Subduction of sediments the special significance of10Be and B

The contribution of a mobile phase, which separated from sediments and altered basaltic crust in the course of subduction and was then transferred into the overlying mantle wedge, is reliably indicated by the 10Be/9Be and B/Be systematics (Table 3.3). Beryllium-10 is mainly generated in the Earth's atmosphere by spallation reactions on O and N nuclei. In the atmosphere, 10Be atoms readily attach to solid and liquid particles, which rain out. Eolian dust is considered as the most significant source of 9Be to the oceans. Thus, the sources and input mechanisms of Be isotopes into the oceans are different. In sea water Be is "particle-reactive"; its residence time in the oceans varies from 100 to 1000 yr (Morris et al., 2002). Because of this short residence time and their different sources, the concentrations of the two Be e143

e143

1200 1000 800

400 200

□ Alaska-Aleutians

i

i o

- o Lesser Antilles

o Marianas

o Nicaragua

_ ♦ Philippines

• Tonga-Kermadec

• •

a n

□ Vanuatu

O* A

" A Kamchatka

„OA

□ t

- t0 ▼

♦ mj

♦ ♦ , T '

P

0.5126 0.5128 0.5130 0.5132

0.5126 0.5128 0.5130 0.5132

isotopes and their ratios vary by a factor — 3 in sea water and in young oceanic sediments (von Blanckenburg et al., 1996), with an average ratio of 10Be/9Be = (8.0 ± 2.5) x 10-8 (Brown et al., 1992b) and a Be concentration in non-carbonate ("undiluted") sediments ~ 1 ppm (Reagan et al., 1994).

Also, the concentration of short-lived 10Be decreases downwards in the sediment column (as the sediments age) and it is generally undetectable below a few hundred metres (Morris etal., 2002). Finally it should be noted that 10Be is mainly attached to leachable phases of ocean sediments, whereas 9Be is incorporated in resistant minerals. Therefore an early dehydration process would extract Be with an enhanced 10Be/9Be ratio, and this ratio would generally (but not always) decrease in residual rocks and late-derived fluids (Bourles et al., 1992).

Boron is highly soluble in sea water, where its residence time is close to 10 Myr. In pelagic (non-carbonate) sediments [B] values vary within a factor — 5, from 30 to 150 ppm with a mean value ~ 70 ppm. The altered oceanic crust is also an important reservoir of B, with [B] ~ 25 ppm, whereas the model DMM and BSE concentrations are quite low, well below 1 ppm (Table 17.1). The above estimates give a mean "initial" ratio of B/Be — 70 in subducting sediments (Fig. 25.3).

If sediments are attached to a hydrous subducting slab, dehydration is inevitable as the temperature and pressure increase. Boron is highly soluble in hydrous fluids whereas Be is relatively immobile. Therefore the B/Be ratios are expected to be fractionated in the course of subduction, being higher in fluid phases and lower in residual phases. This tendency is similar to the evolution expected for the 10Be/9Be ratio. Apart from the labile (surface) siting of 10Be in sediments, another factor in the 10Be/9Be evolution in arcs is dilution, as mantle rocks contain 9Be but no 10Be.

In accord with the above predictions, the 10Be/9Be and B/Be ratios generally correlate and decrease from higher values in fore-arc volcanoes towards the lower values typical for back-arc volcanoes (Fig. 25.3) even though exceptions, e.g. a sharp increase in 10Be/9Be in back-arc volcanics, are also known (Morris et al., 2002). The correlation seen in Fig. 25.3 is interpreted as a mixing trend between two principal end-members (Tera et al., 1986; Reagan et al., 1994). One of these is similar to MORBs or OIBs, with low 10Be/9Be (« 3.5 x 10-11) and B/Be ratios. The other end-member is derived by extrapolation using regression and the average 10Be/9Be value obtained for the bulk pre-subducted sedimentary column. In the case of the Cocos plate the average 10Be/9Be ratio of the "sedimentary end-member" is 460 x 10-11, and the observed regression (Fig. 25.3) gives B/Be in the slab-derived fluid as ~ 1000, exceeding the mean B/Be ratio in subducting rocks (— 70) by a factor > 10. The high B/Be ratio thus indicates that the fluid participated in partial melting in the mantle wedge and the generation of arc magmas. Moreover, co-variations of 10Be/9Be with the activity ratio (230Th/232Th) and the 87Sr/86Sr ratio, and also mass-balance considerations, imply the subduction of almost all the o

