Oxidation Resistance of Ultrathin Silicon Nitride Passivation Layers on Si(100)

  • PDF / 416,407 Bytes
  • 6 Pages / 414.72 x 648 pts Page_size
  • 2 Downloads / 167 Views



(>40A) thermally grown Si3N4 has also been examined [7]. These films are well known to be excellent diffusion barriers under severe oxidation conditions. It is generally observed that initial surface (- tens of A) oxidation of the Si3N4 is very rapid, after which it slows down considerably with the formation of an amorphous silicon oxynitride with a graded N content, increasing toward the substrate-dielectric interface [6]. EXPERIMENT An Integrated CVD-Surface Analysis System (described elsewhere [8]) was used for these experiments. This allowed non-destructive study of the incremental effects of oxidation using XRay Photoelectron Spectroscopy (XPS). The sample preparation and experimental procedure has also been described previously [8]. HF- dipped (H-terminated), Si(100) samples (n type, 9-13 K2cm) were nitrided in 4 Torr NH3 (Scott, 99.999%) between 560-950'C for 10-120s resulting in Si3N4 passivation layers 3-17A thick. Thickness measurements were made by X-Ray Photoelectron Spectroscopy (XPS), using the ratio of the Si 2 psubstrate to the Si 2 poverlayer peak area in the following equation. Parameters were obtained using Si02 and Si3N4 films of known thickness. I sifilm n sifilm •Si2p, film 1-exp(- d/ý,s i, film COS ) ISi,substrate nSi, substrate Si2p,substrate exp(-d1si,)filmC°SO)

where, d= film thickness 0= angle between analyzer axis/sample normal. nSi= Si atomic concentration k--photoelectron mean free path Isi= Si peak intensity from substrate/overlayer These films were incrementally oxidized (typically in 3-5 steps) between 450-800'C in 4 Torr of 02 (Scott, 99.999%) for a total time of 180-360s. After each oxidation step, the film was analyzed using XPS, enabling an accurate determination of the evolution of the bonding structure as a function of process conditions. The definition of oxidation resistance is subjective to some extent, but is important from a point of view of establishing an experimental baseline. It is well known that the separation between the Si 2psubstrate and Si2 poverlayer peaks is very sensitive to the relative amount of 0 in the film. 0 being more electronegative than N, introduces Si oxidation states (Sin+) with higher binding energy, resulting in an increase of the initial Si2p (Si3N4) peak separation (2.4-2.5 eV). A shift of roughly 0.9-1.0 eV /Si-O bond formed is produced, as compared to -0.6 eV / Si-N bond. We define oxidation resistance as no increase in the Si2p peak separation beyond 2.7eV. Additionally, we focus on an oxidation time of 180s as this is adequate to form -60-120A of Ta205 under our RTP conditions (to be reported in a separate paper). This corresponds to an equivalent thickness of 10-20A of SiO2, not including the thickness of the passivation layer. RESULTS AND DISCUSSION We begin with a detailed examination of the oxidation of a 5A Si3N4 film, grown in NH3 at 650'C for 60s. Figure 1 shows a set of Si2p spectra of the film at various intervals of oxidation at 450'C in 4 Torr of 02. The spectra are virtually identical, implying that there is no measurable thi