Context. The [C ii] 158 µm far-infrared (FIR) fine-structure line is one of the most important cooling lines of the star-forming interstellar medium (ISM). It is used as a tracer of star formation efficiency in external galaxies and to study feedback effects in parental clouds. High spectral resolution observations have shown complex structures in the line profiles of the [C ii] emission. Aims. Our aim is to determine whether the complex profiles observed in [ 12 C ii] are due to individual velocity components along the line-of-sight or to self-absorption based on a comparison of the [ 12 C ii] and isotopic [ 13 C ii] line profiles. Methods. Deep integrations with the SOFIA/upGREAT 7-pixel array receiver in the sources of M43, Horsehead PDR, Monoceros R2, and M17 SW allow for the detection of optically thin [ 13 C ii] emission lines, along with the [ 12 C ii] emission lines, with a high signalto-noise ratio (S/N). We first derived the [ 12 C ii] optical depth and the [C ii] column density from a single component model. However, the complex line profiles observed require a double layer model with an emitting background and an absorbing foreground. A multicomponent velocity fit allows us to derive the physical conditions of the [C ii] gas: column density and excitation temperature. Results. We find moderate to high [ 12 C ii] optical depths in all four sources and self-absorption of [ 12 C ii] in Mon R2 and M17 SW. The high column density of the warm background emission corresponds to an equivalent A v of up to 41 mag. The foreground absorption requires substantial column densities of cold and dense [C ii] gas, with an equivalent A v ranging up to about 13 mag. Conclusions. The column density of the warm background material requires multiple photon-dominated region (PDR) surfaces stacked along the line of sight and in velocity. The substantial column density of dense and cold foreground [C ii] gas detected in absorption cannot be explained with any known scenario and we can only speculate on its origins. Key words. ISM:clouds -ISM:individual objects: M43 -ISM:individual objects: M17 -photon-dominated region (PDR) -ISM:individual objects: Horsehead -ISM:individual objects: MonR2 1 At that time, the spectroscopic data were less accurate and the wavelength of the transition was assumed to fall at 157 µm instead of 158 µm.Article number, page 1 of 40 A&A proofs: manuscript no. 34380corr_2 mentum change F=2→1, F=1→0, and F=1→1. The frequencies of the fine structure transitions of both isotopes were determined by Cooksy et al. (1986). The astronomical observations are fully consistent with these frequencies, as was discussed by Ossenkopf et al. (2013), who also noted that the relative strengths of the [ 13 C ii] hyperfine satellites (s F→F , see Table 1) given by Cooksy et al. (1986) are incorrect. We summarize all the relevant [ 12 C ii] and [ 13 C ii] spectroscopic parameters in Table 1, including the velocity offsets of the [ 13 C ii] hyperfine components relative to [ 12 C ii]. The frequency separation of the hyperfi...