the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Increased accuracy and precision in igneous and detrital zircon geochronology using CA-LA-ICPMS
Erin Elizabeth Donaghy
Michael P. Eddy
Federico Moreno
Mauricio Ibañez-Mejia
Abstract. Detrital zircon geochronology by laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) is a widely-used tool for determining maximum depositional ages, sediment provenance, and reconstructing sediment routing pathways. Although the accuracy and precision of U-Pb geochronology measurements has improved over the past two decades, Pb-loss continues to impact the ability to resolve zircon age populations by biasing affected zircon toward younger apparent ages. Chemical abrasion (CA) has been shown to reduce or eliminate the effects of Pb-loss in zircon U-Pb geochronology, but has yet to be widely applied to large-n detrital zircon analyses. Here, we assess the efficacy of the chemical abrasion treatment on zircon prior to analysis by LA-ICP-MS and discuss the advantages and limitations of this technique in relation to detrital zircon geochronology. We show that i) CA does not systematically bias LA-ICP-MS U-Pb dates for thirteen reference materials that span a wide variety of crystallization dates and U concentrations; ii) CA-LA-ICP-MS U-Pb zircon geochronology can reduce, or eliminate, Pb-loss in samples that have experienced significant radiation damage; and iii) bulk CA prior to detrital zircon U-Pb geochronology by LA-ICP-MS improves the resolution of Neoproterozoic to present zircon age populations and the percentage of concordant analyses in Mesoproterozoic and older age populations. The selective dissolution of zircon that has experienced high degrees of radiation damage suggests that some detrital zircon age populations could be destroyed or have their abundance significantly modified during this process. However, we did not identify this potential effect in either of the detrital zircon samples that were analyzed as part of this study. We conclude that pre-treatment of detrital zircon by bulk CA may be useful for applications that require increased resolution of detrital zircon populations.
- Preprint
(1559 KB) - Metadata XML
-
Supplement
(30579 KB) - BibTeX
- EndNote
Erin Elizabeth Donaghy et al.
Status: open (until 20 Oct 2023)
-
CC1: 'Comment on gchron-2023-20', Trystan Herriott, 19 Sep 2023
reply
It’s great to see this submission by Donaghy, Eddy, Moreno, and Ibañez-Mejia, which considers the potential for chemical abrasion (CA) as a means of improving accuracy for LA-ICPMS U–Pb zircon geochronology. Although there are several things that the authors can consider and expand on, I anticipate this study will be welcomed by many folks that rely on (accurate!) LA-ICPMS U–Pb zircon dating for a variety of applications.
This manuscript focuses on Pb-loss as a challenge in obtaining accurate LA-ICPMS U–Pb zircon dates. Because Pb-loss is the major uncertainty source addressed, the authors could consider writing a bit more in the intro text about how they view Pb-loss within the context of high-T/low-T conditions, geologic annealing, alpha-dose, degree of metamictization, young (Phanerozoic? or Meso–Cenozoic?)/old (Precambrian?), fluid flow, and other variables (some of these things are noted in various places in the text). Although the Pb-loss-from-zircon literature is somewhat confusing, it does seem that low-T Pb-loss is a real thing that maybe even happens in near-surface weathering of zircon (e.g., Keller et al., 2019, Geochronology). However, Pb-loss is also sometimes regarded as a mostly high-T phenomenon that isn’t likely to, for example, commonly impact DZ geochronology (e.g., Vermeesch, 2021, Geoscience Frontiers). Either way, it’s a bit hard to know how much of this ostensible low-T Pb-loss is happening in young-ish (e.g., Meso–Cenozoic), low-U (and Th) zircon (e.g., see discussion on alpha-dose and Pb-loss in McKanna et al., in review, Geochronology), but a reader of this paper will benefit if the authors provide some conceptual framing for how the accuracy of LA-ICPMS dates for young, non-metamict zircon that only experienced shallow burial depths can be improved with CA-LA-ICPMS.
