Zhiyu
Han
*ab,
Fei
Wang
ab,
Liekun
Yang
ab,
Wenbei
Shi
ac and
Yinzhi
Wang
ac
aState Key Laboratory of Lithospheric and Environmental Coevolution, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
bCollege of Earth and Planetary Sciences, University of Chinese Academy of Sciences, Beijing, China
cInstitutional Center for Shared Technologies and Facilities, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
First published on 12th March 2025
Accurate dating of submarine volcanic rocks is crucial for understanding many geological processes. However, some submarine volcanic rocks yield inaccurate young 40Ar/39Ar ages. To address this issue, we conducted 40Ar/39Ar, electron microprobe and energy-dispersive X-ray spectroscopy analyses on hornblende, plagioclase and glass samples of the fallout tephra from borehole U1396A of the Integrated Ocean Drilling Program (IODP) 340 Expedition. Based on the 40Ar/39Ar analyses, 3 samples yield reasonable ages for the tephra, whereas the other 6 samples exhibit abnormally young ages. Remarkably, glass is widely present in hornblendes and plagioclases, as shown in the field emission scanning electron microscopy (FESEM) images. These glass samples have much higher K contents than the hornblende and plagioclase samples. Based on the K distribution of the glass, we propose that seawater can significantly increase the K content in some samples containing glass during the post-eruption period. Due to the extra addition of K, the determined 40Ar/39Ar ages of these submarine samples are predicted to be younger and are incorrect. In this study, an effective pretreatment method is proposed to reduce the effect of glass on the 40Ar/39Ar dating of the hornblende samples. The three hornblende samples subjected to our pretreatment method have all yielded more reliable 40Ar/39Ar ages, which suggests that our pretreatment method has the potential to enable more submarine volcanic rock samples to produce trustworthy 40Ar/39Ar age determinations in the future.
The 40Ar/39Ar (K–Ar) method is one of the most commonly used isotopic dating methods due to the multiple types of volcanic products suitable for testing, for example, feldspars, mica, amphibole and volcanic glass.8 This method has the potential to be a good solution for determining the ages of submarine volcanic rocks. However, in practice, when the 40Ar/39Ar (K–Ar) method is used to date submarine volcanic rocks, some samples produce inaccurate young apparent ages.9–13 These erroneous apparent ages are often interpreted as resulting from either the addition of potassium (K) or the loss of argon (Ar) in submarine volcanic rocks due to submarine weathering, because in the K–Ar isotopic system, both the addition of the radioactive parent isotope (K) and the loss of the radiogenic daughter isotope (Ar) during subsequent geological processes can result in an underestimation of the apparent age of samples.8 Unfortunately, in cases where samples have experienced K addition, it is difficult to verify whether Ar loss has occurred. Melson and Thompson (1973) observed that in a submarine environment, the alternation process leads to formation of secondary K-rich smectites within microfractures, which contribute additional potassium to the submarine basalts.14 Consequently, these basalts show young apparent 40Ar/39Ar ages; these young ages result from the addition of parent isotopes. The formation of K-bearing altered minerals appears to be a reasonable cause for some unexpected young ages. Therefore, to avoid the influence of altered minerals on the 40Ar/39Ar analyses, researchers tend to use the least altered fraction for 40Ar/39Ar dating. However, even after careful selection, some analyzed samples continue to display apparent 40Ar/39Ar ages which are difficult to interpret.7,15–18 These samples often show significant perturbations in the 40Ar/39Ar age spectra and possess inconsistent 40Ar/39Ar ages compared with stratigraphic limitations. In some cases, different minerals from the same volcanic rock display distinct ages. In summary, even after excluding the interference of alteration minerals, the 40Ar/39Ar dating results of submarine volcanic rocks still exhibit anomalously young apparent ages. Researchers often attribute these erroneous apparent ages simply to the addition of potassium (K) or the loss of argon (Ar) during the post-eruption phase. While these explanations have their merits, they also raise several questions. For instance, how does the addition of potassium (K) occur in the samples, and to what extent does it affect the ages? Can samples really lose Ar without significant alteration? To address these questions, we conducted a detailed study of submarine volcanic rock samples from the Lesser Antilles Arc (LAA) region, including particle morphology of volcanic ash, along with 40Ar/39Ar dating, electron microprobe analysis (EPMA), electron microscope observation, and energy spectrum analysis (EDS) of hornblende, plagioclase, and volcanic glass samples. This article provides a new perspective on the inaccuracies in 40Ar/39Ar dating ages of submarine volcanic rocks. Furthermore, we developed an effective pretreatment method specifically for hornblende samples in submarine volcanic ash to enable these samples to yield more reliable 40Ar/39Ar ages.
