Ultralarge scale (ULSI) integrated circuits invariably require more than one level of interconnect metal. One of the main challenges in implementing such multilevel interconnect structures is planarization of the intermetal dielectric. Planarity is needed in multilevel interconnect technology because smooth surfaces are required to ensure good metal step coverage, 1 and to provide a flat enough field, within lithographic depth of focus capabilities, such that contact vias and metal wires can be patterned. 2,3 Failure to provide planar surfaces in the dielectric can result in electrical defects, electromigration problems, impairment of device reliability, and yield reduction.Reflow of dielectric films has been effective to a degree in reducing local variations in topography of multilevel interconnections. Phosphorus-doped silicon dioxide, or phosphosilicate glass (PSG), and phosphorus-and boron-doped silicon dioxide, or borophosphosilicate glass (BPSG), reflow processes have been commonly used to obtain insulating films with reasonably flat surfaces. 1,4,5 These films exhibit profiles over steps that get progressively smoother with higher phosphorus and boron concentration, reflecting the corresponding enhancement in viscous flow. However, decreasing thermal budgets and increasing via aspect ratios as well as the requirement of strict control of the dopant concentration have caused a transition to an alternative method for achieving planar surfaces, chemical mechanical polishing (CMP).CMP has been proven to achieve a measure of global planarization not previously possible with spin-on and resist etchback techniques. 6 However, CMP processes are hampered by pattern sensitivities, which cause regions on a chip to have thicker dielectric layers than other regions due to differences in the underlying topography. 7 Also, inconsistencies in reproducing polishing rates have also proven to be a problem. In fact, CMP is performed with empirical polishing rates and timed polishes. This results in a decrease in reliability and yield, and is costly.Conceptually, a glass that could be planarized within thermal budget constraints and still maintain good stress, electrical, and water solubility properties could eliminate the need for some, if not all, CMP steps, especially the CMP step for the first dielectric layer between the devices and the first-level metal, i.e., the poly-metal dielectric (PMD). Boron and/or phosphorus-doped SiO 2 films are unable to incorporate sufficient quantities of the dopant elements to reflow at a sufficiently low temperature before the glasses become unstable. In order to push the reflow temperature lower with the goal of planarizing reflow, an alternative chemical system, such as, GeO 2 -SiO 2 discussed below, is needed.There are several literature reports on the properties of the GeO 2 -SiO 2 system. [8][9][10]11 Only one of these 11 deals with GeO 2 glass films compositions greater than 25 mol %. The limited range of germanosilicate (GeO 2 ) compositions explored in most of these studies is not, ...