October 6, 2004 | David F. Coppedge

How Are Radioactive Dates Determined?

To most of us, the practice of radioactive dating seems like a highly-technical, incomprehensible skill that nevertheless (we are told) yields absolute ages of things.  We don’t know exactly how they arrive at the results, but are led to trust them because very smart people get their answers using hard science with extremely accurate equipment.  It might be helpful to look over their shoulders and see how it’s done.  A couple of recent papers dealt with uranium-lead dating, the kind of method that typically yields ages in the millions of years.

  1. A paper in Science last week1 by an international team of earth scientists discussed evidence for extinct plutonium-244 in Australian rocks dated at 4.2 billion years old.  Plutonium-244 has a half-life of 82 million years.  The authors, Turner et al., begin by assuming Pu-244 was well mixed within the cloud that presumably formed the solar system.  Since the Australian rocks are assumed to be among the oldest on earth, they wanted to determine the ratio of plutonium to uranium (Pu/U) for clues to the early evolution of the earth.  Xenon-136 would have been produced primarily by the more rapidly-decaying plutonium-244 in the early years of the earth, then the slower-decaying uranium-238 would gradually have predominated; but the ratio is so low, .004 to .008, that U tends to overwhelm the contribution from Pu unless the rocks are older than 3.8 billion years, the authors claim.
        They extracted eight tiny zircon crystals just 50-200 millionths of a meter in size, from rocks they claim are up to 4.1 to 4.2 billion years old.  Detecting the xenon in such a small grain – a quadrillionth of a cubic centimeter – is beyond the range of most instruments, “comparable to blank levels and sensitivities of conventional noble gas mass spectrometers” (i.e., the instrument would show no xenon at all).  They developed what they claim is a more sensitive instrument able to get two orders of magnitude below that low detection threshold, and found a few thousand atoms of xenon.  They measured the xenon isotope ratios from the eight zircons, and graphed their results.  Only two of them fell on the expected Pu/U ratio line expected from the age of the rocks, compared with ratios measured in meteorites which presumably predate the formation of the earth.
        The other six were “discordant,” off expectations by 24% to 97%.  Their explanation for these is: “This could be the result of preferential loss of the earlier-formed Pu xenon or the result of chemical fractionation of Pu and U during or before the formation of the zircons.”  How can this be, since they say “Xe is at least as strongly retained as Pb” [lead, the ending fission byproduct]?  Well, lead has also been found to leach out of zircons, and these crystals have been through a long, wild ride: “Nevertheless, Pb loss associated with metamictization is commonly observed in zircons, and, given the antiquity and complex history of the ancient detrital zircons, it is likely that loss of Xe will also have occurred in a portion of our samples.”.  This history could have included “diffusion or recrystallization events” and other metamorphic processes.  Most of the loss would have been early on, when plutonium production of xenon dominated, according to their model, so that explains why the ratio fell short of expectations.  “To be more definitive requires an additional relationship between the time of Xe loss and the degree of loss,” they suggest.  So their study can only claim partial success, and will require more work: “The highest implied Pu/U ratio is within the range of estimates from meteorites, but, in order to quantify a global Pu/U ratio for the early Earth, future work will require an improved understanding of the geochemical behavior of Pu relative to U and the rare earth elements in zircon crystallization.”
  2. A paper in the October issue of Geology1 dated the Devonian-Carboniferous boundary to four significant figures, 360.7 million years, with uranium-lead dating, from zircons in Germany.  A closer look at the German team’s methods of selection and treatment of samples, however, indicates a number of assumptions were made.  First, the target date of the period was established by biostratigraphy, or use of index fossils (see 05/21/2004 headline).  Second, since fossils don’t typically contain uranium, the radiometric dates have to be taken from non-fossiliferous material, like volcanic ash that might be in and around the fossils (as in this case) or removed from them (in many other cases).  Third, the zircons were subjected to air blasts, then heating and soaking in acid solutions for days.  Fourth, anomalous dates were thrown out and only 5 of 13 were kept.  The ones thrown out yielded impossibly old dates, which the team shrugged off as a bit of surprise:

    On the basis of 13 analyses (single zircons or zircon fragments), a younger zircon generation of 5 analyses is distinguished from older zircon generations (Table 1).  The latter, obviously inherited [i.e., formed in earlier periods], yielded 207Pb/206Pb ages of 444 to 2044 Ma (Table 1).  The abundance of Precambrian ages is a remarkable feature; note that no inherited zircons were detected in the study of Claou�-Long et al. (1992).  The error ellipses of the older zircons are clearly separated from a tight concordant cluster of the five youngest zircon analyses, which yield a 206Pb/238U concordia age of 360.5 � 0.8 Ma (Fig. 2A).  This age is interpreted as the crystallization age of the comagmatic zircon population and thus the time of eruption of the ash.  Comagmatic zircons are only a small fraction of the total zircon population.  It is possible that the youngest zircon generation occurs as micrometer-sized rims around inherited zircons as well, but these new growth zones were removed by the air-abrasion procedure prior to the dissolution of the grains.

    It might be surprising to outsiders to see the amount of pretreatment of samples that goes on as standard procedure in radiometric dating:

    Zircons selected for analyses were subjected to air abrasion (Krogh, 1982), and most samples were additionally cleaned for 2 h in concentrated HF-HNO3 (4:1) at 80°C to remove attached impurities.  After washing in 7N HNO3 at 80 �C for 25 min, individual grains were placed in multisample Teflon microcapsules and dissolved for at least 4 days in concentrated HF-HNO3 (4:1) at 180°C [285°F].  Subsequently, dissolved zircons were spiked with a mixed 233U-205Pb tracer solution, dried at 80°C, redissolved in 6N HCl [hydrochloric acid], and equilibrated at 180°C for 1 day.  After drying at 80°C, the samples were loaded on a single Re filament using a mixture of silica gel and 6N HCl-0.25N H3PO4.

    That was just for starters.  The team also “corrected” their measurements; for instance, “For each charge of samples, the maximum Pb blank was assumed to be equivalent to the total amount of nonradiogenic Pb in the analysis of the most radiogenic sample.”  Also, the measurements were done on extremely tiny grains, millionths of an inch in size, with lead masses on the order of tenths of picograms (billionths of a gram): “It was thus necessary to reduce the Pb blank as much as possible… by extremely careful sample handling.”  When measurements were still too low, assumptions were made: “The U blank was too small to be measured and was thus assumed to amount to 20% of the individual Pb blank, based on experience with the analysis of milligram-sized samples” (i.e., they assumed that their samples followed curves established for samples ten million times larger).
        The date concluded for the Devonian-Carboniferous boundary, 360.7 million years, was not calculated directly.  It was interpolated from the ages that remained after the air blasts, acid, heat, and interpretation of selected samples.


1Turner et al., “Extinct 244Pu in Ancient Zircons,” Science, Vol 306, Issue 5693, 89-91, 1 October 2004, [DOI: 10.1126/science.1101014].
2Trapp et al., “Numerical calibration of the Devonian-Carboniferous boundary: Two new U-Pb isotope dilution-thermal ionization mass spectrometry single-zircon ages from Hasselbachtal (Sauerland, Germany),” Geology, Vol. 32, No. 10, pp. 857�860, doi: 10.1130/G20644.1.