I think it’s also critical that the authors directly address matrix effects-related uncertainties for LA-ICPMS. I can’t help but think there’s an impactful matrix effect/reference zircon challenge/issue in LA-ICPMS zircon geochronology that’s got to be solved as we strive toward achieving increased accuracy for LA-ICPMS. The literature on LA-ICPMS U–Pb zircon dating matrix effects and thermal annealing (TA)-LA-ICPMS and full on CA-LA-ICPMS is intertwined. And the authors will need to cite and discuss at least several key references, including Allen and Campbell (2012, Chemical Geology), Marillo-Sialer et al. (2014, JAAS; 2016, Chemical Geology), Solari et al. (2015, Chemical Geology), Sliwinski et al. (2017, Chemical Geology), and Ver Hoeve et al. (2018, Chemical Geology). One thing to keep in mind is that some of these papers highlight the value of TA alone in improving the accuracy of LA-ICPMS dates of zircon by diminishing ablation rate variability among references and unknowns. Our experience is that TA won’t solve all the potential challenges of LA-ICPMS zircon date offsets (e.g., Herriott et al., 2019, Geology), but the benefits of simple TA seems less likely to be burdened by some of the challenges of the full CA protocol (which can impact ablation behavior!) for LA-ICPMS U–Pb zircon geochronology. Mattinson (2005, Chemical Geology) even closed out that pivotal paper by noting the potential of TA to improve microbeam dating of zircon.
Regarding matrix effects and sample–standard bracketing, it's worth noting that by simply switching between which reference zircon is treated as the primary there can be potentially impactful effects on dates for unknowns (e.g., Klötzli et al., 2009, Geostandards and Geoanalytical Research). I’d be curious to see more plots like that of Gehrels et al. (2008, G^3) and Schoene (2014, Treatise on Geochemistry, chapter 4.10) and Pullen et al. (2018, G^3) and Sundell et al. (2021, Geostandards and Geoanalytical Research) of % offsets of “unknowns” dates of reference zircon results that were based on using different zircon primaries. All this relates to the scatter the authors highlight for the Meso–Cenozoic references in their % offset plot and then go on to discuss GHR1 and Fish Canyon Tuff (FCT) offset relations that may be reflecting true geologic dispersion. The geologic population complexity situation is important, although the analytical precision of LA-ICPMS at some point places meaningful limits on resolving autocrystic vs. antecrystic vs. xenocrystic populations (I didn’t re-read Eddy et al. [2019], or Wotzlaw et al. [2013], but the authors should confirm/report whether LA-ICPMS is capable of resolving the geologic populations documented by TIMS for these Cenozoic reference zircon). Either way, seeing young zircon have “issues” like this usually makes me think more along the lines of a matrix effect problem and less so geologic scatter (or Pb-loss, although the ICPMS offsets for FCT and GHR1 are positive here, but all the Pb-loss offset relations are theoretically flipped if CA-LA-ICPMS is compared to ID-TIMS and the zircon lost Pb and all other uncertainties are negligible[ha!]…right?!?). Anyway, maybe this isn’t that big of a deal, but FCT, for example, is nearly –1% offset in Gehrels et al. (2008), has some notable negative offsets in the Schoene (2014) review paper figure, looks a bit better behaved (with minor average negative offset) in the Pullen et al. (2018) paper, and Sundell et al. (2021) have FCT results that are pretty prone to negative offsets at more traditional laser spot analysis durations but have improved overall accuracy (albeit at the trade off for lower precision…not a terrible trade off if that’s what it takes!) at more rapid analysis durations (this is related to matrix effects!). Seeing the positive offsets for FCT in the current paper really caught my eye and seems like a matrix effect effect is going on here. Please share your thoughts on these reference zircon data relations!
I’m not quite sure what the authors can or should do with this, but here’s some additional food for thought: It's probably not yet widely appreciated/highlighted/published, but tandem dating (i.e., paired LA-ICPMS–CA-ID-TIMS analyses) case studies are documenting too-young offsets in LA-ICPMS U–Pb dates of Mesozoic DZ and tephra zircon in the –2% to –3% range (e.g., Herriott, Crowley, et al., GSA Connects 2022, 2023; a couple of published LA-ICPMS–CA-ID-TIMS DZ datasets with average offsets in this range are in Herriott et al., 2019; and Rasmussen et al., 2021, GSA Bulletin). Moving beyond the Mesozoic and a few different basins and a few different U–Pb labs, Howard, Sharman, et al. have a couple recent GSA Connects abstracts (2022, 2023) where they dredged the literature for paired LA-ICPMS–TIMS zircon dates and report offsets for LA-ICPMS of igneous, metamorphic, and detrital zircon typically in the range of –1.5% to –2.5%. A handful of abstracts doesn’t yet make a robust, citable literature trail, but some of this may be of enough interest that it could be noted by the authors with respect to future research/considerations. Again, I think some of this is almost certainly reflecting Pb-loss (and is also benchmarking at least some analytical dispersion in LA-ICPMS dates of single geologic populations as well), but matrix effects/reference zircon-related complexities remain as notable candidates for systematic sources of offset/bias/error that these tandem dating zircon studies are encountering.
