We present radio and infrared (3.6-24um) counterparts to submillimetre galaxies (SMGs) detected in the Extended Chandra Deep Field South with the LABOCA 870-um bolometer camera on the 12-m Atacama Pathfinder Experiment. Using the Very Large Array at
1.4GHz and Spitzer we have identified secure counterparts to 79 of the 126 SMGs (SNR>3.7, S_870>4.4mJy) in the field, 62 via their radio and/or 24-um emission, the remainder using a colour-flux cut on IRAC 3.6- and 5.8-um sources chosen to maximise the number of secure, coincident radio and 24-um counterparts. In constructing our radio catalogue, we have corrected for the effects of `flux boosting, then used the corrected flux densities to estimate the redshifts of the SMGs based on the radio/submm spectral indices. The effect of the boosting correction is to increase the median redshift by 0.2 resulting in a value of z=2.2 (+0.7-0.8) (1-sigma errors) for the secure radio counterparts, in agreement with other studies, both spectroscopic and photometric.
We have produced sensitive, high-resolution radio maps of 12 SMGs in the Lockman Hole using combined MERLIN and VLA data at a frequency of 1.4 GHz. Integrating for 350hr yielded an r.m.s. noise of 6.0 uJy/beam and a resolution of 0.2-0.5. For the fir
st time, wide-field data from the two arrays have been combined in the (u,v) plane and the bandwidth smearing response of the VLA data has been removed. All of the SMGs are detected in our maps as well as sources comprising a non-submm luminous control sample. We find evidence that SMGs are more extended than the general uJy radio population and that therefore, unlike in local ULIRGs, the starburst component of the radio emission is extended and not confined to the galactic nucleus. For the eight sources with redshifts we measure linear sizes between 1 and 8 kpc with a median of 5 kpc. Therefore, they are in general larger than local ULIRGs which may support an early-stage merger scenario for the starburst trigger. X-rays betray AGN in six of the 33 sources in the combined sample. All but one of these are in the control sample, suggesting a lower incidence of AGN amongst the submm-luminous galaxies which is, in turn, consistent with increased X-ray absorption in these dust-obscured starbursts. Only one of our sources is resolved into multiple, distinct components with our high-resolution data. Finally, compared to a previous study of faint radio sources in the GOODS-N field we find systematically smaller source sizes and no evidence for a tail extending to ~4. Possible reasons for this are discussed.
