Investigating ULX accretion flows and cyclotron resonance in NGC 300 ULX1

Methods. We analyzed broadband XMM-Newton and NuSTAR observations of NGC 300 ULX1, performing phase-averaged and phase-resolved spectroscopy. We compared two physically motivated models for the source spectrum: Non-thermal accretion column emission modeled by a power law with high-energy exponential roll-off (AC model) vs multicolor thermal emission from an optically thick accretion envelope plus a hard power-law tail (MCAE model). We combine the findings of our Bayesian analysis with qualitative physical considerations to evaluate the suitability of each model. Results. The low-energy part (<2 keV) of the source spectrum is dominated by non-pulsating, multicolor thermal emission. The (pulsating) high energy continuum is more ambiguous. If modelled with the AC model a residual structure is detected, that can be modeled using a broad Gaussian absorption line centered at ~12 keV. However, the same residuals can be successfully modeled using the MCAE model, without the need for the absorption-like feature. Model comparison, using the Bayesian approach strongly indicates that the MCAE model -- without the absorption line -- is the preferred model. Conclusions. All models considered strongly indicate the presence of an accretion disk truncated at a large distance from the central object, as has been recently suggested for a large fraction of both pulsating and non-pulsating ULXs. The hard, pulsed emission is not described by a smooth spectral continuum. If modelled by a broad Gaussian absorption line, the fit residuals can be interpreted as a cyclotron scattering feature (CRSF) compatible with a ~10^12 G magnetic field. However, the MCAE model describes the spectral and temporal characteristics of the source emission, without the need for an absorption feature and yields physically meaningful parameter values. Therefore strong doubts are cast on the presence of a CRSF in NGC 300 ULX1.