238U/235U Systematics in Terrestrial Uranium-Bearing Minerals

See allHide authors and affiliations

Science  30 Mar 2012:
Vol. 335, Issue 6076, pp. 1610-1614
DOI: 10.1126/science.1215507


The present-day 238U/235U ratio has fundamental implications for uranium-lead geochronology and cosmochronology. A value of 137.88 has previously been considered invariant and has been used without uncertainty to calculate terrestrial mineral ages. We report high-precision 238U/235U measurements for a suite of uranium-bearing minerals from 58 samples representing a diverse range of lithologies. This data set exhibits a range in 238U/235U values of >5 per mil, with no clear relation to any petrogenetic, secular, or regional trends. Variation between comagmatic minerals suggests that 238U/235U fractionation processes operate at magmatic temperatures. A mean 238U/235U value of 137.818 ± 0.045 (2σ) in zircon samples reflects the average uranium isotopic composition and variability of terrestrial zircon. This distribution is broadly representative of the average crustal and “bulk Earth” 238U/235U composition.

The uranium-lead (U-Pb) system is widely used as an isotopic chronometer for geological and meteoritic materials that are less than 1 million to greater than 4.5 billion years old. This system is particularly useful because it has two long-lived isotopes, 238U and 235U, which decay at different rates to 206Pb and 207Pb, respectively, permitting the evaluation of closed-system behavior, and because both decay constants have been determined to relatively high precision (1, 2). Daughter isotope determinations from the two decay systems may also be combined to calculate a 207Pb-206Pb date in concert with an assumed or measured present-day 238U/235U ratio. With recent advances in sample preparation, isotope ratio mass spectrometry, and gravimetric calibration of tracers for isotope dilution methods, the precision of an individual U-Pb or Pb-Pb age determination can exceed 0.1% (2, 3). The U-Pb chronometer has been used to improve the accuracy of other radioisotopic systems such as 40Ar/39Ar (4), Lu-Hf (57), Rb-Sr (8), and Re-Os (9), and to anchor extinct nuclide cosmochronometers that are used to place early solar system events in sequence. Thus, the U-Pb system has far-reaching impacts on the determination of absolute time in geological and meteoric materials.

Historically, the kinetic fractionation of U isotopes was expected to be small because of their high mass; until recently, the present-day 238U/235U ratio of all natural materials was considered invariant. In geo- and cosmochronology, a 238U/235U value equal to 137.88 has been used almost exclusively for the past 35 years (10) and is based on studies of magmatic and sedimentary uranium ore deposits (11). As published, this presumed invariant ratio and its references cannot be traced back to the International System (SI) of Units (12). More recently, IUPAC (13) recommended a value of 137.80 from the analysis of six natural ore deposits (14), confirmed by high-precision isotope ratio analyses using the IRMM-3636 233U-236U double spike (15, 16) whose isotopic composition is traceable to SI units. The invariance of the present-day 238U/235U ratio has been brought into question by studies that have demonstrated U isotopic fractionation in terrestrial materials (1720). Such fractionation occurs during oxidation-reduction reactions (UVI to/from UIV), coordination change during adsorption, or leaching, and is due to thermodynamic or nuclear field shift effects (2123). In extraterrestrial materials, excess 235U may result from α decay of the short-lived 247Cm (24), which has been detected in carbonaceous chondrites and their calcium- and aluminum-rich inclusions (25). Recent cosmochronology studies have highlighted the need for coupled 238U/235U and 207Pb/206Pb data sets in order to determine accurate 207Pb-206Pb dates. Thus, it is crucial to reevaluate (26) the range of natural variation of 238U/235U ratios in U-bearing minerals commonly analyzed for U-Pb age determinations.

We performed 141 238U/235U determinations on a suite of 58 samples of U-bearing accessory minerals that are used for U-Pb geochronology (zircon, monazite, apatite, titanite, uraninite, xenotime, and baddeleyite), spanning the Quaternary to the Eoarchean and covering a diverse range of igneous and metamorphic petrogenetic settings and geographic locations (27). These data are traceable to SI units because they were measured using a gravimetrically calibrated 233U-236U tracer (16); measurement uncertainties are on the order of 70 parts per million (ppm) or better.

