Aluminum control on viscosity and structure of haplogranitic melts: Implications for rhyolitic melt viscosity determination

Accurate determination of melt viscosity near the glass transition temperature (Tg) is critical for modeling volcanic processes, but direct measurements are often compromised by nanostructuration in natural Fe–Ti-bearing systems, especially during experimental manipulation. Differential scanning calorimetry (DSC) offers an alternative method for estimating viscosity via “shift factors” (K), which link enthalpy to shear relaxation (and thus shear viscosity). This is possible because DSC analysis of supercooled melts requires significantly less time than micropenetration viscometry. However, the compositional sensitivity of the shift factors is still debated, particularly in highly polymerized melts.To address this, we investigated the role of Al2O3in controlling melt viscosity and network structure using five Fe–Ti-free haplogranitic compositions: the metaluminous HPG8 base melt and four systematically modified variants with nominal composition of +2, +5, −2 and −5 wt% Al2O3. We combine micropenetration viscometry, DSC, and Raman spectroscopy to examine the rheological and structural response to Al2O3variation.Our results reveal a non-linear viscosity dependence on Al2O3content: peraluminous melts exhibit higher viscosities and Tg, while peralkaline melts are significantly more fluid. Despite these differences, melt fragility remains constant across the compositional series.Calibrated DSC shift factors show no correlation with the network modifier content in peralkaline melts but instead they scale with the infinite-temperature viscosity (log10η∞). Peralkaline melts with log10η∞ > −3.00 show low and constant shift factors, whereas metaluminous and peraluminous melts (log10η∞ < −3.00) yield higher values. These findings establish benchmarks for the application of the DSC shift-factor approach to estimate melt viscosity in natural, silica-rich rhyolitic melts, especially where direct measurements are hindered, thereby improving our ability to model magma rheology and eruption dynamics.