The IODP Expedition 340 drilled boreholes at a series of sites along the LAA to examine the constructive and destructive processes of volcanic edifices in the LAA. Among them, borehole U1396 (16°30.49′N, 62°27.10′W) was drilled on a topographic high location of ∼30 km southwest of Montserrat (Fig. 1). This location was ideal for reconstructing the long-term eruptive history of the northern LAA. The 1396A core stretches from the surface down to a depth of 134.9 m below the seafloor. The core consists of a sequence of hemipelagic sediments, tephra layers and volcanoclastic muds.24,25 Pb isotope data show that the tephra layers in the borehole are mainly from volcanoes in Guadeloupe and Montserrat.26
Briden et al. (1979) reported a K–Ar age of 38–10 Ma in the outer arc (eastern chain), and the most inner arc (western chain) volcanic rocks had K–Ar ages of less than 3 Ma.21 Zami et al. (2014) reported that a Les Saintes sample had a K–Ar age of 2.98 ± 0.04 Ma, and they claimed that this sample formed during the initial subaerial volcanic activity along the western chain of the LAA.27 Harford et al. (2002) conducted 40Ar/39Ar tests on groundmass samples from Montserrat and identified three volcanic centers in Montserrat: Silver Hills, Central Hills and South Soufriere Hills, with 40Ar/39Ar ages ranging from ∼2.6 Ma to the present.28 Samper et al. (2007) conducted the K–Ar Cassignol–Gillot technique on samples from Guadeloupe and obtained a time scale between 2.79 ± 0.04 Ma and 0.435 ± 0.008 Ma.29 In summary, most subaerial volcanic rocks in Guadeloupe and Montserrat are less than 3 Ma in age. However, paleomagnetic records from U1396A indicated that the volcanic activity began ∼4.5 million years ago in the northern LAA;30 therefore, this drill core provides a longer history of arc volcanism in the northern LAA dating back to the early Pliocene. Moreover, the Cenozoic volcanic ashes from core 1396A are ideal for investigating the effects of seawater on the 40Ar/39Ar ages because we can easily assess the accuracy of our dating results by comparing the deposition sequence of the layers and the corresponding paleomagnetic records.
Sample name | Depth CSF-A (m) | Number of analysis particles | Average perimeter (area) | Average aspect ratio (area) |
---|---|---|---|---|
4h773 | 34.33 | 79 | 19.50 | 27.64 |
6h558 | 50.18 | 139 | 57.53 | 194.34 |
7h620 | 60.80 | 136 | 75.87 | 393.31 |
10h1110 | 82.70 | 139 | 39.81 | 65.85 |
10h334 | 84.95 | 97 | 56.36 | 97.77 |
10h482 | 86.91 | 118 | 97.25 | 416.54 |
10h63 | 89.05 | 74 | 61.22 | 223.96 |
11h161 | 91.71 | 84 | 34.36 | 68.74 |
11h480 | 96.32 | 47 | 16.22 | 14.41 |
12h228 | 102.36 | 124 | 47.68 | 146.08 |
14h691 | 124.32 | 82 | 62.75 | 142.14 |
Sample name | Material | Top depths CCSF-A (m) | Interpolated depositional age (Ma) | Total fusion age (Ma) | 39Ar (%) | Plateau age (Ma) | MSWD | Inverse isochron age | 40Ar/36Ar initials |
---|---|---|---|---|---|---|---|---|---|
a Errors are reported at the 2σ confidence level. The top depths CCSF-A and interpolated depositional age are calculated based on data of Hatfield (2015). | |||||||||
First-round analysis results | |||||||||
6h558g | Glass | 56.06 | 2.44 | 2.12 ± 0.05 | 75.66 | 2.20 ± 0.06 | 1.27 | — | 281.5 ± 11.1 |
7h620-p | Plagioclase | 67.82 | 2.73 | 2.61 ± 1.08 | 100.00 | 2.55 ± 1.07 | 1.30 | 2.22 ± 1.39 | 303.2 ± 28.5 |
10h1110h | Hornblende | 92.07 | 3.38 | 1.78 ± 0.85 | 66.10 | 2.37 ± 0.75 | 0.71 | — | 292.6 ± 5.6 |
10h334h | Hornblende | 94.56 | 3.44 | 2.48 ± 0.90 | — | — | — | — | 276.9 ± 12.3 |
10h482-p | Plagioclase | 96.73 | 3.49 | 2.52 ± 1.75 | 85.42 | 1.36 ± 1.47 | 0.96 | — | 302.4 ± 21.4 |
