Modelling effect of sericitization of plagioclase on the 40K/40Ar and 40Ar/39Ar chronometers: implication for dating basaltic rocks and mineral deposits

Abstract The 40Ar/39Ar technique is the most commonly used technique to date basaltic rocks. For basaltic rocks older than about 30 Ma, the dating of plagioclase separates is preferred over groundmass as the latter is susceptible to containing cryptic alteration due to fluid circulations, difficult if not impossible to remove during sample preparation. Alteration under such metamorphic conditions progressively forms K-rich sericite after plagioclase. Owing to its transparency, plagioclase allows a distinction to be made optically between partially–completely altered grains and fresh grains. However, practice shows that grains that contain less than about 1% of sericite are hard to identify under the stereomicroscope. Owing to the high K2O content (c. 10 wt%) of sericite, such compromised grains can have dramatic effects on the age determination of plagioclase. Here, we investigate and quantify the effect of sericite on the 40Ar/39Ar age determination of plagioclase using a numerical model with multiple variable parameters. We show that the most influential parameter is the time difference between the crystallization of plagioclase and the sericitization event. We also show that for some continental flood basalts, even 0.1 wt% of sericite can bias the apparent age of a plagioclase separate by several hundred thousand years. The presence of sericite can be identified using a combination of Ca/K ratios, age spectra, and 39Ar and 37Ar degassing curves obtained during a conventional 40Ar/39Ar step-heating procedure. When the age of the fresh plagioclase and its Ca/K ratio are known, the percentage of sericitization and the age of the alteration event can be estimated. Ultimately, above approximately 65% of sericitization, the apparent age measured on the altered plagioclase is within ±1% of the age of the alteration event, with implications for accurately dating low-temperature metamorphism and mineral deposit formations. Supplementary material: Further details of calculation are available at www.geolsoc.org.uk/SUP18609.

Abstract: The 40 Ar/ 39 Ar technique is the most commonly used technique to date basaltic rocks. For basaltic rocks older than about 30 Ma, the dating of plagioclase separates is preferred over groundmass as the latter is susceptible to containing cryptic alteration due to fluid circulations, difficult if not impossible to remove during sample preparation. Alteration under such metamorphic conditions progressively forms K-rich sericite after plagioclase. Owing to its transparency, plagioclase allows a distinction to be made optically between partially-completely altered grains and fresh grains. However, practice shows that grains that contain less than about 1% of sericite are hard to identify under the stereomicroscope. Owing to the high K 2 O content (c. 10 wt%) of sericite, such compromised grains can have dramatic effects on the age determination of plagioclase.
Here, we investigate and quantify the effect of sericite on the 40 Ar/ 39 Ar age determination of plagioclase using a numerical model with multiple variable parameters. We show that the most influential parameter is the time difference between the crystallization of plagioclase and the sericitization event. We also show that for some continental flood basalts, even 0.1 wt% of sericite can bias the apparent age of a plagioclase separate by several hundred thousand years. The presence of sericite can be identified using a combination of Ca/K ratios, age spectra, and 39 Ar and 37 Ar degassing curves obtained during a conventional 40 Ar/ 39 Ar step-heating procedure. When the age of the fresh plagioclase and its Ca/K ratio are known, the percentage of sericitization and the age of the alteration event can be estimated. Ultimately, above approximately 65% of sericitization, the apparent age measured on the altered plagioclase is within +1% of the age of the alteration event, with implications for accurately dating low-temperature metamorphism and mineral deposit formations.
Supplementary material: Further details of calculation are available at www.geolsoc.org.uk/ SUP18609. 40 Ar/ 39 Ar dating of low-K 2 O basalt has always proved to be a challenging, yet very important, task when it comes to studying the timing of basaltic volcanic events (e.g. large igneous provinces, oceanic island basalts or lunar maria). Many studies have demonstrated a much better reliability and accuracy of the 40 Ar/ 39 Ar technique over 'conventional' K -Ar dating, as only the former approach can unambiguously demonstrate that a given sample is free of alteration and does not contain excess or inherited 40 Ar. Whereas alteration can be circumvented by using relatively young (i.e. few million years old, Ma) rocks, the absence of excess 40 Ar* is impossible to demonstrate using the K -Ar technique alone (e.g. McDougall & Harrison 1999). Dating basaltic rocks of a few tens or even hundreds of Ma is a risky business as alteration can shift the isotopic 'age' by few Ma to tens of Ma compared to the eruption age (see examples in Courtillot et al. 2010). Much more robust is the 40 Ar/ 39 Ar technique as it allows powerful statistical tests to be made of the reproducibility of the age of a sample age using the step-heating approach.
