1529-6466/03/0053-0007$05.00 7 Zircon U-Th-Pb Geochronology by Isotope Dilution ��� Thermal Ionization Mass Spectrometry (ID-TIMS) Randall R. Parrish1,2 and Stephen R. Noble2 1 Department of Geology, University of Leicester 2 NERC Isotope Geosciences Laboratories Keyworth, Notts., NG12 5GG, United Kingdom r.parrish@nigl.nerc.ac.uk INTRODUCTION ID-TIMS is the acronym for Isotope Dilution Thermal Ionization Mass Spectrometry. This refers to the addition of an isotope tracer to a dissolved sample to make a homogeneous isotopic mixture, and the measurement of isotopic composition of the mixture using a thermal ionization mass spectrometer. The method is one of the most accurate and precise methods of isotopic techniques because it is relatively insensitive to chemical yields or mass spectrometric sensitivity. It is a method very widely applied both in earth and many other areas of science involving the measurement of element or isotope concentrations and isotopic ratios. The ID-TIMS technique was first applied to the U-Th-Pb dating of zircon in the 1950s (Tilton et al. 1955, Wetherill 1956, Tilton et al. 1957), exploiting the general availability to academia of enriched uranium isotopes developed in the 1940s and 1950s related to nuclear energy re- search. ID-TIMS has remained the main foundation to zircon geochronology ever since, in spite of the proliferation of other analytical methodologies. In the past 50 years, many improvements have been made and they have contributed to the maturity and reliability of the method. As a method, it was effectively unchallenged until the 1980s when secondary ionization mass spec- trometry (Anderson and Hinthorne 1972) was further developed and applied to zircon geochro- nology by W. Compston and colleagues at the Australian National University (Compston et al. 1984). The instrument developed by the ANU group (SHRIMP, or Sensitive High Resolution Ion Microprobe) and the associated measurement protocols facilitated measurement of Pb/U isotopic ratios within a small region of a single zircon grain, and it proved to be a powerful tool to address complex age structure of multi-component zircons. In the 1990s, laser ablation quadrupole ICP- MS methods came on stream and offered an alternate way to make intra-grain U-Th-Pb isotopic measurements. The advent of double-focusing ICP-MS instruments has offered much better mea- surement precision compared to quadrupole machines, and while still developing, they have yet to make quite the same impact, though many current laboratories are working to close the gap with SIMS methods. There has been an unfortunate tendency for advocates of TIMS on the one hand and SIMS on the other to unfairly criticize each other���s methodology, or at least the ways those methodolo- gies have been applied to zircon geochronology. While this trend has decreased in recent years it is reflected in some of the literature and can give a biased impression of the complementary capabilities of both methods, often ignoring the strengths and weaknesses of the alternative. It is imperative that students of zircon geochronology become more aware of the capabilities of all useful methods, so that when confronted with a problem to solve, they know how to design their measurement experiment. The lack of a comprehensive textbook means we must rely on indi- vidual research and review papers and our own abilities at synthesis to gain a reasonable under- standing of the field. As an alternative, short courses can be very effective at bringing the views of experts together, as this and other volumes (Heaman and Parrish 1991) attempt to do.