X CD

Average B/Be in subducted sediments

Average B/Be in subducted sediments

X CD

B/Be

Fig. 25.3 Beryllium-boron isotopic systematics in Central American arcs (the Cocos plate). The regression indicates mixing between two end-members, an overlying mantle peridotite with quite low 10Be/9Be and B/Be ratios and a slab-derived fluid for which a B/Be ratio, ~ 1000 is obtained via extrapolation of the regression to the bulk ratio 10Be/9Be « 460 estimated for the sedimentary column on the Cocos plate (right-hand plot). The convergence rates in arc systems are generally known from geophysical observations and usually give a few millions of years for the time that elapses from incipient subduction to fluid release from the slab. This allows the amount of 10Be decayed during subduction to be corrected. The decay-corrected input-output inventories of 10Be suggest that the whole sedimentary column is subducted beneath the Central American arcs. After Reagan et al. (1994), © Elsevier Science 1994, reproduced by permission.

O Nicaragua and

El Salvador O Guatemala © Costa Rica • Expected in a fluid at incipient subduction ■ Expected in the mantle wedge

O Nicaragua and

El Salvador O Guatemala © Costa Rica • Expected in a fluid at incipient subduction ■ Expected in the mantle wedge

120 1000

B/Be

Fig. 25.3 Beryllium-boron isotopic systematics in Central American arcs (the Cocos plate). The regression indicates mixing between two end-members, an overlying mantle peridotite with quite low 10Be/9Be and B/Be ratios and a slab-derived fluid for which a B/Be ratio, ~ 1000 is obtained via extrapolation of the regression to the bulk ratio 10Be/9Be « 460 estimated for the sedimentary column on the Cocos plate (right-hand plot). The convergence rates in arc systems are generally known from geophysical observations and usually give a few millions of years for the time that elapses from incipient subduction to fluid release from the slab. This allows the amount of 10Be decayed during subduction to be corrected. The decay-corrected input-output inventories of 10Be suggest that the whole sedimentary column is subducted beneath the Central American arcs. After Reagan et al. (1994), © Elsevier Science 1994, reproduced by permission.

~ 500-m-thick sedimentary column overlying the basaltic crust of the Cocos plate (Fig. 23.1), also pointing to the validity of a sedimentary end-member.

Summarizing, the 10Be, 9Be and B family presents strong evidence not only of slab dehydration and element transfer by a fluid phase but also of the scale of sediment subduction.

Radioactive disequilibria caused by subduction processes

Because 10Be signals trace the expulsion of a mobile fluid phase from young sediments, a comparison of the 10Be/9Be ratios with the U-Th series disequilibria should allow us to determine whether these disequilibria originated by element fractionation in the same fluid segregation; this fractionation could then be dated. An excellent correlation between the 10Be/9Be and 226Ra/230Th ratios was observed for historic lavas from the southern volcanic zone in Chile, where the Nazca plate is subducting beneath the Andes (Fig. 25.4 and 23.1). Radium, whose closest chemical analogue is Ba, readily partitions into a fluid phase while Th is insoluble. Because

(226Ra/230Th)

Fig. 25.4 The 226Ra-230Th-10Be systematics in recent lavas from southern Chile volcanoes. Generally (226Ra/230Th) is not well correlated with 10Be/9Be owing to decoupling between the sediment and fluid signatures but the good correlation here should be considered as "a present from Nature showing how it works": the correlation results from the variable contributions of a slab-derived mobile phase (high ratios) into the mantle wedge, the source of arc melts (low ratios). Similar relationships can also be seen in Fig. 25.5. From Sigmarsson et al. (2002), © Elsevier Science 2002, reproduced by permission.