10h63h | Hornblende | 99.10 | 3.6 | 3.53 ± 1.22 | 84.43 | 3.41 ± 1.07 | 0.55 | 3.31 ± 2.30 | 298.4 ± 56.8 |
11h161-p | Plagioclase | 102.04 | 3.6 | 2.77 ± 0.31 | 95.36 | 2.92 ± 0.29 | 0.57 | — | 277.7 ± 16.5 |
12h228h | Hornblende | 113.83 | 3.8 | 3.15 ± 1.46 | — | — | — | — | 284.3 ± 25.8 |
14h691-p | Plagioclase | 138.15 | 4.2 | 4.04 ± 0.65 | 97.44 | 4.04 ± 0.65 | 0.64 | 4.57 ± 1.34 | 277.7 ± 40.6 |
![]() |
|||||||||
Second-round analysis results | |||||||||
10h1110h-processed | Hornblende | 92.07 | 3.38 | 3.27 ± 1.0653 | 53.81 | 2.79 ± 0.77 | 0.89 | 2.52 ± 0.90 | 308.7 ± 17.5 |
10h334h-processed | Hornblende | 94.56 | 3.44 | 3.34 ± 1.12 | 85.73 | 3.21 ± 1.06 | 0.58 | 3.73 ± 2.33 | 294.6 ± 3.9 |
12h228h-processed | Hornblende | 113.83 | 3.8 | 3.39 ± 0.74 | — | — | — | 3.47 ± 1.63 | 287.1 ± 110 |
All samples were wrapped in aluminum foil and then stacked in quartz vials with the international standard YBCs sanidine32 (29.286 ± 0.045 Ma). The position of each sample was recorded as the distance from the bottom of the vial. The vials were subsequently sealed in vacuo and placed in an aluminum canister. The canister was wrapped with cadmium foil to shield the slow neutrons during irradiation.
For the first round, neutron irradiation was carried out at position B4 of the 49-2 Nuclear Reactor (49-2 NR), Beijing (China), for 4 h in December 2020. For the second round, neutron irradiation was carried out in a high-flux engineering test reactor (HFETR), Sichuan (China), for 5 h in July 2024. The J values for each sample are listed in Table A1, ESI.† Interference reactions during irradiation were checked by using pure CaF2 and K2SO4; these samples produced the following correction factors: [36Ar/37Ar]Ca = 0.000269 ± 0.0000015, [39Ar/37Ar]Ca = 0.000709 ± 0.0000020, and [40Ar/39Ar]K = 0.00165 ± 0.00011 for the first-round irradiation and [36Ar/37Ar]Ca = 0.000230 ± 0.0000078, [39Ar/37Ar]Ca = 0.000660 ± 0.0000010 and [40Ar/39Ar]K = 0.00788 ± 0.00025 for the second-round irradiation.
The 40Ar/39Ar measurements were carried out at the IGGCAS, Beijing. After irradiation, the samples were removed from the vials and then placed on a high-purity copper disc. The disc was sealed in the chamber and maintained at ∼140 °C for 4 days. The samples were heated using a CO2 laser and degassed at 0.4 W for 30 s. We conducted 6 to 8 heating steps for each sample to ensure the collection of a sufficient amount of gas at each step. The gas was purified with two Zr–Al pumps (SAES NP10). The purified gas was transferred to a Noblesse mass spectrometer for isotopic analysis. A decay constant of 5.543 × 10−10 year−1 for 40K was used in the calculation of age.33 The data were processed via ArArCALC software.34 Errors are reported at the 2σ confidence level.
The major element components of glass, hornblende and plagioclase were analyzed using a JXA8230 electron microprobe at IGGCAS, Beijing, China, with a voltage of 15 kV, a beam current of 20 nA, a spot size of 0 μm and a 10–30 s peak counting time. The natural minerals and synthetic oxides that were used for calibration were as follows: albite (Na and Al), bustamite (Mn), diopside (Ca, Si, and Mg), apatite (P), K-feldspar (K), rutile (Ti), and Fe2O3 (Fe). The spectral lines adopted for each element were as follows: Na (129.393 nm), Mn (146.186 nm), K (120.157 nm), Mg (107.399 nm), Si (77.371 nm), Fe (134.625 nm), Al (90.551 nm), Ca (107.907 nm), and Ti (88.751 nm). The detection limits (1 s) were 0.02 wt% for Mg, Al, Cr, K, Si, Mn, Ca and Fe and 0.04 wt% for the P. A program based on the ZAF procedure is used for data correction.
