Geochimica et Cosmochimica Acta, Vol. 62, No. 21/22, pp. 3437–3452, 1998 Copyright © 1998 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/98 $19.00 1 .00 Pergamon PII S0016-7037(98)00255-5 Uranium-thorium-protactinium dating systematics H. CHENG,*,1 R. LAWRENCE EDWARDS,1 M. T. MURRELL,2 and T. M. BENJAMIN2 1 Minnesota Isotope Laboratory, Department of Geology and Geophysics, University of Minnesota, Minneapolis, Minnesota 55455, USA 2 Chemical Science and Technology Division, J514, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA (Received October 21, 1997; accepted in revised form July 24, 1998) Abstract—With precise 234U, 230Th, and 231Pa data available, 230Th and 231Pa ages can now be tested rigorously for concordancy. If the material is not concordant, the isotopic characteristics of this material may be examined in some detail. Here, models similar to those used to describe the U-Pb system are evaluated for use in U-Th-Pa studies, for the case in which initial 230Th and 231Pa concentrations are effectively zero. The systematics of concordia plots in relation to models of variation in d234U, episodic U loss or gain, continuous U loss or gain, and continuous 234U, 230Th and 231Pa gain or loss are considered for the case in which initial U concentration is significant (for example, in many carbonate deposits). We also examine linear U uptake models for the case in which initial U concentration is effectively zero (for example, in teeth and bones). Such models should prove useful in interpreting data from materials that have behaved as open-systems. In particular, these models may help constrain the nature of diagenetic processes, and in some situations it may be possible to determine or constrain true ages with materials that have behaved as open-systems. Copyright © 1998 Elsevier Science Ltd ionization mass spectrometric (TIMS) methods for the measurement of 234U (Chen et al., 1986; 234U half-life 5 244,500 y, Lounsbury and Durham, 1971, DeBievre et al., 1971) and 230 Th (Edwards et al., 1987a,b; Edwards, 1988) were developed, leading to significant increases in analytical precision in 230 Th dating. The use of TIMS for 230Th dating has resulted in the establishment of precise chronologies for sea level changes, continental climate shifts, and an absolute 14C calibration (e.g., Edwards et al., 1987a,b, 1993; Bard et al., 1990a,b; Ludwig et al., 1992; Stein et al., 1993; Gallup et al., 1994). In addition, precise initial d234U values (([234U/238U] 2 1) 3 1000, where brackets indicate activity ratios) of marine samples have been used to test for diagenetic alteration based on the assumption that the d234U value of seawater is constant with location and through time. However, 230Th studies still suffer from uncertainty about possible open-system behavior. For example, many old coral samples have initial d234U values higher than the modern marine value. Thus, it would be desirable to have another chronometer available to test for age concordancy. The 231 Pa method, as first suggested by Sackett (1958), is a likely choice, however, this method does not yield the required precision with decay-counting techniques. We have, therefore, begun using TIMS techniques for 231Pa measurements in carbonates (Cheng et al., 1996; Edwards et al., 1997) using a procedure originally developed at Los Alamos (Pickett et al., 1994). This method provides data with precisions comparable to the 230Th TIMS method, which makes it possible to use the U-Th-Pa system to test for age concordancy and diagenetic processes. Up until now, there has been little motivation to establish the basic systematics for U-Th-Pa dating with the same rigor as seen for the U/Pb system, although these dating systems were both developed during the 1950s to 1960s and share many similarities. A number of previous studies have discussed UTh-Pa concordancy and applications to alpha-counting data from coral, phosphorite, and U ore samples (Allégre, 1964; Ku, 1968; Szabo and Vedder, 1971; Ku et al., 1974; Szabo, 1979; 1. INTRODUCTION 230 Th is the second longest-lived intermediate daughter nuclide (half-life t1/2 5 75383 years, Meadows et al., 1980) in the 238U decay series, and 231Pa is the longest-lived intermediate daughter nuclide (t1/2 5 32760 y, Robert et al., 1969) in the 235U decay series. Because U, Th, and Pa have different valences (typically 41 and 61 for U, 41 for Th, and 51 for Pa) and chemical affinities, significant fractionation can occur between these elements during various geological processes. For example, Th and Pa have very low solubilities compared to U in aqueous systems and are rapidly removed from fluids due to their tendency to hydrolyze and sorb onto sinking particulates (Gascoyne, 1992). Thus, minerals precipitated from waters can have very low initial 230Th and 231Pa contents compared to their parent U. The following return to equilibrium can then be used as a measure of time. As the two daughters have different characteristic timescales of ingrowth, they can be used as independent measures of time, over timescales as long as 550 ka (230Th) and 200 ka (231Pa). U-Th-Pa systematics have been used to obtain mineral formation ages since the 1950s. Barnes et al. (1956) measured coralline 230Th and 238U by alpha-counting. Sackett (1958) first suggested the use of both 231Pa and 230Th methods to date carbonate samples. During the 1960s, the determinations of 230 Th, 231Pa, 234U, and 238U using alpha-counting methods were applied to numerous dating applications (Allégre, 1964; Thurber et al., 1965; Kaufman and Broecker, 1965; Broecker and Thurber, 1965; Sakanoue et al., 1967; Rosholt, 1967; Ku, 1968; Szabo and Rosholt, 1969). In the 1970s and early 1980s, the application of such techniques slowed because the precision of the alpha-counting methods was insufficient to resolve many dating problems, including the evaluation of the diagenetic alteration of carbonate samples. In the late 1980s, thermal *Author to whom correspondence ([email protected]). should be addressed 3437 3438 H. Cheng et al. Roe et al., 1982; Veeh, 1982; Birch et al., 1983; Burnett and Kim, 1986; O’Brien et al., 1986; Veeh and France, 1988; Kaufman and Ku, 1989; Kaufman et al., 1995). In general, the differing effects of diagenesis were not considered in the detail that is now necessary, because, in most cases, the available precision was insufficient to resolve these differences. Before this work, Allegré (1964) studied the equations for episodic U gain for the case where initial [234U/238U] was unity and Kaufmann and Ku (1989) and Kaufmann et al. (1995) studied the case of continuous U gain and loss, also for the case in which initial [234U/238U] was unity. In both cases these models were used to interpret data from U-rich minerals. In a third study, Rosholt (1967) found, and later confirmed (Szabo and Rosholt, 1969), excess 231Pa relative to 235U in a molluscan shell. They proposed a open-system model with two reservoirs of U to explain this 231Pa excess (see Kaufman et al. (1971) for a critical evaluation of this model). Possible open-system behavior of U and Th isotopes has been evaluated with TIMS 230Th data (Edwards et al., 1987a,b; Edwards, 1988; Chen et al., 1991; Bard et al., 1991; Hamelin et al., 1991; Henderson et al., 1993; Gallup et al., 1994). With TIMS 231Pa data now available (Pickett et al., 1994; Cheng et al., 1996; Goldstein et al., 1996; Holden et al., 1996; Pickett and Murrell, 1997; Edwards et al., 1997), it is critical to examine U-Th-Pa systematics in more detail. In this paper, we present several basic models for the U-Th-Pa system using [231Pa/235U] vs. [230Th/238U] or [231Pa/235U] vs. [230Th/234U] diagrams which are analogous to 206Pb/238U vs. 207Pb/235U concordia diagrams. This framework can potentially improve our understanding of U-series dating results and lead to better constraints for Quaternary chronology. It may also help us understand the nature of diagenetic processes in various settings. activity is defined as ai 5 liNi, where li and Ni are the decay constant and the number of atoms for nuclide i, respectively. 