One of the challenges with all of this is how do can we start isolating/identifying/mitigating these offset sources as the zircon dating community continues to strive to improve accuracy for LA-ICPMS. I encourage the authors to contemplate ways to further explore their existing reference and igneous zircon and DZ datasets for potential insights on this question of isolating potential sources of offset that can cumulatively impact accuracy. Maybe some 1:1 plots of ID-TIMS vs. CA-ID-TIMS dates of reference zircon could help identify where we may expect to explicitly resolve a (hopefully minimal…ahh!) Pb-loss issue with CA-LA-ICPMS? (Okay, probably hard or impossible to resolve any of that with ICPMS, but maybe the concept would be useful for discussion?) Perhaps find a way to plot/compare relative offsets for LA-ICPMS vs. CA-LA-ICPMS for the reference zircon treated as unknowns to see those relations more clearly than the Figure 1 offset plot and consider how this compares to Pb-loss mitigation predictions one might make because CA was or was not employed, as well as further characterize the offset impacts of CA. (Could you plot the CA-LA-ICPMS results in an offset plot as their benchmarked-by-TIMS offset minus the LA-ICPMS benchmarked-by-TIMS offset? Then you’d see what CA-LA-ICPMS “did” relative to regular LA-ICPMS…or something like that). Also consider/discuss how CA-LA-ICPMS may render older or younger dates in the context of increased or decreased ablation rates during analysis by the laser.
Please see below for some additional comments tied to lines in the manuscript. Congratulations to the authors on a very nice study, and I will look forward to seeing the final version of this work published in Geochronology.
Regards,
Trystan Herriott
Alaska Division of Geological & Geophysical Surveys–Line 36: Maximum depositional ages are mentioned in the abstract but not here (or anywhere else in the main text). A note or two regarding DZ MDAs in the text could be appropriate.
–Line 57: In listing the first and second benefits of CA, expand the summary to include the initial step and benefits of thermal annealing as part of the full CA pre-treatment.
–Line 87/Table 1: Double check the (non CA) ID-TIMS vs. CA-ID-TIMS superscripts here. Looks like there may be a typo or two. Suggest that here and throughout the paper to refer to non-CA ID-TIMS as “ID-TIMS” and CA-ID-TIMS as “CA-ID-TIMS”. State that this is the usage, stick to it, and any ambiguity will be avoided. (See next comment as well.)
–Line 178/Figure 1: Please clarify how the ID-TIMS vs. CA-ID-TIMS benchmarking was completed here. Just as an example, Mattinson’s (2010) CA-ID-TIMS Temora 2 result is listed as the reference age in the graphic, but the slightly younger ID-TIMS result of Black et al. (2004) is maybe (or maybe not?) used as the benchmark for your regular LA-ICPMS analyses? Perhaps this doesn’t matter a ton for reference zircon treated as unknowns, but some of these potential vagaries may come into play in LA-ICPMS case studies of unknown zircon. In fact, explicit discussion of these considerations would be much appreciated! Also note that the reference ages listed in the ranked date plots of the supplemental are labeled as “CA-ID-TIMS” but a lot of those reference zircon ages are probably ID-TIMS. See comment by Sundell et al. (2021) regarding this.