A high-resolution field emission scanning electron microscope (Zeiss Gemini 450) operating at 15 kV and a beam current of 2 nA was used to observe and image the microstructures of the samples. To eliminate charging and improve the contrast for low-density materials, a carbon coating of approximately 8 nm was applied to the samples using a coater system (Leica EM ACE600) to facilitate high-resolution FESEM imaging. To reveal the elemental distribution of the samples, energy-dispersive X-ray spectroscopy (EDS) data were collected with Oxford Ultim Max 60 mm2 EDS detectors attached to the SEM at 15 kV. The EDS mapping was performed using an acceleration voltage of 20 kV, a beam current of 5 nA, a working distance of 8.5 mm, and a dwell time of 100 μs per point.
![]() | ||
Fig. 5 Grain morphology test results. The dashed line shows the reference line that distinguishes primary tephra fallout and reworked tephra.31 The two samples falling within the range of reworked tephra are 4h773 and 11h480. |
The total fusion ages of the samples for the first-round 40Ar/39Ar analyses are shown in Table 2 and Fig. 6a. The total fusion ages of plagioclase samples 14h691-p (4.01 ± 0.65 Ma) and 7h620-p (2.61 ± 1.08 Ma) and hornblende samples 10h63h (3.53 ± 1.22 Ma) produce an age–depth relationship that is nearly the same as the paleomagnetic record. Moreover, within analytical uncertainties, these samples exhibit concordant plateau ages and inverse isochron ages that agree with the total fusion ages (Fig. 6 and 7b1–d3).14h691h has an age plateau of 4.04 ± 0.65 Ma and an inverse isochron age of 4.57 ± 1.34 Ma, 7h620-p has an age plateau of 2.55 ± 1.07 Ma and an inverse isochron age of 2.22 ± 1.39 Ma, and 10h63 has a plateau age of 3.41 ± 1.07 Ma and an inverse isochron age of 3.31 ± 2.30 Ma. The initial 40Ar/36Ar ratios ((40Ar/36Ar)0) of 7h620-p (302.2 ± 28.5), 10h63h (298.4 ± 56.8) and 14h691-p (277.7 ± 40.6) are indistinguishable from the atmospheric value43 (298.56) within the allowed error range. All evidence mentioned above indicates that the 40Ar/39Ar analyses of 14h691-p, 7h620-p and 10h63h provide reasonable age estimates of the corresponding eruptive activity.
![]() | ||
Fig. 6 Total fusion ages of the first round of 40Ar/39Ar analyses (a) and second round of 40Ar/39Ar analyses (b). In (b), the total fusion ages of the two plagioclases and 1 hornblende samples are also shown for comparison. The errors are reported at the 2σ confidence level. The red dashed line shows the age–depth relationship, which is defined by the paleomagnetic record.30 |
In contrast, the glass samples (6h558g), 3 hornblende samples (10h1110h, 10h334h, and 12h228h) and 2 plagioclase samples (10h482-p and 11h161-p) produce clearly younger total fusion and plateau ages than the previously defined age–depth relationships (Fig. 6a, 7a1–a3 and e1–i3). Several laser heating steps of 10h334h, 10h482h and 12h228h result in a low proportion of radiogenic 40Ar (<5%, Table A1, ESI†). Under these circumstances, a slight fluctuation of the argon blank during testing using the mass spectrometer can lead to intense changes in the age spectrum. This would make the interpretation of the age spectrum highly challenging. A more evident pattern of age change is observed from the age spectra of 6h558g, 10h1110h and 11h161-p. In the first 3–4 laser heating steps of these samples, the stage ages gradually increase from a young age to an older age. The initial 40Ar/36Ar ratios ((40Ar/36Ar)0) of 10h1110h (292.6 ± 5.6), 10h482-p (302.4 ± 21.4) and 12h228h (284.3 ± 25.8) are indistinguishable from the atmospheric value, whereas the initial 40Ar/36Ar ratios of 6h558g (281.5 ± 11.1), 10h334g (276.9 ± 12.3) and 11h161-p (277.7 ± 16.5) are slightly lower than the atmospheric value. The inverse isochron ages of these younger-group samples are not given, and this aspect is further explained in the discussion section.