Notice what Turner’s group did.  First, they assumed what they need to prove: that the rocks were really 4.2 billion years old.  The age of the solar system (4.56 billion years old) and the age of the meteorites was not open to negotiation: these were givens, assumed from the start.  Then notice the extremely minute amounts they had to work with: crystals weighing a few millionths of a gram.  The xenon they were looking for was below the detection threshold of most instruments; how can anyone be sure that their laser instrument, which detected a few thousand atoms in the crystal, did not disrupt the atoms in the process?  (Xenon, after all, is a gas.)  Then notice that only 25% of their 8 samples met expectations, so the rest had to be explained away.  Well, look at the explanation!  The crystals were subject to violent, metamorphic processes of heating and recrystallization, and even though lead is more easily leached from the samples, the lead remained somehow and the xenon was lost.
    These eight tiny zircons were found in detrital deposits.  According to the Merriam-Webster dictionary, detritus is: “(1) loose material (as rock fragments or organic particles) that results directly from disintegration (2) a product of disintegration, destruction, or wearing away.”  How can any geochemist possibly know these itty bitty crystals, after presumably billions of years of plate tectonics, volcanism, erosion and weathering, hark back from the birth of the earth?  How can they know the composition of a presumed solar nebula, and the amount of processing and mixing of elements that occurred before the crust of the earth solidified?  Does any reader feel any confidence that this experiment tells us anything at all about the history of the earth billions of years ago?  Don’t be a sucker.  Zircons exist in the present, not in the past, and they don’t come with dates stamped on them.  To weave a story about what these rocks were doing 4.2 billion years ago requires many assumptions which are impossible to prove.  It also requires ignoring many other well-understood processes that show the earth could not possibly be that old.
    To show that the Turner et al. paper was not an isolated case of cherry-picking data, the second paper in Geology should support the assertion that radiometric dating is fraught with circular reasoning, selective evidence and extrapolation (see also 09/20/2004 where Richard Kerr points out some of the nasty “little details that don’t make it into the literature,” especially the picking and choosing of data they like).  Again, this team tossed out over half the samples that yielded dates too old for their needs.  Some were found to be almost six times as old, which would have put them deep into the Precambrian.  To end up in this volcanic ash deposit, therefore, those older zircons would have had to survive at least one trip through a volcano’s throat, maybe many (after all, a lot can happen in 1.684 billion years, plus or minus 1.683995 billion).  The team just whisked away this difficulty with the statement, “the abundance of Precambrian ages is a remarkable feature.”  OK, let’s hear some more remarks.  In addition, six of the ten samples taken from another boundary bed “are based on pyramids broken off from whole zircon crystals, and these fragments are typically free of inherited core material,” according to more assumptions.  We think readers who hear about “absolute ages” determined from radiometric dating need to see the amount of hand-waving and hocus pocus that goes on in the inner sanctums of the Darwin Party chemistry labs.
    The ratio that counts in any dating method is not the Pu/U ratio or the U/Pb, but the O/A ratio (observations to assumptions).  A conservative dating approach would observe present processes carefully and measure the rate of change, then try to set an upper limit on how long that process could operate, with a minimum of extrapolation: “this phenomenon cannot be more than x years old” (because, sooner or later, the source will run out, or the product will be saturated).  A liberal approach to dating, on the other hand, requires a lot of extrapolation.  It tries to set a lower limit on the age of something: “this phenomenon cannot be less than y years old.”  The conservative approach has a vastly higher observation-to-assumption ratio.  For instance, we’ve only known about radioactive decay for around 100 years.  Conservatively, we can extrapolate backward or forward a little, but should exercise caution beyond one or two orders of magnitude.  Most evolutionary geologists, however, recklessly extrapolate the observed rates of decay by seven orders of magnitude!
    Even if radioactive decay rates could be trusted so far back, knowing that our theories of fundamental physics continue to undergo revolutions (e.g., dark energy, exotic particles, string theory), this paper illustrates that no one can know the initial conditions or subsequent processes that might have altered the samples, without making other assumptions.  Counting atoms and measuring current decay rates may be hard science, but the conclusions are embedded in an assumption-ridden context.  Astrologers were very good observers of the motions of the planets, but the accuracy of their measurements did not justify their assumptions.  Before accepting any conclusion pronounced by the wizards, always separate the observations from the assumptions.  Respect observations; doubt assumptions.

(Visited 31 times, 1 visits today)
Categories: Physical Science, Physics

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.