Parrish and Noble 184 A great range of geochronological problems can be addressed effectively with either ID-TIMS or intra-grain microbeam techniques (SIMS or LA-ICP-MS). While this is the case for many appli- cations, some problems are best-solved using ID-TIMS methods and vice-versa. This chapter de- scribes TIMS methods and it comments on the strengths and weaknesses of method variations, using examples of zircon dating mainly from the literature. Naturally it will pay special attention to those applications where ID-TIMS analysis is preferred, if not essential, and it will portray the field of U-Th-Pb geochronology using ID-TIMS as having a very bright future in modern earth science. METHODS AND DATA PRESENTATION Background Zircon is a refractory mineral, being very difficult to destroy both in nature where it can survive many cycles of sedimentation, metamorphism and melting, and in the laboratory where it is one of the most difficult of all minerals to dissolve for analysis. The crystal structure favors the incorporation of uranium and thorium in typically modest amounts (10-1000 ppm U, 1-100 ppm Th), and it virtually excludes Pb from its structure during crystallization and many other elements (Speer 1982). Its U/Pb ratio upon formation is therefore extremely high. By the particularly useful quirk of nature, two long-lived radioactive uranium isotopes, 238U and 235U, decay to different isotopes of Pb (206Pb and 207Pb, respectively), while 232Th decays to 208Pb. A fourth Pb isotope, 204Pb has no radioactive parent and its abundance in the earth has effectively not changed with time except for addition from bolide impacts. This isotope system provides the frame- work for U-Th-Pb geochronology using U- and Th-bearing accessory minerals, with zircon being the most frequently used due to its widespread occurrence in rocks of the continental crust. The ID-TIMS U-Pb zircon method relies upon measurement of the isotopic composition of U and Pb. Even decades ago using less sophisticated mass spectrometers than are available today, it was possible to make these isotopic measurements to a precision better than 0.1%. Given the nature of the U-Pb isotopic system with its two uranium isotopes of differing rate of decay, this means that it is theoretically possible to measure the age of a zircon to better than ca. ��1-2 Ma throughout all of geological time, which makes the system unique to all isotopic chronometers. Thus, the method of zircon dating has long been applied to geological problems to study the rates of detailed processes, and it has been especially pivotal to the unraveling of the history of the earth in the Precambrian because of the lack of biostratigraphic control. Evolution of analytical methods The first methods of U-Th-Pb isotopic analysis using ID-TIMS on zircon were conducted in the 1950s (Tilton et al. 1955, 1957 Wetherill 1956) and involved the flux-assisted decomposition of zircon, the use of tracers enriched in 208Pb, 230Th, and 235U, and analysis on custom-built mass spectrometers without computer control. The paper by Tilton et al. (1955) is seminal and includes methodology, the first isotopic ages, and additional analyses of sphene and other minerals. In the 1960s and 1970s, improvements to vacuum systems, computer-controlled data acquisition, the very limited availability of 205Pb tracer (Krogh and Davis 1975), the use of sub-boiling Teflon�� stills for acid distillation (Mattinson 1972) and high temperature decomposition of zircon (Krogh 1973) allowed several laboratories to make important progress in methodology. These improve- ments resulted in significant decrease in contamination levels and sample size. A period of major expansion took place in the 1980s when commercially-available mass spectrometers, new synthesis of 205Pb (Parrish and Krogh 1987), and other laboratory improve- ments including air abrasion (Krogh 1982) and multiple ion collection using mixed detector ar- rays (Roddick et al. 1987) came on stream. Together, these improvements allowed the routine analysis of single zircons containing sub-nanogram quantities of Pb, leading to better data quality, greater

Zircon U-Th-Pb Geochronology by ID���TIMS 185 concordance of zircon U-Pb data, and improved interpretations. Other minerals became part of the mainstream (monazite, allanite, baddeleyite, titanite) and served further to improve the understand- ing of U-Th-Pb systematics (Heaman and Parrish 1991). Important milestones in these methods will be discussed below. Sample selection and preparation. There are two important aspects about zircon that set this mineral apart from all other accessory phases. The first was alluded to earlier and is its refractory nature and the resultant difficulty with which it is fully recycled by dissolution in magmatic or metamorphic systems. The enduring aspect of the mineral has led to the frequent preservation of older components within magmatic or metamorphic grains and the formation and preservation of composite zircon grains with multiple age components, reflected in the Concordia diagram as dis- cordant analyses. Examples of the complexity of zircon are abundant in the literature (e.g., Ashwal et al. 1999, Connelly 2000). The second distinguishing aspect of zircon is the paradoxical ease with which it is capable of losing Pb at relatively low temperatures, a non-diffusive process that is related broadly to accumulated radiation damage (Geisler et al. 2002). It can be shown that there are physico- chemical differences between different zircon crystals, and even between different domains within a single zircon, that result in variable degrees of Pb-loss. In some scientific investigations it is desirable to make measurements on materials without any pre-selection of samples, i.e., rigorously avoiding biasing what is analyzed. Conversely, it turns out that being able to bias what one analyzes by ID-TIMS is extremely important in obtaining high- precision and high-accuracy ages because one does not seek to determine the aver- age age and its standard deviation. Clearly, one does not have to be a rocket scientist to realize that the isotopic im- print of the physical complexity of some zircons is likely to be equally complex, and therefore complex zircons require cautious, and often complex interpreta- tion of the U-Pb systematics. The spec- trum of differing degrees of Pb loss is Figure 1. Collage of zircon photo- micrographs. (A) Two columns of zircons illustrating ranges of quality of crystals, from poor to gem quality these crystals were from beach sand adjacent garnet-rich granulite facies paragneiss and orthogneiss. (B) Two columns of Mesozoic igneous zircons showing a comparison of original euhedral shapes with those following a moderately air abrasion treatment. (C) Secondary electron (SE) image of resorbed igneous zircon. (D) Very round metamorphic zircon from paragneiss. (E) Composite zircon grain formed by the welding together of several smaller grains by a metamorphic overgrowth. (F) Euhedral zircon morphology from a migmatitic leucosome. (G) Metamorphic zircon from paragneiss showing traces of crystal facets. (H) Annedral metamorphic zircon from sillimanite-grade paragneiss showing ���casts��� of minerals around which zircon has grown.