The second-round 40Ar/39Ar analysis results for the processed hornblende samples are shown in Fig. 6b and Table 2. Compared with the unprocessed samples, the processed hornblende samples present higher total fusion ages, from 1.78 ± 0.85 Ma to 3.27 ± 1.06 Ma for 10h1120h, 2.48 ± 0.90 Ma to 3.34 ± 1.12 Ma for 10h334h and 3.15 ± 1.46 Ma to 3.39 ± 0.74 Ma for 12h228h. The two processed samples, with defined plateau ages, also increase in age from 2.37 ± 0.75 Ma to 2.79 ± 0.77 Ma for 10h1110h and from no-plateau age to 3.21 ± 1.06 Ma for 10h334h. In addition, compared with their plateau ages, samples 10h1110h-processed and 10h334h-processed yield consistent inverse isochron ages within the allowed error range, and these values are 2.52 ± 0.90 Ma and 3.73 ± 2.33 Ma, respectively.
Notably, almost all samples in this study presented decreasing K/Ca ratios with increasing laser heating temperature (Table A1, ESI† and Fig. 7). This decreasing trend indicates that K and Ca are nonuniformly distributed in the samples.
In glass sample 6h558g, the K2O content has an average value of 0.5137 wt% and shows a small variation range in the inner part of the grain (0.467–0.565 wt%), whereas in the outermost 5 μm portion, the K2O content significantly increases and varies from 0.482 wt% to 0.627 wt% (Fig. 8a–c). For hornblendes, most samples have a K content of ∼0.02 wt%; the K content fluctuates within a small range from the core to the rim, but no clear trend is observed (Fig. 8g–o). Plagioclases have a K2O content of ∼0.08 wt%, which is similar to that of hornblendes; additionally, a K content variation occurs inside the grains, but no clear trend is observed (Fig. 8d–f).
The FESEM images of the hornblende samples show no clear signs of alteration inside the mineral, but they reveal a noteworthy phenomenon. Most hornblende grains exhibit a glass edge surrounding the mineral particles (Fig. 8g, j and m). The thickness of these glass edges is usually a few microns to tens of microns. The EDS data revealed that the K2O contents of these glass edges (∼2 wt%) were several orders of magnitude greater than those of the hornblendes (∼0.1 wt%) (note the difference between the EDS and electron microprobe data). We further tested the K distribution inside the glass edges of the hornblende sample. For each glass edge, we ran the analysis every 1 μm from the linkage between the glass edges and the hornblende to the rim. The results are shown in Table A3, ESI† and Fig. 8.
The K distributions of the glass edges are distinct in sample 10h63h and other hornblende samples. In sample 10h63h, the K2O content has a narrow variation range from 1.3 to 1.5 wt% (Fig. 8g–i), while in samples 10h1110h, 10h334h and 12h228h, the K2O content significantly increases from the innermost region to the rim of the glass edges (Fig. 8j–o). In sample 10h1110h, the K2O content increases from 1.6 wt% to 4.0 wt%. In sample 10h334h, the K2O content increases from 1.5 wt% to 2.1 wt%, and in sample 12h228h, the K2O content increases from 1.8 wt% to 3.7 wt%.
Similar to the hornblendes, no alteration was observed inside the plagioclases. Glass also appears in some plagioclase samples but had different modes of occurrence. In samples 11h161-p and 10h482-p, glass exists mainly in the vitreous inclusions of the plagioclases (Fig. 8p and s). These inclusions normally have a diameter of a few microns to tens of microns, and many vitreous inclusions are in contact with the fractures in the plagioclases. In sample 7h620-p, no vitreous inclusions are observed, and in sample 14h691-p, the vitreous inclusions are small in size (Fig. 8d and e). The K2O contents of the vitreous inclusions are closely related to the fractures in plagioclase. The major element contents of 22 vitreous inclusions show that 10 of these inclusions are in direct contact with the fractures and the other 12 are isolated from fractures (Table A3, ESI†). Interestingly, these two groups of inclusions have distinct K contents. The K2O content of the vitreous inclusions connected to the fractures is as high as ∼3 wt%, whereas the K2O content of the isolated inclusions is as low as ∼1.5 wt% (Fig. 8p–r). For a case of a vitreous inclusion in sample 10h482-p that connects the fractures at its two ends, the K2O content gradually increases from 1.9 wt% in the center to 2.6 wt% in both ends (Fig. 8s–u).
