In this study we present quantitative X-ray computed microtomography measurements (μCT) of retained sub-micron-sized particles in open porous media carried out in a laboratory μCT setup. Due to the polychromatic spectrum of the used X-rays, the tomograms are affected by various non-linear artifacts, which belong to the class of beam hardening artifacts. These artifacts become more dominant, when the amount of retained particles increases and can affect wide areas of the images, making a qualitative and quantitative analysis barely possible. Furthermore, the colloidal depositions show an inhomogeneous distribution inside the filter, making a reliable material discrimination between filter and particle material challenging. We introduce a calibration method, which is capable to sufficiently remove the majority of the artifacts by linearization of the projection data and thus enabling the precise material quantification of the retained colloids in the reconstructed tomograms. While most beam hardening correction routines are only applicable to homogeneous materials, our algorithm takes into account inhomogeneous material distributions and is adapted to multi-material systems. Moreover, the method includes a material discrimination of the colloids and the filter in the raw data domain. Thus, erroneous segmentations at the interfaces between different material fractions are avoided. As a result we present quantitative concentration maps of the particle distribution inside the porous media with a resolution of < 10μm. A series of validation samples was prepared, covering a wide range of different, representative filter loading stages. The accuracy of the particle quantification was evaluated from these samples and the relative deviation of the overall contained particle mass was less than 10% in all cases, partially even less than 1%. The overall image quality due to the artifact removal was significantly improved. The local variation of the particle concentration could be well assessed from the obtained concentration maps.