The byproducts in residue layers produced after etching doped oxide films with anhydrous hydrogen fluoride (HF) in a commercial gas phase wafer processing tool operated at atmospheric pressure and 55 °C have been characterized using transmission Fourier transform infrared (FTIR) spectroscopy. Approximately 4000 Å was etched from the borophosphosilicate glass (BPSG), borosilicate glass (BSG), and phosphosilicate glass (PSG) films, creating a condensed layer on the oxide surfaces during etching. The primary etching products in the condensed layer on BPSG were found to be a mixture of boric acid B(OH)3, phosphoric acid H3PO4, and water. The etching products in the residue on PSG were H3PO4 and water. The reaction of boric oxide (B2O3) crystallites with water to produce B(OH)3 is thermodynamically favorable and should react further to boron trifluoride BF3 in the presence of HF. The formation of B(OH)3 rather than BF3 in the etching residue on BPSG indicates that a kinetic limitation exists due to either the relatively low HF exposure used or the chemistry within the mixed acid film. The etching product in the residue on BSG, which was exposed to a higher HF flux, was primarily boron trifluoride dihydrate BF3⋅2H2O. The condensed layer supports the etching of the Si–O matrix by concentrating the HF and water reactants close to the surface, which explains the enhancement in the etching rate when a condensed layer is formed. The etching products identified by FTIR could also play a direct role in the etching reaction since hydroxyl groups on the acids can activate Si–O bonds similar to water and the Lewis acid BF3 can attach to Si–O via the pair of electrons on the O atom.

An integrated reactor system was built for studying gas phase surface preparation chemistries. The system integrates HF/vapor and UV photochemistry modules with an ultrahigh vacuum deposition reactor and a surface analysis chamber (x-ray photoelectron spectroscopy and Auger) for in situ surface preparation, deposition, and analysis. Each vacuum chamber is mounted on a separate, isolated branch from a main sample transfer tube. The system was designed for samples with variable shapes and thickness, but less than 64mm (212in.) in diameter. This design allows for rapid transfer times between chambers (<5min) and for the simultaneous processing and storage of up to four samples. Use of standard sample transfer and vacuum hardware components minimized initial equipment costs and system maintenance. The capabilities of the clustered reactor apparatus and the importance of surface termination were demonstrated by (1) the removal of a mixed oxide and fluorocarbon residue on silicon, leaving the surface completely terminated with Cl atoms, (2) the removal of copper oxide and copper metal from silicon, (3) the deposition of Ti preferentially on a nonannealed, aqueous-cleaned SiO2 surface relative to an annealed surface, and (4) the use of complementary surface analysis techniques to chemically identify hydrogen-bonded silanol groups on a silicon surface after HF/vapor etching. Gas phase cleaning and surface termination utilized a combination of HF/vapor (100Torr, 27°C for 200s) and UV∕Cl2 (10SCCM Cl2, 90°C for 15min) steps. The results demonstrate that integrated processing provides a means to clean thin layers of organic, oxide, and metal contaminants from semiconductor surfaces and to control the terminating atom or chemical group.

No abstract

No abstract

No abstract