Exploring the wonders of geology in response to young-Earth claims...

Never been here? Please read my guidelines and background posts before proceeding!

Wednesday, March 2, 2011

Inventing the isochron: Steve Austin, Andrew Snelling, and the Cardenas Basalts of the Grand Canyon

At the beginning of the year, I reviewed an article that cited anomalously old radiometric dates for historical lava flows to argue against the validity of the Potassium-Argon (K-Ar) dating method. Therein, I proposed that young-Earth authors ubiquitously employ the following approach in their discussions of radiometric dating:

1) Remind readers that several assumptions are inherent to radiometric dating methods.
2) Provide a case-in-point where at least one of those assumptions was falsified.
3) Extrapolate the proven uncertainty to the rest of geochronology without qualification.

Earlier this month, Answers in Genesis (AiG) reposted a study by Dr. Steve Austin and Dr. Andrew Snelling, entitled “Discordant Potassium-Argon Model and Isochron “Ages” for Cardenas Basalt (Middle Proterozoic) and Associated Diabase of Eastern Grand Canyon, Arizona”. The study was originally published in 1998, but one may assume from the recent posting that the authors consider the information sufficiently up-to-date. With that being said, I believe this article provides good opportunity to 1) test my proposal regarding the young-Earth approach, 2) discuss the validity of the K-Ar dating method, and 3) determine whether the young-Earth geologists offer a valid explanation for the results of radiometric dating.

How old are the Cardenas Basalts?

More than a mile of sedimentary rock is present below the Phanerozoic (less than 542 m.y. old) strata of the Grand Canyon—the latter characterized by their near-horizontal, “pancake-layer” stratigraphy. The Precambrian-age sediments were deposited, tilted, and subsequently eroded to form the Great Unconformity. The Cardenas Basalt unit is actually comprised of several lava flows, which overly the Dox Formation and cap the Unkar Group as a whole. Overlying the weathered surface of the Cardenas Basalt is the Nankoweap Formation—a coarse-grained sandstone unit. For those interested in detailing the Precambrian history of the Grand Canyon, the date of the Cardenas Basalt lava flows is of upmost importance, because it provides a minimum age for the rocks below and a maximum age for the rocks above. McKee and Noble (1976) reported a Rb-Sr isochron age for the lava flows of 1090±70 Ma, which was later refined by Larson et al. (1994) to 1103±66 Ma. The latter date has been used by subsequent researchers as the most accurate age of the formation.

Also preserved within Precambrian sediments are mafic dikes and sills, which cut through the Unkar Group. Elston and McKee (1982) provided a Rb-Sr isochron age of 1070±30 Ma for the mafic dikes/sills, which is concordant with the isochron age for the Cardenas Basalts. Although no field evidence is available for a genetic relationship between the intrusive rocks (dikes/sills) and the Cardenas Basalts, the similar age (Elston and McKee, 1982; Timmons et al., 2001) and mineralogical composition (Hendricks and Lucchitta, 1974) corroborate that possibility.

Discordant radiometric dates using the Potassium-Argon (K-Ar) method

A few years after the initial Rb-Sr age was published, Elston and McKee (1982) provided whole-rock and mineral K-Ar ages for the Cardenas lava flows and associated dikes. Their analysis yielded a range of ages (790–958 Ma), which were significantly younger than the 1090 Ma age published by McKee and Noble (1976). The result is not entirely unexpected, however, because the K-Ar method assumes perfect retention of argon—a noble gas—since the minerals cooled below the blocking temperature. This condition is not always met, as subsequent geological disturbances, such as heating events or deep burial, can easily promote the loss of argon and cause the samples to appear younger. Elston and McKee (1982) interpreted the ages to reflect partial loss of argon during subsequent rifting in the Neoproterozoic, when temperatures climbed in the deeply buried lava flows and dikes.

