Context. Though Arp 220 is the closest and by far the most studied ULIRG, a discussion is still ongoing on the main power source driving its huge infrared luminosity. Aims. To study the molecular composition of Arp 220 in order to find chemical fingerprints associated with the main heating mechanisms within its nuclear region. Methods. We present the first aperture synthesis unbiased spectral line survey toward an extragalactic object. The survey covered the 40 GHz frequency range between 202 and 242 GHz of the 1.3 mm atmospheric window. Results. We find that 80% of the observed band shows molecular emission, with 73 features identified from 15 molecular species and 6 isotopologues. The 13 C isotopic substitutions of HC 3 N and transitions from H 18 2 O, 29 SiO, and CH 2 CO are detected for the first time outside the Galaxy. No hydrogen recombination lines have been detected in the 40 GHz window covered. The emission feature at the transition frequency of H31α line is identified to be an HC 3 N molecular line, challenging the previous detections reported at this frequency. Within the broad observed band, we estimate that 28% of the total measured flux is due to the molecular line contribution, with CO only contributing 9% to the overall flux. We present maps of the CO emission at a resolution of 2.9 × 1.9 which, though not enough to resolve the two nuclei, recover all the single-dish flux. The 40 GHz spectral scan has been modelled assuming LTE conditions and abundances are derived for all identified species. Conclusions. The chemical composition of Arp 220 shows no clear evidence of an AGN impact on the molecular emission but seems indicative of a purely starburst-heated ISM. The overabundance of H 2 S and the low isotopic ratios observed suggest a chemically enriched environment by consecutive bursts of star formation, with an ongoing burst at an early evolutionary stage. The large abundance of water (∼10 −5 ), derived from the isotopologue H 18 2 O, as well as the vibrationally excited emission from HC 3 N and CH 3 CN are claimed to be evidence of massive star forming regions within Arp 220. Moreover, the observations put strong constraints on the compactness of the starburst event in Arp 220. We estimate that such emission would require ∼2−8 × 10 6 hot cores, similar to those found in the Sgr B2 region in the Galactic center, concentrated within the central 700 pc of Arp 220.