C yr BP xlOOO

FIGURE 3.16 Simulations of the effect of differences between radiocarbon age and calendar year age.The top diagram shows a set of calendar years, and below are the equivalent l4C ages, derived from the CALIB (ver. 3.03) program of Stuiver and Reimer (1993). In case A, the set of uniformly distributed calendar years leads to a set of clustered radiocarbon ages. In B, a cluster of dates around a mean of 11,000 calendar yr (and a standard deviation of 500 yr) is compared to the equivalent radiocarbon distribution. A bimodal distribution apparent in the l4C data does not reflect reality (from Bartlein et al., 1995).

solar activity, it has also been argued that 14C variations are inversely related to worldwide temperature fluctuations (Wigley and Kelly, 1990). This implies that solar activity, radiocarbon variations, and surface temperature are all related, perhaps through fundamental variations in the solar constant (i.e., low solar activity = high 14C production rate = low temperature). If so, then the 14C record itself, as a proxy of solar activity, may provide important information on the causes of climatic change. However, this is a controversial topic; several authors have shown that the correlations between radiocarbon variations and paleotemperature records are very poor when the records are examined in detail (Williams et al., 1981). This may be because atmospheric 14C is only a small part of the global 14C inventory and climate-related changes in ocean circulation and deep water formation may overwhelm the effect of solar activity changes. In this regard, 10Be may be a better proxy of solar activity (Beer et al., 1994, 1996). On the other hand, high resolution data from the Greenland ice sheet show a very strong 11-yr signal in 8lsO and spectral analysis reveals periodicities associated with those known from the spectra of radiocarbon variations (Stuiver et al., 1995; Sonnett and Finney, 1990). Indeed the amplitude of the 8lsO signal is so large (~1.5%o) that it is very hard to imagine how such small irradiance changes (0.05%) could be amplified within the climate system to produce such a strong signal. There is some evidence from modeling experiments that larger (0.25% or more) reductions in irradiance that seem possible for extended periods of reduced solar activity, like the Maunder Minimum, can influence temperatures on a global scale (Rind and Overpeck, 1993; Lean, 1994). Still, it is not yet clear that radiocarbon variations or 10Be can be used as an index of irradiance and at this stage, therefore, the evidence relating solar activity and radiocarbon variations to surface temperatures remains equivocal, an intriguing but so far unproven possibility.

3.2.2 Potassium-Argon Dating (40K/40Ar)

Compared to radiocarbon dating, potassium-argon dating is used far less in Quaternary paleoclimatic studies. However, potassium-argon and argon-argon dating have indirectly made major contributions to Quaternary studies. The techniques have proved to be invaluable in dating sea-floor basalts and enabling the geomagnetic polarity timescale to be accurately dated and correlated on a worldwide basis (Harland et al., 1990; see also Section 4.1.4). Potassium-argon dating has also been used to date lava flows which, in some areas of the world, may be juxtaposed with glacial deposits. In this way, limiting dates on the age of the glacial event may be assigned (Lôffler, 1976; Porter, 1979).

Potassium-argon dating is based on the decay of the radioisotope 40K to a daughter isotope 40Ar. Potassium is a common component of minerals and occurs in the form of three isotopes, 39K and 41K, both stable, and 40K, which is unstable. The 40K occurs in small amounts (0.012% of all potassium atoms) and decays to either 40Ca or 40Ar, with a half-life of 1.31 X 109 yr. Although the decay to 40Ca is more common, the relative abundance of 40Ca in rocks precludes the use of this isotope for dating purposes. Instead, the abundance of argon is measured and sample age is a function of the 40K/40Ar ratio. Argon is a gas that can be driven out of a sample by heating. Thus, the method is used for dating volcanic rocks that contain no argon after the molten lava has cooled, thereby setting the isotopic "clock" to zero. With the passage of time, 40Ar is produced and retained within the mineral crystals, until driven off by heating in the laboratory during the dating process (Dalrymple and Lanphere, 1969). Unlike conventional 14C dating, 40K/40Ar dating relies on measurements of the decay product40Ar; the parent isotope content (40K) is measured in the sample.