More recently, better accuracy of plagioclase over groundmass 40 Ar/ 39 Ar dating has been demonstrated, at least for rocks older than few tens of Ma, due to the unavoidable cryptic and pervasive alteration of these rocks (Hofmann et al. 2000;Jourdan et al. 2007a, b). Contrary to groundmass, plagioclase allows the optical separation of altered grains from fresh grains due to its transparency ( Fig. 1, but see the discussion below). Nevertheless, owing to its very low content in K 2 O (often ,0.1 wt%), plagioclase is extremely sensitive to alteration and, in particular, to sericitization, where sericite is a replacement phase introduced during open-system alteration. Plagioclase replacement by sericite occurs during the fluid alteration of magmatic rocks under low-pressure -lowtemperature metamorphism; that is, under zeolitelower greenschist facies conditions (i.e. from 100 to 300 8C, ,10 kbar). Owing to the high content of sericite in K 2 O (c. 10 wt%), only minute quantity of this phase is required to overprint the 40 Ar/ 39 Ar age signal of the plagioclase and so compromise the validity of the age.
In this study, we provide theoretical models showing: (1) how sericitization modifies the K -Ar and 40 Ar/ 39 Ar systematic dating of plagioclase, and, ultimately, how it can affect the accuracy of plagioclase (and groundmass) ages; (2) how and why the sericite signature can be identified using the 40 Ar/ 39 Ar age spectrum; and (3) how completely sericitized plagioclase can be used for accurately dating greenschist facies metamorphic events and hydrothermal fluid systems relevant to the study of the timing of ore deposit formations.

Petrological features of the sericitization
Sericitization is a process of alteration in lowtemperature conditions (,300 8C) encountered commonly during hydrothermal alteration in a epithermal context or during greenschist metamorphism. Sericite KAl 3 Si 3 O 10 (OH,F) 2 replaces plagioclase by mineralogical substitution and/or fills microfractures within the plagioclases (Fig. 1). Sericitization can correspond to a punctual instantaneous event, a long-term episode or a multistage process. But, unfortunately, these processes are often impossible to differentiate in thin section.
Strongly sericitized (.50%) plagioclase grains commonly have a milky-white appearance (cloudy) under a binocular stereomicroscope (Fig. 1), and are easily discernible. Although major sericite replacement is easily identified in thin section (Fig. 1), it is very difficult to evaluate the minimum amount of sericite that can be detected with the human eye using a stereomicroscope. For instance, it is possible that a sericitization of less than 1-2% cannot be detected during hand-picking, even for those with an experienced eye, and especially when sericite is located within small cracks (Fig. 1).
Considering the relatively low potassium content of plagioclase (often K 2 O ,0.1 wt%), fine alteration, hardly discernible during hand-picking, could have significant effects on the measured K -Ar and 40 Ar/ 39 Ar ages, yielding anomalously young ages, often disturbed age spectra and also, in extreme cases, well-defined 'alteration plateau' ages (e.g. Jourdan et al. 2003;Baksi 2007a;Pati et al. 2010).

Two-component mixing model
In order to evaluate the age reduction of the sericitized plagioclase, we propose a mathematical model based on a two-component mixture between unaltered plagioclase and sericite, bearing in mind that sericitization occurs in a single brief stage. Note that the proposed model does not include any thermaly activated diffusive loss of 40 Ar* from plagioclase caused by the hydrothermal circulation or lower greenschist conditions. Whereas diffusive loss inevitably occurs in many cases, most of the age reduction is generally controlled by sericite formation rather than by Ar loss.
Using the K -Ar age equation, for a given K content in plagioclase ( 40 K Pl ), radiogenic 40 Ar in sericite-free plagioclase with an age t is: For a plagioclase that is sericitized at t ′ (t ′ , t), with X% of sericite in the mixture, for a given K We obtain: 40 Ar tot = l e l 40 K Pl (e lt − 1) We can define the apparent age t ′′ of the twocomponent mixture (sericitized plagioclase) as follow: 40 Ar tot = l e l 40 K tot (e lt ′′ − 1).
Previous equations may be combined to give : We deduce that the apparent K-Ar age t ′′ of the mixing is: where 40 K Pl is the 40 K content of fresh plagioclase, 40 K Ser is the 40 K content of sericite, 40 K tot is the 40 K content of the mixture, t is the K -Ar magmatic age of the plagioclase, t ′ is the K -Ar age of the sericitization event and l is the total 40 K decay constant (here 5.544 × 10 210 : Steiger & Jäger 1977).
