

All such models for the isotopic change with time from the formation of the Earth's crust up to a time p of mineralisation involve the hypothesis that the lead spent the entire time interval (w - p) in a milieu characterized by a constant value of m . Curiously, phenomena such as metamorphism, which occurred frequently through geological time, do not appear to have influenced the m -values of the mother material of lead minerals appreciably. There is also a small dispersion in the values of x.

It is important to add that there arose some serious objections against the Holmes model in regard to anomalous leads. Common leads which provide meaningful model dates are termed 'ordinary', in contradistinction to these which yield meaningless values. In fact, the number of ore deposits, which have been found to contain ordinary lead is small as a result of the probability that the lead in most such deposits underwent a more complex history than, is envisaged in the Holmes model. It may be stated that, in order to conform with this model, the relevant lead must have had a single-stage history which can be confirmed if the model dates agree reasonably with isotopic dates obtained using other minerals from the ore and if the isotopic ratios of Pb from a particular deposit are constant within experimental error (Bowen, 1988).

The basic assumption regarding the development of the isotope constitution of lead is the addition of the radiogenic isotopes ²⁰⁶Pb, ²⁰⁷Pb and ²⁰⁸Pb from the decay of U and Th with which it is associated, characterized by the values of m and xm , these having remained constant during the time interval (w - p).

It might be expected that the time p calculated, should coincide with the time of mineralisation, which is the geological age of the lead mineral. However, it became evident that such a model age p often does not so coincide, the geological age as revealed by fossils in the beds containing veins of lead ore being different. In other instances, absurd results were obtained and these led to negative ages so that common lead ores now in existence would form in the far future, say up to 2 X 10⁹ a from then.

Two types of anomalies we re recognised, namely B and J. The B-type anomalies are those in which the model age p may exceed the actual geological age of the rock containing the lead ores, whereas the J-type anomalies contain excess radiogenic lead and give model ages less than the age of the ore deposit.

B-type anomalies are named after Bleiberg in Austria and the explanation for their origin is rather simple. If a lead mineral has formed at a time p, then its model age p will give its correct age. However, processes such as pneumatolysis may redistribute lead minerals in ore bodies or veins and thus produce a situation in which g < p, where g is the geological age.

J-type anomalies are named after leads such as those occurring in Joplin, Sudbury, Ontario, Canada. Here, application of the Holmes model equation produces negative model ages. Nevertheless, there do exist cases of apparently ordinary common leads, which may be J-types, which give a model age, which is more or less inferior to their ages of mineralisation. Such cases are extremely difficult to recognise and if they can be detected it is by means of their anomalous x-values. An explanation of the occurrence of J-type anomalies is straightforward according to the basic hypothesis of the Holmes model is of a constant value of m during the total time (w - p). This cannot be regarded as more than, at best, a first approximation of the actual history of the Pb sample. For instance, if purely common lead were mixed with a small quantity of radiogenic lead, an isotopic composition would result which would resemble a J-type anomalous lead. Such an event can have taken place in nature at any time in the time interval (w - p) before the uranium- and thorium content of the lead sample fell to practically nil. It must be emphasised that a single analysis of an anomalous lead of J-type can give almost no information on the date of addition of radiogenic lead at any time between w and p, or on the actual isotopic constitution and thus the age. More complete evidence results from analyses of many samples from a region (Bowen, 1988).

Obviously, lead contained in igneous and metamorphic rocks will have an isotopic composition which relates to a multistage history from which it is clear that such lead will have had associations in the past with a number of systems processing different U/Pb and Th/Pb ratios. Consequently, interpreting the isotopic compositions of such multistage leads necessitates proper application of approaches outlined previously in this chapter.

Turning to the more complex situation of a sample of lead which has undergone transference through two uranium-bearing systems which were characterized by different ²³⁸U/²⁰⁴Pb ratios designated as m₁ and m₂, the a -value of the sample is given by:

When lead is transferred from the first system to the second, this probably usually entails a physical removal process, for instance through the formation of magma in the mantle and its later emplacement in the crust of the Earth. In such a manner, a lead sample, which was connected formerly with a system in the first environment, becomes associated with a second, which has a different ²³⁸U/²⁰⁴Pb ratio. Subsequently, the lead in question may recrystallise as galena in an ore deposit. It is also possible that the transition from system 1 to system 2 may take place without such a physical removal occurring, ie. by changes only of the U/Pb ratio of the initial system. This can be the result of a loss or addition of uranium, or a fraction of the lead may be lost. Such changes would have affected the ²³⁸U/²⁰⁴Pb ratio and hence the isotopic composition of the lead which is ultimately incorporated into an ore deposit. Another, rarer, possibility is that of the gain of lead; in this case, the isotope composition of the lead will alter due to the different isotope composition of the added lead. The above arguments regarding a two-stage history apply also to three or more stages and also it is possible to provide equations for the other isotope ratios for such multistage leads.

The isotopic history of lead has been studied, together with strontium, in basalts and young volcanic rocks from various localities. The analyses of samples from Ascension and Gough Islands in the Atlantic Ocean really initiated this line of investigation (plumbology). The researches were effected by Gast et al. (1962) and the results demonstrated important differences between isotopic ratios of lead within suites of rocks from each island as well as differences between the islands.³³ In fact, these findings were difficult to reconcile with the idea that leads originating from the upper mantle ought to plot near the point of intersection of the growth curve for conformable ore leads and the geochron. The leads from the islands contained radiogenic lead, however, and are anomalous. In fact, later studies showed that lead both in continental and oceanic basalts cannot be explained using the single-stage approach.

From this, it may be inferred that conformable ore leads do not originate in the upper mantle, their uniform lead growth pattern perhaps arising from the extensive mixing of lead derived from pelagic sediment which was subducted under volcanic island arc systems.³⁵ Interestingly, volcanic rocks from St Helena plot far to the right of the geochron and are enriched in ²⁰⁶Pb with depletion in ²⁰⁷Pb. Variations in the lead isotope ratios here and from samples taken in other islands must mean that the lava flows involved had no common source, but originated in diverse regions with different U/Pb ratios. Consequently, the upper mantle is heterogeneous everywhere and the isotope ratios of Pb from oceanic islands may form secondary or higher-order isochrons which would imply multistage histories. An inference is that the U/Pb and Th/Pb ratios in the upper mantle may have altered either from time to time or continuously through geological time, the actual rates probably having varied appreciably in the mantle. There are various other explanations for the variability of the isotope ratios of lead in volcanic rocks. One is contamination of single-stage mantle-originated leads by variable quantities of lead deriving from granitic rocks of the continental crust. However, this has an inherent problem, namely the absence of such sialic material in the ocean basins. Nevertheless, the proposal still stands as possibly valid in the case of continents and island arcs (Bowen, 1988).

