In Parts I 1 and II 2 of this paper, it has been shown that both undoped and boron/phosphorus-doped germanosilicate thin films can be planarized over structures at significantly lower temperatures than other glasses, such as borophosphosilicate glass. However, variations in several film properties based on composition, such as water solubility, electrical leakage, and mechanical stress, must also be addressed. For instance, it is well known that germanium dioxide and germanosilicates rich in germanium oxide are water soluble. 3-6 Germanium dioxide is also reported to have high leakage currents when functioning as an insulator. 7-10 Mechanical stress is also an issue, since the thermal coefficient of expansion (TCE) of GeO 2 is greater than that of silicon, 11,12 while the TCE of SiO 2 is less than that of silicon. 13 Therefore, changes in the germanosilicate composition may have a profound effect on stress, or at least on the extrinsic (thermal expansion) component of stress.This paper examines the water solubility, electrical leakage, breakdown characteristics, and the mechanical stress of both the undoped and boron/phosphorus-doped germanosilicate glasses in both the as-deposited and the argon annealed states. Moreover, these properties are to be examined following the "composition changing" forming gas and steam anneals, as discussed in Parts I and II, to ascertain the benefits or disadvantages incurred by the forming gas, steam, or argon-steam two-step processes. ExperimentalSample preparation.-Germanosilicate films were deposited on single-crystal silicon substrates by the plasma-enhanced chemical vapor deposition (PECVD) method reported elsewhere. 14 Following deposition, reflow experiments were conducted in either argon, steam, forming gas, or a two-step process with argon and either steam or forming gas. These experiments were all conducted at atmospheric pressure. Temperatures for argon treatments ranged from 550 to 1050ЊC. Steam treatments were conducted with stoichiometric combinations of hydrogen and oxygen in a 2:1 ratio as well as with 10% excess oxygen. Excess oxygen was used to consume any excess hydrogen in the steam pyrolysis reaction. Steam treatments were limited to temperatures at and above 750ЊC to ensure complete combustion of the hydrogen/oxygen mixtures. Experiments were actually performed at both 750 and 800ЊC. Forming gas experiments were conducted with 3.4, 8.9, and 9.6% H 2 with the balance being N 2 . Various percentages of hydrogen were used to study the effects of different partial pressures of hydrogen on the films during the anneals. Forming gas reflow experiments were also performed at 750 and 800ЊC for direct comparison with the steam treatment results.Samples for water solubility tests were cleaved into 1 ϫ 1 cm dies prior to thermal treatments. Wafers for electrical testing were left whole in the undoped experiments, but they were cleaved into quarters prior to thermal treatment in the doped experiments. Stress measurement wafers were left intact, as required by the stress me...

Present isolation technology in microelectronic applications uses chemical mechanical polishing (CMP) in order to achieve planar surfaces for dielectric films used in trench isolation and interlayer dielectric structures, among others. Reflowed glasses have not played an important role in planarization because of the high temperatures required to achieve a planarized glass layer. 1 However, germanosilicate glasses (combinations of SiO 2 and GeO 2 ) with or without B 2 O 3 and/or P 2 O 5 dopants have shown promise for reduced reflow temperatures as compared to traditional borophosphosilicate glasses. [2][3][4][5] The purpose of the present work is to examine bulk glasses comprised of SiO 2 , GeO 2 , B 2 O 3 , and/or P 2 O 5 via differential thermal analysis (DTA).Glasses experience a substantial change in properties when passing through a specific temperature region, sometimes referred to as the "anomalous" region for the particular glass. The temperature at which the effects begin is dubbed the glass transition temperature, T g . 6,7 Several properties of a glass change nonlinearly near the glass transition temperature, including thermal coefficient of expansion (TCE), 7 specific heat, 7,8 electrical conductivity, elastic modulus, 7 and dielectric constant. 8 Perhaps the most important change at T g with respect to reflow is the reduction in the glass viscosity upon heating up to and beyond T g . 8 Since glass viscosity is reduced markedly at the glass transition temperature, reflow should occur at lower temperature for glasses that have lower glass transition temperatures.According to the literature, DTA can be used to determine glass transition temperatures. 6,[9][10][11][12][13] The present work utilizes DTA in order to examine the thermal behavior of glass formers, namely, SiO 2 , GeO 2 , B 2 O 3 , and P 2 O 5 , as well as several pseudo-binary, pseudoternary, and pseudo-quaternary mixtures of these materials. Such information may provide guidelines for choosing compositions and anneal temperatures when attempting reflow of doped or undoped germanosilicate glasses. O 5 as well as selected pseudo-binary, pseudo-ternary, and pseudo-quaternary compositions of these compounds have been examined for glass transitions by differential thermal analysis (DTA) in the context of reflow of doped germanosilicate glasses. SiO 2 does not exhibit a glass transition to temperatures above 1135ЊC. GeO 2 appears to exhibit a glass transition around 578ЊC, while B 2 O 3 appears to exhibit a glass transition in the range of 257-268ЊC. Although the glass transition temperature of P 2 O 5 could not be determined, the transition and melting behavior of the H, O, and OЈ phases have been reevaluated. Moreover, a new mechanism for conversion from H phase to O phase is presented. Namely, the melting of H phase followed by the spontaneous recrystallization of the resulting liquid to form the O phase was observed by DTA. Germanosilicate mixtures exhibited no glass transition, but the germanoborates' glass transition temperatures increased ...