The percentage of K -Ar age reduction (AR) is: The 40 K Pl / 40 K tot and 40 K Ser / 40 K tot ratios in equation (6) can be developed taking into account the chemical composition of plagioclase and sericite; that is, %K 2 O, %CaO and the Ca/K ratio (identified in the 40 Ar/ 39 Ar technique by 37 Ar Ca / 39 Ar K ). Detailed equations are reported in the Appendix.
Finally, we obtain the following equations that quantify the two theoretical ratios reported in equation (6): The variables are X (percentage of sericite in the mixing), M Pl (molecular weight of plagioclase), (Ca/K) Pl and %CaO Pl . M Pl , (Ca/K) Pl and %CaO Pl are interdependent (see the Appendix for further explanation). A and B are known values related to the chemical composition of sericite and the molecular weight of implied oxides.

K-Ar age perturbation during the sericitization process
Using equations (1) -(3), we can calculate an age reduction (AR) for a given altered plagioclase. The variable parameters are: (1) the age of the plagioclase, t; (2) the age of the sericite (alteration event, t ′ ); (3) the Ca/K ratio of the plagioclase; and (4) the amount of alteration. In Figure 2, as an exercise, we plotted three curves showing the AR v. percentage of sericitization reported for an An 70 and for t ′ being 10% younger than the fresh plagioclase. The three curves correspond to three different chemical compositions of plagioclase; that is, three different Ca/K ratios. For all values of theoretical t, ranging from 100 ka to 1 Ga, AR values display imperceptible variations (and, thus, not shown). For a common basaltic plagioclase (An 70 ), our model confirms that a few percentage of fine alteration, hardly discernible during the handpicking, has a dramatic effect on the measured K -Ar ages of the plagioclase (Fig. 2). The age reductions (AR) of the sericitized plagioclase are correlated with the Ca/K ratio of the fresh feldspar. If sericite is 10% younger than plagioclase and represents more than 10% of the total plagioclase aliquot, the K -Ar age decrease can range from 7 to 10%, depending on the Ca/K ratio of plagioclase. Furthermore, a sericitization of less than 1% can also significantly modify the K -Ar ages of the feldspar. For instance, a plagioclase with a Ca/K ratio of 36 and 100 Ma in age provides an apparent K -Ar age of 98 Ma with minute sericitization of only 0.5% (using an age of alteration of 90 Ma: Figs 2 & 3). In extreme cases, plagioclase with an age of 500 Ma that underwent partial alteration (1%) at 0 Ma would have a shift in its apparent age by 170 Ma (not shown). Figure 2 shows that the higher the Ca/K ratio of the unaltered plagioclase, the lower is the age reduction for a same amount of sericitization. For instance, a K-rich An 70 displays an age reduction of less than 2% with a sericitization of 1%, whereas analysis of a K-poor An 70 almost reached the age of the sericite for the same degree of alteration (Fig. 3). If we consider a Ca-poor (and, hence, K-very poor) plagioclase (i.e. albite end member), the effect of sericitization on the age reduction is more important. The curves of the decreasing K -Ar ages are quite similar for a K-poor An 70 (Fig. 3a) and a K-rich An 05 (Fig. 3c) plagioclase (Fig. 3).

Influence of the time span between plagioclase and sericite crystallization
The age difference between magmatism and alteration is the main parameter influencing the calculated K -Ar age reduction (Fig. 4a). When the alteration is contemporaneous with the time of magmatic emplacement (i.e. syn-eruptive sericite: e.g. Jourdan et al. 2009a), the decrease in the K -Ar age is obviously equal to zero whatever the amount of sericitization. When the alteration is contemporaneous with the present day, the K -Ar decrease curve displays the maximal values for a given percentage of sericitization. For a difference in age of 5% between crystallization and alteration, the related AR is 2% for a low percentage of alteration of 1% (Fig. 4b). In short, the younger the sericitization event, the greater is the decrease in the K -Ar age of plagioclase, for a given degree of alteration.

