1 Introduction

Very young massive stars are still deeply embedded in their parental molecular cloud cores and can only be observed at infrared and radio wavelengths (see, e.g., Garay & Lizano 1999; Henning et al. 2000a; Kurtz et al. 2000). Compact luminous infrared sources, hot molecular cores, and ultracompact H II regions (UCH IIs) are tracers of on-going massive star formation. How these stars form is largely unknown, because few observations with high spatial resolution at infrared, sub-millimetre, and radio wavelengths are available. Such observations are required in order to resolve the often complex and distant regions of massive star formation into individual sources and to characterize their local environment. We have started a comprehensive attempt to close this observational gap by near-infrared adaptive optics (AO) observations supplemented with multi-wavelength data (Feldt et al. 1998, 1999).

To continue our study, we selected the two UCH IIs G11.11-0.40 and G341.21-0.21 which were observed with the ALFA and ADONIS AO systems, respectively. Both objects are associated with IRAS sources (G11.11-0.40 with IRAS18085-1931 and G341.21-0.21 with IRAS16487-4423) which have colours typical of ultracompact H II regions (Wood & Churchwell 1989). The CS(2-1) survey performed by Bronfman et al. (1996) demonstrated that the sources are located in dense molecular cloud cores.

The object G11.11-0.40 was identified as an ultracompact radio source (Kurtz et al. 1994; Becker et al. 1994). The source velocity fits the kinematics of the 3 kpc arm. Kurtz et al. adopted a distance of 5.2 kpc which gives an IRAS luminosity of $\sim$ $7.6\ensuremath{~ 10^{4}}$ (Walsh et al. 1997). G11.11-0.40 is also associated with high-velocity gas seen in the ¹²CO (J=1-0) line (Shepherd & Churchwell 1996).

The other object, G341.21-0.21, is associated with an H₂O maser detected at the position of the IRAS source (Filho & Escalise 1990), which is generally taken as evidence for massive star formation. In addition, an OH maser was detected by Caswell (1998). In this paper, we adopt a kinematic distance towards this source of 3.7 kpc. Note that this distance corresponds to the near solution of the kinematic model, whereas Forster & Caswell (1999) prefer the far solution of 15.3 kpc. That far distance would, however, yield a total luminosity of more than 10⁶ , a value that seemed unrealistically large to us. For our chosen distance, the luminosity resulting from the IRAS fluxes is $\sim 6.1\ensuremath{~ 10^{4}}$ .