Considering Pb-Pb isochrones of igneous and metamorphic rocks, if a magma volume involving homogeneous material crystallises, a rock suit results which contains different U/Pb and Th/Pb ratios. Another scenario would entail a volcanic and volcanic rock assemblage in which lead is isotopically homogenised following a high-grade region metamorphic episode. Either way, the Pb afterwards evolves along a set of divergent and curved trajectories, which correspond to different values of m in each specimen. If such growth proceeds without interruption until the present time, the leads lie along an isochrone (assuming that the rocks had the same initial isotope ratios of lead formed at the same time and remained closed as regards U, Th and Pb after closure). Recent alterations in the concentrations of these elements are allowable if the isotope compositions of the lead are unaffected. The equation for the isochrone may be derived, and its slope is given by:

Application was made to a suite of metamorphic rocks from the Lewisian basement complex in north-west Scotland by Moorbath and Welke (1969). The rocks found to be depleted in uranium compared with the crustal average; concentrations of U and Pb were determined to be 0.24 ppm and 7.9 ppm respectively and the average ²³⁸U/²⁰⁴Pb ratio was 1.76. The isotope ratios of lead were scattered about a straight line interpreted as a secondary isochrone. Instead of utilizing this in order to calculate the age of the samples, however, a date was found from the point of intersection of the isochrone with the primary lead growth curve; this curve had a value for m of 8.68, which was near, but slightly lower than, the value for conformable lead ores. The result obtained was (2900 ± 100) X 10⁹ a, taken to be the time of variable uranium-loss during pyroxene-granulite metamorphism of the ancestral Lewisian rocks.

Of course, not all-linear patterns of isotope ratios on lead evolution diagrams are isochrones, because mixing of leads of different isotope composition during petrogenesis may cause them. In fact, Moorbath and Welke (1969), cited an instance of this phenomenon from the igneous rocks of the Isle of Skye in north-west Scotland. This region is especially interesting since it includes a wide spectrum of igneous rocks, both intrusive and extrusive, of Tertiary age. As regards their petrogenesis, controversy arose regarding the origin of the felsic rocks which may have been formed either by fractional crystallization of basic magma or by partial melting of underlying Lewisian basement. It was demonstrated that the initial ⁸⁷Sr/⁸⁶Sr isotope ratios of the felsic and granitic rocks are consistently higher than those of the basement rocks and it was concluded that the granitic rocks resulted, at least to some extent, from partial melting of the antique Lewisian gneiss which are shallow in occurrence here. Later, it was established that the isotope ratios of Pb extracted from basic and acidic igneous rocks of Skye, after correction for in situ decay since they were formed some 60 Ma ago, fit a straight line on the lead evolution diagram. This constitutes a chord intersecting a primary lead growth curve (m = 8.92) at two points corresponding to a mixture of two times, namely (3100 ± 50) X 10⁹ a and 60 Ma ago, which was interpreted as follows: the chord in question is a mixing line of lead derived from the Lewisian basement complex with single-stage lead which evolved in the upper mantle until incorporated into basaltic magma approximately 60 Ma B.P. It is important to mention that this accords with an earlier interpretation of the strontium data. Thus, the isotope compositions of both Sr and Pb imply that the igneous rocks of Skye contain significant quantities of both of these elements derived from the Lewisian basement complex, and also that at least some of the granitic magma may have originated in partial melting of this basement complex (Moorbath and Welke, 1969).

The difficulty in understanding zircon U-Pb systems stems from the fact that even in rapidly-cooled igneous rocks the crystals of zircon commonly vary greatly in their contents of trace elements, including the radioactive elements U and Th. Complicating matters further, individual crystals generally are strongly zoned, with internal differences in trace element contents of two to three orders of magnitude over distances of a few micrometres not being unusual. With the passing of time, the decay of the radioactive elements in such crystals not only generates a correlated zonation in the concentration of radiogenic Pb but also, if the temperature remains below that at 'which zircon anneals, creates a zonation in the extent to which the crystal structure becomes damaged by radiation. The effect of that damage is to cause a differential expansion of the crystal lattice (Holland & Gottfried 1955), resulting in the fracturing of the lower Th-U, less damaged zones and a physical weakening of the whole crystal. If the radiation damage is sufficiently severe, all trace of the zircon crystal lattice can be destroyed; the zircon is metamict.

This varied structural damage within the zircon crystals from any one rock means that the response of the zircon U-Pb system to events causing isotopic disturbance differs from grain to grain. This feature was recognised and exploited very early in the history of zircon U-Pb geochronology. Within the population of zircon crystals from a single rock there tends to be a correlation between mean radioactivity and U-Pb discordance (Silver 1963). There also tends to be a correlation between radioactivity and other physical properties of individual grains, for example grain size and magnetic susceptibility. Those properties have been utilised to subdivide zircon populations according to their radioactivity levels and hence relative discordance, thereby defining the pattern of discordance and allowing the concordant (undisturbed) isotopic composition to be inferred (Silver & Deutsch 1961,1963; Silver 1963).

This approach to dating zircon has changed little in nearly 30 years. Certainly the physical techniques for separating low radioactivity, undamaged crystals from the remainder of the zircon population have been improved, particularly with the development of the air abrasion and high gradient magnetic separation techniques by Krogh (1982a, 1982b). However, little real progress has been made in understanding how the zircon U-Pb system behaves under different geological conditions, and in exploiting that behaviour not only to improve the accuracy (as distinct from the precision) of zircon age determinations but also to extract more chronological information from "disturbed" isotopic systems.

One of the reasons for the slow progress is the very popularity of the zircon U-Pb method as a geochronometer. In many laboratories the push to isolate the most concordant zircon in order to measure geological ages with maximum precision takes precedence over studies of discordance, which all too often give only ambiguous results. However, a far more fundamental reason is that, as advanced as the techniques of microchemistry are, the smallest samples analysed are generally a single zircon crystal, or at best a fragment of a crystal, weighing in the order of a microgram. Given the micrometre scale of zircon zoning, most such analyses are as much an average of a variety of compositions as are analyses of whole zircon populations. With the scale of the analysed sample being several orders of magnitude larger than the scale of the isotopic heterogeneity within individual crystals, the detailed study of that heterogeneity is made almost impossible (Williams, 1992).