–Line 186: The issue of xenocrysts and antecrysts for GHR1 and FCT are valid points (although analytical resolution limits of LA-ICPMS may come into play here as noted above), but the CA variable and TA-LA-ICPMS and CA-LA-ICMS literature and matrix effects should be discussed a bit more here. Also, the abstract states “CA does not systematically bias LA-ICP-MS U-Pb dates”, but that’s not necessarily a conclusion I’d quickly make from looking at Figure 1. This is a central point of the manuscript and should be further evaluated beyond Figure 1. As noted above, consider making a figure with a quantitative comparison between the LA-ICPMS offsets and CA-LA-ICPMS offsets relative to each other to further help the reader see the (delta!) differences. And consider plotting all the individual dates rather than (or in addition to) the weighted means.
–Figure 6: The (non-CA) LA-ICPMS vs. CA-LA-ICPMS DZ sample results should be further compared quantitatively with common methods (e.g., see similarity/dissimilarity metrics of papers by Vermeesch and Saylor and Sundell and Sharman and so on). The metrics may indicate super-similarity between the LA-ICPMS vs. CA-LA-ICPMS results…or not…either way, it’s worth reporting (add necessary discussion to main text as well).
–Line 380 and Figure 8: Please expand on “similar and indistinguishable”. Not sure how this impacts the interpretation, but indistinguishable isn’t exactly what comes to mind for Figure 8.
–Line 406: Some folks (including me) may be interested in seeing a few references regarding how Pb-loss occurs in zircon at low-T…
–Line 477: Do McKanna et al. (2023) mention LA-ICPMS? I may have missed this…
–Lines 487-491: Consider noting possible avenues for future work that could further examine and resolve outstanding questions for improving accuracy for LA-ICPMS zircon geochronology. Couple/few things to mention/consider here: For folks that remain reluctant to plow ahead with bulk CA for DZ for provenance work, you might consider recommending alternative approaches, including follow-up/multiple LA-ICPMS spot analyses of targeted dates/populations of notable interest with respect to (possibly discriminating) source areas as an option. This has been done for DZ MDA work (e.g., Spencer et al., 2014, GSA Bulletin; Herriott et al., 2019)…not sure if this has been done for DZ provenance work (I don’t know that literature especially well). A similar approach could be to target a certain date range/population for follow-up with CA-ID-TIMS (e.g., Holland, Mohr, et al., GSA Connects 2022). And consider making a plug for the marked potential of CA-LA-ICPMS to be employed for DZ MDA work (see Donaghy, Eddy, et al., GSA Connects 2023)!
Citation: https://doi.org/10.5194/gchron-2023-20-CC1 -
RC2: 'Reply on CC1', Marcel Guillong, 20 Sep 2023
reply
Dear Trystan,
With interest I read your comments to this manuscript and I agree with most of your statements, however, one thing catched my attention:
"Our experience is that TA won’t solve all the potential challenges of LA-ICPMS zircon date offsets (e.g., Herriott et al., 2019, Geology), but the benefits of simple TA seems less likely to be burdened by some of the challenges of the full CA protocol (which can impact ablation behavior!) for LA-ICPMS U–Pb zircon geochronology"
This seems to indicat that you observed ablation behavour change between a TA only zircon and an annealed and leached (CA) zircon. In my understanding the leaching only affects the not healed zircon crystal and inclusions and not the intact or due to TA healed zircon. So I do not see why the leaching should have an influence on the ablation behavior. Can explain your observations and share some insights or is there a Publication dealing with this?
Thank you,
MarcelCitation: https://doi.org/10.5194/gchron-2023-20-RC2 -
CC2: 'Reply on RC2', Trystan Herriott, 20 Sep 2023
reply
Marcel, thank you so much for this comment!
My only direct experience with regular LA-ICPMS vs. TA-LA-ICPMS is in Herriott et al. (2019, Geology). In that study we compared DZ results for two samples collected from the same bed (Figure 2, third panel from the top). One sample was a legacy result with typical LA-ICPMS data and the other was new TA-LA-ICPMS data. The comparison between the results is complicated for a variety of geologic and analytical reasons, but the TA-LA-ICPMS dates were improved overall in the sense that they were a bit older (and it was pretty clear we were having too-young “issues” with ICPMS dating). Our follow-up CA-ID-TIMS dating clearly documented a too-young bias even in TA-LA-ICPMS results for the tandem dated DZ; some logic would suggest that the LA-ICPMS dates would have had even more significant too-young offsets, but we did not have TIMS benchmarks for those results.
