A non-LTE analysis of C i lines at 1.068–1.069 \mu m was carried out for selected 46 halo/disk stars covering a wide metallicity range ( -3.7 \ltsim [ Fe/H ] \ltsim + 0.3 ) , based on the spectral data collected with IRCS+AO188 of the Subaru Telescope , in order to examine whether and how these strong neutral carbon lines of multiplet 1 can be exploited for establishing stellar carbon abundances , especially for very metal-poor stars where CH molecular lines have been commonly used .
These C i lines were confirmed to be clearly visible for all stars down to [ Fe/H ] \sim - 3.7 , from which C abundances could be successfully determined .
The resulting [ C/Fe ] vs. [ Fe/H ] diagram revealed almost the same trend established from previous studies .
When the results for individual stars are compared with the published data collected from various literature , while a reasonable agreement is seen as a whole , a tendency is observed that our abundances are appreciably higher than those from CH lines especially for very metal-poor giants of low gravity .
Since the abundances of these C i lines are subject to rather large non-LTE corrections ( typically by several tenths dex ) whose importance progressively grows as the metallicity is lowered , attention should be paid to how the collisional rates ( especially due to neutral hydrogen ) are treated in non-LTE calculations .