Osg Process
The OSG process swaps fluorine for carbon and replaces FSG's dense PECVD oxide with an engineered porous film — Applied Materials' Black Diamond family (and Novellus/Lam's CORAL line) deposits an organosilicate glass matrix, with methylsilane or TMCTS precursor chemistry carrying methyl groups into the Si-O network, then uses a UV-cure step to burn out a sacrificial porogen and leave nanoscale voids behind, pushing dielectric constant down to roughly 2.5-2.7 — a full point below FSG's 3.5-3.7 ceiling and the successor that finally let copper BEOL hold RC delay in check once pitches needed k well under 3.0.
The chemistry is a straight substitution of carbon for fluorine, but the k reduction mechanism is almost entirely different. FSG lowers dielectric constant by swapping in a more electronegative atom that polarizes less under field; OSG instead builds a methylsilane or trimethylsilane backbone whose methyl (CH3) groups already drop k into the 2.8-3.0 range just from reduced network density and polarizability, and then goes further by co-depositing a sacrificial organic porogen alongside the matrix and burning it out with a UV cure, leaving behind a sponge of nanoscale air gaps whose near-1.0 dielectric constant pulls the bulk film down to 2.5-2.7 across the Black Diamond II and III generations. Air is the best dielectric available, so every percent of porosity buys a real k reduction — the catch is that air also carries zero mechanical load, so the same porosity that helps k directly erodes elastic modulus, cracking resistance, and adhesion, the three properties CMP and packaging stress care about most.
| Film | k | Porosity | Modulus (GPa) | Moisture risk |
|---|---|---|---|---|
| FSG | 3.5-3.7 | ~0% (dense) | ~60-70 | Low below ~4-6 at% F, HF risk above it |
| Black Diamond I | 2.9-3.0 | ~0-5% | ~10-12 | Low |
| Black Diamond II | 2.5-2.7 | ~15-25% | ~6-8 | Moderate, open pore network |
| Black Diamond III | 2.4-2.5 | ~25-30% | ~4-6 | High without post-cure plasma seal |
That modulus penalty is what sets the practical ceiling on how far any single OSG generation can push porosity. Black Diamond I stayed dense enough to need no special handling, but Black Diamond II and III rely on genuinely porous networks with open or semi-open pore structures, and an open-pore low-k film is both mechanically softer under CMP downforce and chemically exposed to ambient moisture and plasma-etch byproducts along every interior pore wall, which is why fabs add a post-deposition plasma repair step to re-terminate exposed silicon sites with hydrophobic Si-CH3 groups before the film ever sees a subsequent metal or barrier deposition. None of this trade-off existed for FSG, whose failure mode was dopant-driven corrosion rather than porosity-driven softness — the two films fail for opposite reasons even though both are chasing the same number.
Read the OSG process as carbon and porosity finishing the job fluorine started and USG never attempted: USG held k near 4.0-4.2 by refusing any dopant risk at all, FSG bought a modest drop to 3.5-3.7 by accepting a bounded corrosion risk, and OSG's Black Diamond generations buy a much larger drop to 2.4-2.7 by trading that corrosion risk for a mechanical one instead — porosity, not chemistry, is now the knob, and every further k reduction below roughly 2.4 has to come from engineering pore structure and interface strength rather than from any new dopant chemistry at all.