One of the most widely used methods is the single zircon Pb-Pb evaporation technique. For single zircon evaporation only the highest-quality zircons (idiomorphic, absolutely clear, no visible inclusions, no visible overgrowth or dissolution phenomena) are used. The zircons used for evaporation analyses are washed in warm high-purity HNO₃, rinsed with sub-boiled water and dried. Zircon evaporation analysis follows modified procedures (Kober, 1987). A double Re-filament arrangement in a thermal-ionisation mass spectrometer is commonly used. Single zircon crystals are encased in the evaporation filament. The zircons are heated stepwise in order to strip off common-Pb and radiogenic-Pb components weakly bound to metamict zircon domains or inclusions within the crystal. Progress of stripping is monitored with an ion counter. Certain masses are additionally monitored during this procedure in order to recognise isobaric overlaps during low-temperature steps. As soon as no "low-temperature" Pb is present (approximately 0 min. -2 hours) the zircon temperature is raised by about 20° C and, depending on ion-beam intensities, the Pb isotopic composition is either measured directly or the evaporated Pb is deposited on the cold ionisation filament. After a deposition step (15-45 min.) the Pb is analysed using either static Faraday or dynamic ion-counter data acquisition procedures. Depending on crystal quality, size, age, U and Pb content of the zircons, 2 to 8 evaporation analysis cycles can be made. To check whether the Pb evaporation really is completed or whether the zircon has just jumped off the evaporation filament, the evaporation filament can be slowly raised (Klötzli and Parrish, 1996). The procedures and analysis ranges can differ from sample to sample.

Over the last decade, an alternative method of zircon analysis has been developed with the express intention of reducing the sample size to such an extent that it becomes possible to analyse the interior of single crystals at a scale approaching that on which the internal isotopic heterogeneity occurs. The technique uses a sensitive, high mass-resolution ion microprobe (SHRIMP) built for the purpose (Compston et al. 1982) and is based on an analytical approach developed by researchers such as Andersen and Hinthorne (1972) and Hinthorne et al. (1979).

The SHRIMP is a large mass spectrometer with which small areas (about 30m m in diameter) on the surface of a target crystal can be analysed. The crystal is bombarded by a finely focused ion beam and the ionised portion of the material eroded is transferred directly to the mass spectrometer for determination of its chemical and isotopic composition. The interiors of crystals are accessed by sectioning them by polishing before analysis (Williams, 1992).

Ion microprobe analysis has been accepted only slowly by geoscientists at large. There are a number of reasons for this, some valid concerns, some misconceptions and some unfounded prejudices.

One valid concern is that of the accuracy of isotope ratio measurements. The mass spectrum produced by an ion probe is a complex mixture of atomic and molecular species. Some of the more ambitious of the early ion probe studies (e.g. Lovering et al. 1981) suffered from the low mass resolution of the probes then available, which necessitated large, difficult corrections for the interferences between species that could not be separated by mass. SHRIMP largely bypasses this problem by using high mass resolution to reduce interferences to levels where the corrections required are minimal. As for other mass spectrometers, the corrections for machine-induced mass dependent mass fractionation are determined by analysing standards.

A second concern is the problem of measuring interelement ratios. Although ion probes, like other mass spectrometers, measure isotopic ratios directly, the ratios between elements can only be determined relative to those in standards. Calibration procedures have been refined to the extent that some element ratios in some minerals can now be measured to a precision of about 2% per determination. However, those procedures are sensitive to the exact mineralogy of the target and are therefore not necessarily equally accurate over the whole compositional range even of one mineral. This sensitivity, and the limited homogeneity of available standards, mean that it will be some time yet before ion probe analysis is as precise or accurate as isotope dilution, but the present accuracy is nevertheless more than adequate to solve some very basic scientific problems.

It is through the common tendency to confuse precision and accuracy that the principal misconception regarding ion probe analysis has arisen. Because an ion probe analysis consumes about one thousand times less zircon than the smallest conventional analyses, individual ion probe analyses have larger analytical uncertainties, i.e. are less precise, than conventional zircon analyses, generally by about a factor of ten. It does not follow, however, that the ion probe analysis is any the less correct, i.e. accurate. In fact, a suite of relatively imprecise ion probe analyses on a single zircon crystal, distinguishing differences between core and rim ages and regions of isotopic discordance, is likely to be far more informative than a single high precision conventional analysis of the average composition of the whole grain.

The small-scale isotopic analyses made possible by the SHRIMP have led to a much clearer understanding of the way zircon and zircon U-Pb systems can behave under different geological conditions. On the one hand, this understanding has resulted in the development of much more effective strategies for the microscale sampling of the record of provenance, age and metamorphic history that commonly is preserved within a zircon population. On the other, it has indicated a number of ways in which the interpretation of conventionally-determined zircon discordance patterns might be improved (Williams, 1992).

Among recent advances are studies of zircons from a tonalitic Amîtsoq gneiss in the Godthåb district of southern West Greenland carried out by P. D. Kinny, and work on zircons from the Isua supracrustal belt in West Greenland by Kinny with W. Compston et al., also reported in 1986. In both exercises, a sensitive high mass-resolution ion microprobe (SHRIMP) was used, following its employment in 1984 in obtaining an U-Pb geochronology of zircons from Lunar Breccia 73217 (Comptons et al., 1984).

The modifications in SHRIMP for the terrestrial researches involved the following factors. A primary O₂ ion current of ca 2 mA was maintained on a 30 m m target area. Positive sputtered secondary ions from the sample were collected successively on the ion counter by cyclic field stepping through Zr₂O (2 s count time), Pb isotopes and back round (10 s each), ²²⁸U (5 s), ThO (5 s), ²³⁸U (2 s). Peaks were centred on the collector slit using computer-controlled electrostatic beam deflection and the collector slit was moved automatically along the beam path so as to obtain the optimum focal point for each mass. Data were combined after seven cycles and two such data sets collected per analysis. Raw counts were corrected for 21 ns effective deadtime in the pulse counter/discriminator system and rated to a fraction of the total secondary ion current measured simultaneously at an aperture before the entrance slit to the mass analyser to compensate for minor instability and drift in the primary beam intensity during the analysis. SHRIMP was operated at a mass resolution of 6500 (M/D M, 1% valley) with sensitivity for lead isotopes of 15-25 cps/ppm. All known isobaric interferences were resolved from the peaks of interest, with the possible exception of PbH⁺ from Pb⁺. However, the possible occurrence of PbH⁺ is restricted by both the lack of observation of ²⁰⁸PbH⁺ at mass 209 and by an assessment of ion-probe results for standard targets of known isotopic composition. The mean radiogenic ²⁰⁷Pb/²⁰⁶Pb measured for the 555 Ma old standard Zr SL was 0.O5014 ± 21 (compared with 0.5842 previously measured by SHRIMP and thermal ionisation. Assuming this slightly higher result to be due entirely to the presence of ²⁰⁶PbH⁺ under ²⁰⁷Pb⁺ (ie. ignoring possible differences in mass fractionation), the same level of unresolved hydrides in analyses of early Archaean zircons (with ²⁰⁷Pb/²⁰⁶Pb> 0.3) would cause an overestimation of age of only 3 Ma. Therefore, a small residual PbH⁺ component in the mass spectrum would cause no significant bias.