2.2. Concordia Diagrams Given the above conditions, the basic age equations for a closed U-Th-Pa system are as follows (modified from Bateman, 1910; see also Ivanovich and Harmon, 1992 for a general discussion of these equations): F G 231 Pa 5 1 2 e2l231xt (231Pa age equation) U (1) 235 F G 234 U 5 11 U 238 SF G D 234 238 U U 2 1 0 3 e 2l2343t ~ 234U/238U age equation) F G 230 Th 5 12 e2l2303t 1 U 238 SF G D S 234 238 U U 21 3 D l 230 l 230 2 l 234 3 ~1 2 e2(l230-l234)3 t)(230Th/238U age equation) F G F G S S D 230 Th 5 U 234 238 234 (2) (3) S F GD U 3 1 2 e2l2303t)1 1 2 U 238 234 U U l230 3 ~1 2 e2(l2302l234)3t! l2302l234 3 ~ 230Th/234U age equation! (4) 2.1. Assumptions and Conventions YF F G F G SF G D S G In the following models, a number of initial conditions and conventions are adopted. (1) Initial 230Th and 231Pa are assumed to be zero. This is an important assumption that simplifies our theoretical treatment but also limits the use of our models to materials for which this assumption is valid. Studies of initial 231Pa levels, how one might correct for initial 231Pa, relationships among initial 231Pa, initial 230Th, and 232Th concentrations, as well as the theoretical basis for interpreting U/Th/Pa isotopic data for materials with significant initial 230 Th and 231Pa represent a class of problems which remain to be studied and are beyond the scope of this contribution. Although many materials may contain significant initial 230Th or 231Pa (Kaufman and Broecker 1965; Ku et al., 1979; Schwarcz, 1980), many calcite, aragonite, phosphorite, and tooth or bone samples have been shown to contain trivial initial 230 Th or 231Pa (e.g., Edwards, 1988; Edwards et al., 1988; Cheng et al., 1996; Goldstein et al., 1996; Edwards et al., 1997). Our analysis aims to aid in the interpretation of data from the latter types of materials. (2) 231Th, 234Th and 234Pa are assumed to be in secular equilibrium at the time of sample formation. (3)li 2 l238 5 li; li 2 l235 5 li; e2l238t 5 e2l235t 5 1, where li is the decay constant for nuclide i and t is time. (4) Ratios in square brackets are activity ratios where Where the subscript 0 denotes initial values. From these equations, it is apparent that the U-Th-Pa system is similar in many ways to the U-Pb system. In both cases, the parents of the two decay schemes are U. Although the two daughters, 231Pa and 230Th, are not the same element as in the U-Pb system (daughters both Pb), they have similar chemical properties in aqueous systems. It follows that in addition to 231 Pa and 230Th ages, it is also possible to determine 231Pa/ 230 Th ages (Eqn. 5) with systematics analogous to 207Pb/206Pb ages for the U/Pb system. A significant difference between the U-Th-Pa and the U-Pb systems is that the temporal change in the 230Th/238U and 230 Th/234U values depends on the initial d234U values of the sample. Thus, different concordia curves exist for different initial d234U values. Figure 1 gives [231Pa/235U] vs. [230Th/ 234 U] and [231Pa/235U] vs. [230Th/238U] concordia diagrams for different initial d234U values. In the [231Pa/235U] vs. [230Th/ 238 U] diagram, the differences in the concordia curves are more significant for different d234U values. This is because 238U is essentially constant over these timescales and higher 234U/238U ratios result in a relatively greater change in [230Th/238U] with time (Fig. 1a). In contrast, the [231Pa/235U] vs. [230Th/234U] diagram displays smaller differences between curves. This is 2. MODELS 231 235 230 238 Pa 5 Th U 3 ~1 2 e2l2313t! U 234 1 238 ~1 2 e2l2303t! D U l 230 21 3 U l 230 2 l 234 3 ~1 2 e2(l2302l234)3t! (231Pa/230Th age equation) (5) U-Th-Pa dating 3439 Fig. 1. [230Th/238U] vs. [231Pa/235U] (a) and [230Th/234U] vs. [231Pa/235U] (b) concordia diagrams using half-lives of 75,383 years and 32,760 years for 230Th and 231Pa, respectively. The curves describe the evolution of the U-230Th-231Pa system with initial activity ratios of [230Th/238U] (or [230Th/234U]) 5 0 and [231Pa/235U] 5 0, contoured with different initial d234U values ranging from 0 to 1500. The horizontal lines are isochrons and the associated numbers indicate the ages in thousands of years (ka). Diagram (a) is much more sensitive to initial d234U values than diagram (b). For evaluating variations in d234U, diagram (a) may be more useful. For minimizing the effect of variations in d234U on age determinations, diagram (b) may prove more useful. because the more similar half-lives for 234U and 230Th result in smaller changes in the 230Th/234U ratios (Fig. 1b). Both diagrams have distinct advantages. For example, if the effects on d234U due to diagenesis are of interest, Fig. 1a will maximize the sensitivity to variations in d234U. Alternatively, in order to minimize the effect of initial d234U values on age determinations, Fig. 1b may prove useful. It will allow for age estimates of marine samples using a single concordia curve, because the initial d234U value for marine samples should be close to 150, and the possible variations are indistinguishable in this diagram. Uncertainties in the decay constants of 230Th and 231Pa lead to uncertainties in the location of concordia. Direct measurements of the decay constants currently yield the following 2s fractional errors: for 230Th, 68‰ (Meadows et al., 1980), and for 231Pa, 67‰ (Robert et al., 1969). Recent measurements on materials assumed to be in secular equilibrium (Cheng et al., 1997) give a 230Th decay constant within error of that reported by Meadows et al. (1980), but constrain the error to 63‰. The propagation of errors in decay constants depends on the types of standards used for each nuclide. For instance, if gravimetric U and Th standards are used, the fractional error in 230Th age due to error in the 230Th decay constant is less than the fractional error in the decay constant for ages significantly less than 105 y, but greater than the fractional error in the decay constant for ages significantly larger than 105 y (Edwards et al., 1987). However, if materials assumed to be in secular equilibrium are used for U and Th standards, the fractional error in age due to error in the 230Th decay constant is similar to the fractional error in the decay constant regardless of age (Ludwig et al., 1992). For U and Th, both gravimetric and secular equilibrium standards are used by the scientific community today. However, for Pa only secular equilibrium standards are used as, to the best of our knowledge, gravimetric Pa standards do not exist. For this type of standardization, the fractional error in 231Pa age due to error in the 231Pa decay constant is exactly equal to the fractional error in the decay constant regardless of age. In Fig. 2, the thickness of the curve reflects the maximum uncertainty in the position of the concordia curves due to the uncertainties in the decay constants of 230Th and 231Pa, given that analytical measurements of both abscissa and ordinate are based on secular equilibrium standards. In the y-direction ([231Pa/235U]), the error in the position of concordia varies from 67‰ (the fractional error in the 231Pa decay constant) at zero age to less than 62‰ at .100 ka. In the x-direction ([230Th/238U] or [230Th/234U]) the error in the position of concordia ranges from 63‰ (the fractional error in the 230Th decay constant) at zero age to less than 61‰ at .250 ka. In both abscissa and ordinate, the error in the position of concordia approaches zero as age approaches infinity and as the activity ratios approach unity. Figure 3 is analogous to Fig. 2, except that the error in the x-direction is calculated assuming standardization to gravimetric Th and U standards. As the standardization is essentially to a known atomic ratio of Th to U (as opposed to a known activity ratio of Th to U as illustrated in Fig. 2), we use atomic ratios on the abscissa of Fig. 3. In these diagrams, the uncertainties in the position of concordia in the ordinate ([231Pa/ 235 U]) are as same as in Fig. 2. In the abscissa, uncertainties in the position of concordia, increase from less than 1‰ at ages ,60 ka to 3‰ at ages . 