While a relationship between ballast and carbon in sedimenting particles has been well-documented, the mechanistic basis of this interaction is still under debate. One hypothesis is that mineral ballast protects sinking organic matter from degradation. To test this idea, we undertook a laboratory experiment using the diatom Skeletonema marinoi to study in parallel the dissolution of one of the most common mineral ballasts, biogenic silica (bSiO2), and the associated degradation of organic matter. Three different models were applied to our results to help elucidate the mechanisms driving bSiO2 dissolution and organic compound degradation. Results of this modeling exercise suggest that the diatom frustule is made up of two bSiO2 phases that dissolve simultaneously, but at different rates. In our experiments, the first phase was more soluble (View the MathML source) and made up 31% of the total bSiO2. In this phase, bSiO2 was mainly associated with membrane lipids and the amino acids glutamic acid, tyrosine, and leucine. The second phase was more refractory (View the MathML source), and contained more neutral lipid alcohols and glycine. Until it dissolved, the first bSiO2 phase effectively protected much of the organic matter from degradation: particulate organic carbon (POC) degradation rate constants increased from 0.025 to 0.082 d−1 after the total dissolution of this phase, and particulate organic nitrogen (PON) degradation rate constants increased from 0.030 to 0.094 d−1. Similar to POC and PON, the total hydrolyzable amino acids (THAA) degradation rate constant increased from 0.054 to 0.139 d−1 after dissolution of the first bSiO2 phase. The higher THAA degradation rate constant is attributed to a pool of amino acids that was produced during silicification and enclosed between the two silica phases. This pool of amino acids might come from the incorporation of silica deposition vesicles into the diatom wall and might not be directly associated with bSiO2. In contrast, most lipid degradation was not prevented by association with the more soluble bSiO2 phase, as the average lipid degradation rate constant decreased from 0.048 to 0.010 d−1 after 17 d of degradation. This suggests that most lipids were associated with rather than protected by silica, except pigments that appeared resistant to degradation, independently from silica dissolution. When the only organic compounds remaining were associated with the second bSiO2 phase, degradation rate constants decreased greatly; concentrations changed only slightly after day 25.