My comment that chemically abrading zircon can impact ablation behavior was somewhat conceptual in the sense that thermal annealing is (probably partially) structurally repairing radiation damage in zircon (which should be a good thing for microbeam dating) while the pre-treatment leaching step of full CA is chemically removing parts zircon that retain damage (just as a recent example, see McKanna et al., 2023). I think it’s fair to say that CA is likely to have varying impacts on references and unknowns such that, again conceptually, these variable responses to partial HF dissolution could result in increased variability in ablation behavior among references and unknowns. I have some concern that chemically abraded zircon may in some cases be porous or spongy or pitted (e.g., see Figure 10a of McKanna et al., 2023) and could cause some “issues” during laser ablation, if for no other reason than the aforementioned likelihood that responses to full CA will vary. I’m not sure how/where an LA-ICPMS spot could be placed on the zircon of the McKanna et al. figure noted above that would not have ablation behavior directly influenced by CA. However, the CA-LA-ICPMS studies of Q. Crowley et al. (2014) and von Quadt et al. (2014) and Ver Hoeve et al. (2018) generally report benefits for full CA for LA-ICPMS dating of zircon, as does this submission by Donaghy et al., although Ver Hoeve et al. (2018) did end up concluding that “Acid leaching of the annealed grains appears to have little effect on the ablation behaviour of the analyzed zircon beyond the effects of annealing.” So seems full CA isn’t always needed, which has significant implications for DZ studies because I think folks are going to be quite concerned about potentially losing entire zircon populations during bulk CA.
Digging a bit deeper, however, Q. Crowley et al. (2014) do note that “chemical abrasion of zircons affects their physical response to ablation, which in turn influences laser-induced fractionation.” But it turned out that “porous texture” on zircon post-CA led to perhaps diminished coupling with the laser, reducing spot pit depths and thus reducing downhole fractionation (see Figure 2 therein and associated discussion). That’s not necessarily where my intuition was leading me, but I’m also not a geochronologist. All of this makes me think of the comment by Sundell et al. (2021) that “variations in downhole fractionation and compositional heterogeneity are expected to become less important with shorter acquisition rates, which may also contribute to better analytical results in some cases.” Interesting!
I trust that Donaghy et al. may benefit from this exchange as much as I did. I also wanted to note that there are at least a couple of CA-SIMS studies of zircon that are relevant and could potentially be noted in the current submission: Kryza et al. (2012, Gondwana Research) and Watts et al. (2016, Chemical Geology).
Thanks again, and regards,
TrystanCitation: https://doi.org/10.5194/gchron-2023-20-CC2
-
CC2: 'Reply on RC2', Trystan Herriott, 20 Sep 2023
reply
-
RC2: 'Reply on CC1', Marcel Guillong, 20 Sep 2023
reply
-
RC1: 'Comment on gchron-2023-20', Marcel Guillong, 20 Sep 2023
reply
The application of the CA protocol for Zircons prior to dating, successfully introduced to high precision ID-TIMS dating is increasingly also tested and applied in less precise LA-ICP-MS methods. This manuscript uses the CA method successfully on a igneous Zircons with high Pb loss and on 2 DZ samples. While the improvement in the igneous sample is substantial and well documented the improvement on DZ is less obvious and needs better investigation or clarification. On 13 different zircon reference materials, all having minimal to no Pb loss there is no real improvement but also no obvious offset introduced using the CA treatment. Effect (matrix effects) of thermal annealing and leaching might be smaller than the general uncertainty that is associated with U-Pb dating of Zircons by LA-ICP-MS, but a better understanding and further investigations are desirable in this field but likely beyond the scope of this manuscript.
Main points to consider:
The title is kind of misleading. Consider a different title:
e.g. “Increased date accuracy and precision in Pb loss affected igneous and detrital zircons using CA-LA-ICPMS”
Accuracy and precision of reference materials analysed using CA and non CA seems not improved. Accuracy and precision is only improved if Pb loss occurred due to radiation damage of the zircon crystal structure ore high amount of common Pb rich inclusion were analysed. See also comment to Figure 1.
Some wording is inaccurate:
Rank order plots are not ranked. (Also, in the supplementary.)
