Spectral energy distribution of the HIFI data (in grey). The data is unsmoothed, what looks like noise are actually individual spectral lines. The best fit towards Sgr B2(M) and N are represented by the solid orange line. The synthetic intensity maps have been convolved to the frequency dependent beam of the 3.5 m Herschel telescope. The fit includes dust and free-free emission. The surface, where the dust optical depth equals one is plotted in olive, the corresponding axis is shown on the right. The z-axis points towards the observer, Sgr B2(M) is located at z = 0 and Sgr B2(N) is located at z = 7 × 105 au. A tau surface datapoint at −22.5 pc indicates that the dust is optically thin at the corresponding frequency.