500 ka. The error in the location of concordia due to the uncertainty in the 230Th decay constant vanishes as age approaches zero. As age approaches infinity, 3440 H. Cheng et al. Fig. 2. [230Th/238U] vs. [231Pa/235U] (a) and [230Th/234U] vs. [231Pa/235U] (b) concordia diagrams similar to Figs. 1a and 1b, showing the effect of the half-life uncertainties for 230Th and 231Pa. The initial d234U is assumed to be 150. The thickness in the curve represents the maximum variation in the concordia curve positions due to the half-life uncertainties. The white circles around 50 ka in (a) and (b) represent the uncertainty at this age caused by the uncertainties in the half-lives of 230Th (75383 years with 2s error of 63‰) and 231Pa (32,760 years with 2s error of 67 ‰). 2301, 2311, and 2302, 2312 indicate 230Th and 231Pa half-lives 3‰ and 7‰ higher than the nominal values, and 230Th and 231Pa half-lives 3‰ and 7‰ lower than the nominal values, respectively. For example, using 2301 and 2312, the concordia curve will be shifted to the upper left; using 2302 and 2311, the curve will be shifted to the lower right. When 2301 and 2311 or 2302 and 2312 are used, the position of concordia curve is close to the position of a curve using the nominal 230Th half-life of 75383 y and 231 Pa half-life of 32760 y; however, ages along the curve will be shifted to the lower left or the upper right, respectively. The arrows indicate the direction of these variations. The uncertainty of the concordia positions in ordinate ([231Pa/235U]) decreases from 7‰ (the fractional error in the 231Pa decay constant) at zero age to less than 2‰ for ages .100 ka, and in abscissa ([230Th/238U] or [230Th/234U]), from 3‰ (the fractional error in the 230Th decay constant) at zero age to less than 1‰ for ages .250 ka. the fractional error in the location of concordia in the x-direction approaches the fractional error in the 230Th decay constant. As the error in the position of concordia as well as the error in assigning times to points on concordia (see captions for Figs. 2 and 3) are in some cases comparable to analytical errors, it may be necessary to consider these sources of uncertainty in applications using concordia plots. Theoretically, all measured 231Pa and 230Th ages represent apparent mean ages of samples which have been deposited over some period of time. These apparent ages are a little younger than the true mean ages. This phenomenon occurs because the rates of 231Pa and 230Th accumulation slow with time. Thus, ages calculated from average 231Pa/235U and 230Th /238U (or 230 Th /234U) ratios do not correspond exactly to true mean ages. Furthermore, the offset between the 231Pa and true mean age and that between the 230Th and true mean age are in general different. Measured mean 231Pa ages are younger than corresponding 230Th ages. The 231Pa and 230Th mean age differences are generally much smaller than the analytical errors. However, if the sample has been deposited over a large (.50 ka) time interval (for example, if the sample growth rate is extremely low), the difference in mean ages may need to be considered for concordia problems. d234U vs. [231Pa/235U] and d234U vs. [230Th/238U] (or 230 [ Th/234U]) are other examples of concordia diagrams for the U-Th-Pa system (Fig. 4a,b). The d234U vs. [230Th/238U] dia- gram has been widely used to evaluate possible diagenesis for coral samples (i.e., Edwards, 1988; Ku et al., 1990; Bard et al., 1991; Chen et al., 1991; Gallup et al., 1994). With the availability of TIMS data for 231Pa, the d234U vs. [231Pa/235U] plot can now also be used to check for diagenesis. Although the 231Pa dating method has a shorter time range (;200 ka) than that of the 230Th dating method (;550 ka), it still can be used to test for concordia and evaluate diagenesis processes for samples older than 200 ka. For samples older than about 200 ka, the [231Pa /235U] ratio must equal one, within present analytical errors, if the sample has behaved as a closedsystem. Values other than one suggest gain or loss of either U or 231Pa (see following discussion). 2.3. Episodic Uranium Loss or Gain Relative to and 231Pa 230 Th Allégre (1964) first provided an episodic model for the U-Th-Pa system for the case in which [234U/238U] equals one. This is similar to the episodic model for the U-Pb systematics introduced by Wetherill (1965a,b). The basic assumption is that the sample experienced open-system conditions for a very short period (compared to true age of the sample) which resulted in Pb loss or U gain. Minerals of the same age but experiencing different degrees of Pb loss or U gain will fall on a straight line with the upper intercept with the concordia curve correspond- U-Th-Pa dating 3441 Fig. 3. 230Th/238U vs. [231Pa/235U] (a) and 230Th/234U vs. [231Pa/235U] (b) concordia diagrams. All notations and legends are as same as those in Fig. 2, except the abscissas, 230Th/238U and 230Th/234U, which are atomic ratios instead of activity ratios. As opposed to Fig. 2, which illustrates how errors in decay constants propagate if secular equilibrium standards are used for both Pa/U and Th/U, this figure illustrates how errors in decay constants propagate if secular equilibrium standards are used for Pa/U, but gravimetric standards are used for Th/U. The uncertainty of the concordia positions in ordinate, ([231Pa/235U]) is as same as Fig. 2. However, in this case, the error in the position of concordia in the abscissa (230Th/238U or 230Th/234U), increases from less than 1‰ for ages ,60 ka to up to 3‰ (the fractional error in the 230Th decay constant) for ages . 500 ka. ing to the time of crystallization and the lower intercept corresponding to the time of the secondary open-system event. The U-Th-Pa system has very similar features. The following discussion refers to episodic U loss or gain, for the case where U isotopes are not fractionated during loss and where any gained U has the same isotopic composition as the sample. In the discussion, we refer to U loss or gain, but these processes are mathematically equivalent (and equally applicable) to Pa Fig. 4. Plots of d234U vs. [231Pa/235U] (a) and d234U vs. [230Th/238U] (b), contoured in 231Pa age and 230Th age (vertical and subvertical lines in (a) and (b), respectively) and initial d234U (subhorizontal curves in (a) and (b)). Diagram (b) has been used previously to evaluate diagenesis. With 231Pa TIMS data now available, a similar plot for [231Pa/235U] may be useful (a). 3442 H. Cheng et al. and Th gain or loss, subject to the condition that Pa and Th are gained in the same proportion as they are present in the sample or that Pa and Th are lost without fractionation. If U loss or gain occurs over a very short period (tD years ago, D for diagenesis), samples of the same primary age (tP years, P for primary) with varying degrees of U loss or gain will also fall on a straight line in a [231Pa/235U] vs. [230Th/238U] or [231Pa/235U] vs. [230Th/ 234 U] concordia plot. The upper intercept with the concordia curve corresponds to their true primary age (tP). A major difference between this system and the U-Pb system is that the lower intercept does not theoretically correspond to the exact age of the secondary event (tD) unless initial d234U is zero (i.e., [234U/238U] 5 1). The basic equations for the episodic model for U-Th-Pa evolution are: F G 231 Pa 5 ~1 2 e2l2313(tP2tD)) 3 e2l2313tD U 235 3 F 1 ~1 2 e2l2313tD! F G F 230 Th 5 ~1 2 e2l2303(tP2tD)! 