A simple evaluation of the influence of glass on the 40Ar/39Ar age can be performed. We assume that the hornblende is a regular cylinder and that the glass edges are cylindrical shells (Fig. 9). Assuming a hornblende grain in the shape of a cylinder with a diameter of 200 μm and a height of 800 μm, where only one-quarter of the hornblende is enclosed by a cylindrical glass shell with a thickness of 5 μm, and considering the typical K contents of glass (0.5 wt%) and hornblende (0.02 wt%), it can be calculated that the K content at the glass edge constitutes 40% of the total K content. Despite this inaccurate assumption, the role that glass plays in 40Ar/39Ar dating cannot be disregarded.
![]() | ||
Fig. 9 Schematic diagram of the calculation method for estimating the K content of the glass edge of hornblende. |
As mentioned above, the K content has extensively changed in the glass edges of the hornblende samples of 10h1110h, 10h334h and 12h228h. This variation can be attributed to an increase from the innermost region to the rim or a decrease from the rim to the innermost region. The latter variation means that K has diffused from glass to minerals; Ozima et al. (1977) reported a diffusion model to explain the co-reduction of K content and K–Ar age of submarine basalts from the surface to the interior.12 He assumed that K had diffused from seawater into the basalt. If this process were to occur at the boundary between glass and hornblende as well, given that the potassium (K) content in the glass is much higher than that in hornblende, we would expect to observe a significant increase in potassium (K) content at the edges of the hornblende. However, we did not observe such a phenomenon (Fig. 8g–i). Hence, a more likely explanation is that K gradually increases from the inner region to the rim of the glass edge. The elevated K in the rim of glass sample 6h558g also supports this deduction (Fig. 8a–c). The extra addition of K after eruption significantly decreases the total 40Ar/39Ar fusion age. In contrast, no extra K is added to the glass edge in sample 10h63h, and the consistent total fusion age with the interpolated depositional age indicates that the glass and hornblende share similar ages. Thus, the 40Ar/39Ar age of 10h63h can represent the eruption time.
The K distribution in plagioclase samples can better clarify the possible source of this K variation. In samples 11h161-p and 10h482-p, the K contents of the vitreous inclusions tend to be higher when they are connected to the fractures of plagioclases (Fig. 8p–u). This relationship demonstrates that the fractures of plagioclase can provide pathways for seawater ion exchange with the vitreous inclusions, resulting in an increased K content in these samples and leading to a younger total fusion 40Ar/39Ar age. Dymond and Windom (1968) found plagioclase dates 25–50% younger than coexisting biotite in a dredged Pacific basalt sample; they attributed this difference to argon loss of plagioclase, although they could not rule out the possibility of potassium addition.9 Regan et al. (2019) reported that some plagioclases showed slightly younger 40Ar/39Ar age (∼47 Ma) compared to the coexisting pyroxene (∼51 Ma); they interpreted 47 Ma to be a reheating age.7 However, according to the observation of this study, the vitreous inclusions embedded in plagioclase are very difficult to eliminate (see more discussion in 5.3); the young apparent 40Ar/39Ar age may be caused by the vitreous inclusions with varying K content. In samples 7h620-p and 14h691-p, the vitreous inclusions are uncommon; thus, glass only has a slight effect on the 40Ar/39Ar age.
In Fig. 6b, we showed the total fusion age of processed and unprocessed samples. The unprocessed samples, 10h63h, 7h620p and 14h691p, exhibit only a minor effect from altered glass, and we assumed that the influence of glass on the 40Ar/39Ar age of processed samples (10h1110h, 10h334h and 12h228h) can be negligible. If significant Ar loss occurs in hornblende and plagioclase during seawater alteration, then they should give 40Ar/39Ar ages which are significantly younger than the stratigraphic reference age. However, our results showed that 5/6 samples gave 40Ar/39Ar total fusion ages which were highly consistent with the age–depth relationships which are defined by the paleomagnetic record. At least according to our results, Ar loss is not the main factor that causes the young apparent 40Ar/39Ar age.