Our data set has a >5.4 per mil (‰) range in 238U/235U (Fig. 1). The lowest measured value is 137.743 from the pegmatite-derived Moacyr monazite, and the highest is 138.490 for the Fish Canyon Tuff titanite, erupted in a large-volume silicic ash flow. Two other samples yield 238U/235U values greater than 138: BLR-1 titanite (138.068) and Table Cape zircon (138.283). The Miocene Table Cape from Tasmania may be derived from a unique, isotopically heavy reservoir more subtly expressed by Pliocene Bullenmerri (137.862) and Miocene Mornington (137.855) from Victoria, Australia, at the higher end of the main zircon 238U/235U population. These three zircon samples are likely to be mantle-derived and are sourced from regional alkaline volcanic fields. Six monazite samples have 238U/235U values from 137.743 to 137.856. Most monazite samples are sourced from pegmatites, a lithology with the potential to contain high proportions of low-temperature redox-fractionated protoliths. Resolvable 238U/235U variation between different accessory phases from the same sample—such as 01RP1 zircon and monazite (65 ppm), Mud Tank zircon and apatite (225 ppm), and Fish Canyon Tuff zircon and titanite (4.78‰)—indicates crystal-chemical and/or petrogenetic control on 238U/235U fractionation processes that operate at magmatic temperatures.

Fig. 1

(A) 238U/235U mineral summary plot including the 44 samples used to define our recommended zircon composition (represented by the yellow band). Solid and open boxes for each sample represent 2σ measured and total uncertainties, respectively. (B) One additional zircon sample and two additional titanite samples shown relative to the data in (A), highlighting the total range of sample compositions observed. Sample solid boxes represent 2σ total uncertainties.

Of 45 zircon 238U/235U measurements, 44 of them are within a range of ~1‰, from 137.772 (Zim265) to 137.908 (168952). There is resolvable variation between samples, but no first-order correlation with age, petrogenetic setting, or geographic location. All five samples of uraninite, apatite, xenotime, and baddeleyite fall within the compositional range of zircon. The resolvable 238U/235U differences between samples could arise from multiple processes, including incorporation of uranium from a protolith with fractionated 238U/235U into parental magma and isotopic fractionation associated with magmatic or mineral crystallization processes. Samples of similar genetic affinity typically show agreement in 238U/235U values, suggesting isotopic homogenization within some magmatic systems (e.g., zircon from Yellowstone’s Lava Creek, Mesa Falls, and Huckleberry Ridge ash-flow tuffs; Hungarian volcanic tuffs 97JP32 and 97JP33; Californian tonalites 81P-131 and 81P-209; Ontarian pegmatites Bancroft and Cardiff; Minnesotan rhyolite and anorthosites MS99-30, FC1, and AS3; Greenland tonalites 492118 and 492120; and monazite from British Columbian pegmatites FC-1 and 01RP1).

Of the zircon samples measured, 44 of 45 define an approximately normally distributed population with a mean of 137.818 and standard deviation of 0.022, with population parameters calculated using (28), which corrects for the expected additional dispersion from analytical uncertainties. We propose that this average zircon value and its associated variability (137.818 ± 0.045/0.050, 2σ), which is traceable to the SI system of units, is applicable for the majority of U-Pb determinations and, in the absence of an independently determined 238U/235U value, should be adopted for future use in U-Pb geochronology of zircon. The first uncertainty reported reflects the variability found in nature; the second additionally incorporates systematic uncertainties in the isotopic composition of the tracer. Other phases, such as monazite and titanite, require further assessment of their 238U/235U variability.