As the abundance ratios of the isotopes of potassium are known, the 40K content can be derived from a measurement of total potassium content, or by measurement of another isotope 39K. Because of the relatively long half-life of 40K, the production of argon is extremely slow. Hence, it is very difficult to apply the technique to samples younger than -100,000 years and its primary use has been in dating volcanic rocks formed over the last 30 million years (though, theoretically, rocks as old as 109 years could be dated by this method). Dating is usually carried out on minerals such as sanidine, plagioclase, biotite, hornblende, and olivine in volcanic lavas and tuffs. It may also be useful in dating authigenic minerals (i.e., those formed at the time of deposition) such as glauconite, feldspar, and sylvite in sedimentary rocks (Dalrymple and Lanphere, 1969). Problems of 40K/40Ar Dating

The fundamental assumptions in potassium-argon dating are that (a) no argon was left in the volcanic material after formation, and (b) the system has remained closed since the material was produced, so that no argon has either entered or left the sample since formation. The former assumption may be invalid in the case of some deep-sea basalts that retain previously formed argon during formation under high hydrostatic pressure. Similarly, certain rocks may have incorporated older "argon-rich" material during formation. Such factors result in the sample age being overestimated (Fitch, 1972). Similar errors result from modern argon being absorbed onto the surface and interior of the sample, thereby invalidating the second assumption. Fortunately, atmospheric argon contamination can be assessed by measurement of the different isotopes of argon present. Atmospheric argon occurs as three isotopes, 36Ar, 38Ar, and 40Ar. As the ratio of 40Ar/36Ar in the atmosphere is known, the specific concentrations of 36Ar and 40Ar in a sample can be used as a measure of the degree of atmospheric contamination, and the apparent sample age appropriately adjusted (Miller, 1972).

A more common problem in 40K/40Ar dating is the (unknown) degree to which argon has been lost from the system since the time of the geological event to be dated. This may result from a number of factors, including diffusion, recrystalliza-tion, solution, and chemical reactions as the rock weathers (Fitch, 1972). Obviously, any argon loss will result in a minimum age estimate only. Fortunately, some assessment of these problems and their effect on dating may be possible. 40Ar/39Ar Dating

One important disadvantage of the conventional 40K/40Ar dating technique is that potassium and argon measurements have to be made on different parts of the same sample; if the sample is not completely homogeneous, an erroneous age may be assigned. This problem can be circumvented by 40Ar/39Ar dating, in which measurements are made simultaneously, not only on the same sample, but on the same precise location within the crystal lattice where the 40Ar is trapped. Instead of measuring 40K directly, it is measured indirectly by irradiating the sample with neutrons in a nuclear reactor. This causes the stable isotope 39K to transmute into 39Ar; by collecting both the 40Ar and 39Ar, and knowing the ratio of 40K to 39K (which is a constant) the sample age can be calculated. Further details are given by Curtis (1975), McDougall and Harrison (1988), and McDougall (1995).

Actually, 40Ar/39Ar dating has no advantages over conventional 40K/40Ar dating for samples that have not been weathered, subjected to heating or metamorphism of any kind since formation, or are free of inherited or extraneous argon. In such cases, dates from 40K/40Ar methods would be identical to those from 40Ar/39Ar methods. In practice, however, there is no way of knowing the extent to which a sample has been modified or contaminated; hence the 40Ar/39Ar method has significant advantages over 40K/40Ar because it is often possible to identify the degree to which a sample has been altered or contaminated, and thus, to increase confidence in the date assigned. Furthermore, several dates can be obtained from one sample and the results treated statistically to yield a date of high precision (Curtis, 1975).

The advantages stem from the fact that the 40K, which yields the 40Ar by decay, occupies the same position in the crystal lattice of the mineral as the much more abundant 39K that produces the 39Ar on irradiation. Heating of the sample thus drives off the argon isotopes simultaneously. Any atmospheric argon contaminating the sample occurs close to the surface of the mineral grains, so it is liberated at low temperatures. Similarly, loss of radiogenic argon by weathering would be confined mainly to the outer surface of a mineral. In such cases the 40Ar/39Ar ratios on the initial gas samples would indicate an age that is too young (Fig. 3.17b). At higher temperatures, the deeper-seated argon from the unweathered, uncontaminated interiors of the crystals will be driven off and can be measured repeatedly as the temperature rises to fusion levels. If such gas increments indicate a stable and consistent age, considerable confidence can be placed in the result. By contrast, conventional 40Ar/39Ar dating on a sample such as that shown in Fig. 3.17b would yield a meaningless age, resulting from a mixture of the gases from different levels.

