I want to know whether the reflective spectrum from the sample should be considered? And what is the most important caution to consider in such calibration?
Time after time it is necessary to calibrate monochromator. Usually we used the well-known parameters of irradiation (laser beam, xenon lamp, Hg lamp etc). We measured spectra of irradiation of this source with the minimal size of slits and shift spectra to correct lines position.
This is the Standard Procedure, I mean, monochromator calibration in practice is done by acquiring a radiation spectrum from a calibrated source/lamp.
In absence of a calibrated lamp/source, monochromator calibration can also be done using the well-known emission peak of a standard/certified sample (for example Raman monochromators using diamond/Si-wafer Raman peak).
Literature advices the utilization of narrow-spectral peaks (transmission, scattering, reflection) from known standard/certified samples for monochromator calibration purpose, in absence of a calibrated lamp/source.
In terms of visible and infrared wavelength range, we'd like to choose flat response pyroelectric detector to do it. The most important is that the gratings changing would shift a little bit the curve. So the grating calibration is also important sometimes.