Avoiding the use of the term standards when referring to reference materials.( https://www.geoanalyst.org/glossary/ ) there are no zircon standards, only zircon reference materials change all “standards” to reference material (RM).CA changes the mean measured TE concentrations in strong metamict samples as MIGU-02. Please mention that after CA no representative or accurate TE composition can be analysed in these Zircons.
While I agree that for strongly metamict Zircons CA improves precision and accuracy, the comparison for MIGU-02 is not entirely fair as for the CA about 150 zircons were used (80-85 % completely dissolved 23 analysed) vs non CA only 35 grains were analysed. If you would analyse 150 non CA Zircons (same number as were initially used in the CA experiment) you would get 4-5 times more “good”, concordant analyses.
In the end it is the user who has to make a decision to do CA or not: if you have separated 150 Zircons from a strongly metamict sample, there is also a chance of not only dissolve 80-85% as in the presented example but 90-99% and when all zircons are dissolved, there is nothing left to analyse. If the pb loss occurred at one specific age and the common Pb is not too much a problem it is possible to get (not ideal) age information from 150 non concordant pb loss affected analyses. This possibility should be mentioned / discussed.Figure 1: All investigated RM show no improvement of CA analyses over non CA analyses with respect to precision and accuracy. Uncertainties and age offsets are very similar.
Figure 2 this is not a rank order plot as the data seems not ranked. If it is ranked, please indicate ranked by what (normally ranked order plots are ranked by the date)
Figure 3A. choose different scaling so that the ellipses are better visibel. Consider to also plot the discarded analyses, to show the effect of Pb loss (+ comm Pb).
Figure 4, 8 and 9: Generally, how was the U concentration quantified? With internal standard? Semiquantitative? With which RM? Is the RM homogeneous in U?
Figure 5 and 6: To show the differences in DZ between CA and non CA seems quite qualitatively, and partly inconclusive: in Rora Med, the 2120 peak seems broader than with no CA where there are two well resolved peaks at 2115 and 2190. Which is the opposite of what the authors describe (CA makes peak sharper and more resolved). Overall, the improvement arguments are quite weak, except the clearly better concordance which is definitively an advantage.
Is there an explanation, why in Sample Rora Med peak hights in for 1890 and 2100-2200 inverse between CA and non CA? With the high number of analysed zircons (n=1000) this should not be the case? Is the 1890 peak much more metamict (rather not, no peak shift indication or younging due to Pb loss (as nicely shown for the 2675 peak). As Suggested by Trystan Herriott a quantitative approach comparing the two DZ spectra could help.Figure 9: would scaling the x axis logarithmic improve readability? Interestingly in Rora Med the CA sample seems to have higher U content, in contrast to the finding in the igneous MIGU-02 sample. Could this be a quantification artefact from the heterogenous RM?
Line 73-74: Annealing zircons heals the crystal structure making the ablation rate little smaller then without annealing (see e.g.: Sialer et al 2016: https://doi.org/10.1016/j.chemgeo.2016.05.014, Solari 2015, https://doi.org/10.1016/j.chemgeo.2015.09.008, and others see comment by Trystan Herriott) In my experience the acid etching itself has little to no influence on the ablation behaviour in comparison to thermal annealing only, as the zircons were mounted and polished?
Line 187-189: the scatter in age offset is not improved by CA but generally for Proterozoic and Archean aliquots.
Line 350: Are you sure it’s a 207Pb/238U age? (Rather a typo)
Line 354-355: There is no improvement shown in precision and accuracy for 13 RM, so a general statement that CA improves precision and accuracy in LA-ICP-MS U-Pb zircon analyses is not correct. Accuracy and precision is improved when there is Pb loss and possibly common Pb in inclusions that is not excluded during data reduction.
Tables: e.g. S14 please provide data in a way that it can be easy accessed by the interested reader for plotting etc. (e.g. excel file)
Citation: https://doi.org/10.5194/gchron-2023-20-RC1
Erin Elizabeth Donaghy et al.
Erin Elizabeth Donaghy et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 268 | 101 | 16 | 385 | 24 | 2 | 4 |
- HTML: 268
- PDF: 101
- XML: 16
- Total: 385
- Supplement: 24
- BibTeX: 2
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1