鋰與隱沒沉積物：海岸山脈的制約

Abstract

鋰同位素被認為具有示蹤殼幔循環的潛力，不過對於控制鋰同位素在隱沒帶演化的因子仍存在爭議，早期研究基於鋰同位素只在地表低溫的環境下分化，加上易進入流體相的特性，認為鋰同位素能示蹤隱沒帶的脫水作用；但近期研究發現，隱沒沉積物的輸入也是影響島弧火山岩鋰同位素的重要因子，因此本研究選擇岩漿成分中混入了沉積物的海岸山脈火山岩，細部探討隱沒沉積物如何影響島弧火山岩鋰同位素。 由於海岸山脈火山岩被認為受到地殼混染，我們發現以繼承鋯石比例結合鍶、釹同位素資料可以建立一套系統，排除明顯受地殼混染影響的樣本。剩餘的樣本中，挑選新鮮的海岸山脈火山岩，在經過酸溶、層析法純化後，以Agilent 7700x ICP-QMS與Nu Plasma II MC-ICP-MS分別測量鋰濃度與鋰同位素比值。 由於用來代表岩漿鋰濃度的Li/Y，以及鋰同位素比值 (δ7Li) 受部分熔融或結晶分異的影響不顯著，樣本數據主要反映岩漿源區的鋰組成變化。分析結果發現海岸山脈火山岩的Li/Y比值 (0.4 ~ 1.4) 與δ7Li (-1.3 ~ +7.8‰) 數值多變且明顯不同於中洋脊玄武岩 (Li/Y = 0.2; δ7Li = +3.7 ± 1.0)，由於Li/Y與流體指標 (e.g., Ba/Th, Sr/Th, and U/Th) 不具相關性，卻隨著時間演進有Li/Y比值上升，同時沉積物熔體指標 (e.g., Th/Ce, Th/Nb, and Th/La) 增加和ℇNd(T) 下降的現象，可推知岩漿源區的鋰組成受到Li/Y較高的沉積物熔體加入而影響；此外，結合鋰、釹同位素計算的岩漿混合模式顯示，ℇNd(T) 較高的樣本主要受到沉積物流體以及換質的海洋地殼所釋出的流體影響，而出現較大的δ7Li變化 (-1.3 ~ +7.8‰)；低ℇNd(T) 的樣本則明顯有沉積物熔體加入源區，因此而限縮了δ7Li的變化範圍 (+2.9 ~ +4.1‰)。 海岸山脈火山岩在鋰濃度與鋰同位素上得到的結論也可適用於其他島弧，岩漿成分中同樣混入沉積物且已報導鋰同位素資料的馬提尼克島，火山岩Li/Y的上升也同樣伴隨著沉積物熔體指標上升與ℇNd(T) 下降的變化；鋰、釹同位素的岩漿混合模式也分別展現出沉積物熔體、流體的影響。承襲前人認為隱沒沉積物能影響島弧火山岩鋰同位素的想法，本研究藉由馬提尼克與海岸山脈的例子進一步發現，隱沒沉積物的熔體及流體兩者都是影響島弧火山岩鋰組成的重要因子。

Regarding lithium has high solubility in liquid phase and its two stable isotopes, 6Li and 7Li, significantly fractionate in the near-surface water-rock interaction, lithium isotopic ratios of arc volcanic rocks have the potential as a tracer of fluid in crust-mantle recycling. However, recent studies suggested that the input of sediments into the mantle wedge can also play a critical role in the Li abundances of arc volcanic rocks. The Coastal Range is the northernmost part of the Northern Luzon arc which has long been demonstrated as one of the arc systems with involvement of sediments in their magma sources, and thus its volcanic rocks are studied here to further examine the effects of sediments on behavior of Li in arc volcanic rocks. Remarkable crustal contamination in the Coastal Range volcanic rocks has long been known. The integration of zircon U-Pb ages and Sr-Nd isotopic data of the Coastal Range volcanic rocks is first proposed in this study to successfully divide samples into two groups and to rule out samples in the group with relatively higher 87Sr/86Sr comparing with Nd isotopic values, resulted from crustal contamination. Then, Li concentrations and isotopic ratios, represented as δ7Li, of the remaining samples were analyzed by ICP-QMS (Agilent 7700x) and MC-ICP-MS (Nu Plasma II), respectively. The variations of both Li/Y and δ7Li reflect the geochemistry of magma sources in the mantle wedge because these two parameters are not significantly modified during fractional crystallization and partial melting. Li/Y (0.4 ~ 1.4) and δ7Li (-1.3 ~ +7.8‰) in the Coastal Range volcanic rocks are significantly variable and different from those of MORB (Li/Y = 0.2; δ7Li = +3.7 ± 1.0). Although Li/Y ratios conventionally are treated as a tracer of fluid in arc magmatism, they show no correlation with other fluid proxies (e.g., Ba/Th, Sr/Th, and U/Th) in our samples. On contrary, Li/Y ratios become higher not only with time, but also increasing values of sediment-melt proxies (e.g., Th/Ce, Th/Nb, and Th/La) and decreasing ℇNd(T). This demonstrates the involvement of sediment melt, with higher Li/Y than that of MORB, in magma source of the Coastal Range arc volcanic rocks. Li isotopic values of our samples do not vary with their corresponding Li contents. Mixing models of Li-Nd isotopes suggest that δ7Li values in samples with higher ℇNd(T) were essentially influenced by slab fluids derived from sediments and altered oceanic crust, while those in samples with lower ℇNd(T) show a relatively limited δ7Li range and were dominated by melt of subducted sediments. The results of this study can also be applied to those of the arc magmatism in the Martinique island arc, the only reported case that so far has the input of sediments to the mantle wedge and also has Li isotopic data reported. We further suggest that both the melt and fluids from subducted sediments can play key roles in the evolution of Li in subduction zone.