Larson et al. (1994) complemented the work of Elston and McKee (1982) with two new K-Ar ages from low-K samples (1.2–1.3 wt% K2O). The apparent ages of 957 Ma and 1013 Ma are significantly closer to the crystallization age of the basalts (1090 Ma), and caused Larson et al. (1994) to consider the relationship between potassium concentration and apparent K-Ar age. An inverse exponential relationship between weight percent K2O and K-Ar age was consistent with the hypothesis that burial alteration caused higher argon-loss in the more felsic, potassium-rich samples, which contained more glassy groundmass. Moreover, their results were consistent with Elston and McKee’s (1982) hypothesis that a Neoproterozoic heating event caused the partial loss of argon, and thus a partial resetting of K-Ar clocks.

Below is a plot of K-Ar ages of the Cardenas Basalts against K-content, including new data from Austin and Snelling (1998). With the exception of three outliers (in green), the logarithmic relationship explains most of the variance, reflecting the fact that K-rich rocks are more susceptible to alteration that undermines the retentivity of argon. The relationship is not perfect, however, because the rocks are spread over a wide geographic region and stratigraphic position, and may have undergone varying degrees of alteration to begin with.

Relevant studies since the publication of Austin and Snelling (1998)

Dating of the Cardenas Basalt and related magmatic events is not only important to stratigraphers. Paleogeographers also use the magnetic signatures of these rocks (along with their ages) to plot the movement of the continents throughout time. Timmons et al. (2001) provided 40Ar/39Ar dates for the mafic dikes and Cardenas Basalts. Interpreted age spectra range from 771–988 Ma, suggesting non-uniform loss of Ar during burial (2–3 km depth) and heating (<250°C) as a result of rifting between 800–742 Ma. Individual steps during 40Ar/39Ar analysis indicate one of the dikes is at least 1050 Ma, however, consistent with the Rb-Sr isochron ages reported earlier. Weil et al. (2003) reported 40Ar/39Ar dates from individual biotite minerals in two mafic dikes. The analysis revealed the presence of excess argon in the initial steps, but the authors interpreted an age of 1090.6±4.5 Ma from the concordant, higher temperature steps. Finally, Timmons et al. (2005) reported a 40Ar/39Ar age of 1104±2 Ma for an individual biotite within a mafic sill (most precise age yet for the intrusive rocks associated with the Cardenas Basalts), along with U/Pb and 40Ar/39Ar ages (1120–1270 Ma) for detrital muscovite and zircons within the underlying Dox Formation.

Overall, the range of K-Ar and 40Ar/39Ar ages accurately reflects the geologic history of Precambrian strata in the Grand Canyon. While their discordance presents a challenge to researchers in the field, adequate familiarity with the assumptions and limits of the K-Ar and 40Ar/39Ar dating methods allows one to interpret the dates consistently within the conventional framework.

K-Ar isochron of Austin and Snelling (1998)

If you are not already familiar with the use of isochron dating, it may be helpful to review the process (Chris Stassen provided a technical, but well illustrated, explanation here). The isochron method is applicable to cogenetic (i.e. sourced from the same magma) igneous rocks, in which the initial concentration of radioactive parent element (40K) varied between samples, but the ratio of radiogenic daughter element (40Ar) to its stable isotope (36Ar) did not. If the assumptions of the method are met, one can determine the age of the sample as well as the initial concentration of radiogenic daughter (40Ar). In other words, the isochron method does not need to assume that no radiogenic daughter was initially present, and is superior to the traditional K-Ar method in this regard.

As mentioned, loss of argon subsequent to crystallization causes samples to appear younger when dated by the K-Ar method. If the loss of argon was consistent between samples (for example, if all samples lost all their argon), a K-Ar isochron can be used to infer when argon was lost, because such an event would effectively reset the isochron age. If the former hypothesis holds—that Neoproterozoic rifting caused argon loss as a result of heating—a K-Ar isochron should reveal an age near 800–742 Ma (coincident with rifting; Timmons et al., 2001).

Sample quality

To Austin and Snellings’ (1998) own admission, most of the rock samples are highly altered. Moreover, several of the samples (especially those high in potassium) contain abundant glass. When volcanic glass is altered, its ability to retain argon (an assumption of the K-Ar method) is significantly lowered. Faure and Mensing (2005, p. 121) put it this way:

“Samples that have been altered or that contain devitrified glass...and xenoliths or xenocrysts should be avoided...the Ar retentivity of devitrified or hydrated glass is questionable.”

