The origin of mafic fiamme in the late-erupted Carpenter Ridge Tuff, CO: Insights into the construction of a zoned magma chamber

The origin of compositional zoning in ignimbrites has long been a central question of volcanic petrology. The Carpenter Ridge Tuff (CRT) erupted ~27.5 Ma in the Southern Rocky Mountain volcanic field (SRMVF) and is one of the largest (~1,000 km3) of such zoned deposits. The source of late-erupted, dacitic fiamme found above the dominant crystal-poor (5-15% crystals) rhyolite zone of the CRT remains controversial. These fiamme can have extremely high whole-rock Ba, Zr and Sr contents compared to the host rhyolite, and were, therefore, considered to be derived from a more primitive highly alkaline magma that mixed with a melt-rich rhyolitic body before eruption. However, the fiamme contain abundant low-temperature phases (sanidine and biotite) despite having a pre-eruptive temperature up to 100°C hotter than the rhyolite, and display REE patterns that are complementary to the rhyolitic host (positive Eu anomaly). In addition, there are no other eruptive deposits in the SRMVF with such high Ba and Zr contents and those found elsewhere (North Qiangtang Terrane, Paraná-Etendeka Province, and the Leucite Hills, WY) are from extremely high-K systems, which the CRT is not. In order to better define the link between these fiamme and the co-erupted rhyolite, we have focused on the mineral record of each end-member. Crystal contents are variable in the fiamme (~ 10-33 %) and the mineral assemblage is similar to the rhyolitic portion of the tuff (plagioclase, sanidine, biotite, with Fe-Ti oxides and traces of zircon, pyroxene and amphibole). Biotite and sanidine in the fiamme are characterized by highly variable Ba contents (from <1 to >4 wt.% BaO in biotite and <1 to 9 wt.% in sanidine). Zoning patterns in these two low temperature phases indicate early growth from an evolved melt followed by resorption and subsequent growth from a less evolved, high Ba melt. We propose that these fiamme represent fragments of a previously highly crystalline (> 50 vol %) cumulate residua at the base of an upper crustal reservoir, which was remobilized by heating and dissolution of a significant portion of the mineral content (leaving ~10-33 vol% crystals) following a mafic recharge. They are, therefore, co-genetic to the host rhyolite rather than being derived from an unrelated alkaline magma. An origin by crystal accumulation for these fiamme would explain the high Ba (sanidine, biotite), Zr (zircon) and Sr-Eu (plagioclase) contents in whole-rock analyses, while there is no concurrent elevation in other incompatible element concentrations (Rb and LREE). Late reheating by recharge and partial melting of the lowest temperature minerals (sanidine, low An plagioclase) would further account for (1) the high Ba rims in sanidine and biotite (produced by regrowth of these minerals from high Ba melt zones following sanidine dissolution), (2) the presence of mixing/mingling textures and trace amounts of high-P pyroxenes and amphiboles (typical for SRMVF mafic recharge), (3) highly variable phenocrysts contents ranging from ~10-33% and (4) that a crystal cumulate was able to erupt despite having likely reached high crystallinities (> 50 vol%).