1 U 238 3 e2l2343(tP2tD) 3 S SF G D 234 238 D U U 21 0 l 230 3 ~1 2 e2(l2302l234)3(tP2tD)! l 230 2 l 234 SF G D 234 3 e2l2303tD 3 F 1 ~1 2 e2l2303tD! 1 234 238 230 Th 5 U 234 234 230 Th U 238 Th U 238 234 238 238 U U G 21 0 l230 3 ~1 2 e2(l2302l234)3tD! l2302l234 U 511 U 230 5 S U U 238 D F G SF G D F G F GYF G F G Y F SF G D 3 e2l2343tP 3 (6) 2 1 3 e2l2343tP (7) (8) 0 U U 234 11 238 U U 2 1 3 e2l2343tP 0 G (9) Where at time tD, U loss or gain follows: Uold/Unew 5 F. Where Uold and Unew are U atoms per unit sample before and after tD, respectively. tP is the sample’s primary age. F is related to the fraction of U lost (.1) or gained (,1) during the diagenetic event at time, tD. Using these equations, it can be shown that the derivatives d[231Pa/235U]/d[230Th/238U], d[231Pa/235U]/d[230Th/234U], 234 238 231 235 d[ U/ U]/[ Pa/ U] and d[234U/238U]/[230Th/238U] are all independent of F. Therefore, samples of the same age with different degrees of U loss or gain occurring at tD will plot along straight lines in diagrams of [231Pa/235U] vs. [230Th/ 238 U] (Fig. 5a), [231Pa/235U] vs. [230Th/234U] (Fig. 5b), d234U vs. [231Pa/235U] (Fig. 6a), and d234U vs. [230Th/238U] (Fig. 6b). The upper intersection of the line with the concordia curve corresponds to F 5 1 (no U loss or gain) and indicates the sample’s primary age (tP). If d234U 5 0, the lower intersection of the line with the concordia curve corresponds to F 5 0 (all 230 Th and 231Pa lost or infinite addition of U) and indicates the time of open-system behavior (tD). When d234U is negative, the point corresponding to tD and F 5 0 will actually lie to the upper right of the lower intercept. When d234U is positive, this point will lie to lower left of the lower intercept. The larger the initial d234U offset from 0, the larger the distance between the point corresponding to tD and F 5 0 and the lower intercept. However, in many cases the age of most interest is the true age (tP, the upper intercept point). Furthermore, if initial d234U is close to 0, for example between 0 and 200 (the case for marine samples), the lower intercept will be indistinguishable from the F 5 0 point in a [231Pa/235U] vs. [230Th/234U] diagram (Fig. 5b). Other graphical methods common to the episodic model for the U-Pb system can also be applied to the U-Th-Pa system. For example, the distance of the data points from the upper intercept point represents the fractional U loss or gain relative to 230Th and 231Pa. For the case of relative U loss, the different ages have the following relationship: 231Pa age . 230Th age . 231Pa/230Th age . true age. For the case of relative U gain, 231Pa age , 230Th age , 231Pa/230Th age , true age. Thus, in these types of open systems, if analytical imprecision is negligible, the 231 Pa/230Th age is closest to the true age. In practice, constraints from 231Pa/230Th ages are most useful for samples older than about 10 ka years as 231Pa/230Th ages are significantly less precise than either 231Pa or 230Th ages for samples younger than 10 ka (because the fractional change in 231Pa/ 230 Th ratio with time is very small in this interval). We can estimate a sample’s 230Th, 231Pa, and 231Pa/230Th ages through graphical methods using concordia plots (Fig. 1). Note that if a sample has been altered these ages will not in general be identical to the true age of the sample. In either a [231Pa/235U] vs. [230Th/238U] plot (Fig. 1a) or a [231Pa/235U] vs. [230Th/234U] plot (Fig. 1b), the 231Pa age is given by the intersection between concordia and a horizontal line through a sample point. For estimating 230Th and 230Th/231Pa ages, the initial d234U must be known, either independently or by using the estimated age from a concordia plot to calculate an initial d234U from a measured value. One can then select the concordia curve that corresponds to the initial d234U value. Given this curve in either a [231Pa/235U] vs. [230Th/238U] plot (Fig. 1a) or a [231Pa/235U] vs. [230Th/234U] plot (Fig. 1b), the 230Th age is given by the intersection between concordia and a vertical line through a sample point and the 231Pa/230Th age is given by the intersection between concordia and a line between a sample point and the origin. [231Pa/235U] vs. [230Th/234U] plots are generally the most useful for graphically estimating 230Th and 231 Pa/230Th ages as the position of concordia is not strongly dependent on initial d234U value in this type of plot. 2.4. Continuous Uranium Loss or Gain Relative to and 231Pa 230 Th If U loss or gain occurs as a continuous process with no change in the U isotopic composition, the behavior of the U-Th-Pa system is graphically similar to the diffusion model for U-Pb systematics (Nicolaysen, 1957; Tilton, 1960; Wasserburg, 1963). Kaufmann and Ku (1989) and Kaufmann et al. (1995) examined this process for the case that the initial d234U value and the d234U value of added U were both zero. We examine this process with initial d234U as a variable, but stipulate that gained or lost U may not change the U isotopic U-Th-Pa dating 3443 Fig. 5. [230Th/238U] vs. [231Pa/235U] (a) and [230Th/234U] vs. [231Pa/235U] (b) concordia diagrams showing the effect of episodic U loss or gain relative to 230Th and 231Pa 20 ka ago (tD) for a suite samples with the same crystallization age of 100 ka (tP). Upon initial precipitation 100 ka ago, the whole suite of samples lie on the origin. Over the next 80 ka, the isotopic composition of the suite is assumed to evolve through closed-system radioactive decay and ingrowth, following concordia to the 80 ka point. At that time (20 ka ago, tD), the suite of samples undergoes episodic diagenesis. Some samples gain U (F , 1) and some lose U (F . 1). The squares represent the isotopic composition of the suite of samples immediately after the open-system episode 20 ka ago (tD). The square on concordia at 80 ka represents a sample in the suite that has not gained or lost any U. The squares above concordia show the isotopic compositions samples that have lost U (20% for the first square above concordia (F 5 1.2), 40% for the second square above concordia (F 5 1.4). The squares below concordia show the isotopic composition of samples that have gained U. The first square below concordia shows the isotopic composition of a sample, which after the episode has 80% primary U and 20% diagenetic U (F 5 0.8). The second square below concordia shows the isotopic composition of a sample that has 60% primary U and 40% diagenetic U (F 5 0.6), and so on. The square that plots on the origin has 0% primary U (infinite addition of diagenetic U, F 5 0). The dashed line in each diagram connects the squares and is mathematically linear. After this episode of U loss and gain, the suite is assumed to again behave as a closed-system. The square points evolve by radioactive decay and ingrowth along the dotted lines to the corresponding circles over a period of 20 ka. At all times in the intervening 20 ka, the points lie on a line that intersects concordia at a time that represents the primary age of the suite (upper intercept with concordia) and a time that is generally close to the time of the diagenetic episode (if initial d234U is reasonably close to zero; see text, lower intercept with concordia). The circles represent the isotopic composition of the suite of samples after the intervening 20 ka of closed-system behavior. In each diagram, the fine solid line connects the open circles and intersects concordia at the time of primary crystallization and close to the time of episodic diagenesis. In this model, samples that plot above concordia, represent samples that lost U relative to both 230Th and 231Pa at tD, and samples that plot below concordia, gained U relative to both 230Th and 231Pa at tD. The distance between a sample point and upper intercept point is proportional to the degree of its U loss, for samples that plot above concordia, or to the fraction of diagenetic U contained in the sample, for samples that plot below concordia. composition of the material. In our model, we refer to U loss or gain, however, these processes are mathematically equivalent to gain of both 231Pa and 230Th or loss of both 231Pa and 230Th if the gain or loss causes no shift in the 231Pa/230Th ratio in the sample. The equations for our model are given in the next section. Samples of the same age will lie along a curve, instead of a straight line, in [231Pa/235U] vs. [230Th/238U] and [231Pa/235U] vs. [230Th/234U] diagrams (Fig. 7a and 7b). However, this curve is nearly linear near its upper intercept with concordia. In plots of d234U vs. [231Pa/235U] and d234U vs. [230Th/238U], samples of the same age will lie on lines (Fig. 8a and 8b). In diagrams of [231Pa/235U] vs. [230Th/ 238 U] and [231Pa/235U] vs. [230Th/234U], the lower intercept now has no temporal significance; however, the upper intercept still corresponds to the primary age. Because the curve is indistinguishable from a straight line near the upper intercept (Fig. 7a,b), it may be possible to estimate the primary age by graphical methods. In this model, the 230Th, 231Pa and 231Pa/230Th ages have the same relationship as in the episodic model described above for both U loss and gain cases. 