Consequences for K-Ar dating of Large Igneous Provinces (LIPs)
Plagioclase has been widely used for the dating of major flood basalt volcanism, such as the Deccan, Ethiopian or Siberian traps, or the Karoo or Central Atlantic Magmatic Province, in order to determine the timing and duration of such huge volcanic activities (Deckart et al. 1997;Hofmann et al. 1997Hofmann et al. , 2000Marzoli et al. 1999Marzoli et al. , 2011Courtillot et al. 2000;Hames et al. 2000;LeGall et al. 2002;Knight et al. 2003;Jourdan et al. 2004Jourdan et al. , 2005Jourdan et al. , 2009aVerati et al. 2005Verati et al. , 2007. According to the brevity of most of these events (,1-2 Ma), Curves are defined with a theoretical An 70 plagioclase, and an age difference of 10% between alteration and plagioclase crystallization events. An AR ¼ 0 means that the K-Ar chronometer in plagioclase is not perturbed by the sericitization event. All curves show that a sericitization of more than 10% strongly affects the K-Ar age of plagioclase, which tends rapidly to the value of the alteration age.
the time resolution required to study these events necessitate that the internal precision of the ages must be, at very least, better than 1% (2s) but ideally approaching 0.5 -1 Ma to be considered useful. We thus investigate when and how the sericitization could disturb this required time precision. In Figure 5, we report, for several flood basalt provinces, on the value of sericitization that causes 1% and 10% of AR. For all of the volcanism events, 1% of sericitization implies an age reduction of more than 1%, and thus more than the minimum time precision required. For the Deccan traps (with a common K content for plagioclase, with a mean Ca/K ratio of 20), only 0.35% of sericite is required to bias the measured age by 1%, whereas an even smaller degree of alteration (,0.05%, definitely imperceptible) is sufficient in the case of the Rajahmundry traps (K-poor plagioclase, with a mean Ca/K ratio of 200). Using a classical K -Ar method for volcanic rocks older than about 20 -30 Ma is clearly unsuitable, given that undetectable alteration might easily overprint the magmatic age of plagioclase (and even more for the groundmass) and that the presence of such alteration is not directly recognizable using the K -Ar technique.

Effects of the plagioclase -sericite replacement in Ar/Ar results
A better way to test whether plagioclase recorded a true emplacement age or if it is affected by alteration   Fig. 3. AR parameter v. % sericitization ranging from 0 to 1% (detail of Fig. 2). Curves for An 70 are the same as those reported in Figure 2, with the three different Ca/K ratios. Curves for An 05 plagioclase (Ca-poor, Na-rich) are also reported for comparison.
is by using the 40 Ar/ 39 Ar method. The 37 Ar Ca / 39 Ar K ratio, related apparent ages and degassing curves of individual argon isotopes are three useful ways of identifying the presence of sericite in plagioclase.

Detection of sericitization in degassing curves
In Figure 6, we reported unpublished data of plagioclases from the CAMP that have more than 2% sericite, as identified in thin section. Related degassing curves represent the relative abundance of argon degassed (mV) for each argon isotope, normalized to the increment in temperature (DT8) for each step to avoid dependency on the heating schedule.
The complex degassing history can be interpreted as sericite microfractures or 40 Ar* diffusion loss in the early part of the age spectrum, and sericite patchy replacements in the mid -high temperature steps. Indeed, the sericite degassing curve for 39 Ar (one sharp peak at 1050-1200 8C with a high activation energy of c. 67 kcal/mol: Harrison et al. 2009) is different to the relatively smooth curve of pure plagioclase identified by the 37 Ar Ca (two peaks at around 900 and 1300 8C, and a lower activation energy of c. 40 kcal/mol: Cassata et al. 2009). A perceptible sericitization in thin section (X . 2%) provides a clear 39 Ar peak between (and clearly resolved from) the two main 'smooth' peaks of plagioclase. This is also clearly identified by an age reduction for these steps (Fig. 6) Ar Ca , and 39 Ar K and 37 Ar Ca are no longer covariant, this probably indicates that most of the 39 Ar K now comes from the sericite during this particular phase of the degassing process, ultimately causing an apparent age reduction. However, owing to the facts that (1) sericite is directly intergrown with plagioclase, and (2) the variable sericite grain sizes have variable closure temperatures, the argon diffusion of sericite will be intimately linked with the argon diffusion of plagioclase during the entire step-heating process. Therefore, sericite will contribute to variably reduce the 40 Ar/ 39 Ar age along the entire age spectrum. In other words, no step can reach the true age of the volcanism. Furthermore, diffusive 40 Ar loss from plagioclase due the rise in temperature associated with a given alteration event will also contribute to reduce the individual step ages, although the latter effect is not included in our current model. As a result, only a minimum qualitative age for the eruption can be estimated from the oldest steps of the age spectrum.
For example, Figure 6b shows a minimum age of ≥190 Ma, whereas the true emplacement age is known to be approximately 199 Ma, based on neighbouring well-dated lava flows.

