Although graphene oxide has shown encouraging results in membrane field, it has a major stability issue when coated on ceramic supports, particularly those formed of geomaterials. This work aims to prepare stable graphene oxide layer deposited on a low-cost rich-silica support. The graphene oxide membrane was prepared using grafting method, which included surface activation and modification followed by the graphene oxide layer coating. Siloxane functional groups were activated using either hydrochloric acid or piranha solution to determine which of them provided the highest membrane stability. The support surface was chemically modified using 3-glycidoxypropyltrimethoxysilan followed by the deposition of graphene oxide nanosheets using evaporation-assisted assembly. According to Fourier-transform infrared spectroscopy, the support activated with piranha solution has more silanol functional groups than the support activated with hydrochloric acid while the appearance of 3-glycidoxypropyltrimethoxysilan band on the modified support, proves the accomplishment of grafting. The stability test of the membrane demonstrated a highly stable deposited layer for the graphene oxide membrane activated with piranha solution in different pH contrary to the non-grafted graphene oxide membrane and the graphene oxide membrane activated with hydrochloric acid that peeled off from the support. On the support activated with piranha solution, 3-glycidoxypropyltrimethoxysilan concentration was investigated and an optimal value of 0.1 mol L−1 was determined. The filtration performance of the membrane was assessed by the removal of anionic and cationic dyes. For anionic dyes, the membrane rejection reached a value of 90 % for direct red 80 and 87 % for methyl orange. For the rhodamine B considered as cationic dye, the membrane rejection reached a value of 81 %. The membrane may be considered sustainable since it is inexpensive and emits at least 9.76 kg CO2eq m−2 less than other ceramic membranes during its preparation.