摘要： Raman scattering has become a powerful method in analytic spectroscopy and label-free imaging to study the chemical composition of a sample. Especially coherent Raman scattering (CRS) techniques such as stimulated Raman scattering (SRS) and coherent anti-Stokes Raman scattering (CARS) are developed further and show a great potential, due to their amplified signal in comparison to spontaneous Raman scattering. In CRS the interaction of light pulses with molecular vibrations are investigated, typically resulting in a characteristic molecular spectrum. However, especially CARS spectra are effected by the non-resonant contribution to the Raman scattering signal, resulting in a non-specific background and a distortion of the lineshape. SRS is free of a non-resonant background but is affected by two-photon absorption, cross-phase modulation (XPM) [1], and thermal scattering (temperature-induced refractive index change) [2]. The resulting artefacts in SRS become relevant for high pulses energies and ultra-short pulses. One example, where high pulse energies and ultra-short pulses are needed, is the field of spatial resolution enhancement in CRS. All techniques to overcome the diffraction limit by Abbe are based on depleting one of the scattering resources, as for example the pump or probe photons [3]. High pulse energies in the order of up to 200-300 nJ are needed, leading to also unwanted spectral distortions through XPM [4]. In this paper, we present the avoidance of spectral XPM artefacts in femtosecond stimulated Raman scattering, based on a systematic study varying the pump pulse duration in the high pulse energy regime. To study the effects of different pump pulse durations on the FSRS spectra a setup as shown in Fig. 1 (a) was realized, which consists of a folded Fourier filter, to freely choose the pump pulse duration, and a supercontinuum generation in a gadolinium vanadate crystal (GdVO4) working as broadband probe pulse source. After combining the two beams at the dichroic mirror (DM) measurements with different peak powers of the pump pulse, time delays between pump and probe and varying pump pulse durations were performed. For pump pulses with a duration between 2 ps and 3 ps the Raman spectra are nearly unaffected by XPM, whereas pulses shorter than 1 ps show a significant XPM induced distortion of the stimulated Raman lineshape and a reduction of the Raman resonance maximum, becoming visible as a negative stimulated Raman loss (SRL) signal on the high wavenumber side of the Raman resonances. Thus, it is favourable to choose pump pulses with a longer pulse duration in the regime of high peak power to reduce the impact of cross-phase modulation in stimulated Raman scattering.