For Pb isotopes in zircons, the mass discrimination a per mass-unit SHRIMP associated with ion extraction from sputtered material and other causes is not well known as yet. After subtraction of common lead, the relationship between ²⁰⁷Pb/²⁰⁶Pb (R) and a measured R_m, is

where h = ²⁰⁶PbH/²⁰⁶Pb. A positive value of a is taken to indicate light isotopic enrichment. For extremes of zero and 0.0005 for h, the data for SL3 standard gave corresponding limits of +0.34% and –0.51% for a with standard error ± 0.3% in each due to error of measurement and uncertainty regarding true age.

Ion-microprobe U-Th-Pb analyses of zircons from the conglomeratic metavolcanic unit of the Isua supracrustal belt showed a magmatic age of 3807 ± 1 Ma and did not detect any older zircon cores or xenocrysts. Low-uranium areas within zircons at a 30 m m scale lost only small amounts of radiogenic Pb and only in the recent past, whereas high uranium areas within grains lost a much greater proportion of their lead both recently and as early as the Archaean. Some areas in the zircons have concordant U-Pb and Th-Pb ages (Compston et al., 1984).

SHRIMP measurements of the U-Th-Pb isotopic composition of 30 m m areas within individual grains demonstrated that components of magmatic zircon from the original protolith are at least 3822 ± 5 Ma old, i.e. slightly, but significantly, older than those from the felsic volcanic unit of the Isua supracrustal belt measured similarly, and also older than all ages previously determined by mineral and wholerock isotopic techniques for Amîtsoq gneisses at Godthåb and at Isukasia (Kinny, 1986). Overgrowths of younger zircons formed at about 3630 Ma, coinciding with an episode of major lead loss from the older grains. These data indicate that previous whole-rock isochrone ages refer to later isotopic redistribution and/or that other phases of the Amîtsoq gneisses intruded during this later period and dominate the isotopic systematics. Comparison of the ion probe results with previous conventional U-Th-Pb data for the same zircon sample demonstrates that there is a strong interference of discordant high-uranium parts of zircon on the mean isotopic composition of multigrain samples.

Instrumentally, lead isotope ratios were directly measured and assigned errors determined from counting statistics. A minor correction for common lead was based upon the non-radiogenic ²⁰⁴Pb⁺ isotope. The measured ²⁰⁴Pb⁺ was dominated by a surface lead contaminant introduced during sample preparation, so the common lead composition was assumed to be that of Broken Hill ore. The Amîtsoq zircons, regardless of their uranium content, were found to be very low in intrinsic common lead. Each spot analysed was preburned to minimise the contribution of surface lead to which analyses of low-uranium spots were particularly sensitive. The worst case was spot 11-5 with a ²⁰⁶Pb/²⁰⁴Pb ratio of ~500 (corresponding to 4.2% of the total ²⁰⁶Pb being common).

Pb/U ratios were determined using the standard zircon SL and utilizing the observed correlation between UO/U and Pb/U in the secondary ion spectrum that is now approximated by a quadratic function. Reproducibility of Pb/U for the standard analyses during Kinny's study was 1.55% per set, which translated to a minimum coefficient of variation of ²⁰⁶Pb/²⁰⁸U of 1.10% for the combined data-set pairs of the unknowns.

Such ion-probe data reveal a more complex pattern of isotopic compositions than that indicated by the orthodox zircon data. Baadsgaard's result for sample 110 999 yielded an apparent age approximating the time of high-grade metamorphism in the early ' Archaean because of the preponderance in the bulk population of high uranium metamorphic zircon and zircon which lost most or all of its accumulated radiogenic lead at that time. This shows a hazardous aspect of bulk zircon analyses where a minority of uranium-rich grains prone to radiogenic lead loss can strongly influence the mean U-Pb isotopic composition of the composite sample. Whereas some grains, especially type 2, suffered major lead loss about 3600 Ma ago, others were comparatively little affected.

Possibly the contrasting responses of the different grains to the same event is related to their position in the rock fabric. Th-Pb isotopic data gave no evidence of any differential movement of thorium with respect to uranium either into or out of the zircons and it is justifiable, therefore, to assume that the presently observed low uranium and thorium contents are close to original levels. From this, it may be inferred that only minor structural damage was generated by the in situ decay of actinide elements in the 200-Ma time interval between the formation and recrystallization processes and that, around 3600 Ma ago, radiogenic lead was flushed out of the essentially pristine zircon structures. A clue to the main mechanism of lead loss may lie in the association of secondary fluid inclusions in the more disturbed grains. These appear to be remnants of fracture planes which, if present at the time of metamorphism, would have effectively increased the surface area of grains and potentially facilitated the wholesale leaching of radiogenic lead from the zircons by the metamorphic fluid phase (Bowen, 1988).

Baadsgaard et al. (1984) discussed the zircon geochronology of the Akilia association and Isua supracrustal belt, both being interpreted as deriving from a single volcano-sedimentary sequence since they occupy the identical position in local stratigraphic sequences and also show a similar association of lithologies as well as being intruded by geologically and geochronologically correlated orthogneisses. Zircons from Akilia gave a concordia intersection age of 3587 ± 38 Ma and from the Isua belt an age of 3813 _-14⁺²¹ Ma was derived. The latter is close to the original age of crystallization of the zircons according to these workers while the former result is ascribed to the resetting of the zircon date by granulite facies metamorphism close to 3600 Ma ago.

Compston et al. (1985), using the SHRIMP, carried out an investigation of zircon xenocrysts from a part of the world far removed from Greenland, namely western Australia. There, samples were taken from the Kambalda volcanics in order to review age constraints as well as direct evidence for older continental crust below the Kambalda Norseman greenstones. The Hanging Wall basalt at Kambalda contains zircons shown by ion microprobe analysis to have very high uranium and thorium contents and a wide variety of crystallization ages. Nearly all are most probably xenocrysts and a few may relate to intrusive veinlets. The age of the youngest is 2693 ± 50 Ma; this is stated to demonstrate that the eruptive age of the basalt cannot exceed 2743 Ma, confirming that the apparent Sm-Nd isochrone giving 3000 Ma for Kimbaldan mafic and ultramafic rocks is a mixing line between unrelated components enriched and depleted in light rare earth elements (REE). Mixing probably occurred at depth by erosion of 3200 Ma to 2500 Ma-old felsic crust from the walls of the HWB conduits. The zircon xenocryst ages are claimed to represent the first direct evidence for the presence of very old felsic crust in the eastern Yilgara Block. This implies that the Kalgoorlie-Norseman greenstone sequences were formed in a continental rather than an oceanic environment.