From the outset, there is no reason to expect that the whole-rock samples of Austin and Snelling (1998) met the conditions assumed by the K-Ar method. If anything, they should be analyzing mineral isochrons, where some quality control is practical. Instead, we must expect that altered volcanic glass sufficiently retained the argon over time. If it did not, however, we might expect to see significant scatter in the isochron plots. In fact, that is exactly what we find.

K-Ar isochron for Cardenas Basalt samples

Austin and Snelling (1998) combined published K-Ar dates along with thirteen K-Ar dates from their own samples (submitted to various laboratories), and reported the data in Table 1 of their paper. Figure 5 contains their results plotted as a K-Ar isochron. Along with their new data, they plotted a “reference isochron” of 1100 Ma to emphasize strong discordance between the two, before any discussion of the physical reasons behind the discrepancy. Unfortunately, most of their readers are likely to have lost interest by this point, and might conclude that the study offers evidence against the conventional interpretation (or that accelerated nuclear decay offers a valid explanation).

To evaluate the claims of Austin and Snelling, it is helpful to analyze their data in steps. [Note: a least-squares regression is used for simplicity and lack of variance data, but illustrates the points sufficiently well.] Below is a K-Ar isochron using only data published prior to Austin and Snelling (1998; listed there in Table 1). Treating sample LPM:4E as a statistical outlier, the good fit (R2 = 0.998), intercept (698.11), and estimated age (781 Ma) are all consistent with systematic loss of argon, or equilibration of argon isotopes in an aged system, around the onset of Neoproterozoic rifting.

When new data from Austin and Snelling (1998) are plotted, the result is a poorly fitting isochron (R2 = 0.902) with a much younger age (554 Ma), and much higher intercept (1904). Moreover, the slope is heavily influenced by samples A:C-16 and A:C-19, which are extremely rich in potassium and altered volcanic glass (i.e. they have a low retentivity of argon, represent highly fractionated end-members, and may have been enriched in potassium during alteration). Excluding both samples from the isochron results in a greater age (873 Ma), but poorer correlation (0.84) and unrealistic intercept (-1049). In both cases, there is good reason to believe these samples did not meet the physical requirements of an isochron, which assumes: 1) equal values of 40Ar/36Ar at t = 0; 2) similar magma source; and 3) systematic loss of argon between all samples, if such occurred.

When all data are combined, the resulting “isochron” age of the Cardenas Basalt unit is 657 Ma (below), but with a relatively poor fit to the data (R2 = 0.895). Austin and Snelling (1998) use a weighted regression to calculate an age of 756 Ma (see Fig. 5, Austin and Snelling, 1998), but the poor fit of the linear equation is apparent from their plot as well. If taken as a meaningful indicator of age, the plot suggests that excess argon (40Ar/36Ar = 787) was present at crystallization, or that isotopic equilibration took place some time after (as a result of heating/alteration). But there is more to the story.

For a K-Ar isochron, 40Aro is the initial daughter (Do), and while the method does not assume a zero initial concentration, any enrichment must have been uniform throughout the samples. In their discussion on the application of isochrons, Faure and Mensing (2005, p. 59) note “experience indicates that, in some cases, lava flows erupted by the same volcano in a short interval of time have different values of Do.” Lava flows that now comprise the Cardenas Basalts vary widely in their chemistry and mineralogy, and reflect varying degrees of crustal assimilation (Larson et al., 1994). If excess argon was a factor (both isochrons—as well as data from Weil et al., 2003—suggest that it was), it is highly unlikely the enrichment was uniform between flows, and hence very likely that the isochron has no “age” significance.

Although there is good statistical and geological basis to reject the isochron as having little to do with crystallization age (or even an argon-resetting age), the new data add nothing to the geochronological discussion of the Cardenas Basalts. The reason is that the isochron, if real, indicates that argon loss took place near 756 Ma, similar to the hypothesized age of alteration (see above; Timmons et al., 2001).

