Astron. Astrophys. 325, L17-L20 (1997)

3. Results

A monotonic trend was observed in the MECS light curve (1.5-10 keV) with the intensity decreasing by during the pointing. No associated spectral variability was detected.

3.1. The continuum shape above 0.8 keV

We fitted a simple power-law model, absorbed by a neutral column density , separately to the spectra of the four detectors. The results are summarized in Table 1. The column density was constrained to the Galactic value (, Savage et al. 1993). Hereafter statistical uncertainties are 90% confidence level for one interesting parameter ().

Table 1. Power-law fits to single detectors spectra. is the photon spectral index. Column density has been held fixed to the Galactic value . N/ is the relative normalization at @ 1 keV from the simultaneous fit of all the instruments.

This simple parameterization is a fairly good representation of the spectral shape in the MECS, HPGSPC and PDS. Spectral indices are consistent with each other within the statistical uncertainties. In the LECS data, instead, a deviation from the simple power-law behaviour is present below keV, and is responsible for a rather high ; spectral behaviour in this band will be discussed in . The LECS spectral slope is flatter that the MECS one in the overlapping energy band (). This effect has been revealed in several sources observed by BeppoSAX   so far and is probably due to a combination of penetration in the driftless LECS gas cell (see Parmar et al. 1997) and obscuration by a window support structure rib. If the LECS data are fitted in the 0.8-4 keV energy range, the spectral index is consistent the MECS slope, within the statistical uncertainties (see Table 1).

We then fitted the power-law model simultaneously to the data of the four detectors (0.8-200 keV), restricting the LECS data to the 0.8-4 keV band. The resulting normalization factors are different in the four instruments: their values with respect to the MECS ( in Table 1) are consistent with those obtained in the calibration performed on the Crab nebula (Cusumano et al. 1997). The fit is acceptable, , (see Figure 1), the spectral photon index is and the unabsorbed flux in 2-10 keV is .

Fig. 1. Broadband 3C273 spectrum (upper panel) and the residuals (lower panel) when a simple absorbed () power-law model is applied in the 0.8-200 keV energy range.

3.2. Spectral features below 0.8 keV

As mentioned in the previous section, strong deviations from a simple power law model are present in the 3C273 spectrum below 0.8 keV. When a power law is fitted to the LECS data (0.12-4.0 keV), fixing the photon index to the average broadband slope () and the cold absorber to the Galactic value, the is unacceptable ( /d.o.f = 259/196). The inspection of the residuals clearly shows a photon deficit around E keV and an excess emission below 0.3 keV (Fig.2, left panel).

Fig. 2a and a. An absorption feature and a soft excess are clearly evident in the LECS residuals when a power law (, = ) is fitted the the data (left panel). Data are well parameterized by a broken power-law plus a notch (right panel)

We first parameterized the continuum with a broken power-law () and then added an absorption feature using an absorption edge or a rectangular trough profile (model NOTCH in XSPEC). We fixed the notch covering fraction equal to 1 since it was not possible to simultaneously determine meaningful constraints on its depth and width. The broken power law alone can not adequately reproduce the LECS data (see Table 2). Adding either an edge or a notch yields a statistically significant improvement (Fig.2, right panel) in both cases. Equally acceptable fits are obtained when a power law () or a blackbody ( eV) are used to describe the soft excess. Similar absorption best-fit parameters are obtained. In Table 2 the best-fit parameters are quoted for the broken power-law case (hereafter energies are quoted in the source rest frame).

Table 2. Best fit parameters of the soft emission when a broken power-law model + a feature in absorption are applied to the LECS data in the 0.1-4 keV band. was fixed to the best-fit value 0.30 keV to calculate the statistical uncertainties. , (fixed).

3.3. Iron Line Emission

In the MECS spectrum there is evidence of an emission line around . If a Gaussian profile is added to the power-law model in the MECS spectrum, the is reduced by , with best-fit (rest frame) parameters: keV and eV. In order to test the influence of minor miscalibration around the nearby Xenon edge (), we have divided the MECS 3C273 spectrum by the Crab spectrum and inspected the residuals around the line energy. The line structure is still present and a Gaussian fit yields an . The line can be naturally explained as fluorescence from neutral or mildly ionized iron, in agreement with the outcome by GINGA (Williams et al. 1992) and ASCA (Cappi & Matsuoka 1996).

© European Southern Observatory (ESO) 1997

Online publication: April 28, 1998

helpdesk.link@springer.de