2 Observations

The majority of the observations presented in this paper were obtained in an observing run at the 1.23 m telescope in Calar Alto, using the near-infrared camera MAGIC, between 25 August and 7 September 2001. A first, small subset of observations was also obtained during a summer school held at Calar Alto in July 2000 (http://www.iap.fr/eas/neon.html), using identical instrumentation and setup. We used a grism covering both the K and most of the H bands in a single setup, in combination with a slit yielding a width of 2 pixels on the detector. This setup allowed us to obtain spectra covering the 1.50-2.40 $\mu$m range at a resolution of $R \simeq 240$. Besides the spectra of the Cygnus OB2 targets we also obtained spectra of the G5IV star HD 190771 at approximately one-hour intervals. This star, assumed to be essentially featureless at the resolution employed, lies near Cygnus OB2 in the sky and was used to monitor telluric absorption. For each source we obtained six spectra at positions evenly spaced along the slit, each spectrum being a stack of individual 6 s exposures. The number of individual exposures at each position varied between 10 and 20, depending on the cataloged Kmagnitude of the target. Each frame was dark-subtracted and flat-fielded, and individual spectra were extracted and wavelength-calibrated using OH airglow lines (Oliva & Origlia 1992). The individual, wavelength-calibrated spectra were then combined separately for each program object and for the telluric reference star, and the spectrum of each program star was ratioed by that of the telluric reference star obtained closest in time. To achieve relative flux calibration the resulting spectrum was finally multiplied by a blackbody of 5700 K temperature, which should adequately represent the spectral energy distribution of the telluric reference star in the near infrared.

A few targets of special interest because of their spectra or their association with sources at longer wavelengths were imaged through the broad-band JHK_M filters with the same telescope and instrument. In imaging mode, the detector yields a field of view of $4'9 \times 4'9$ at a scale of 1$\arcsec$15 per pixel. Individual images of the field around each of these stars were obtained using a dither pattern of 9 points on a $3 \times 3$ points grid, adding up 12 integrations of 5 s each at every grid position. A sky frame was constructed for each field and filter by median-filtering its component frames uncorrected for the motions of the telescope between consecutive grid points, so as to remove stellar images. The sky frame was subtracted from each image in the dither, and the resulting dark-subtracted, flat-fielded, and sky-subtracted images were then aligned and coadded.

Figure 1: (H-K), K diagram of all the 2MASS sources detected within 1 degree of the adopted center of Cygnus OB2. The nearly vertical line on the left is the position of the unreddened main sequence at the distance of 1.7 kpc, and the dashed lines indicate the locus expected for stars of different spectral types at that same distance reddened by different amounts.

Figure 2: (H-K), (J-H) diagram of all the 2MASS sources detected within 1 degree of the adopted center of Cygnus OB2 and above the limiting line for spectral type B0 in Fig. 1. The solid lines at the lower left of the figures mark the expected positions for unreddened main sequence (lower branch) and giant (upper branch) stars.