On the other hand, if one assumes my interpretation that sample LPM:4E is a statistical outlier, and that samples A:C-16 and A:C-19 should not be included in the calculation (high-K, high glass, highly altered), the resulting isochron is highly concordant. A least-squares regression line (see below) yields an R2 value of 0.985 and an intercept of 308.7 (very near atmospheric value of 295.5). If the physical conditions were met, the plot suggest a complete loss of argon from the suite of rocks at 794 Ma—exactly when Neoproterozoic rifting was in full swing, the geothermal gradient was high (i.e. the rocks were reheated), and movement along the Butte fault promoted hydrothermal alteration of the glass-rich rocks, which undermined the argon retentivity of K-rich samples. This interpretation is not conclusive, being contingent upon field and petrographic inspections of the samples, but offers a valid explanation for the data of Austin and Snelling (1998) within a conventional geological framework.

Isochron age for Precambrian dikes and sills of the Grand Canyon

Austin and Snelling (1998) also constructed a K-Ar isochron using only the seven mafic dike/sill samples (although one sample, EM:(Tap), was excluded as an outlier), and calculated an age of 676±35 Ma, using a weighted regression line. Visual inspection of their plot (Fig. 6 in Austin and Snelling, 1998) reveals the poor fit of their isochron. For example, sample A:DI-10 appears as a complete outlier—the line is drawn outside of 2σ uncertainty for that point—even though it’s included within the age calculation! If A:DI-10 is treated as the only outlier, the resulting age (717 Ma) and intercept (257.9) are more reasonably within the confines of a meaningful isochron.

Despite the past discussion of argon loss during burial alteration and heating (especially around 800 Ma), the authors again plot a ‘reference isochron’ of 1100 Ma to show the ‘strong discordance’. For any geologist working in Precambrian geochronology of the Grand Canyon, these results would not be unexpected and, in fact, only corroborate previous hypotheses regarding the discordance. While the isochron age is broadly consistent with total loss of argon during a heating event ~800-742 m.y. ago, however, a few details should be considered.

Evolving magma chamber (source)
First, as the authors point out, the slope (and thus age) of the isochron is strongly determined by the high-potassium granophyre samples. The mineralogy and chemistry of the granophyre is due, however, to extensive fractional crystallization in the magma chamber, which may also have been accompanied by crustal assimilation (Larson et al., 1994). An isochron date, using the K-Ar method, assumes that each sample had an equal 40Ar/36Ar ratio at t = 0, but it is unlikely that this condition was met if crustal material was incorporated (the ratio is much higher in rocks surrounding the dikes/sills) or if the intrusive magma varied in viscosity and retentivity of argon during cooling. Excluding both granophyre samples from the isochron increases the estimated age to ~823 Ma, but with higher uncertainty.

Statistical significance of the isochron, and physical meaning of its intercept
The coefficient of determination (R2) for the isochron of Austin and Snelling (1998) is not reported, but appears to be less than desirable, as mentioned above. A least-squares regression line calculated from their reported data (treating A:DI-10 as an outlier) is 0.93, which, although high, is on the low end for most published isochrons. Moreover, R2 is highly affected by the granophyre samples, which, if excluded based on the discussion above, reduces the coefficient to 0.88. In other words, it is very likely that the isochron reflects neither a true crystallization age nor age of complete resetting.

Austin and Snelling (1998) reported the intercept of the isochron as 453, which would suggest the presence of excess argon during crystallization. Given the poor fit of the isochron, however, more likely scenarios are: 1) excess argon was present in each sample, but not to the same degree, or 2) argon was lost from each sample, but not to the same degree. One means to test this independently is through the use of 40Ar/39Ar dating, which can account for excess argon or reveal differential argon loss. Reported 40Ar/39Ar ages for the mafic dikes range from 771–988 Ma (Timmons et al., 2001) and age spectra are “highly disturbed”, corroborating the latter possibility.