Determination of the sericitization age
If we know the age of fresh plagioclase, and its Ca/K ratio, and if the latter is constant within the plagioclase, we then can estimate the percentage of sericitization by taking into account the integrated Ca/K of the mixture (Ca/K) mel (see the section on 'Determination of the sericitization value (X ) in 40 Ar/ 39 Ar spectra' in the Appendix). In practice, this exercise is rendered difficult by the fact that the Ca/K ratio is variable at the single grain level. Nevertheless, for homogenous plagioclase, we are able to estimate the alteration age t ′ (see the section on 'Estimate of the age t ′ of the alteration in Ar/Ar spectra' in the Appendix) and the percentage of alteration. For the 40 Ar/ 39 Ar experiment on CAMP basalts reported in Figure 6, we obtain approximately 3% for the value of alteration, with an estimated age of sericitization of 174 Ma, in accordance with some alteration plateau ages obtained from some strongly sericitized plagioclase (Jourdan et al. 2003).
When the sericitization is less apparent in the 40 Ar/ 39 Ar spectra, we still can provide a rough estimate of the sericite/plagioclase content. Plagioclase samples from the Karoo Province (Fig. 7 (Knight et al. 2003); Ethiopian and Siberian traps (Hofmann 1997); Central Atlantic Magmatic Province (Deckart et al. 1997;Marzoli et al. 1999Marzoli et al. , 2011Verati et al. 2005Verati et al. , 2007; Karoo Province (LeGall et al. 2002;Jourdan et al. 2004Jourdan et al. , 2005Jourdan et al. , 2008; and Deccan traps Hofmann et al. 2000). We see that an alteration of 1% always implies a K-Ar age reduction more than the required time precision needed for all of these volcanism events.
yielded a plateau age (essentially mimicking a plateau age) of 179.9 + 0.9 Ma (2s; but note that the error boxes in Fig. 7 are given at 1s) with 100% of cumulated 39 Ar. However, some individual steps show lower apparent ages correlated to slight 37 Ar Ca / 39 Ar K depletion, probably associated with some sericite present in the system. If we remove these steps from the plateau age calculation, the new corrected plateau age (180.5 + 0.6 Ma) is now 0.3% older. According to equations (A10) -(A15) (see the sections on 'Determination of the sericitization value (X ) in 40 Ar/ 39 Ar spectra' and 'Estimate of the age t ′ of the alteration in Ar/Ar spectra' in the Appendix), such a value corresponds to 0.16% of sericitization, related to an alteration event calculated at 172 Ma. This model age is indistinguishable from the 40 Ar/ 39 Ar ages measured for a large-scale oceanization process that affected the East African region at that time (Jourdan et al. 2007a). Although this 'correction' could represent the finest we can reasonably make, one should bear in mind that the presence of sericite might have affected the age of the lower and higher temperature regions of the age spectrum to some degree (cf. the discussion above), hence preventing the true age of the eruption being established. Therefore, this kind of approach is strongly discouraged, especially when one tries to establish a precise and accurate age history for a given set of eruptions.

40
Ar/ 39 Ar results obtained for plagioclase from basaltic rocks from large igneous provinces (LIPs) can only be understood if one recognizes and takes into account the effect of alteration of plagioclase. Previously, we have shown that a tilda-shape pattern observed on a plagioclase 40 Ar/ 39 Ar age spectra and/or a Ca/K plots, as well as a decoupling of the 39 Ar and 37 Ar degassing curves (Fig. 6), signal the presence of sericite in the plagioclase separate. Thanks to the ability of the 40 Ar/ 39 Ar system to recognize disturbances, the main consequence of sericitization of plagioclase will be the departure of individual step age from a statistically concordant population; that is, from a well-defined plateau age. In addition to the tilda-shape pattern, sericitization from a non-synchronous alteration event can be easily recognized as step ages will fail to pass a simple x 2 -test, leading to an abnormally high mean square weighted deviate (MSWD) and absolute P values lower than 0.05 (e.g. see the detailed description by Baksi 2007a andJourdan et al. 2009b), indicating discordant results. It has also been proposed that altered plagioclase exhibits an unusually large amount of 36 Ar, which can be quantified by the alteration index proposed by Baksi (2007a).