Adoption of the average 238U/235Uzircon value of 137.818 ± 0.045 for use in zircon geochronology will decrease 207Pb-206Pb, 207Pb-235U, and 206Pb-238U dates relative to those calculated using the conventional 238U/235U value of 137.88 (12, 18). For 207Pb-206Pb dates, the 238U/235U ratio is implicit in the age equation and the magnitude of the difference is largest, changing gradually from ~1 million years for samples dated 100 million years ago (Ma) to ~700,000 years for samples dated 4 billion years ago (Ga) (Fig. 2A). The observed variability in 238U/235Uzircon may limit precision for >1 Ga zircon samples with no independent 238U/235U constraint. For 207Pb-235U and 206Pb-238U dates (Fig. 2, B and C), the mineral 238U/235U is used in tracer subtraction and fractionation correction calculations for tracers enriched in 235U that are commonly used in high-precision U-Pb geochronology (supplementary text). For typical sample/tracer 238U/235U ratios close to unity, the biases are <500,000 years for 207Pb-235U dates and <30,000 years for 206Pb-238U dates younger than 4.4 Ga. For Phanerozoic zircons, the change in 206Pb-238U dates is <4000 years.

Fig. 2

(A to C) Plots of absolute differences between dates calculated with the consensus 238U/235U value of 137.88 and the newly defined value of 137.818 ± 0.045 for (A) zircon 207Pb-206Pb dates, (B) zircon 207Pb-235U dates, and (C) zircon 206Pb-238U dates. Gray bands represent the 2σ uncertainty from the variability in zircon 238U/235U determined in this study. The difference is calculated by subtracting the Pb-Pb or U-Pb date calculated using 238U/235U = 137.818 ± 0.045 from the Pb-Pb or U-Pb date calculated using the 238U/235U value of 137.88. U-Pb dates are modeled using typical analytical parameters (sample/tracer 238U/235U = 1) to illustrate the magnitude of differences.

High-precision U-Pb isotope analyses of closed-system zircon and xenotime have also been exploited to derive a more precise 235U decay constant (λ235U) value (2, 3, 29). In this approach, the systematic bias between 206Pb-238U dates and 207Pb-235U dates is minimized by solving for a new value of λ235U relative to the more precisely determined λ238U (1). These studies have used an assumed 238U/235U value of 137.88 (2, 3, 29). Mattinson (2) discussed the effects of intermediate daughter disequilibrium, mass isotopic fractionation, tracer calibrations, and 238U/235U on the accuracy and precision of U-Pb analyses. Our data set includes a subset of samples dated in these previous studies and allows us to better evaluate the impact of a more accurate 238U/235U on uranium decay constant intercalibration. Accepting the published U-Pb data (2, 3) and the samples’ unique 238U/235Uzircon values determined in this study yields a recalculated λ235U = 0.98531 per billion years (fig. S10), intermediate between the Jaffey et al. (1) counting experiment value and the closed-system U-Pb reevaluations using 238U/235U = 137.88 (2, 3). However, a robust λ235U can only be determined with U-Pb analyses using a tracer calibration that is traceable to SI units and free of other potential sources of bias, so we refrain from suggesting that this value be adopted at present and urge caution in abandoning the Jaffey et al. (1) λ235U determination until such a data set has been generated and evaluated.