Alternative explanations of Austin and Snelling

Once the geochronological data is properly evaluated, there is little room for discussion on ‘discordance’ of K-Ar isochrons. Nonetheless, the authors evaluated the possibility of argon resetting as the result of the young K-Ar isochron age. Strangely, they used a 516 Ma isochron age (Fig. 3, which is not really an isochron) as the “best” apparent age for the Cardenas Basalts, even though the plot contains several outliers (beyond 2σ from the regression line) that were included in the calculation. Also, the plot cannot account for extraneous argon. Without mention of the real isochron age suggested by their data, they argue as follows:

“...an age of 516±30 Ma is Upper Cambrian, not Proterozoic. The conventional interpretation of Grand Canyon Upper Cambrian stratigraphy at Basalt Canyon is that the Tapeats Sandstone (Upper Cambrian) was sitting level on the floor of an ocean just above the Great Unconformity and just a few hundred meters above 12° dipping Cardenas Basalt. No trace of the assumed Upper Cambrian heating event is seen in the Tapeats Sandstone at Basalt Canyon. Therefore, the reset model has incorrect timing and fails to explain the K-Ar data within conventional geology.”

Of course, such an age is not warranted by their data, so the argument is not worth considering. Austin and Snelling continue to describe argon leakage, inheritance, and mixing models—all of which are unnecessary explanations for data that don’t form an isochron!

Accelerated nuclear decay—steaks ‘well done’, anyone?

Not surprisingly, Austin and Snelling present at least one completely absurd scenario: that the rate of radioactive decay was significantly higher in the past, and more so for the Rb-Sr system than for K-Ar. They pose the following question:

“Could 87Rb decay and/or 40K decay constants be altered to make the data be interpreted in a concordant way? The mathematics simply requires a change to one or both decay constants and they could be concordant.”

And the answer is: no. The mathematical fix is not as simple as they propose and is falsified quickly when all data are considered. If we assume that Rb decayed at ~twice the rate of K (to force concordance on the 1103 Ma and 516 Ma isochron ages, respectively), then the 40Ar/39Ar biotite ages of 1102 Ma (Timmons et al., 2005) and 1090 Ma (Weil et al., 2003) are no longer concordant from any perspective. Furthermore, magmatic events around 1100 Ma occurred throughout the whole North American continent and have been dated concordantly by the K-Ar, Rb-Sr, and U-Pb methods (Larson et al., 1994; Timmons et al., 2005). These data simply make no sense if decay rates changed to different degrees at any point in Earth history.

Of course, the best reason to reject outright any notion of accelerated nuclear decay is the heat problem. In essence, radioactive decay releases heat, and proposing that ~1 billion years worth of decay occurred in a matter of years would cause significant problems for any living creatures in the universe at that time (the heat released would be enough to melt the Earth several times over). Young-Earth researchers are aware of this problem, but have offered no valid solution. For more information, Dr. Larry Vardiman of ICR summarizes the problem here, while the American Scientific Affiliation has provided several technical discussions here on their website.


Within their paper, Austin and Snelling (1998) remind their readers of the assumptions behind the K-Ar dating method (for example, that it “assumes no radiogenic 40Ar was present when diabase and lavas cooled to form rocks”). They proceed to demonstrate that K-Ar ages range over several hundred million years, and that “wide variation in model ages remains unexplained” (contrary to published literature elsewhere on the topic). Without any specific discussion of the meaning behind isochron ages—what it might represent or how one is recognized—they plot their results as evidence of discordance between the K-Ar and Rb-Sr isotope systems. For the uninformed reader, it may seem as though a rough, linear trend in data points constitutes a valid isochron.

Differential loss of argon during burial alteration, and even interaction with hydrothermal fluids during movement along the Butte fault, can account for the discrepancy in K-Ar dates, relatively poor statistical significance in the isochron, and disturbed 40Ar/39Ar age spectra. Thus Austin and Snellings’ new K-Ar data are not inconsistent with previous hypotheses regarding the geologic history of these rocks and their deviance from the accepted age of 1103 Ma. When the most potassium-rich samples (which contain abundant volcanic glass and plot as outliers) are removed from the calculation, the resulting isochron is statistically significant and suggests resetting of the argon system near 794 Ma—inline with the age of thermal disturbance to the region.

