Future infrared space missions will undoubtedly employ passively cooled focal planes (T similar to 30K), as well as passively cooled telescopes. Most long-wave detector arrays (e.g. Si:As IBC) require cooling to temperatures of similar to 6 - 8K.(1) We have been working with Rockwell to produce 10 mum cutoff HgCdTe detector arrays that, at temperatures of similar to 30K, exhibit sufficiently low dark current and sufficiently high detective quantum efficiency to be interesting for astronomy. In pursuit of these goals, Rockwell Scientific Company has delivered twelve 256 x 256 arrays (several of them engineering arrays), with cutoff wavelengths at 30K between 7.4 and 11mum for characterization at Rochester. Seven of these arrays utilize advanced structure diodes with differing capacitances arranged in rows (banded arrays), and the materials properties of the HgCdTe also vary significantly from array to array. Of ultimate interest to astronomy is the fraction of pixels with dark current below the target value of similar to 100e(-)/s with 10-60mV of actual reverse bias across the diodes at T similar to 30K. These arrays were developed for the purpose of selecting diode architecture: we use this fraction as one criterion for selection. We have determined from these experiments the optimal diode architecture for future array development. Measurement of the dark current as a function of reverse bias and temperature allows us to ascertain the extent to which trap-to-band tunneling dominates the dark current at this temperature. We present the results for one representative array, UR008.