Plots of "apparent age" calculated from the ratios of 40Ar to 39Ar at airrerent temperatures can indicate much information about the past history of the sample, including whether the sample has lost argon since formation or whether the rock was contaminated by excess radiogenic argon at the time of formation. Such interpretations are discussed further by Curtis (1975) and by Miller (1972). Thus 40Ar/39Ar dating possesses considerable advantages over conventional 40K/40Ar dating methods by providing more confidence in the resulting dates.

40Ar/39Ar dating has been used to assess the age of the major geomagnetic polarity reversal in the Quaternary — the Brunhes-Matuyama (B/M) boundary. Early 40K/40Ar studies had placed the age at -730 ka B.P. but this was questioned by Johnson (1982) and Shackleton et al. (1990), who found that a consequence of tuning the marine oxygen isotope record to maximize coherence with Milankovitch orbital frequencies was to push the B/M boundary back to -780 ka B.P. They therefore argued rather boldly that the hitherto accepted age of 730 ka was probably incorrect. Several subsequent studies deriving 40Ar/39Ar dates from lava flows have supported this assertion (Spell and McDougall, 1992; Baksi et al., 1992; Izett and Obradovich, 1994) or at least demonstrated that the uncertainties in both approaches span the interval 730-780 ka B.P., making the two estimates statistically indistinguishable (Tauxe et al., 1992).

FIGURE 3.17 Schematic plots of *°Ar/39Ar data. Each point in (a) and (b) indicates the age obtained for that increment of argon released as the temperature is increased in steps from 0 °C to the fusion point (1000 °C). In (a) the data show uniform ages for all increments, the plateau indicating a precise age determination. In (b) the ages appear to be progressively older as the temperature rises, indicating loss of argon after original crystallization of the sample so that a precise age cannot be determined; even the oldest age obtained is probably too young (Curtis, 1975).

3.2.3 Uranium-series Dating

Uranium-series dating is a term that encompasses a range of dating methods, all based on various decay products of 238U or 235U. Figure 3.18 illustrates the principal decay series nuclides and their respective half-lives; some intermediate products with very short half-lives (in the order of seconds or minutes) have been omitted. The main isotopes of significance for dating are 238U and 235U, 230Th (also known as ionium) and 231Pa. The ultimate product of the uranium decay series is stable lead (206Pb or 207Pb).

In a system containing uranium, which is undisturbed for a long period of time (~106 yr), a dynamic equilibrium will prevail in which each daughter product will be present in such an amount that it is decaying at the same rate as it is formed by its parent isotope (Broecker and Bender, 1972). The ratio of one isotope to another will be essentially constant. However, if the system is disturbed, this balance of production and loss will no longer prevail and the relative proportions of different isotopes will change. By measuring the degree to which a disturbed system of decay products has returned to a new equilibrium, an assessment of the amount of time elapsed since disturbance can be made (Ivanovich and Harmon 1982). Isotopic decay is expressed in terms of the activity ratios8 of different isotopes, such as

NUCLIDE uranium-238


thorium-230 (ionium)




4.51 x 109 years uranium-235 7.13 x 108 years

2.5 x105 years protactinium-231 3.24 x 104 years

7.52 x 104years thorium-227 18.6 days radium-226



1.62 x 10 years radium-223 11.1 days

3.83 days 22 years lead-207

polonium-210 138 days lead-206 stable

FIGURE 3.18 Decay series of uranium-238 and uranium-235.


8 Concentrations of radioactive isotopes are reported in units of decays per minute per gram of sample. In U-series dating, these rates are considered relative to each other and are referred to as activity ratios.

230Th/238U and 231Pa/235U; for the former, the useful dating range is from a few years to -350,000 B.P., and for the latter 5000-150,000 B.P. (Fig. 3.19). Thermal ionization mass spectrometry (TIMS) revolutionized uranium series dating in the mid-1980s, making very precise analyses of small samples routine; subsequent refinements have made possible dates that are even more precise, enabling corals of last interglacial age to be dated to within ± 1 ka (2 ct error) (Edwards et al., 1987b; 1993; Gallup et al., 1994). Furthermore, in view of the inconstancy of atmospheric (and oceanic) radiocarbon content, 230Th dating of corals can provide results that exceed the accuracy of 14C dates, and have comparable accuracy to the counting of annual growth bands (Edwards et al., 1987a).

