부산대학교 도서관

The maturing of rapid single flux quantum (RSFQ) circuits into a VLSI complexity technology has focused the need for advanced design and analysis capabilities. For example, the temperature dependence of RSFQ circuits has created a need for a thermal analysis methodology, including an accurate thermal model and related partitioning algorithm. The Josephson critical current density and the superconductive properties of the interconnects become compromised at elevated temperatures. Heating of Josephson junctions (JJs) and niobium interconnects results in reduced margins and/or functional failure. A methodology for evaluating the thermal properties of RSFQ integrated circuits, targeting large-scale systems, is presented here. This methodology comprises a thermal model and a multistage partitioning algorithm. The algorithm, based on a layout of the IC, partitions the circuit into blocks. An average thermal model is applied to the partitioned structure, producing a netlist for thermal simulation. The hot spots are also determined with a threshold temperature indicating functional failure. A peak thermal model is presented to detect hotspots based on the threshold temperature. The model is evaluated at several block complexities and validated using a numerical solver. An error of less than 1% between the model and numerical simulations is achieved. The algorithm is applied to the AMD2901 benchmark circuit composed of more than 300000 heating elements. The thermal profile for AMD2901 is generated in under 68 min.