2.5. Continuous Addition or Loss of and 231Pa 234 U, 230 Th Gallup et al. (1994) formulated a 230Th and 234U continuous input model to explain U-series results obtained for Barbados corals. We present here a general model for continuous 234U, 230 Th, and 231Pa input together with continuous loss or gain of U. The basic differential equations for this model are: d 238U 5 2l238 3 238U 1 Ru 3 238U dt (10) d 235U 5 2l235 3 235U 1 Ru 3 235U dt (11) 3444 H. Cheng et al. Fig. 6. Plots of d234U vs. [231Pa/235U] (a) and d234U vs. [230Th/238U] (b). (See caption for Fig. 4). The episodic assumptions, symbols and lines are same as in Fig. 5. Each of the squares and open circles in this figure correspond to an analogous square or circle in Fig. 5. The intercept of the horizontal fine solid line with the concordia curve (initial d234U 5 150) gives the true crystallization age for a suite of samples of the same age. Points that lie to the right of the concordia curve indicate U loss (F . 1) or gain of both 230Th and 231Pa. Points that lie to the left of concordia curve indicate U gain (F , 1) or loss of both 230Th and 231Pa. No information about the time of U loss or gain can be obtained in diagrams such as these. In addition, the initial d234U must be known in order to select the correct concordia curve to be used for a suite samples. d 231Pa 5 l235 3 235U 2 l 231 3 231Pa 1 R231 dt d (12) 234 Ut 5 234 U 5 l 238 3 238U 2 l 234 3 234U 1 Ru 3 234U 1 R234 dt d 230Th 5 l 234 3 234U 2 l 230 3 230Th 1 R230 dt 2 (13) Tho 5 Pao 5 0, Tht 5 2 U0 5 137.88 3 235Uo, 238 F G 238 U U 235 231 5 constant, Ut 5 238Uo 3 e(Ru2l238)3t 5 238Uo 3 eRu3t (15) Ut 5 235Uo 3 e(Ru2l235)3t 5 235Uo 3 eRu3t (16) S 2l2313t 3e D R231 1 l231 0 l 238 3 238U . l234 U U 3 0 l238 3 238Uo l234 l238 3 238Uo l 234 R234 3 2 l234 2 Ru l234 Ru 2 l234 1 l230 l 238 3 238Uo 1 (Ru 1 l230 234 238 U U o l 238 3 238Uo l238 3 238Uo R234 2 2 l234 l234 2 Ru l234 3 l 234 l234 3 R234 R230 1 1 Ru 2 l234 1 l230 l230 l230 3 (l234 2 Ru) 3 e2l2303t 1 (17) 234 238 3 0 l235 3 235Uo R231 l235 3 235Uo Pat 5 3 eRu3t 2 1 Ru 1 l231 l231 Ru 1 l231 3 FF G G SF G F D l238 3 238Uo 3 eRu3t 1 l234 3 e2(l2342Ru)3t 2 we obtain the following analytical solutions to Eqns. 10214: 238 U 238 U (18) (14) Using the following initial and boundary conditions, 234 FF G G 234 l238 3 Uo R234 R234 3 e2(l2342Ru)3t 1 2 l234 2 Ru l234 l234 2 Ru 238 234 230 l238 3 238Uo 3 eRu3t 1 l234 l234 3 R234 R230 1 l230 l230 3 (l234 2 Ru) G (19) Where l230, l231, l234, l235, and l238 are the decay constants (with units of inverse time) for 230Th, 231Pa, 234U, 235U, and 238 U. R230, R231, and R234 are input rates (with units of atoms/ U-Th-Pa dating 3445 Fig. 7. [230Th/238U] vs. [231Pa/235U] (a) and [230Th/234U] vs. [231Pa/235U] (b) concordia diagrams representing continuous U loss or gain. Suites of 150 ka and 80 ka age samples with different U loss or gain rates plot along the fine solid and dashed curves, respectively. The upper intercept of each curve with concordia represents the true age of the suite. Samples that follow this model and plot on the portion of each dashed or fine solid line above concordia have undergone continuous U loss relative to Pa and Th and samples that plot on the portion of each curve below concordia have undergone continuous U gain relative to Pa and Th. The rate of U gain is greatest for samples that plot on the portion of each curve near the origin, whereas the rate of U loss is greatest for samples that plot on the portion of each curve furthest from the origin. The diagram is graphically similar to the diffusion model for the U-Pb system. It is important to note that the curve near the upper intercept is close to a straight line, which may allow the use of graphical methods to obtain the true age of a set of samples of the same age by extrapolation to its upper intercept with concordia. However, the extrapolation of the near linear portion of each curve to its lower intercept with concordia will yield a meaningless age, as has been demonstrated for the U-Pb system (Tilton, 1960). time) of 230Th, 231Pa, 234U from outside of the system. Ru is the fraction (per unit time) of U loss (negative) or gain (positive) relative to 230Th and 231Pa, and it is assumed to be much greater than l235 and l238 (Ru .. l235 and Ru .. l238). t is the age of sample. 230Th, 231Pa, 234U, 235U, and 238U represent atoms of 230Th, 231Pa, 234U, 235U, and 238U. Subscripts t and o indicate time t and t 5 0, respectively. If R231 5 R230 5 R234 5 0 and Ru Þ 0, the equations describe the continuous U loss (negative Ru) and gain (positive Ru) model (see Figs. 7 and 8). If R230 ' R234, and Ru and R231 5 0, the equations describe the 230Th and 234U continuous input model of Gallup et al., 1994. If R231, R230, and R234 Þ 0, and Ru 5 0, the equations describe the 234U, 230Th and 231Pa continuous input model shown in Figs. 9 and 10. Data for samples of the same age will lie along a curve. The intercept between this curve and the concordia curve is the primary age of the sample set. The portion of curve near the upper intercept with concordia is very close to a straight line with a slope dependent on the R231, R230, and R234 values. In this continuous addition model, the 231Pa age will be greater than the true age, the 230Th age and the 231 Pa/230Th age could be greater or less than true age, depending on the input ratios of R230/R234 and R231/R230. Here, the 231 Pa age provides the best upper limit on the true crystallization age. This model can be extended to models for 231Pa, 230Th, and 234 U continuous losses (R231, R230, and R234 are negative) as shown in Figs. 9 and 10 (the very short curves from concordia curve down to lower left). For this case, the 231Pa age will be less than the sample’s true age. As with the addition model, the 230 Th and the 231Pa/230Th ages also could be greater or less than the true age, depending on the relative loss rates, i.e., R230/R234 and R231/R230. 2.6. Linear Uptake of Uranium, with No Initial Uranium Ikeya (1982) suggested two U-accumulation models based on electron spin resonance (ESR) analysis of teeth, which start with essentially no U, but pick up U diagenetically. Such models could be equally applicable to other materials such as mollusks and bones, which also start with essentially no U. The two models are (1) early uptake (EU) of U in which U accumulated within a short time span after it was buried, (2) linear uptake (LU) of U in which U addition was continuous and constant. He favored the LU model. Ku (1982) outlined the 230 Th age, 231Pa age and true age differences for the LU model using a plot of the true age vs. the average apparent age. Subsequently, the LU model has been discussed and applied to many tooth and bone U-series and ESR dating studies (e.g., Grün et al., 1988; Chen et al., 1994; Grün and McDermott 1994; Swisher et al., 1996). These studies focused mainly on ESR dating and corrections. Kaufmann et al. (1995) examined the LU model with respect to U-Th-Pa systematics, under the assumption that d234U is constant and equal to zero. Here, we describe the basic behavior of the U-Th-Pa system in the LU model, with the d234U of added U as a constant not equal to 3446 H. Cheng et al. Fig. 8. Plots of d234U vs. [231Pa/235U] (a) and d234U vs. [230Th/238U] (b) (see caption for Fig. 4), showing continuous U loss or gain relative to 230Th and 231Pa. Suites of 150 ka and 80 ka age samples with different U loss or gain rates lie along the fine solid and dashed horizontal lines, respectively. Each of the dotted curves illustrates how the isotopic composition of a sample with a specific U loss or gain rate changes with time. zero. The d234U of the material is a variable, which is allowed to change as a function of time following mass balance and radioactive decay constraints. The basic differential equations are d 238 d 235 U 5a dt 231 Pat 5 234 Ut 5 d U 5b dt (21) Pa 5 l2353235U 2 l 231 3 231Pa dt (22) 230 Th 5 l2343234U 2 l 230 3 230Th dt (24) At t 5 0: 238 Tho 5 Pao 5 Uo 5 235 Uo 5 234 Uo 5 0 U 5 137.88 3 235U (or a 5 137.88 3 b) S D 234 238 U U 5 c input we obtain the following analytical solutions to Eqns. 20224: Ut 5 a 3 t (25) Ut 5 b 3 t (26) 238 235 S S D l 238 3 a