Systematic leaching using a dilute acid (e.g. HF, HNO 3 ) during plagioclase preparation, especially more commonly in use during the last 10-15 years, can help in removing part of the alteration products within the plagioclase, while leaving the latter intact. Nevertheless, the extreme sensitivity of plagioclase to minute amount of sericite renders plagioclase dating a hard task. As a result, many plagioclase 40 Ar/ 39 Ar data in the literature do not reflect the true crystallization age of the rock (e.g. Baksi 2007b;Nomade et al. 2007). Using some of these criteria, some authors (e.g. Hofmann et al. 2000;Jourdan et al. 2005;Nomade et al. 2007;Baksi 2007b;Evins et al. 2009;Reichow et al. 2009;Cucciniello et al. 2013) have reinvestigated the 40 Ar/ 39 Ar age database of several LIPs. The main filtering parameters used were: (1) the selection of 40 Ar/ 39 Ar age data measured exclusively on plagioclase for relatively old LIPs (≥50 Ma); and (2) removal of dates obtained on data that did not satisfy the x 2 -test and, thus, did not show welldeveloped plateaus. Baksi (2007b) also used the alteration index to test the freshness of the mineral post-analysis, although the presence of alteration usually resulted in disturbed age spectra failing the x 2 -test anyway. In all cases, the filtered ages showed a much more restricted duration of activity for a given type of magmatic unit. For example, Jourdan et al. (2007a) showed that the 1.4 kmthick Lesotho lava pile was emplaced in a time span shorter than ,0.8 Ma, below the resolution of the 40 Ar/ 39 Ar technique (at least when using plagioclase). These results when compared to other magmatic events within the Karoo provinces suggested an overall prolonged activity over several Ma, with different short-lived volcanic centres active at different times (Jourdan et al. 2008). Similarly, Marzoli et al. (2011), using the most recent 40 Ar/ 39 Ar database of CAMP, showed that the onset and main peak volcanic activity was nearly synchronous over the entire province, although different geochemical signatures from different sub-provinces could be correlated with resolvable differences in the times of eruptions. These authors also suggested that diachronous magmatism could be related to the northwards rift -drift transition during the break-up of Pangea. Rigorous statistical filtering and selection of the most robust data can significantly reduce the number of available ages and even leave some provinces with only a few valid 40 Ar/ 39 Ar ages (e.g. Kalkarindji: Evins et al. 2009). However, such an approach allows the removal of all dates that do not strictly represent crystallization ages of plagioclase, and therefore produces more coherent and internally consistent results. In most cases, much more precise time resolutions of different magmatic events allows key points inherent to LIP investigations to be studied,  such as the timing of the onset and end of the LIP magmatism, the duration of the peak activity of the volcanism, identifying magmatic centre migrations, and testing true synchronicity between LIPs and mass extinctions. At the other end of the spectrum, it is worth noting that completely sericitized plagioclase yielding robust plateau ages can provide time constraints on major hydrothermal events, such as continental rifting.
As our knowledge of the effect of alteration on plagioclase (and other minerals and groundmass) is improving, so is our understanding of what 40 Ar/ 39 Ar data of a particular plagioclase sample truly means. Only by applying rigorous filtering criteria on what is an acceptable 'age', will 40 Ar/ 39 Ar dating of plagioclase reach its full potential in the study of LIPs.

Implications for zeolite/pumpelleyite/ greenschist facies dating
Low-degree metamorphism is difficult to date because few related metamorphic minerals are available for geochronology. Yet, it has an important application to date important events, such as the formation of many ore deposits. As far as the 40 Ar/ 39 Ar technique is concerned, most analyses attempted so far have been carried out on sericite, adularia, alunite, celadonite and clays, and will be briefly described below.  In principle, typical K-rich zeolite minerals could also be used as potential low-temperature dateable minerals (WoldeGabriel et al. 1996;Hall et al. 1997;Parry et al. 2001), but low retentivity of argon, and saturation of these minerals by active gases, makes zeolite difficult to date in practice. Clay dating by 40 Ar/ 39 Ar has also been attempted using encapsulation to prevent 39 Ar recoil loss (e.g. Dong et al. 1995;Haines & Van der Pluijm 2008). However, in this case, the emplacement age of the clay is best defined by the total gas age of the experiment, making the step-heating approach irrelevant, and where similar results can be obtained using the conventional K -Ar technique (Clauer et al. 2008). Few K -Ar and 40 Ar/ 39 Ar analyses of celadonite are available, and whether this mineral can provide meaningful ages remains to be seen (Oliveros et al. 2008a;Jourdan unpublished data). Adularia has been used in numerous 40 Ar/ 39 Ar dating studies to constrain the age of fluid flow and associated mineralization (e.g. Oliveros et al. 2008b, c;Marton et al. 2010;Verati et al. this volume, in press), and generally provides robust plateau ages.
