egs_kerma - dose scored for "closed" geometry

[Cukhen]

Hi all,

I've created a simple geometry for egs_kerma

:start geometry definition:
    :start geometry:
        library = egs_cones
        type = EGS_ConeStack
        name = probe
        axis = 0 0 -0.1250025 0 0 1
        :start layer:
                thickness = 0.0000025
                top radii = 0.5 1
                bottom radii = 0.5 1
                media = mylar PEEK
        :stop layer:
        :start layer:
                thickness = 0.25
                top radii = 0.5 1
                bottom radii = 0.5 1
                media = blood_icrp PEEK
        :stop layer:
        :start layer:
                thickness = 0.0000025
                top radii = 0.5 1
                bottom radii = 0.5 1
                media = mylar PEEK
            :stop layer:
        :stop geometry:

    :start geometry:
        library = egs_cones
        type = EGS_ConeStack
        name = surrounding_air
        axis = 0 0 -50 0 0 1
        :start layer:
            thickness = 100
            top radii = 100
            bottom radii = 100
            media = air_dry_nearsealevel
        :stop layer:
     :stop geometry:

    :start geometry:
        library  = egs_gunion
        name = experiment
        geometries = probe surrounding_air 
    :stop geometry:

    simulation geometry = experiment
:stop geometry definition:

When I exclude all of the regions surrounding the scoring volume 2 (excluded regions = 0 1 3 4 5 6 ) from dose scoring, I still get Kerma values > 0 for low energy x-rays ~15 keV. It's arround 5 % of the Kerma without excluding this regions.
Any idea where this contributions are coming from?