Very recently, work has been published on zircon fractions and single grains together with fractions of monazite from Tibet, this referring to the ages of Himalayan leucogranites⁴⁷. These are found in the High and Tethys Himalayas and are especially interesting because of their compositional distinctness from other Himalayan plutons such as the calc-alkaline Gangdese belt, and also because they have a particular emplacement context which shows a strong relationship to both high-grade metamorphism and large-scale thrusting. The favoured model originates the granites by crustal melting during intraplate subduction within the northern border of India reacting to continuing postcollisional northward drift of the sub-continent into Eurasia. However, the exact ages of some of them are unknown because Rb-Sr dating is uncertain in view of a lack of equilibrium in this isotopic system at whole-rock scales. Also, even rather good linear arrays of whole-rock samples may be fortuitous isochrones, which define quite meaningless geological ages. A good instance of this is given by the Makalu granite for which precise U-Pb ages of 24·0 ± 0·4 and 21·9 ± 0·2 Ma showed that a 92·7 ± 9·4 Ma `isochrone' from the same granite represented an accidental linear array.

While Rb-Sr and K-Ar dating on minerals of various granites and country gneisses in the Himalayas define ages between 17 and 1 Ma, most of these have been interpreted as cooling ages, thus fixing mineral dependent minimum ages for granite emplacement and regional metamorphism. Cooling ages are probably cooling-rate dependent and this could result in age differences up to several Ma for contrasting cooling histories of the same mineral. This constitutes a significant potential handicap for precise dating of Cenozoic rocks using Rb-Sr and K-Ar analyses on minerals. For this reason, Schärer et al., (1986) chose to use U-Pb dating, claiming that this geochronometer is generally accepted as better approximating the time of primary rock crystallization due to the much higher blocking temperatures of this system in accessory phases. They cited an interesting example, namely the Palung granite which is located south of the High Himalayas. Zircon and monazite from samples of this granite gave primary Palaeozoic ages, but the Rb-Sr system in micas from the same samples proved to have been reopened and reset completely during the Alpine metamorphism.

Precise rock ages are desirable in order to establish that the plutonism and related high-grade metamorphism took place in fact in post-collisional times after about 40 Ma ago, and to determine time constraints on the thermal regime during the Himalayan orogeny (important in modelling anatectic granite genesis) as well as to facilitate a temporal-spatial reconstruction of the thrust events (break-of) in the Indian margin (that is to say the formation of the Himalayas). An additional aim was to investigate the nature and age of the relevant magma sources as traced by inherited radiogenic lead in zircons and to compare data with U-Pb systematics in country gneisses (the possible source rocks of the granites).

The analytical method utilised was to dissolve zircon (in 50% HF) and monazites (in 33% HCl), in Teflon bombs at 220° C for some 100 h. In the case of Th-U analyses of whole-rock powders, the method described by Allegre and Condomines (1976) may be used, and Manhes et al. (1978) have given the chemical procedure for U-Pb dating.

For common lead isotopic measurements in potassium feldspars, the leaching technique of Gariépy et al. (1985) may be followed. In the work on the leucogranites, isotopic compositions and concentrations were determined on Thomson mass spectrometers with thermal ionisation sources and mixed spikes of ²⁰⁵Pb-²³⁵U for mineral dating and ²³⁰Th-²³⁵U for Th-U concentrations of rock powders. Except for Pb isotopic compositions in potassium feldspars, all measurements were carried out on secondary electron multipliers (the lead isotopic compositions being handled by Faraday cup collectors). Correction for incorporated common lead in zircon and monazite was effected with the isotopic compositions of Pb in the potassium feldspars from the same sample. The concentration and isotopic composition of blank Pb were evaluated from a set of some 20 independent measurements giving 30 ± 15 pg of blank Pb with the following isotopic composition:

Schärer et al. (1986) referred to particular problems of U-Pb dating and mentioned two significant constraints which must be taken into account in the case of dating minerals which are younger than around 100 million years. Firstly, they indicated that the trajectories of lead loss are virtually coincident with the concordia curve (which is practically a straight line for the segment from zero to ca 100 Ma). One result is that all data give concordant ages and so demonstrate a closed-system type of behaviour even for very perturbed minerals. Another disadvantage of the coincidence is that it precludes the determinations of ages by interception of the concordia curve with regression lines through discordant data (upper and lower intercept ages). The matter can be resolved only by repeated analyses of differing fractions of the same mineral which demonstrate whether or not the mineral actually behaved as a closed system in respect to U, Th and Pb. Secondly, it was pointed out that thorium-rich minerals such as monazite must be corrected for excess ²⁰⁶Pb originating from original excess ²³⁰Th.⁵² Such an excess arises from thorium enrichment in the mineral relative to the Th/U ratio of the magma so that the excess ²⁰⁶Pb does not come from ²³⁸U, causing a secular disequilibrium in the system which necessitates a correction for the ²⁰⁶Pb/²³⁸U ages. This correction is reached by determining the Th/U ratio of the rock and assuming that the whole-rock powder still reflects the Th/U ratio of the magma. However, it cannot be obtained if uranium was leached from the rock after the crystallization of monazite. Then the ²⁰⁷Pb/²⁰⁶Pb ages remain negative, even after adjustment for the undercorrection due to wholerock Th/U ratios that are too high consequent on such leaching of uranium. In a manner similar to ²³⁰Th for the ²³⁸U decay series, ²³¹Pa might produce a small excess of ²⁰⁷Pb in the ²³⁸U decay chain. The degree of such an initial ²³¹Pa disequilibrium can only be guessed, and direct measurement in rocks older than about one million years is not feasible because of the lack of a stable Pa isotope. In the case of monazites analysed in the leucogranite study, the possible ²⁰⁷Pb excess is not significant because the extent of initial Pa enrichment is much lower than that for thorium.

