Long-term flow-through column experiments and their relevance to natural granitoid weathering rates
Four pairs of fresh and partly-weathered granitoids, obtained from well-characterized watersheds—Merced River, CA, USA; Panola, GA, USA; Loch Vale, CO, USA, and Rio Icacos, Puerto Rico—were reacted in columns under ambient laboratory conditions for 13.8 yrs, the longest running experimental weathering study to date. Low total column mass losses (<1 wt. %), correlated with the absence of pitting or surface roughening of primary silicate grains. BET surface area (S BET ) increased, primarily due to Fe-oxyhydroxide precipitation. Surface areas returned to within factors of 2 to 3 of their original values after dithionite extraction. Miscible displacement experiments indicated homogeneous plug flow with negligible immobile water, commonly cited for column experiments. Fresh granitoid effluent solute concentrations initially declined rapidly, followed by much slower decreases over the next decade. Weathered granitoid effluent concentrations increased modestly over the same time period, indicating losses of natural Fe-oxide and/or clay coatings and the increased exposure of primary mineral surfaces. Corresponding (fresh and weathered) elemental effluent concentrations trended toward convergence during the last decade of reaction. NETPATH/PHREEQC code simulations indicated non-stoichiometric dissolution involving Ca release from disseminated calcite and excess K release from interlayer biotite. Effluent 87 Sr/ 85 Sr ratios reflected a progressive weathering sequence beginning and ending with 87 Sr/ 85 Sr values of plagioclase with an additional calcite input and a radiogenic biotite excursion proportional to the granitoid ages. Effluents became thermodynamically saturated with goethite and gibbsite, slightly under-saturated with kaolinite and strongly under-saturated with plagioclase, consistent with kinetically-limited weathering in which solutes such as Na varied with column flow rates. Effluent Na concentrations showed no clear trend with time during the last decade of reaction (fresh granitoids) or increased slowly with time (weathered granitoids). Analysis of cumulative Na release indicated that plagioclase dissolution achieved steady state in 3 of the 4 fresh granitoids during the last decade of reaction. Surface-area normalized plagioclase dissolution rates exhibited a narrow range (0.95 to 1.26 10 -13 moles m -2 s -1 ), in spite of significant stoichiometric differences (An 0.21 to An 0.50 ). Rates were an order of magnitude slower than previously reported in shorter duration experiments but generally 2 to 3 orders of magnitude faster than corresponding natural analogs. CrunchFlow simulations indicated that more than a hundredfold decrease in column flow rates would be required to produce near-saturation reaction affinities that would start to slow plagioclase weathering to real-world levels. Extending simulations to approximate long term weathering in naturally weathered profiles required additional decreases in the intrinsic plagioclase dissolution and kaolinite precipitation rates and relatively large decreases in the fluid flow rate, implying that exposure to reactive mineral surfaces is significantly limited in the natural environment compared to column experiments.