In natural systems, disturbance of the decay series is common because of the different physical properties of the intermediate decay series products. Most important of these is the fact that 230Th and 231 Pa are virtually insoluble in water. In natural waters these isotopes are precipitated from solution as the uranium decays, and collect in sedimentary deposits. As the isotope is buried beneath subsequent sedimentary accumulations, it decays at a known rate, "unsupported" by further decay of the parent isotopes (234U and 235U, respectively) from which it has been separated. This is known as the daughter excess or unsupported dating method (Blackwell and Schwarcz, 1995). In sediment that has been deposited at a uniform rate, the 230Th and 231Pa concentrations decrease exponentially with depth. Providing that this initial concentration of the isotopes is known, the extent to which they have decayed in sediment beneath the surface can be related to the amount of time elapsed since the sediment was first deposited (Fig. 3.20).

10' 105 10' SAMPLE AGE (YEARS) FIGURE 3.19 Temporal changes in activity ratios of 23lPa/23SU, 226Ra/230Th, 230Th/23',U, and 234U/238U (Broecker and Bender, 1972).




FIGURE 3.20 (a) Excess 230Th concentrations and (b) excess 23lPa concentrations vs depth in Caribbean coreVI2-122. As the original amounts of 230Th and 23'Pa in freshly deposited sediment can be estimated, the extent to which they have been reduced with depth gives a measure of time since the sediment was deposited. Sedimentation rates are obtained from slopes of the best-fitting regression lines and a knowledge of the decay rate of each isotope (Ku, 1976).


FIGURE 3.20 (a) Excess 230Th concentrations and (b) excess 23lPa concentrations vs depth in Caribbean coreVI2-122. As the original amounts of 230Th and 23'Pa in freshly deposited sediment can be estimated, the extent to which they have been reduced with depth gives a measure of time since the sediment was deposited. Sedimentation rates are obtained from slopes of the best-fitting regression lines and a knowledge of the decay rate of each isotope (Ku, 1976).

This procedure is an example of dating based on the physical separation of the parent and daughter isotopes, with age calculated as a function of the decay rate of the unsupported daughter isotope (Ku, 1976). Another method relies on the growth of an isotope that is initially absent, towards equilibrium with its parent isotope;

this is known as daughter deficiency dating (Blackwell and Schwarcz, 1995) and is most commonly applied to carbonate materials (corals, molluscs, speleothems). It is based on the fact that uranium is co-precipitated with calcite or aragonite from natural waters that are essentially free of thorium and protactinium. Initial values of 230Th and 231Pa in the carbonates are thus negligible. Providing that the carbonate remains a closed system, the amounts of 230Th and 231Pa produced as the 234U and 235U decay will be a function of time, and of the initial uranium content of the sample. In the growth of corals, for example, uranium is co-precipitated from seawater to form part of the coral structure, but thorium concentrations are essentially zero. As the 234U/238U ratio in ocean water is constant at -1.14 (and studies show it has been almost constant over long periods of time) the build-up of 230Th in the coral as the uranium isotopes decay thus provides chronometric control on the time since the coral formed. Providing the coral has not undergone recrystallization (thereby incorporating anew more uranium) this approach can provide useful dating control from a few years to -350,000 yr B.P. with extremely high precision (Edwards et al., 1987b). New Guinea (Huon peninsula) coral samples demonstrate 2a errors of only 30-80 yr on late Glacial/early Holocene materials, considerably smaller than the errors associated with AMS 14C dates on the same samples (Edwards et al., 1993). Indeed, U-series dating has been used to calibrate the radiocarbon timescale back to >25 ka B.P. (Bard et al., 1993; see Section The method has been widely used to date raised coral terraces and hence to provide a chronologically accurate assessment of glacio-eustatic changes of sea level, with broad implications for paleo-climatology (Bard et al., 1990; Edwards et al., 1993; Gallup et al., 1994).