Parrish and Noble 186 not surprising when one considers the wide range of crystal quality (Fig. 1a). Since the first U-Th-Pb zircon measurements of Tilton et al. (1955), it was clear that the two U-Pb decay systems do not always reveal the same age���that is, they are discordant. Furthermore, in many early studies the Th-Pb system appeared to often be somewhat decoupled from the U-Pb system and more difficult to interpret. The simplest explanation was that Pb loss had taken place at some time in the past, and this phenomenon was thought to be associated with the accumulation of radiation damage within the crystal structure. Silver and Deutsch (1963) realized that magnetic susceptibility could be used in some cases to distinguish between relatively high- and low-U zir- cons. Relatively low-U zircons have less radiogenic Pb, but are better suited for analysis because there is less radiation damage and generally less Pb-loss. This latter attribute outweighs the greater difficulty in analyzing low levels of U and Pb on a mass spectrometer and is effective in improving the concordance of analyses, but not eliminating it. Following Silver and Deutsch���s study it was generally accepted that pre-selection of the best crystals from a large population was a necessary component of ID-TIMS work. While the two coupled U-Pb decay systems allowed one to work out the original age of crystallization from discordant zircon analyses (Wetherill 1956), the Th-Pb system often failed to provide additional insight. For these reasons as well as the difficulty of Th mass spectrometry relative to U arising from poorer thermal ionization, the 232Th-208Pb system was rarely measured. In 1982 Krogh (1982) made another major improvement in ID-TIMS geochronology whereby a method of air abrasion designed to remove the outer portion of grains was presented. The notion was that the Pb loss was likely to be more abundant in the outer portions of grains exposed over eons to fluids. Krogh showed in this and later papers how to reduce or eliminate discordance (arising mainly from low temperature Pb loss) with air abrasion of high quality, crack-free, visu- ally clear zircons. Figure 1b shows examples of un-abraded and abraded crystals, the latter giving results consistent with closed system behavior. This abrasion technique has been available now for 20 years, having become standard proce- dure in nearly all ID-TIMS zircon geochronology laboratories, and the improvement in results is undeniably impressive. This is one of if not the most important technique development in zircon geochronology, in spite of the improvements in computer and mass spectrometer wizardry. The attention to detail of the sample characteristics���the quality of grains and the quality of the abra- sion���made a huge difference to results. While the presence of internal multi-age complexity can be daunting to sort out, the careful preparation of high quality abraded grains for analysis allows their complex compositions to be established, notwithstanding subsequent Pb loss. Air abrasion has been a key procedure in allow- ing ID-TIMS to address the dating of complex crystals, a task that is clearly well suited to alterna- tive micro-beam techniques that can sample single components within grains. A further development in sample preparation took place by Mattinson (1994) whereby zircon was subjected to an aggressive HF���HNO3 acid-leaching procedure to partially dissolve relatively soluble portions of grains. The enhanced solubility was thought to be related to radiation damage and other crystal imperfections, and has a relationship to uranium and thorium concentration and of course age. Experiments clearly showed in many cases that the first stages of leachate contained evidence of more soluble radiogenic Pb, and that if further steps were performed, the residue was likely to be more concordant than the original crystal. Several recent detailed studies showed that the procedure could in certain cases produce artifacts (Corfu 2000, Davis and Krogh 2000, Chen et al. 2002). Sample-tracer equilibration has been shown to be a problem in the dissolution of fluorides from the leachate, and in some samples 206Pb appears to be more soluble than 207Pb, potentially linked to the variable energies of decay for each radioactive parent isotope. The HF leaching proce- dure is a less mature alternate preparation technique than air abrasion it has been only occasionally adopted (Mundil et al. 2000). An example of this will be mentioned below. Finally, an example of a thoughtful, but common sense approach has again come from Krogh