a3c )2 l234 l2342 (28) D l238 3 a a3c 3 e2l2343t 2 l234 3 (l230 2 l234) (l230 2 l234) 1 F 1 l238 3 a l238 3 a a3c 2 3 e2l2303t 1 l230 3 l234 l230 l230 a3c l238 3 a l238 3 a 2 1 l230 2 l234 l234 3 (l230 2 l234) l2302 3t1 Using the following initial and boundary conditions, 238 Tht 5 (23) d 234U 5 l238 3 238U 2 l 234 3 234U 1 a 3 c dt 231 D (27) l238 3 a a 3 c l238 3 a 2 3 e2l2343t 1 l2342 l234 l234 3t1 231 230 S (20) 230 d b 3 l235 3 (e2l2313t 1 l231 3 t 2 1) l2312 G l238 3 a l238 3 a a3c 2 2 l230 l230 3 l234 l2302 S D (29) Where a and b are constant rates of 238U and 235U accumulation; c is the 234U/238U atomic ratio of the input U; t is the true depositional age of the sample. If U uptake has been linear, the standard 231Pa, 230Th and 231 Pa/230Th ages, which are equivalent to EU ages, will all be younger than the true age of a sample. In [231Pa/235U] vs. [230Th/238U], [231Pa/235U] vs. [230Th/234U] plots, d234U vs. [231Pa/235U], and d234U vs. [230Th/238U] plots (Figs. 11 and 12), the data points lie below the concordia curves. This tendency becomes large when true ages are greater than about 20 ka. This deviation from concordia can be used to check the U-Th-Pa dating Fig. 9. [230Th/238U] vs. [231Pa/235U] (a) and [230Th/234U] vs. [231Pa/235U] (b) concordia diagrams. When 230Th, 231Pa and U are added to (or lost from) samples continuously, the data for samples of the same age will lie along a curve (the short solid or dotted curves). The intercept of this curve with concordia will indicate the true crystallization age. These curves are indistinguishable from straight lines near the intercepts, and their slopes depend on the relative addition rates of 230Th, 231 Pa and 234U (R230, R234, and R231). In these two diagrams, two examples are given: R230 5 R234 5 20 * R231 (short dotted curves) and R230 5 R234 5 50 * R231(short solid curves). These ratios were chosen because 20 is close to the production ratio of 230Th to 231Pa and would be the approximate isotope ratio of a young carbonate (,10 ka). 50 is close to the 230 Th/231Pa ratio of secular equilibrium materials. The upper end and lower end of each short line represent 200,000 atom/year/g gain and 100,000 atom/year/g loss of 230Th for a sample with a 3 ppm 238U concentration, a typical concentration for a coral. For reference, a coral with this 238U concentration produces about 1.3 million atoms of 230 Th/year/g by radioactive decay. Graphical methods can be used to obtain the true age for a suite of the samples of the same age. 234 Fig. 10. Continuous 230Th, 231Pa and 234U addition models for plots of d234U vs. [231Pa/235U] (a) and d234U vs. [230Th/238U] (b). The symbols are same as in Fig. 8. The dashed curves (b) represent the condition of R230 5 R234. All of the thin dashed, dotted, and solid curves are indistinguishable from straight lines. It is possible to get information for a suite of samples of the same age using graphical methods; however, the initial d234U values must be known. In this model, the curves are generally not subhorizontal, which is different from simple U loss and gain models. 3447 3448 H. Cheng et al. Fig. 11. [230Th/238U] vs. [231Pa/235U] (a) and [230Th/234U] vs. [231Pa/235U] (b) concordia plots. Solid curves are concordia curves (initial d234U 5 150 and 1500) as in Fig. 1. Dashed curves are concordia plots assuming linear U (LU) uptake (initial d234U 5 150 and 1500). The ages on LU model curves correspond to the true ages of samples for which uptake of U has been constant with time. 230Th and 231Pa ages calculated assuming linear U uptake are older than those calculated assuming closed-system behavior for the same measured [230Th/238U], [230Th/234U], and [231Pa/235U] values. Diagrams such as these can be used to test for LU model behavior (isotopic compositions should lie on the dashed linear uptake concordia curves). applicability of the LU model. In addition, Eqns. 20, 23, and 24 can be used to solve for a, c, and t; and Eqns. 21 and 22 will provide b and t. In this way, age and input rate can be checked for concordancy. If the sample follows the LU model, the different standard ages of a sample have the following pattern: true age $ apparent 231Pa/230Th age . apparent 230Th age $ apparent 231Pa age. When the true age is less than 20 ka, the true age is nearly twice that of the apparent 231Pa or 230Th ages (Figs. 11 and 12) and about 1.5 ; 1.6 times that of the apparent 231Pa/230Th age. Fig. 12. Plots of d234U vs. [231Pa/235U] (a) and d234U vs. [230Th/238U] (b). The concordia curves of the LU model (initial d234U 5 150) are shown as dashed curves. They are below close-system concordia curves with the same initial d234U value (solid curves). The offset increases with the age. U-Th-Pa dating 3449 Fig. 13. [230Th/238U] vs. [231Pa/235U] (a) and [230Th/234U] vs. [231Pa/235U] (b) concordia plots for last interglacial corals from Barbados (Gallup et al., 1994; Edwards et al., 1997). The solid ellipses represent 2s errors and the labels are sample numbers. The heavy lines are best-fit lines calculated using the least squares method (y 5 0.926x 1 0.211 and y 5 1.462x 2 0.087 in diagrams (a) and (b), respectively). r is the correlation coefficient. Other symbols are as same as in Fig. 9. The continuous U loss or gain lines, similar to Fig. 9, are plotted for comparison. The best-fit lines intersect concordia at ages of 131 ka in diagram (a) and 126 ka in diagram (b). Of the two ages, the 126 ka age is the best estimate of the true age for two reasons: (1) the effect of uncertainty in initial d234U on the age of the intercept is insignificant in diagram (b) and (2) the angle of intersection between the best-fit line and concordia is much steeper in diagram (b) than in diagram (a). The differences between true and apparent ages increase with age (Figs. 11 and 12). 3. AN EXAMPLE Here we present an example of the use of these concordia plots using some of the first high resolution U-Th-Pa data. The example uses five analyses of four last interglacial samples (two subsamples of FU-1, Figs. 13 and 14) from Barbados (see Gallup et al., 1994 for information about samples and the original U/Th data and Edwards et al., 1997 for more recent U/Th data and the U/Pa data). These samples should have similar ages (the last interglacial), if they have behaved as closed-systems. However, some of them, have been altered during diagenesis as shown by the discordant 231Pa and 230Th ages (FU-3 and UWI-16). If the diagenesis process follows one of the models described above (Figs. 5210), then these data will lie on a line in each plot (i.e., in Figs. 5210). The least squares method was used to obtain best fit lines. The correlation coefficients are between 0.9865 and 0.9983 for the data in four types of plots ([231Pa/235U] vs. [230Th/238U], [231Pa/235U] vs. [230Th/234U], d234U vs. [230Th/238U], and d234U vs. [231Pa/ 235 U]. In the first two plots, all the data lie within analytical error of a best-fit line and the correlation coefficients (0.9983 and 0.9924, Fig. 13) are very high. For the two plots with d234U as the ordinate the correlation coefficients are also very high (0.9913 and 0.9865), but slightly lower than in the other two plots. Also, in the latter two plots, the data are not co-linear within analytical error. The data are consistent with a 230Th, 231 Pa and 234U continuous input model, or a model of continuous loss of U coupled with continuous increase in d234U. As the U concentrations of the discordant samples are similar those of primary corals (see Min et al., 1995), the addition model would appear to be more appropriate. The parameters that best fit the data in a continuous input model are a 230Th/234U input ratio is close to 1 (see Gallup et al., 1994) and a 230Th/231Pa input ratio of about 35. Young carbonate material (,10 ka) has a 230Th/231Pa ratio of about 20 (the production ratio for material with d234U close to zero), and old material (secular equilibrium material) has a 230Th/231Pa ratio of about 50. Thus, the calculated 230Th/231Pa ratio of the added material is intermediate between the two extremes, suggesting that net fractionation of 230Th from 231Pa during the dissolution of the ultimate source material and incorporation of