The zircons selected for analysis possessed the following characteristics. They were small, needle-like, essentially newly grown crystals having little inheritance, larger and short prismatic zircons containing on average more inherited radiogenic Pb. They were transparent and colourless indicating low uranium concentration with a lesser amount of post-crystallizational lead loss and lower contents of common lead. Highly translucent dark-coloured inclusion-bearing or altered grains were avoided. The monazite fractions were chosen on the basis of grain size alone. The zircon and monazite data from the leucogranites are similar. All of the former contains components of inherited radiogenic lead, including the needle-like ones with no microscopically distinguishable cores. The inherited components originated from the source rocks and confirm the anatectic source of all the granites and migmatites according to Schärer et al. (1986). The inherited lead demonstrates that the material in the magma source regions is quite old, defining apparent minimum ages between 400 and 2200 Ma. As the individual zircon analyses (except for a single grain) were grain fractions, each point in the diagrams might represent a mixture of different zircon generations, and these minimum ages may be considered as wide estimates for the age of the source rocks. However, the inherited lead indicates Precambrian material to be a major component in the magma sources. Crustal residence times of such length for the source materials are quite consistent with the highly radiogenic isotopic composition of Pb in potassium feldspars revealing ²⁰⁷Pb/²⁰⁴Pb ratios up to 16.04. Nd model ages from some of these granites suggest mean source ages of the order of 1100-2300 Ma and are consistent with ages inferred from inherited lead in zircons. By contrast, the monazites showed no inherited components, which permitted useful inferences to be made regarding granite crystallization. The ages obtained were 15.1 ± 0.5 Ma for the Lhagoi Kangri granite of the Tethys Himalaya belt, and 9.8 ± 0.7 and 9.2 ± 0.9 Ma respectively for two varieties of the Maja granite. For the High Himalayan granites, the monazites define ages of 14.3 ± 0.6 Ma for a granite near Mt Everest and 16.8 ± 0.6 Ma for a migmatitic granite north of Nialam. The approximately 9 Ma age for the Maja granite is compatible with, if slightly older than, preliminary ³⁹Ar-⁴⁰Ar dating ages on minerals from the same granite (Maluski, 1984). Ages yielded were 7.2 ± 0.6 and 7.2 ± 0.4 Ma on biotite and muscovite respectively.

From the Th-U data from whole rocks, concentrations of thorium and uranium in the relevant powders implies, if compared with Th/U ratios in monazites, that uranium has been leached from the rock, falsifying the initial Th/U ratio of the magma from which the minerals crystallised. Because most of the ²⁰⁷Pb/²⁰⁶Pb ages of monazite remain negative even after correction for excess ²⁰⁶Pb, it seems that leaching of uranium is a widespread phenomenon in the Himalayas. Curiously, a migmatitic granite showed an uncommonly high thorium concentration (about 37 ppm), implying a late exchange with a very thorium-rich fluid which on a priori grounds (U is much more mobile than Th during hydrothermal alteration) would appear to be improbable (Stuckless et al., 1981).

Both monazite and xenotime are pale green to green transparent yttrium phosphate minerals, but the former; rare earths mineral is relatively enriched in thorium. Nevertheless, they are difficult to distinguish during individual grain hand-picking. Monazite analysis of the sample XGS-117. shows through the lower atomic content of ²⁰⁸Pb that some xenotime grains may well have been mixed in with the monazite. However, for the majority of rocks, monazite can be separated from cogenetic xenotime Schärer et al. (1986).

Some implications of this fascinating work on the Himalayan leucogranites of South Tibet emerge. One of the most significant is that, together with previous U-Pb ages of 21.9 ± 0.2 and 24.0 ± 0.4 Ma for two phases of the High Himalayan Makalu granite, the later ages of 16.8 ± 0.6, 15.1 ± 0.5, 9.8 ± 0.7 and 9.2 ± 0.9 Ma demonstrate that the High and Tethys Himalaya leucogranites postdate the collision between the Lhasa block (Eurasia) and India which took place approximately 40 Ma ago. The inherited radiogenic leads found in all zircons from these leucogranites substantiate a basically anatectic origin which is confirmed by field observations. Evidence comes from isotopic tracers such as Sr, Nd, Pb and O also give crustal signatures and permit the inference of long crustal residence times of the magma source materials. The occurrence of large parts of Precambrian material in the magma source area is demonstrated by the average minimum ages of the inherited lead in zircons, ranging from ca. 500 and 2200 Ma. The U-Pb ages in addition show that the leucogranites of both the Tethys and High Himalayan belt constitute intrusions of age about 14-17 Ma; this leads one to query the separation of the two suggested by the actual pattern of outcrops in the relevant region.

Some matters remain to be clarified. One is whether 22-24 Ma old granites, e.g. that at Makalu, were also emplaced in the Tethys Himalayan (Lhagoi Kangri) belt. Another is whether approximately 9 Ma old granite such as Maja occurs also in the High Himalayan range. In addition, there is the open question regarding the precise age relationship between high-grade metamorphism and granite emplacement. In this connection, it is noteworthy that the ca 17-Ma-old migmatitic granite north of Nialam which is emplaced following the main planar structures of regional deformation implies that in some areas the metamorphism is younger than early (22-24 Ma old) granites. However, the only 9 Ma old granite shows that igneous rock of this type formed after major phases of both tectonic and metamorphic activities occurred for instance some 13 Ma ago at Kangmar in the east of the Lhagoi Kangri belt. There is field evidence that deformation in the cases of the 24 Ma old and 16 Ma old granites in the Tethys Himalaya do not follow the alternation of regional tectono-metamorphism and granite emplacement and High Himalaya belts. The granites crosscut the dominant Alpine foliations in the country rocks. It is inferred from this that the various phases of strong deformation and metamorphism must have taken place in alternation with the emplacement of granites, ie. they occurred before and after discrete pulses of granite formation.

In sum, the U-Pb ages determined to date imply pulses of plutonism around 24, 16 and 9 Ma ago, and strongly contrasting heat-flow measurements and zones with high magnetotelluric conductivity data suggest that a fourth phase of plutonism may be active today. The inference is that shortening (intracontinental subduction in the Indian margin after collision with Eurasia) occupied four phases separated by periods of 7-9 Ma. If this is true, then the large-scale thrustings juxtaposing the various terranes must have been rapid, lasting only a few million years at maximum between pulses of generation of magma. Such an episodic alternation of plutonism and thrusting could explain in addition the means by which granites and high-grade gneisses were brought to the surface, i.e. that the initial granite-high-grade gneiss terranes were upthrust during a younger phase which itself promoted metamorphism and granite formation, and so on. It will be necessary to effect additional precise dating, especially of metamorphism in the country rocks, in order to consolidate such a model (Bowen, 1988).

Several studies from different parts of the Earth has been done on zircons up to now, including conventional U-Pb studies, zircon Pb-Pb evaporation and SHRIMP technique often combined with zircon typology. The case studies which will be discussed in this section mostly are about SHRIMP model.

