Official URL: https://doi.org/10.1051/0004-6361/202245633

Context. The Medium Resolution Spectrometer (MRS) is one of the four observing modes of JWST/MIRI. Using JWST in-flight data of unresolved (point) sources, we can derive the MRS absolute spectral response function (ASRF) starting from raw data. Spectral fringing, caused by coherent reflections inside the detector arrays, plays a critical role in the derivation and interpretation of the MRS ASRF. The fringe corrections implemented in the current pipeline are not optimal for non-extended sources, and a high density of molecular features particularly inhibits an accurate correction. Aims. In this paper, we present an alternative way to calibrate the MIRI/MRS data. Firstly, we derive a fringe correction that accounts for the dependence of the fringe properties on the MIRI/MRS pupil illumination and detector pixel sampling of the point spread function. Secondly, we derive the MRS ASRF using an absolute flux calibrator observed across the full 5–28 µm wavelength range of the MRS. Thirdly, we apply the new ASRF to the spectrum of a G dwarf and compare it with the output of the JWST/MIRI default data reduction pipeline. Finally, we examine the impact of the different fringe corrections on the detectability of molecular features in the G dwarf and K giant. Methods. The absolute flux calibrator HD 163466 (A-star) was used to derive tailored point source fringe flats at each of the default dither locations of the MRS. The fringe-corrected point source integrated spectrum of HD 163466 was used to derive the MRS ASRF using a theoretical model for the stellar continuum. A cross-correlation was run to quantify the uncertainty on the detection of CO, SiO, and OH in the K giant and CO in the G dwarf for different fringe corrections. Results. The point-source-tailored fringe correction and ASRF are found to perform at the same level as the current corrections, beating down the fringe contrast to the sub-percent level in the G dwarf in the longer wavelengths, whilst mitigating the alteration of real molecular features. The same tailored solutions can be applied to other MRS unresolved targets. Target acquisition is required to ensure the pointing is accurate enough to apply this method. A pointing repeatability issue in the MRS limits the effectiveness of the tailored fringe flats is at short wavelengths. Finally, resulting spectra require no scaling to make the sub-bands match, and a dichroic spectral leak at 12.2 µm is removed.