the 230Th and 231Pa in the corals was small. If fractionation was negligible, the data are consistent with a source material of about the same age as the corals. Thus, a plausible source for the added Th and Pa is dissolution of carbonates of the same age as the coral. If the initial d234U is 150, the ages of intercepts between best-fit lines and concordia curves are 131 ka, 126 ka, 116 ka, and 121 ka for plots of Figs. 13a,b, and 14a,b, respectively. Among these plots, the [231Pa/235U] vs. [230Th/234U] diagram provides the best upper intercept age (126 ka), largely because it is not affected by small variations in initial d234U value. For example, the upper intercept age in this example only changes 0.1% when the initial d234U value changes by 65‰ around 150 (from 145 to 155). However, for the other three plots, the upper intercept ages shift 1.322.8% for the same shift in initial d234U value. Furthermore, for the plots with d234U on the ordinate, the correlation coefficients, although high, are smaller than for the plots with [231Pa/235U] on the ordinate. Considering the latter two plots, in the [231Pa/235U] vs. [230Th/238U] diagram, the angle of intersection between concordia and a best-fit line is much shallower than in the [231Pa/235U] vs. [230Th/234U] diagram. Thus, errors in the location of the best-fit 3450 H. Cheng et al. Fig. 14. Plots of d234U vs. [231Pa/235U] (a) and d234U vs. [230Th/238U] (b) for the last interglacial samples. The symbols are as same as in Figs. 10 and 13. Best fit lines are y 5 428.9x 2283.9 and y 5 396.0x 2192.3 in diagrams a and b, respectively. The best-fit lines intersect the concordia curves (d234U 5 150) at an ages of 116 ka (diagram a) and 121 ka (diagram b). We consider both of these age estimates to be less reliable than the estimate from Fig. 13b because (1) data points in both plots lie off of a best-fit line, and (2) the error in age due to uncertainty in initial d234U is significant in both plots. line translate into larger errors in time for the [231Pa/235U] vs. [230Th/238U] plot. For all of these reasons, the intersection of concordia with the best-fit line in the [231Pa/235U] vs. [230Th/ 234 U] provides the best estimate of primary age for this particular example. 4. CONCLUSIONS With precise 230Th and 231Pa TIMS data now available, testing for U-series concordancy should become common. Samples with concordant 230Th and 231Pa ages will be likely candidates for closed-system histories. For nonmarine samples, such as speleothems, teeth, and bones, 231Pa TIMS methods might provide the only possible high-precision check for 230Th dating. For marine samples, the initial d234U value can also be compared to the modern marine value (about 150). If samples give concordant 230Th and 231Pa ages but values of initial d234U different from the modern marine value, this could indicate a different initial d234U value for past seawater, or suggest a type of diagenesis which resulted in the 234U change without disturbing the 230Th and 231Pa ages, for example, early 234 U input or 231Pa, 230Th, and 234U input in just the right proportions (Figs. 9 and 10). When a single sample has discordant 230Th and 231Pa ages, it can generally be considered to have experienced open-system behavior (assuming no significant initial 230Th and 231Pa). To provide some constraints on its true age, a [231Pa/235U] vs. [230Th/234U] diagram might be used in an effort to reduce the effects of disturbed d234U values. If the isotopic composition lies above the concordia curve in a [231Pa/235U] vs. [230Th/ 234 U] diagram, it is likely that the dominant diagenetic process was U loss or gain of both 230Th and 231Pa. If the isotopic composition lies below the concordia curve, then U gain or loss of both 230Th and 231Pa is probable (Figs. 12Fig. 10). In the former case, its true age will be younger than its apparent 230Th and 231Pa ages. In the later case, its true age will be older. 231 Pa/230Th ages should be closest to the true age in such cases. The use of U-Th-Pa systematics can also provide checks on the LU or EU models, important to many tooth and bone dating studies relating to archaeology. When the true age is greater than 20 ka, in the LU model, the measured data should lie below the concordia curve. Thus, this approach can be used to check the veracity of model 231Pa and 230Th ages based on different assumed modes of U uptake. Furthermore, such data may be useful in establishing constraints on U uptake histories for specific samples that have aged in particular diagenetic environments. Using these sample specific constraints on U uptake, it may then be possible to constrain the true age of such an open-system material. As noted, the use of concordia diagrams has been a unique strength of U-Pb dating. A single discordant U-Pb date is not very meaningful, instead, the upper intercept point of a series of samples becomes important for an age determination. A similar approach can be used in the U-Th-Pa dating of discordant samples. For example, a suite of samples of the same age (for example, corals from the same terrace and location) can be used to provide a series of points for a concordia diagram. If these samples experienced different degrees of one of the diagenetic processes described by the above models, the data will fall U-Th-Pa dating along a straight line or a curve that is very close to a straight line near the upper intercept, in a concordia diagram. The upper intercept will be the true crystallization age. Acknowledgments—We thank J. A. Hoff for discussions on this research. This work was supported by NSF grants OCE-9402693, OCE9500647, EAR-9512334, EAR-9406183, and EAR-9702137 to RLE, the Donors of the Petroleum Research Fund, administered by the American Chemical Society, and a grant to MTM from the Geosciences Research Program, Office of Basic Energy Sciences, U.S. Department of Energy. REFERENCES Allègre M. C. (1964) De l’extension de la méthode de calcul graphique Concordia aux mesures d’âges absolus effectués à l’aide du déséquilibre radioactif. Cas des minéralisations secondaires d’uranium. Note (*) de. C. R. Acad. Sci. Paris, 259, 408624089. Bard E., Hamelin B., and Fairbanks R. G. (1990a) U/Th ages obtained by mass spectrometry in corals from Barbados: Sea level during last 130,000 years. Nature 346, 4562458. Bard E., Hamelin B., Fairbanks R. G., and Zindler A. (1990b) Calibration of the 14C time-scale over the past 30,000 years using mass spectrometric U-Th ages from Barbados corals. Nature 31, 4052409. Bard E., Fairbanks R. G., Hamelin B., Zindler A., and Hoang C. T. (1991) Uranium-234 anomalies in corals older than 150,000 years. Geochim. Cosmochim. Acta 55, 228522390. Barnes J. W., Lang E. J., and Potratz H. A. (1956) The ratio of ionium to uranium in coral limestone. Science 124, 1752176. Birch G. F., Thomson J., McArthur J. M., and Burnett W. C. (1983) Pleistocene phosphorites off the west coast of South Africa. Nature 302, 6012603. Broecker W. S. and Thurber D. L. (1965) Uranium series dating of corals and oolites from Bahaman and Florida Key limestones. Science 149, 58260. Burnett W. C. and Kim K. H. (1986) Comparison of radiocarbon and uranium-series dating methods as applied to marine apatite. Quat. Res. 25, 3692379. Chen J. H., Edwards R. L., and Wasserburg G. J. (1986) 238U, 234U and 232 Th in seawater. Earth Planet. Sci. Lett. 80, 2412251. Chen J. H., Curran H. A., White B., and Wasserburg G. J. (1991) Precise chronology of the last interglacial period 234U-230Th data from fossil coral reefs in the Bahamas. Geol. Soc. Amer. Bull. 103, 82297. Chen T., Yang Q., and Wu E. (1994) Antiquity of Homo sapiens in China. Nature 368, 55256. Cheng H., Edwards R. L., Murrell M. T., and Goldstein S. J. (1996) Pa-231 Dating of Carbonates Using TIMS Techniques. EOS Trans. AGU, 1996 Spring Meeting, S168. Cheng H., Edwards R. L., and Wang Y. (1997) U/Th and U/Pa dating of Nanjing Man. EOS Trans. AGU, 1997 Fall Meeting, F787. De Bievre P. et al. (1971) The half-life of U-234. In Proc. Intl. Conf. Chem. Nucl. Data, Measurement and Applications, Canterbury (ed. M. L. Hurrell), pp. 2212225. Inst. Civil Eng. Edwards R. L. (1988) High precision thorium-230 ages of corals and the timing of sea level fluctuations in the late Quaternary. Ph.D thesis, California Institute of Technology. Edwards R. L., Chen J. H., and Wasserburg G. J. (1987a) 238U-234U230 Th-232Th systematics and the precise measurement of time over the past 500,000 y. Earth Planet. Sci. Lett.. 81, 1752192. Edwards R. L., Chen J. H., Ku T.