Klötzli and Parrish (1996) reported the mean values of at least 20 measured ²⁰⁷Pb/²⁰⁶Pb ratios per zircon. According to these authors, these ages are interpreted as concordant and thus geologically meaningful because of the following reasons: (1) 3 or more individual zircon ages (rims, cores or total crystals) give within 2 std. errors of the mean the same mean age and there are no data between the observed age clusters. This clustering of ages argues strongly for the fact that the observed ages are not caused by single- or multi-stage episodic or diffusive lead loss, (2) the observed very recent lead loss does not affect the ²⁰⁷Pb/²⁰⁶Pb ages significantly, (3) evaporation data and conventional data for the Carboniferous ages are either concordant (338 Ma) or concordant within error (353 and 352 Ma respectively), (4) no obvious relation between evaporation temperature and ²⁰⁷Pb/²⁰⁶Pb ages can be observed. Klötzli and Parrish (1996) analysed also 5 different fractions of one sample from the central part of the pluton were analysed with conventional U/Pb techniques.

Klötzli and Parrish (1996) detected the evaporation ages > 1206 Ma are interpreted as being derived from inherited cores in single zircon crystals. These can only be interpreted as minimum ages with more or less unknown discordance. Thus they must be interpreted as minimum age estimates for magmatic or metamorphic zircon growth during the Proterozoic and Archean. It cannot be decided whether these old zircons belong to distinct Proterozoic or older rocks or whether they must be attributed to metasedimentary (basement-) sequences. The upper intercept age of 2005±12 Ma is interpreted as a minimum mean age for the Lower Proterozoic crust.

Klötzli and Parrish (1996) observed two distinguishable zircon populations by typological investigations and distinguished at least 4 different zircon-forming events. These events are inherited cores with ages around 623 ± 22 Ma and single ages > 1206 Ma from type 1 zircons imply the reworking of rocks derived from Cadomian and Proterozoic to Archean crust. Ages around 353 ± 9 Ma from type 1 zircons are interpreted as timing a first magma formation or the onset of a long-lasting magma-generating event during the Variscan plutonism in the South Bohemian pluton. The actual intrusion of the granodioritic magma into the middle crust took place around 338 ± 2 Ma (type 2 and rims of type 1 zircons). Only type 1 zircons are found as inclusions in large K-feldspar phenocrysts providing evidence that these phenocrysts have grown before the 338 Ma event and may be as old as 353 Ma.

Pidgeon and Compston (1992) report the results of an ion microprobe study of individual zircons from four Caledonian granites from the Scottish Highlands. The inherited zircon often forms nuclei for growth of magmatic zircon, and direct investigation of the isotopic systems of such complex zircon grains is not possible using conventional zircon analyses. This limitation has been overcome by the development of the ANU ion microprobe SHRIMP, which provides a means of measuring the U-Pb isotopic characteristics of 20-40 m m diameter areas on the polished internal surfaces of individual zircon grains (Compston et al., 1984). Their initial objective was to use this spatial resolution of SHRIMP to determine the ages of inherited zircon cores. These cores are generally encased in magmatic zircon rims. Chemical changes across the rims reflect evolving magma conditions, and their second objective was to examine the potential for using the spatial resolution of SHRIMP to identify the Th-U-Pb isotopic and chemical patterns across well-developed zircon rims as potential monitors of magmatic processes. This initial study was made on well-developed, concentric, zircon shells in zircons from the Glen Kyllachy granite. They have also applied SHRIMP techniques to date zircons from the Ben Vuirich granite, to provide an independent age determination for this granite. The Ben Vuirich granite intrudes the Dalradian rocks of the Scottish Highlands and has until recently been considered to be Ordovician in age.

Middle Proterozoic age components were found in inherited zircons from all four granites. Late Proterozoic (900-1.100 Ma) components have been identified in zircons from the Glen Kyllachy and Ben Vuirich granites in the Grampian Highlands. A Late Archaean age has only been detected in one zircon from the Glen Kyllachy granite. The distribution of inherited components in the granite zircon populations could refiect fundamental divisions in the age composition of granite source rocks; however, detailed assessment of this possibility must await further ion microprobe analyses on zircons from many more granites.

SHRIMP isotopic and U, Th and Ph analyses were made by Pidgeon and Compston (1992) on successive shells of zoned zircon surrounding inherited cores from the Glen Kyllachy granite to monitor chemical changes during magmatic zircon growth. Results show that zircon shells have characteristic but significantly different Th, U and Pb concentrations. Magmatic zircon from the Vagastie Bridge granite also forms as clearly defined oscillatory zoned shells around unzoned nuclei of inherited zircon. However, the distinction between magmatic and inherited zircon in zircons from the Inchbae granite is less clear. Zircons from the Ben Vuirich granite occur as euhedral, magmatic zircons, or as rounded, subhedral, inherited zircon grains.

Concordia plots can be obtained as a result of SHRIMP studies. A SHRIMP age of 597 ± 11 (2s ) Ma for euhedral magmatic zircon from this granite is identical, within the uncertainty, to the conventional multigrain zircon age of 590 ± 2 (2s ) Ma reported by Rogers et al. (1989) and confirms the conclusions of those authors that sedimentation of the Dalradian sequence took place in the Precambrian.

According to Pidgeon and Compston (1992), it is generally not possible to determine specific ages of inherited zircon using conventional multigrain U-Pb isotopic analyses technique, as the analysed multigrain samples are complex mixtures of inherited and magmatic zircon, but SHRIMP analysis spots can be targeted on to the inherited cores themselves and ages measured directly, without contamination by magmatic zircon or other generations of inherited zircon. This has opened up new possibilities for examining the distribution of ages of inherited zircon in granites, with implications for the nature of the source rocks, and provides a means of investigating the internal stability of the U-Th-Pb system within individual inherited zircons. Following their initial crystallisation, inherited zircons may have survived a range of geological conditions such as weathering, transport, deposition, metamorphism and melting of their host rock, exposure to a granite melt, recrystallisation and possibly later metamorphism. As a consequence, it is not unexpected that analysed areas of some inherited zircon have been isotopically disturbed during one or more of these events (Pidgeon and Compston, 1992).

Clemens et al. (1986) argued on textural grounds that zircon was one of the first minerals to crystallise in the magma and was entirely melt precipitated. The fact that zircon was present in all their experimental runs and they were not able to locate the zircon saturation boundary was cited as supporting evidence. Using the Zr solubility data of Watson and Harrison (1983) they calculated that the zircon saturation temperature for the magma would be about 940° C.

In this context, the zircon from the Watergums Granite was analysed by Williams (1992) the expectation that it would be of uniform age and undisturbed isotopically by postemplacement processes. This expectation was based also on the morphology of the zircons. Most of the crystals consist of {100} prisms terminated by simple {101} pyramids, similar to type D of Pupin (1980), a type of zircon that rarely contains inheritance. A concordia plot of the SHRIMP U-Pb analyses of 32 areas on 29 grains shows that these expectations were largely fulfilled; all but four of the analyses are clustered close to the concordia at an age of about 400 Ma.