Zircon U-Th-Pb Geochronology by ID���TIMS 187 (1998) in an application to the very old rocks of West Greenland that have seen so much attention from isotope geochemists and geochronologists (e.g., Whitehouse et al. 1999, 2001 Nutman et al. 2000, 2001). In this procedure, he first used etching (NaOH HF can also be used) to induce partial dissolution in order to ���screen��� grains for their relative damage from radiation accumulation. The least affected grains were then selected and air abraded to fully remove the leached portions. These abraded grains then yielded mainly concordant data that fell into several age clusters, rather than the more diffuse and discordant pattern of ages of many previous studies. This appears to be a particularly useful approach that should be widely applicable to studies of older rocks. The study of zircon grains using optical and/or scanning electron (back-scattered, secondary, and cathodoluminescence) microscopy is paramount to a correct interpretation of U-Pb analyses (Vavra 1990, Paterson et al. 1992, Hanchar and Miller 1995). Many studies have highlighted the variable inter- nal and external morphological characteristics produced during geological process. These include mag- matic growth and resorption, metamorphic growth and resorption in the solid state, occlusion of grains within other crystals, reaction textures and intergrowths with other phases, and the partial dissolution of external surfaces. Magmatic morphologies in themselves are quite varied and in part indicative of geo- logical environments, residence time, and temperature of crystallization. This subject is beyond the scope of this chapter, but Figure 1 provides the reader with a sampling of the variety of textures. Needless to say, the geological setting of the sample (state of deformation and metamorphism, for example) is equally vital to linking the isotopic measurements to a sensible interpretation. Mass spectrometry Thermal ionization mass spectrometers have been used for making Pb and U isotope mea- surements for over 50 years. The long-lived viability of this analytical technique is the product of a number of key attributes. These are: reasonable ionization efficiency, a simple mass spectrum, excellent signal to noise characteristics, relatively small-order mass fractionation, negligible Pb and U contamination of samples by the instrument, and the lack of reliance on mineral standards in the calibration process. Remarkably, modern instruments are all founded upon the design prin- ciples of Alfred Nier���s early mass spectrometers (e.g., Nier 1940). Nevertheless, significant ad- vances have been necessary in achieving the current state-of-the-art performance. A crucial advance for U-Pb geochronology has been the use of silica gel and phosphoric acid as an exceptionally efficient and stable Pb+ (and UO2+) ion emitter (Cameron et al. 1968). The silica gel method has only evolved modestly over the past 35 years with most of the improvements relating to lowering of U and Pb contents in the gel and increasing ionization efficiency by up to an order of magnitude (Gerstenberger and Hasse 1996). Emission stability is crucial because it facilitated the easy use of peak-switching of U and Pb isotopes on single collector mass spectrometers, a practice still widely applied. The advent of commercially available mass spectrometers with multiple arrays of fara- day detectors in ca. 1981-82, usually supplemented by an axial electron multiplier detector, per- mitted the more rapid acquisition of isotopic ratios of increased precision, with a more efficient duty cycle (Roddick et al. 1987). Some laboratories also routinely used simultaneous collection of isotopes using both faraday and electron multiplier detectors, allowing large increases in data production. For several laboratories these improvements, in conjunction with miniaturization of chemistry and using a mixed 205Pb-235U��233U tracer resulted in an increase of productivity by up to 5-10 fold within a few years. The differences between the measurement state of the art now and that in 1985-90 are relatively unimportant, mainly being a reduction in Pb blank levels from 5-20 picograms to less than 1 picogram in the chemical procedure. Advances in chemical procedures and tracers Of Teflon and tracers. The modern era of U-Pb zircon dating was ushered in by developments of T. Krogh and J. Mattinson at the Carnegie Institute of Washington. Mattinson adapted FEP and TFE Teflon��-ware to the distillation and production of ultra-pure acids that resulted in a large de-