The sericitization process occurs at the same P-T8 conditions for all of these minerals and, hence, when sericite is dated using 40 Ar/ 39 Ar, it can provide the age of the associated metamorphic and/ or hydrothermal and/or mineralization events, with the advantage that sericite occurs in a K-rich alteration phase associated with moderately hot fluids and that the isotopic system is generally closed to subsequent alteration. Of course, the latter statement is true provided that the rock does not subsequently reach temperature conditions higher than the closure temperature of sericite (Jourdan et al. 2009a).
For plagioclase (that contains more than 1% of sericite), the time span between magmatism and metamorphism is the main parameter that affects the plagioclase K -Ar age. Although the K-Ar age of sericite can only be found experimentally with 100% of sericitization, a degree of sericitization of only about 65% allows an age within 1% of the true alteration age to be measured (here calculated using a plagioclase Ca/K value of 40, and an age difference of 10% between crystallization and alteration), which appears reasonable when one tries to constrain the age of an alteration event. Nevertheless, the true age of alteration can only be established when ≥85% sericitization occurred, which should ultimately be tested at first using electron probe and X-ray diffraction (XRD) techniques and then confirmed by the [Ca/K] Ar ratio. In theory, [Ca/K] Ar ratios of pure sericite phase should tend towards 0 but, in practise, it is possible that some Ca is left behind during the sericitization process, yielding to a non-null ratio. For example, Ca/K ratios reported for completely sericitized plagioclase purposely selected show values generally below ≤1 (Oliveros et al. 2008b, c;Jourdan et al. 2009a), whereas sericite directly deposited by fluid flow yielded Ca/K values well below ≤0.1 (e.g. Jian-Wei et al. 2006;Chi et al. 2008). A clear Ca/K cut-off value for what constitutes .85% sericitization (cf. the discussion above) is therefore hard to estimate at this stage, and more data are required before such a value can be established. Nevertheless, an alteration origin and the high K content of sericite allows the K -Ar system to be insensitive to further perturbation, making it an excellent tool for dating alteration events.
As said previously, because the 40 Ar/ 39 Ar technique enables the production of 39 Ar age spectra, we can quantify the amount and the age of a sericitization event if we know the formation age of the fresh plagioclase along with its Ca/K ratio, even if the 37 Ar Ca / 39 Ar K and age spectra are strongly affected. In practice, a limiting factor is related to the understanding of the Ca/K and the age of plagioclase before the metamorphism event. In different geological contexts, we are often able to have a good understanding of the nature of the protolith, particularly if it only suffered a low degree of metamorphism. For instance, acid plutonic rocks commonly have Na-rich plagioclases (with low Ca/K), whereas basaltic rocks display Ca-rich plagioclases (with high Ca/K). The magmatic age of the protolith can be measured using a radio-isotopic method resistant to low-degree metamorphism, such as zircon U -Pb dating.
At the present stage, further studies are necessary to refine our understanding of the behaviour of the K -Ar system during plagioclase sericitization, including: (1) phase mixing experiments (e.g. Van-Laningham & Mark 2011) between plagioclase and muscovite (in progress); and (2) modelling of the diffusion of different sericite -plagioclase argon reservoirs during the step-heating process.

Conclusion
In this study, we investigate the effect of sericitization on the 40 Ar/ 39 Ar age determination of plagioclase using a numerical model including multiple-variable parameters. The most important conclusions are as follows. † Although sericitization of plagioclase below 1% is hard to detect using a stereomicroscope, it can have dramatic consequences on the apparent 40 Ar/ 39 Ar age measured on a plagioclase separate. We show that, in some realistic cases, even ≤0.1-0.2% of sericite can shift the apparent age by several hundred ka and even several Ma. † The most influential parameter affecting the apparent 40 Ar/ 39 Ar age of an altered plagioclase is the age difference between the age of crystallization of the plagioclase and the age of the sericitization event. † When the Ca/K ratio and the crystallization age of fresh plagioclase are known, and the Ca/K ratio of the plagioclase is constant, the amount of sericitization of an altered aliquot of plagioclase and the age of the alteration event can be estimated. † The presence of sericite within plagioclase can be identified using a combination of [Ca/K] Ar ratios, age spectra, and 39 Ar and 37 Ar degassing curves, all produced during a conventional 40 Ar/ 39 Ar step-heating procedure. † Ultimately, above 65% of sericitization of plagioclase, the 40 Ar/ 39 Ar age measured on the plagioclase -sericite mixture will be within +1% of the age of the alteration event. The two ages will be identical above 85 wt% of sericitization. Dating highly sericitized plagioclase is therefore an important tool for dating low-temperature metamorphic events and mineral deposits, especially since the K -Ar system of the sericite is generally immune to further alteration. † Rigorous statistical filtering of plagioclase 40 Ar/ 39 Ar data from large igneous provinces allows the effect of alteration on the age database to be largely minimized if not completely removed, and allows the true crystallization ages of dated magmatic units to be accessed.