-L., and Wasserburg G. J. (1987b) Precise timing of the last inter-glacial period from mass spectrometric analysis of 230Th in corals. Science 236, 154721553. Edwards R. L., Taylor F. W., and Wasserburg G. J. (1988) Dating earthquakes with high precision thorium-230 ages of very young corals. Earth Planet. Sci. Lett. 90, 3712381. Edwards R. L. et al. (1993) A Large Drop in Atmospheric 14C/12C and Reduced Melting in the Younger Dryas, Documented with 230Th Ages of Corals. Science 260, 9622968. Edwards R. L., Cheng H., Murrell M. T., and Goldstein S. J. (1997) 3451 Protactinium-231 dating of carbonates by thermal ionization mass spectrometry: Implications for Quaternary climate change. Science 276, 7822786. Gallup C. D., Edwards R. L., and Johnson R. G. (1994) The timing of high sea levels over the past 200,000 years. Science 263, 7962800. Gascoyne M. (1992) Geochemistry of The Actinides And Their Daughters. In Uranium-series Disequilibrium: Applications to Earth, Marine, and Environmental Sciences (ed. M. Ivanovich and R. S. Harmon), pp. 34261. Clarendon Press. Goldstein S. J., Murrell M. T., Lea D., Chakraborty S., and Kashgarian M. (1996) 231Pa and 230Th dating of deep sea coral. EOS Trans. AGU, 1996 Spring Meeting, F291. Grün R. and McDermott F. (1994) Open-system modeling for U-series and ESR dating of teeth. Quat. Geochronol. Quat. Sci. Rev. 3, 1212125. Grün R., Schwarcz H. P., and Chadam J. M. (1988) ESR dating of tooth enamel, Coupled correction for U-uptake and U-series disequilibrium. Nuclear Tracks 14, 2372241. Hamelin B., Bard E., Zindler A., and Fairbanks R. G. (1991) 234U/238U mass spectrometry of corals: How accurate is the U-Th age of the last interglacial period? Earth Planet. Sci. Lett. 106, 1692180. Henderson G. M., Cohen A. S., and O’Nions R. K. (1993) 234U/238U ratios and 230Th ages for Hateruma Atoll corals: Implications for coral diagenesis and seawater 234U/238U ratios. Earth Planet. Sci. Lett. 115, 65273. Holden P., Lundstrom C. C., Zachos J., and Jacobson A. D. (1996) Combined 234U-230Th and 235U-231Pa dating of a central California speleothems: implications for improved temporal resolution in paleoclimatology. EOS Trans. AGU, 1996 Spring Meeting, F384. Ikeya M. (1982) A model of linear uranium accumulation for ESR age of Heidelberg, Mauer and Tautavel bones. Japanese J. Appl. Phys. 21, L6902L692. Ivanovich M. and Harmon R. S. ed. (1992) Uranium Series Disequilibrium: Applications to Earth, Marine, and Environmental Sciences. 2nd ed. Oxford Univ. Press. Kaufman A. and Broecker W. S. (1965) Comparison of 230Th and 14C ages for carbonate materials from Lakes Lahontan and Bonneville. J. Geophys. Res. 70, 403924054. Kaufman A., Broecker W. S., Ku T.-L., and Thurber D. L. (1971) The status of U-series methods of mollusk dating. Geochim. Cosmochim. Acta 35, 11552 1183. Kaufman A. and Ku T.-L. (1989) The U-series ages of carnotites and implications regarding their formation. Geochim. Cosmochim. Acta 53, 267522681. Kaufman A., Ku T.-L., and Luo S. (1995) Uranium-series dating of carnotites: Concordance between Th-230 and Pa-231 ages. Chem. Geol. 120, 1752181. Ku T.-L. (1968) Protactinium-231 method of dating coral from Barbados Island. J. Geophys. Res. 73, 227122276. Ku T.-L. (1982) Progress and Perspectives. In Uranium-series Disequilibrium: Applications to Earth, Marine, and Environmental Sciences (ed. M. Ivanovich and R. S. Harmon), pp. 3972506. Clarendon Press. Ku T.-L., Kimmel M. A., Easton W. H., and O’Neil T. J. (1974) Eustatic sea level 120,000 years ago on Oahu Hawaii. Science 183, 9592962. Ku T.-L., Bull W. E., Freeman S. T., and Knauss K. G. (1979) 230 Th/234U dating of pedogenic carbonates in gravelly deserts of Vidal Valley, Southwestern California. Bull. Geol. Soc. Amer. 90, 106321073. Ku T.-L., Ivanovich M., and Luo S. (1990) U-series Dating of Last Interglacial High Sea Stands: Barbados Revisited. Quat. Res. 33, 1292147. Lounsbury M. and Durham R. W. (1971) The alpha half-life of U-234. In Proc. Intl. Conf. Chem. Nucl. Data, Measurement and Applications, Canterbury (ed. M. L. Hurrell), pp. 2152219. Inst. Civil Eng. Ludwig K. R. et al. (1992) Mass-Spectrometric 230Th-234U-238U Dating of the Devils Hole Calcite Vein. Science 258, 2842287. Meadows J. W., Armani R. J., Callis E. L., and Essling A. M. (1980) Half-life of 230Th. Phys. Rev. C 22, 7502754. Min G. R., Edwards R. L., Taylor F. W., Recy J., Gallup C. D., and Beck J. W. (1995) Annual cycles of U/Ca in corals and U/Ca thermometry. Geochim. Cosmochim. Acta 59, 202522042. 3452 H. Cheng et al. Nicolaysen L. O. (1957) Solid diffusion in radioactive minerals and the measurement of the absolute age. Geochim. Cosmochim. Acta 11, 41259. O’Brien G. W., Veeh H. H., Cullen D. J., and Milnes A. R. (1986) Uranium-series isotopic studies of marine phosphorites and associated sediment from the East Australian continental margin. Earth Planet. Sci. Lett. 80, 19235. Pickett D. A. and Murrell M. T. (1997) Observations of 231Pa Disequilibrium in Volcanic Rocks. Earth Planet. Sci. Lett. 148, 2592271. Pickett D. A., Murrell M. T., and Williams R. W. (1994) Determination of Femtogram Quantities of Protactinium in Geologic Samples by Thermal Ionization Mass Spectrometry. Anal. Chem. 66, 104421049. Robert J., Miranda C. F., and Muxart R. (1969) Mesure de la période du protactinium-231 par microcalorimétrie. Radiochim. Acta 11, 1042108. Roe K. K. Burnett W., Kim K. H., and Beers M. J. (1982) Excess protactinium in phosphate nodules from as coastal upwelling zone. Earth Planet. Sci. Lett. 60, 39246. Rosholt N. J. (1967) Open system model for uranium-series dating of Pleistocene samples. Radioactive Dating Methods Low-Level Counting, Proc. Symp. Monaco, IAEA. Proc. Ser. Publ., SM-87/50. Sackett W. M. (1958) Ionium-uranium ratios in marine deposited calcium carbonates and related materials. Ph.D. Thesis, Washington Univ. Sakanoue M., Konishi K., and Komura K. (1967) Stepwise determinations of thorium, protactinium, and uranium isotopes and their applications in geochronological studies, in: Radioactive Dating Methods Low-Level Counting, Proc. Symp. Monaco, IAEA. Proc. Ser. Publ., SM287/28. Schwarcz H. P. (1980) Absolute age determination of archaeological sites by uranium series dating of travertines. Archaeometry 22, 3224. Stein M. et al. (1993) TIMS U-series dating and stable isotopes of the last interglacial event in Papua New Guinea. Geochim. Cosmochim. Acta 57, 254122554. Swisher III C. C. et al. (1996) Latest home erectus of Java: Potential contemporaneity with home sapiens in Southeast Asia. Science 274, 187021874. Szabo B. J. (1979) 230Th, 231Pa and open system dating of fossil corals and shells. J. Geophys. Res. 84, 492724930. Szabo B. J. and Rosholt J. N. (1969) Uranium-Series Dating of Pleistocene Molluscan Shell from Southern California-An open-system Model. J. Geophys. Res. 74, 325323260. Szabo B. J. and Vedder J. D. (1971) U-series dating of some Pleistocene marine deposits in Southern California. Earth Planet. Sci. Lett. 11, 2832290. Thurber D. L., Broecker W. S. Blanchard R. L., and Potratz H. A. (1965) Uranium-series ages of pacific atoll coral. Science 149, 55258. Tilton G. R. (1960) Volume diffusion as a mechanism for discordant lead ages. J. Geophys. Res. 65, 293322945. Veeh H. H. (1982) Concordant 230Th and 231Pa ages of marine phosphorites. Earth Planet. Sci. Lett. 57, 2782284. Veeh H. H. and France R. E. (1988) Uranium-Series Ages of Corals and Coexisting Phosphate Deposits on Pelsaert Reef Complex, Houtman-Abrolhos Islands, Western Australia. Quat. Res. 30, 2042209. Wasserburg G. J. (1963) Diffusion processes in lead-uranium systems. J. Geophys. Res. 68, 482324846. Wetherill G. S. (1956a) An interpretation of the Rhodesia and Witwatersrand age patterns. Geochim. Cosmochim. Acta 9, 2902292. Wetherill G. S. (1956b) Discordant U-lead ages. I. Trans. Amer. Geophys. Union 37, 3202326.
© Copyright 2026 Paperzz