An emerging 238U/235U data set for a wide range of rocks, minerals, and meteorites is now available (17, 18, 25, 3033) (Fig. 3). Given that natural 238U/235U variation has been demonstrated up to ~0.13% (17, 18), it might be expected that a corresponding variation be observed in the U-bearing mineral data set, because 238U/235U fractionated material from low-temperature environments is incorporated into higher-temperature systems through crustal recycling processes. A first-order observation from the compiled 238U/235U data is that materials formed in near-surface environments (e.g., chemical precipitates) record a wider range than crustal rocks and minerals formed in higher-temperature magmatic environments (17, 18). This suggests that uranium in magmatic and derived crustal reservoirs (e.g., siliciclastic sediments) is isotopically well mixed relative to uranium in materials formed in near-surface environments, and that the low-temperature materials with highly fractionated 238U/235U constitute volumetrically minor reservoirs that are continually and efficiently homogenized via crustal recycling processes. Second, modern seawater and Quaternary seawater precipitates are systematically lower than the “bulk Earth” 238U/235U composition, indicating 235U enrichment in the marine reservoir. Seawater enrichment in 234U (34) by ~147‰ relative to radioactive secular equilibrium is a well-known consequence of radioactive α-recoil processes and the preferential release of the non–lattice-bound 234U daughter nuclide into the hydrological environment (35, 36). Previous studies demonstrated a broad positive correlation between 234U and 235U depletion in near-surface environments (17), but an α-recoil–related mechanism cannot account for 235U/238U fractionation, as both are lattice-bound. Zircon acid leaching experiments carried out in this study also recorded a systematic enrichment in 235U in the leachate (27), which suggests preferential leaching of lattice-bound 235U, and similar fractionation has been detected in euxenite leaching experiments (17). By analogy, we suggest leaching of lattice-bound 235U during long-term chemical weathering of exposed crustal rocks as a viable mechanism to explain 235U enrichment in seawater.

Fig. 3

Compilation of 238U/235U data obtained on a wide variety of geological and extraterrestrial materials. The “bulk Earth” field (yellow band) is based on the terrestrial/high-temperature data set, consistent with eucrites and ordinary chondrites. Seawater and related precipitates show a systematic enrichment in 235U relative to the “bulk Earth” field. Data are from literature sources (17, 18, 25, 3033, 3941) and this study (zircon and other U-bearing minerals).

Uranium-bearing accessory minerals from a wide range of crustal and mantle-derived rock types record a restricted (0.07%) range of 238U/235U values that encompasses nearly all published 238U/235U values determined on high-temperature (i.e., magmatic) rocks and minerals including granites, dunite, and basalts (Fig. 3). The overlap in 238U/235U values for crust and upper mantle–derived lithologies indicates no resolvable fractionation between the terrestrial reservoirs sampled. Furthermore, despite 238U/235U values of meteoritic material recording excess 235U derived from extant 247Cm (24, 25, 37), ordinary chondrites, eucrites, and the upper limits for calcium/aluminum inclusions and carbonaceous chondrites overlap with the field delineated by terrestrial crust and mantle materials (30, 32, 33). This agreement suggests that a uniform 238U/235U was achieved relatively early during planetary accretion and that U isotope compositions in the high-temperature terrestrial crust and upper mantle are also likely to apply to the lower mantle, thereby defining the bulk silicate Earth isotope composition. In light of the agreement of terrestrial and meteoritic isotope compositions and current Earth accretion models (38), this average 238U/235U value of 137.818 ± 0.050 would also represent the isotope composition of “bulk Earth.”

Supporting Online Material

Materials and Methods

Supplementary Text

Figs. S1 to S10

Tables S1 to S6

References (4282)

Databases S1 to S5

References and Notes

  1. See supporting material on Science Online.
  2. Contact Dan Condon ( for data.
  3. Acknowledgments: Supported by the UK Natural Environment Research Council (NERC) (fellowship NE/C517909/1, NERC Isotope Geosciences Facilities Steering Committee award IP/1028/0508, and recurrent support to the NERC Isotope Geoscience Laboratory), the European Community’s Seventh Framework Programme (FP7/2007-2013) grant agreement 215458, and NSF award EAR 0451802 (the EARTHTIME project). We thank N. Atkinson, M. Horstwood, and N. Roberts for assistance in the laboratory; J. Aleinikoff, J. Boyce, D. Chew, P. Fitzgerald, D. Kimbrough, C. Kirkland, J. Mattinson, R. Parrish, P. Renne, N. Reyner, G. Rossmann, M. Schmitz, K. Sircombe, M. Wingate, and J. Woodhead for provision of samples; GEUS (Denmark) for permission to publish data on 492118/20; S. Bowring, R. Parrish, B. Schoene, M. Schmitz, J. Mattinson, and others for discussions; and three reviewers for helpful comments.
View Abstract

Navigate This Article