Even apart from the model above, there is good geological and statistical foundation on which to reject the isochron from Austin and Snelling (1998) as an indicator of the rocks’ age (crystallization or metamorphism). Coincident magmatic events are recorded on the North American continent, and date near 1100 Ma using several methods (Rb-Sr, U-Pb, and K-Ar). Furthermore, several good 40Ar/39Ar dates (with undisturbed age spectra) are available for the dikes/sills, and agree with the Rb-Sr isochron age of 1103 Ma (Weil et al., 2003; Timmons et al., 2005). For the argument of Austin and Snelling (1998) to have any relevance, they must be able to account for this data. Instead, they propose the unrealistic “model” of accelerated nuclear decay (i.e. change in decay rates) to account for the apparent discordance, despite the fact that it would contradict regional K-Ar and Rb-Sr data already available to them (e.g. Larson et al., 1994). Despite their in-depth, technical discussion of the isotope geochemistry and petrology of Grand Canyon samples, the conclusions of Austin and Snelling (1998) are the result of bad scientific practice and a propagandist effort to dissuade uninformed readers from lending any credibility to geochronology.

References Cited:
Austin, S.A., and Snelling, A.A., 1998, Discordant Potassium-Argon Model and Isochron “Ages” for Cardenas Basalt (Middle Proterozoic) and Associated Diabase of Eastern Grand Canyon, Arizona: Proceedings of the Fourth International Conference on Creationism, p. 35-51.

Elston, D.P., and McKee, E.H., 1982, Age and correlation of the late Proterozoic Grand Canyon disturbance, northern Arizona: Geological Society of America Bulletin, v. 93, p. 681-699.

Faure, G., and Mensing, T.M., 2005, Isotopes: Principles and Applications (Third Edition), Wiley, 897 p.

Hendricks, J.D., and Lucchitta, I., 1974, Upper Precambrian igneous rocks of the Grand Canyon, Arizona, in Karlstrom, T.N.V., Swann, G.A., and Eastwood, R.L., eds., Geology of northern Arizona, Part 1—Regional studies: Geological Society of America Field Guide, Rocky Mountain Section, p. 65–86.

Larson, E.E., Patterson, P.E., Mutschler, F.E., 1994, Lithology, chemistry, age, and origin of the Proterozoic Cardenas Basalt, Grand Canyon, Arizona: Precambrian Research, v. 65, p. 255–276.

McKee, E.H., and Noble, D.C., 1976, Age of the Cardenas Lavas, Grand Canyon, Arizona: Geological Society of America Bulletin, v. 87, p. 1188-1190.

Timmons, J.M., Karlstrom, K.E., Dehler, C.M., Geissman, J.W., Heizler, M.T., 2001, Proterozoic multistage (ca. 1.1 and 0.8 Ga) extension recorded in the Grand Canyon Supergroup and establishment of northwest- and north-trending tectonic grains in the southwestern United States: Geological Society of America Bulletin, v. 113, p. 163-181.

Timmons, J.M., Karlstrom, K.E., Heizler, M.T., Bowring, S.A., Gehrels, G.E., Crossey, L.J., 2005, Tectonic inferences from the ca. 1255–1100 Ma Unkar Group and Nankoweap Formation, Grand Canyon: Intracratonic deformation and basin formation during protracted Grenville orogenesis: Geological Society of America Bulletin: v. 117, p. 1573-1595.

Weil, A.B., Geissman, J.W., Heizler, M., Van der Voo, R., 2003, Paleomagnetism of Middle Proterozoic mafic intrusions and Upper Proterozoic (Nankoweap) red beds from the Lower Grand Canyon Supergroup, Arizona: Tectonophysics, v. 375, p. 199-220.

1 comment:

  1. I ran an Ar-Ar mass spec for some time, and dated many different samples, from relatively young (<20Ma) to much older (>1Ga). We did step heating, as well as core-rim ages, by means of a laser-ablation system.

    Ar-loss is very common, but could be most often accounted for within the first couple of steps. Excess Argon defintely disturbed not a few analyses, but it was quite obvious when thta happened. What you look for are repeatable plateau ages - a plateau is a continuous set of analyses in one grain, during the step-heating process, comprising at least 40% of all 39Ar released by the grain, and with results within at least 1-sigma error margin. A couple of plateau ages from different grains that are within error of each other could potentially be considered a credible result.

    Compared to the rigour of this method I find the simplicity of Snelling et al quite shocking. I met Snelling myself briefly in SA just as I started working in the Geochron lab - his comments then indicated very little understanding, and it appears that general ignorance has not progressed much.