Numerical application for the 40 K Pl . Unlike the sericite, which has a fixed chemical composition, plagioclases have various Ca, Na and, thus, K contents. Thus, we have to determine the percentage of K 2 O in plagioclase with regard to the various Na and Ca contents.
The structural formula of plagioclases can be expressed as: For each x and y value -that is for each plagioclase composition ranging from pure end-member of albite (An 0 ) to anorthite (An 100 ) -we can calculate the proportion of K related to those of Ca and Na. The K content in common plagioclase rarely exceeds 5% of the K-rich end-member (see Fig. A1). We can realistically consider that the stoichiometric index, y, ranges from 0 to 0.05.
Thus, we can model various composition of plagioclases, based on their structural formula which gives us the related (1) molecular weight, (2) the %K 2 O and (2) the Ca/K of the plagioclase (see Table A1 and Figs A2 & A3: see also the section on 'Relationship between %wt K 2 O, %wt CaO and Ca/K in plagioclase' later in this Appendix for detailed explanations). For example, y values ranging from 0.001 to 0.05 implies that related K 2 O contents for classical plagioclases range from 0.0017 to 0.87% using an An 50 value for this plagioclase (Fig. A2). For an An 05 , calibrated values are shown in Figure A3.
By knowing the molecular weight and related %wt K 2 O for all of the theoretical plagioclases, we can apply the previous equations (A1)-(A3) (see Table 1A).
Detailed equations for the determination of ( 40 K Ser / 40 K tot ) and ( 40 K Pl / 40 K tot ) in the mixture Using equations (A4) and (2), 40 K tot = (1 − X) 40 K Pl + X 40 K Ser , we determine the two ratios for the mixture as following: and also:  because it is related to the 37 Ar Ca / 39 Ar K ratio measured during the degassing process of analysed minerals. If we apply equations (A1) and (A2) for Ca in plagioclase: We can determine the (Ca/K) 1 mass ratio or the (Ca/ K) 2 molar ratio as following: Consequently, we can write taking into account the molar ratio: If we use equation (A7) in equations (A5) and (A6), we obtain the final equations used in our model:   The %wt CaO for plagioclase is related to the stoichiometric indices or Ca in the structural formula. For our model mixing, we fixed first on An 70 , of An 10 , or An 30... for a given K content (y stoichiometric index) and we calculated the related %wt CaO, Ca/K ratio and, thus, the induced molecular weight of the considered plagioclase (see the example in Table 1A). We input these values into equations (A8) and (A9).
Determination of the sericitization value (X) in 40 Ar/ 39 Ar spectra With the 37 Ar Ca / 39 Ar K data from a real experiment, we approximate the Ca/K ratio of the mélange (Ca/K) mel (hereinafter defined as "s" parameter) with respect to Ca/K and the related percentage of 39 Ar of each temperature step : where s represents the experimental (Ca/K) mel , i is the number of steps and Ca/K is 1.83 × 37 Ar Ca / 39 Ar K (after McMaster Reactor Facilities). We can also define the theoretical value of (Ca/K) mel considering the (Ca/K) Pl of the unaltered plagioclase: Using equations (7) and (8), we can replace K Ser /K Pl with : where A and B are already defined in (7) and (8); C is related to plagioclase chemistry with C ¼ (%CaO) Pl × (K/Ca) Pl × M Pl . Finally, we obtain: We use these equations for the 40 Ar/ 39 Ar data reported in Figures 6 and 7.
For the CAMP plagioclase (Fig. 6), we find a degree of sericitization of approximately 3% (see Fig. A4). For the Karoo sample, we find a slight sericitization of 0.16% (Fig. 7).
Estimate of the age t ′ of the alteration in Ar/Ar spectra Using equation (1) in the text, and considering previous the X value in equations (2) and (3), we can write for t ′ : where t is the age of the unaltered plagioclase, and t mel is the age of the mélange and thus the total gas age during the entire degassing process (for i temperature steps):