For background calculations the electron beam was started at 1 cm in front of the carbon target that simulates the diamond crystal. The input (electron momentum distribution) was taken from the beam transport calculations. Multiple scattering in the target was found to be the dominating process which determines the beam divergence. Initially, the background calculations assumed a configuration with a distance of 123 cm between the crystal and the center of the dipole magnet which separates the electron beam from the photons. The full beam transport without consideration of electron scattering favours this solution but the electron tracking by GEANT shows too much bremsstrahlung production at the entrance of the dipole magnet. A significant reduction of the bremsstrahlung background in the GEANT simulations was achieved by increasing the beam tube diameters from 40 to 63 mm and the gap of the magnet from 60 to 90 mm. This way, the bremsstrahlung could be reduced to the level of the unavoidable contribution from the target. Secondary neutron production from these components has been treated by multiplying the bremsstrahlung photon distribution with photo-neutron cross sections using the method described in [4]. In the optimized layout, the contamination of the X-ray dose delivered to the cell layer by neutrons is expected to be less than 1 %. The biological sample is assumed to consist of a monolayer of fibroblast cells (4 mm thick) placed on a 10 mm cellulose membrane and covered with a 8 mm thick medium layer.

Fig. 2 Dose distributions obtained by EGS 4 simulations for the described biological sample irradiated by bremsstrahlung (dots) and channeling radiation (thick line). The dotted and dashed lines are the corresponding calculations without the 70 mm Al absorber.

Fig. 2 shows the dose distribution in this sample produced by the photon beam after passing a 100 mm Be vacuum window calculated with the EGS 4 code [5]. We studied both tracking and energy deposition in such thin layers. A thickness greater than 1 mm was found to be necessary to perform dose calculations with EGS 4 in the photon energy range from 1.5 keV to 30 keV. The dose provided by the discrete channeling lines exceeds the bremsstrahlung background by at least one order of magnitude. But the integral dose of the bremsstrahlung is comparable to that of the channeling radiation. The introduction of an additional 70 mm Al absorber allows to reduce the low energy bremstrahlung contribution below 10 keV. It shows, however, that energy differential precision measurements of doses in biological samples require monochromatization of the photon beam, e.g. by reflection at a curved pyrolytic graphite crystal.