Attempts have also been made to date molluscs in the same way as coral but the results are generally inconsistent (Szabo, 1979a). The main problem is that molluscs appear to freely exchange uranium post-depositionally (i.e., they do not constitute a closed system) with the result that fossil molluscs commonly have higher uranium concentrations than their modern counterparts. Hence, the resulting thorium and protactinium concentrations are not simply a function of age (Kaufman et al., 1971). 230Th/234U dates on bone have also been attempted (Szabo and Collins, 1975) but similar problems have been encountered. Repeated checks with different dating methods suggest that accurate dates have been obtained on only 50% of shell and bone samples to which 230Th/234U and 231Pa/234U dating methods have been applied (Ku, 1976). "Open system" models have been developed to compensate for post-depositional exchange problems (Szabo and Rosholt, 1969; Szabo, 1979b) but many assumptions are required that reduce confidence in the resultant dates. Indeed, Broecker and Bender (1972) categorically rejected dates obtained on any kind of molluscs and concluded that only corals can give reliable U-series dates. However, other work has shown that U-series dates on Arctic marine molluscs can provide valuable minimum age estimates when considered in relation to amino-acid data on the same samples (Szabo et al., 1981).

Much more confidence can be placed in uranium-series dates obtained on carbonate samples from speleothems (stalactites and stalagmites). Such deposits are dense and not subject to post-depositional leaching. Because 230Th is so insoluble, the water from which the speleothem carbonate is precipitated can be considered to be essentially thorium-free. Hence, providing that the initial uranium concentration is sufficient, measurement of the 230Th/234U ratio will indicate the build-up of 230Th with the passage of time (Harmon et al., 1975). The main problem is to determine reliably the initial 234U/238U ratio, and to ensure that detrital 230Th has not contaminated the sample, thereby negating the assumption that the initial thorium content is zero (see Section 7.6.2).

On a much shorter timescale, unsupported 210Pb may also be used as a chronological aid. The 210Pb is derived from the decay of 222Rn following the decay of 226Ra from 230Th (see Fig. 3.18). Both 226Ra and 222Rn escape from the Earth's surface and enter the atmosphere, where the 210Pb is eventually produced. The 210Pb is then washed out of the atmosphere by precipitation, or settles out as dry fallout, where it accumulates in sedimentary deposits and decays (with a half-life of 22 yr) to stable 206Pb (Fig. 3.21). Assuming that the atmospheric flux of 210Pb is constant, the decay rate of 210Pb to 206Pb with depth can be used to date sediment accumulation rates (Appleby and Oldfield, 1978, 1983). It is of value only in dating sediments over the last -200 yr, but this may be of particular value in confirming that laminated sediments are true varves (i.e., annual) or in confirming that core-tops are undisturbed, enabling floral and faunal contents to be calibrated with instrumental climatic data to derive accurate transfer functions for paleoclimatic reconstructions. The 210Pb has also proved useful in dating the upper sections of ice cores and hence allowing estimates of long-term accumulation rates to be made, though nowadays this is rarely carried out (Crozaz and Langway, 1966; Gaggeler et al., 1983). Problems of U-series Dating

The major problems in U-series dating have already been alluded to briefly. First, an assumption must be made as to the initial 230Th/234U, 234U/238U, and/or 231Pa/235U ratios in the sample. In the deep oceans this may not be a significant problem, as modern oceanic ratios are known to have been relatively constant over long periods of time, but in terrestrial environments such as closed inland lakes, this assumption is far less robust. The second problem concerns the extent to which the sample to be dated has remained a closed system through time. Recrystallization of aragonitic carbonate to calcite may provide some guidance, but as discussed in Section this is not always reliable. At present only carbonate dates on coral seem to be consistently reliable. Reliability can be checked by obtaining activity ratios for different isotopes from the same sample. If the sample has remained "closed" the dates should all cross-check and be internally consistent (see Fig. 3.19).

3.2.4 Luminescence Dating: Principles and Applications

Luminescence is the light emitted from a mineral crystal (mainly quartz and feldspars) when subjected to heating or when exposed to light. Light emitted in response to heating is referred to as thermoluminescence or TL, light emitted in response to radiation in the visible or infrared parts of the spectrum is termed optically stimulated luminescence (OSL), or infrared stimulated luminescence (IRSL), respectively. In each case, the quantity of light emitted is related to the

210p b t

Rain V«V Scavenging Xv.V-,

0 0

Post a comment