Most samples in this study showed a decrease in the K/Ca ratio as the laser heating temperature increased (Fig. 7). In calcium-rich minerals, the redistribution of 37Ar and 39Ar during irradiation can lead to a gradual decrease in K/Ca ratios with increasing step-heating temperatures.54 However, according to Jourdan and Renne's (2014) study, such redistribution effects are more pronounced in fine-grained minerals (<5 μm).54 In contrast, the majority of our mineral grains exceed 200 μm in size. Additionally, if the observed gradual decrease in K/Ca were caused by isotopic redistribution of 37Ar and 39Ar, it would typically coincide with progressively younger ages in the step-heating age spectra.54 Such a trend, however, is not evident in our samples. In summary, we propose that the nuclear recoil of 37Ar and 39Ar during irradiation likely has negligible effects on K/Ca ratios and age determinations. Given that glass exhibits a higher K/Ca ratio (∼0.2) than hornblende (∼0.02) and plagioclase (∼0.01), the observed decreases in K/Ca are more likely to reflect a diminished contribution of glass with increased heating temperature.
In the preceding discussion, we noted that the glass rim of sample 6h558g exhibits elevated potassium (K) content and yields younger total fusion and plateau ages. We propose that this sample effectively illustrates the contribution of glass within hornblende or plagioclase. Fig. 7a1–a3 display the age spectrum, K/Ca ratios, and radiogenic 40Ar release pattern of glass 6h558g. Notably, the radiogenic 40Ar release pattern remains relatively stable with increasing step-heating temperatures, indicating that the glass continues to degas even at high-temperature steps. Another distinct feature is the gradual decline in K/Ca ratios as step-heating temperature increases. The glass rim, which experienced K addition, initially shows higher K/Ca values. The observed decrease in K/Ca likely reflects a diminishing contribution of the K-enriched rim to the gas release. Furthermore, the K-enriched rim, compared to the glass interior, would yield lower 40Ar/39Ar ages. The hypothesis of reduced glass rim contribution at higher temperatures aligns with the progressively older step-heating ages observed. This inverse correlation between K/Ca and step heating age is particularly pronounced in the low-temperature steps, suggesting that K-enriched sites within the glass are preferentially degassed during early heating stages. However, we caution against dismissing glass rim effects at high-temperature steps, as the plateau ages at these stages remain younger than the reference age (Table 2).
The preferential release of Ar from altered glass during low-temperature steps is similarly evident in plagioclase and hornblende samples. Due to the low radiogenic 40Ar content (40Arr < 5%) and low resolution (these samples tend to release most of the 40Arr in 1 or 2 discontinuous steps) of the age spectra of 10h334h, 10h482-p and 12h228h, the correlations between K/Ca and the age spectra of these samples were not clear (Fig. 7g1–i3). A more distinct correlation in samples 10h1110h and 11h161-p was observed. In the first 5 stages for 11h161-p, the stage age gradually increased from 0.5 Ma to 3.14 Ma, and the K/Ca ratio decreased from 0.355 to 0.060. For hornblende sample 10h1110h, the first four-stage age increased from 0.41 Ma to 3.11 Ma, and the K/Ca ratio decreased from 0.596 to 0.355 (Fig. 7e1–f3). The covariation trend between step-heating ages and K/Ca ratios closely mirrors that observed in glass sample 6h558g during low-temperature steps, suggesting that the contribution of altered glass to argon release progressively diminishes with increasing temperature. This finding is further supported by our observation of a progressive increase in radiogenic 40Ar release during low-temperature steps in both samples. Because, compared to glass, hornblende and plagioclase have a higher total content of radiogenic 40Ar based on the assumption we made in Section 5.1.1. However, in sample 10h63h, the decrease in the K/Ca ratio was not accompanied by any increase in the step heating age (Fig. 7b1). This discrepancy was potentially caused by the distinct K distribution in the glass. For sample 10h63h, the glass edges showed no sign of K addition; glass and hornblende should have an identical age. Thus, the contribution proportion of glass change did not cause a fluctuation in the age spectrum. The sample 7h620-p exhibits similar change patterns to 10h63h. Compared to 11h161-p, sample 14h691-p exhibits a less pronounced trend of age variation with K/Ca ratios. Although these two samples still show decreased K/Ca ratios with increased step-heating temperature, the magnitude of decrease is significantly lower than that observed in other plagioclase samples (Fig. 7c1–d3), because 7h620-p and 14h691-p exhibit a limited effect of glass as demonstrated in Section 5.1.1.