It is the few analyses that are not in the cluster that initially are of most interest from the point of view of discordance. Two of those analyses have the same ²⁰⁷Pb/²⁰⁶Pb age as the main population, but are significantly lower in ²⁰⁶Pb/²³⁸U. The other two have very high ²⁰⁷Pb/²⁰⁶Pb ages, 1,811± 60 and 2,481± 40 Ma respectively. These older ages were obtained on the centres of grains in which there was no obvious core, just a faint irregular shadow in transmitted light to suggest a break in the grains' growth histories. Analyses of the zircon near the margin of both grains plot within the main population close to 400 Ma (Williams, 1992).

According to Williams (1992), this is the manner in which inherited zircon commonly occurs in a granite, not as discrete older grains but as nuclei on which zircon has later precipitated from the melt. The cores and overgrowths are rarely distinguishable optically, particularly in thin section, making visual inspection of zircon crystals a totally inadequate means of assessing whether or not inheritance is present, let alone its abundance. On the other hand, the structural breaks within zircons that commonly accompany inheritance are splendidly revealed in backscattered electron or cathodoluminescence images of sectioned grains (Paterson et al. 1992; Vavra 1990).

The two Watergums Granite zircon cores with high ²⁰⁷Pb/²⁰⁶Pb ages are strongly discordant, indicating a major disturbance to their Pb/U. From the fact that the ²⁰⁶Pb/²³⁸U age of the core with lower ²⁰⁷Pb/²⁰⁶Pb is less than 400 Ma, at least part of the Pb/U disturbance occurred after the pluton was emplaced. This being so, it is invalid to estimate the primary age of the other core by assuming single stage Pb loss at 400 Ma. All that can be inferred with reasonable confidence is that the cores are at least as old as their ²⁰⁷Pb/²⁰⁶Pb ages, and even that conclusion is not necessarily correct. Had the cores been larger, the question of their ages might have been resolved by obtaining additional analyses.

The two strongly discordant analyses from ~400 Ma-old grains are unequivocal evidence for very young disturbance of the zircons' Pb/U; the disturbance must have been at least as late as the youngest ²⁰⁶Pb/²³⁸U age, 115 Ma (Williams, 1992).

Following the work of Silver (1963) it has become generally accepted that discordance in zircon is principally a function of its radioactivity. As more single crystal and ion probe work is done, however, this is found to be true only in the most general sense, so that on a grain by grain, zone by zone scale, discordance obviously is controlled by other factors as well.

The relationship between radioactivity (approximated by U content) and discordance (measured by ²⁰⁶Pb/²³⁸U age) for the Watergums Granite zircons, excluding the two inherited cores, was studied by Williams, 1992. Radioactivity does not control discordance in this case. The control instead appears to be Th/U; even though both the Th and U contents of the discordant areas are close to the averages for the population, both have unusually high Th/U ratios. They also have very low Pb-Th ages. This illustrates a phenomenon that has been found, through ion probe analysis, to be quite common in zircons; high Th, and high Th/U in particular, significantly increases a zircon's susceptibility to isotopic disturbance. Whether this is a characteristic of the zircon as such or related to the presence of small amounts of strongly discordant thorite has yet to be investigated (Williams, 1992).

Williams (1992) also detected the discordant U/Pb ages in the Watersmeet gneiss, in northern Michigan, N America (discordance associated with amphibolite-grade metamorphism), in the Mount Sones gneiss from Enderby Land, Antarctica (discordance associated with granulite-grade metamorphism), in the Kameruka Granodiorite, Bega Batholith, Australia (discordance in an I-type granite containing significant inheritance), in the Dalgety Granodiorite, southeastern Australia (discordance in an S-type granite) in his work. Depending on these studies, Williams (1992) stated that, several generations of zircon can co-exist within one population or crystal, each affected differently from an isotopic viewpoint by its post-crystallisation history. The melt-precipitated zircon in many igneous rocks is dominantly a closed U-Pb isotopic system. The small fraction that is isotopically disturbed usually has suffered severe Pb loss. All else, being equal, zircon's susceptibility to disturbance increases with increasing damage to its crystal structure. This varies on a sub-micrometre scale, increasing, for example, with increasing U content (but becoming significant only above a threshold concentration of a few hundred parts per million that is different from rock to rock), increasing Th/U, increasing age and increasing time since the zircon was last annealed.

For relatively young zircons (younger than mid Proterozoic) the most accurate measure of age is generally obtained from radiogenic ²⁰⁶Pb/²³⁸U. This is least susceptible to errors in the determination of radiogenic ²⁰⁷Pb/²⁰⁶Pb, such as arise from errors in estimating the amount or composition of the initial Pb or unsuspected inheritance. In older zircons, domains with undisturbed Pb/U are scarce, but more common are weakly disturbed domains in which an accurate record of age is still preserved by the radiogenic Pb isotopic composition. In rare cases even the upper limit provided by this age is demonstrably unreliable due to the presence of radiogenic Pb unsupported by U. It is simplistic to treat conventionally-defined discordance trends as dominantly two-component systems. Even in simple metamorphosed rocks, the combined effects of new zircon growth and isotopic disturbance invalidate the use of such discordance lines to infer either the age of the protolith or of the metamorphism in any but the most general terms In multiply-metamorphosed rocks, conventional analyses yield little information at all. The same applies to zircon populations rich in inheritance. Individual inherited zircon cores commonly preserve an accurate isotopic record of their age, but the range in analyses of populations obtained by mixing these cores, plus the melt-precipitated zircon, plus the variable discordance in both, generate mixing arrays that define neither the age of magmatism nor the age of any one inherited component (Williams, 1992).

The presented work include the short review of the age dating techniques on zircons, and further attention is made on the SHRIMP technique which developed in recent years, and found much wider use.

As Williams (1992) stated that the future in U-Pb geochronology, of zircon and many other minerals, lies in microscale analyses. Only in this way will mixed isotopic systems be adequately deconvoluted, the reasons for and implications of discordance be understood, and the full potential of the U-Pb technique be realised. Thus, the data obtained by SHRIMP techinique combined with zircon typological investigations should apply for reliable ages on zircons. By microscale analyses core-rim structures of zircons, discordance patterns and inheritance problems can be solved, and petrogenetical implications can be made more properly depending on these analyses.

This study was prepared through the Term Project Paper of Isotope Geology Course (Instructor: Prof.Dr. Nilgün GÜLEÇ).



\|HOME\|	\|INTRO\|	\|BASICS\|	\|TYPOLOGY\|	\|RESEARCH\|	\|TURKEY\|	\|ME\|