(226Ra/230Th)

Fig. 25.4 The 226Ra-230Th-10Be systematics in recent lavas from southern Chile volcanoes. Generally (226Ra/230Th) is not well correlated with 10Be/9Be owing to decoupling between the sediment and fluid signatures but the good correlation here should be considered as "a present from Nature showing how it works": the correlation results from the variable contributions of a slab-derived mobile phase (high ratios) into the mantle wedge, the source of arc melts (low ratios). Similar relationships can also be seen in Fig. 25.5. From Sigmarsson et al. (2002), © Elsevier Science 2002, reproduced by permission.

the 10Be is sourced in subducting sediments, the correlated 226Ra most probably originated from the same source. Also, a shift of 10Be/9Be towards 0 is the signature of mantle melts, for which a (226Ra/230Th) activity ratio ~ 1 (secular equilibrium) is expected and is actually observed (Fig. 25.4), thus validating the parameters for the mantle end-member used in the previous section.

The Chilean volcanics further show a good correlation between the activity ratios (226Ra/230Th) and (238U/230Th) (Fig. 25.5), and similar trends are also observed in several other arcs. This is expected for fluid segregation under oxidizing conditions, because U is highly fluid-soluble in the hexavalent state. Note that if the oxygen fugacity were to vary significantly between or within subduction zones this could have a major effect on U-series fractionation. Most observations, however, indicate that fluid phase transport causes fractionation of the Ra-Th-U trio, whereby the fluid carries Ra and U preferentially to Th from the downgoing slab into the overlying mantle.

The observed disequilibrium between 226Ra and 230Th gives an upper time limit for the combined process of mobile phase transfer from a source-slab segment to the melting zone in the mantle wedge followed by melt transport to the Earth's surface. As t226 = 1.6 x 103 yr (Fig. 24.6), these two species equilibrate after about 8 x 103 yrs, and so the transfer time scale must be shorter than this value. If the slab is assumed to be at about 80-100 km depth below the volcanic front, the

V 1985

V 1971

V 1971

V 1985

Fig. 25.5 The 238U-230Th-226Ra systematics in recent lavas from southern Chile volcanoes. Correlations between 226Ra/230Th and U/Th, two fluid-indicative indices, are common for arc volcanics, indicating recent fluid addition into the arc melt source. After Sigmarsson et al. (2002), © Elsevier Science 2002, reproduced by permission.

above time scale gives a minimum average ascent velocity of ~ 10 m yr-1 for fluid and melt (Sigmarsson et al., 2002).

A lower time limit for the fluid + magma segregation and ascent is given by the 228Ra-232Th systematics (t= 6 yr). Even in very young arc lavas this ratio is in secular equilibrium, indicating that 226Ra - 230Th fractionation occurred at least 30 years ago, which is comparable with the value of ~ 50 years derived for mid-ocean ridge magmatism from (210Pb/226Ra) disequilibria (Fig. 24.7(a)). The 30 yr lower time limit gives a maximum average ascent velocity of ~ 2000 m yr-1 (Turner et al., 2004b).

In contrast with the enhanced abundance of fluid-sensitive isotopes discussed above (Fig. 25.5), 235U is less abundant than its fluid-insensitive daughter, 231Pa. Therefore (231Pa/235U) > 1 indicates that not only fluid-related fractionation processes operate in arc magmatism (see Section 25.7 below).

As discussed above, arc magmatic rocks contain enhanced abundances of fluid-soluble (e.g. Rb, K, Ba, Pb, Sr) and sediment-sourced (B, Th, 10Be) elements and isotopes. These observations, along with some revealing isotopic system-atics (10Be/9Be, Section 25.3) and "input-output" correlations (i.e. between the abundances of these species in the slab and those transferred by arc magmas to the

0 0

Post a comment