We have shown the nonuniform distribution of K in the glass in some samples; due to the continuous ion exchange between the seawater and glass, the K atoms inside the sample have different decay times, resulting in a nonuniform distribution of 40Ar in the samples. The Ar isotope correlation analyses of the inverse isochrons in these samples yield meaningless inverse isochron ages due to the poor linearity between the 36Ar/40Ar and 39Ar/40Ar ratios. However, the influence of glass in the Ar isotope correlation diagrams of 6h558g, 11h161-p and 10h334h can still be observed (Fig. 7a3, f3 and g3). These samples exhibit initial 40Ar/36Ar ratios ((40Ar/36Ar)0), which are lower than the atmospheric values. If the samples remained in a “closed system” after eruption, they should have initial 40Ar/36Ar ratios ((40Ar/36Ar)0), which are similar to the atmospheric values; however, the increased K values in these samples lead to higher 39Ar/40Ar values at early laser heating stages, cause the isochron to shift upward, and finally produce a higher initial 36Ar/40Ar ratio of these samples.
The glass occurs mostly as edges encircling the hornblende. Here, we designed a pretreatment method to remove the glass edges of hornblende. The detailed method is described in the Methods section above. Fig. 4b and c display photographs of the samples before and after pretreatment. The images demonstrate a noticeable reduction in glass content following the pretreatment process. The total fusion ages and age spectra of the 3 unprocessed hornblendes and 3 processed hornblendes are shown in Fig. 6. All three samples showed varying degrees of increase in total fusion age. The samples 10h1110h-processed and 10h334h-processed yielded total fusion ages of 3.27 ± 1.06 Ma and 3.34 ± 1.12 Ma, respectively; these ages were highly consistent with the age–depth relationships defined using 7h620-p, 10h63h and 14h691-p. Sample 12h228h-processed still provided a slightly younger total fusion age (3.39 ± 0.74 Ma); however, this age estimation was improved with respect to the unprocessed sample. Although we could not define a reliable plateau age for sample 12h228h-processed, samples 10h1110h-processed and 10h334h-processed yielded plateau ages of 2.79 ± 0.77 Ma and 3.21 ± 1.06 Ma, respectively; these ages were consistent with their total fusion ages and inverse isochron ages (2.52 ± 0.90 Ma for 10h1110h-processed and 3.73 ± 2.33 Ma for 10h334h-processed) within the allowed error range (Fig. 10a–d). These results indicated that our pretreatment method could effectively remove the glass edges of hornblende and yield high-quality 40Ar/39Ar data.
![]() | ||
Fig. 10 Age spectra and inverse isochron diagrams of the following samples: 10h334-h-processed (a and b), 10h1110-h-processed (c and d) and 12h228-h-processed (e and f). |
For plagioclase samples, the influence of glass was difficult to avoid because glass mostly was present inside plagioclase, and the majority of vitreous inclusions had a diameter of a few microns. To remove the vitreous inclusions, the plagioclase needed to be crushed into fine grains that could be removed. However, this process would cause serious 39Ar loss from the nuclear recoil effect during irradiation in a reactor before 40Ar/39Ar dating.8,55,56 Therefore, for submarine plagioclase samples, detailed observations are needed to determine whether plagioclase contains vitreous inclusions before 40Ar/39Ar dating is used.
2. Based on FESEM images, glass was widely present in the hornblendes and plagioclases but in different forms. The glass was predominantly found in the edges surrounding the hornblendes, whereas the glass existed inside plagioclase, mainly as vitreous inclusions.
3. Based on electron microprobe and energy spectrum analyses, the K content of glass was several orders of magnitude greater than that of hornblende and plagioclase and provided a considerable contribution to radiogenic 40Ar.
4. Based on the K distribution in the different samples, we proposed that the abnormally high K content in the glass resulted from alternation with seawater and could cause inaccurate young 40Ar/39Ar ages for some hornblendes and plagioclases and there is no significant Ar loss that occurred in the hornblende and plagioclase during the post-eruption period.
5. Based on the age spectra and K/Ca ratios of samples, altered glass tends to have a greater impact on the age of hornblende and plagioclase during low-temperature stages.
6. We designed a pretreatment method for the submarine hornblende sample dating to reduce the effect from the glass. The second-round 40Ar/39Ar analyses confirmed the validity of this method. We also deduced that not all submarine plagioclase samples were suitable for 40Ar/39Ar dating.
Footnote |
† Electronic supplementary information (ESI) available: Table A1 shows the detailed results of 40Ar/39Ar analysis of 12 samples. Table A2 shows the major element content of hornblende, plagioclase and glass which were tested using an electron microprobe. Table A3 shows the major element contents of glass inside the minerals which were tested by energy-dispersive X-ray spectroscopy analyses. See DOI: https://doi.org/10.1039/d5ja00045a |
This journal is © The Royal Society of Chemistry 2025 |