The nucleation and growth mechanisms during high temperature oxidation of liquid Al−3% Mg and Al−3% Mg−3% Si alloys were studied with the aim of enhancing our understanding of a new composite fabrication process. The typical oxidation sequence consists of an initial event of rapid but brief oxidation, followed by an incubation period of limited oxide growth after which bulk Al2O3/Al composite forms. A duplex oxide layer, MgO (upper) and MgAl2O4 (lower), forms on the alloy surface during initial oxidation and incubation. The spinel layer remains next to the liquid alloy during bulk oxide growth and is the eventual repository for most of the magnesium in the original alloy. Metal microchannels developed during incubation continuously supply alloy through the composite to the reaction interface. During the growth process, a layered structure exists at the upper extremity of the composite, consisting of MgO at the top surface, MgAl2O4 (probably discontinuous), Al alloy, and finally the bulk Al2O3 composite containing microchannels of the alloy. The bulk oxide growth mechanism appears to involve continuous formation and dissolution of the Mg-rich oxides at the surface, diffusion of oxygen through the underlying liquid metal, and epitaxial growth of Al2O3 on the existing composite body. The roles of Mg and Si in the composite growth process are discussed.

The oxidation of liquid Al–Mg–Si alloys at 900–1400 °C was studied by thermogravimetric analysis (TGA). The development of a semi-protective surface layer of MgO/MgAl2O4 allows the continuous formation of an Al2O3-matrix composite containing an interpenetrating network of metal microchannels at 1000–1350 °C. An initial incubation period precedes bulk oxidation, wherein Al2O3 grows from a near-surface alloy layer by reaction of oxygen supplied by the dissolution of the surface oxides and Al supplied from a bulk alloy reservoir through the microchannel network. The typical oxidation rate during bulk growth displays an initial acceleration followed by a parabolic deceleration in a regime apparently limited by Al transport to the near-surface layer. Both regimes may be influenced by the Si content in this layer, which rises due to preferential Al and Mg oxidation. The growth rates increase with temperature to a maximum at ∼1300 °C, with a nominal activation energy of 270 kJ/mole for an Al−2.85 wt. % Mg−5.4 wt. % Si alloy in O2 at furnace temperatures of 1000–1300 °C. An oscillatory rate regime observed at 1000–1075 °C resulted in a banded structure of varying Al2O3-to-metal volume fraction.

No abstract

Wafer warpage is common in microelectronics processing. Warped wafers can affect device performance, reliability and linewidth control in various processing steps. We proposed in this paper an in-situ approach for estimating wafer warpage profile during the thermal processing steps in microlithography process. A programmable multizone thermal processing system is developed to demonstrate the approach. Early detection will minimize cost and processing time. Based on first principle thermal modeling and system identification techniques, we are able to estimate the profile of a warped wafer from available temperature measurements. Experimental results demonstrate the feasibility and repeatability of the approach. The proposed approach is applicable to other semiconductor substrates.

Wafer warpage can affect device performance, reliability and linewidth control in various processing steps in microelectronics manufacturing. Early detection will minimize cost and processing time. We have previously demonstrated an on-line approach for detecting wafer warpage and the profile of the warped wafer. The proposed approach demonstrates that the profile of the wafer can be computed during thermal processing steps in the lithography sequence. However, the approach is computationally intensive and information is made available at the end of the thermal processing step. Any attempts at real-time correction of the wafer temperature are thus not possible. In this paper, we proposed an in situ approach to detect wafer warpage and its profile midway through the thermal process. Based on first principles thermal modelling, we are able to detect and estimate the profile of a warped wafer from available temperature measurements. The proposed approach can be implemented on conventional thermal processing systems. Experimental results demonstrate the feasibility and repeatability of the approach. A 75% improvement in computational time is achieved with the proposed approach.