It is typically deposited using Plasma-Enhanced Chemical Vapor Deposition (PECVD) or Low-Pressure Chemical Vapor Deposition (LPCVD) at low thermal budgets to prevent the premature diffusion of implanted S/D dopants .
In advanced planar CMOS technologies, Stress Memorization Technology (SMT) is integrated into the Middle-of-Line (MOL) flow to selectively boost nMOSFET channel mobility . Positioned directly after the source/drain (S/D) extension and deep junction implants (PIO S/D Impla
nt 2 and 3) and photoresist stripping, the SMT Oxide Deposition step marks the beginning of the SMT module . The primary integration purpose of this oxide layer is to act as a conformal stress-buffer, screening layer, and highly selective chemical etch-stop liner before the deposition of the highly-stressed SMT Nitride capping layer . This temporary stack remains on the wafer during the subsequent SMT Rapid Thermal Anneal (RTA) and is selectively stripped during the SMT Nitride Removal step prior to contact formation . By introducing this oxide, the underlying active silicon, newly formed S/D junctions, and fragile gate spacer structures are protected from direct mechanical damage, physical sputtering, and chemical attack during the nitride deposition and stripping steps .
The physical mechanism of SMT relies on solid-state recrystallization, thermal expansion, and viscoelastic stress coupling . Prior to SMT deposition, a pre-amorphization implant (PAI) using heavy species such as germanium or arsenic amorphizes the upper region of the polycrystalline silicon (poly-Si) gate . During the subsequent high-temperature RTA, the poly-Si undergoes solid-phase epitaxial recrystallization and grain growth under the physical constraint of the temporary oxide-nitride capping stack . As the grain structure reforms, the local volume changes and thermal expansion generate significant mechanical stress, which is governed by anisotropic linear elastic relations ($\sigma_i = C_{ij}\varepsilon_j$) and quasi-static force balance . This mechanical deformation is transferred through the thin oxide and "frozen" into the poly-Si gate and the underlying channel . Upon removal of the capping layers, the residual strain persists due to geometric constraints , creating a vertical compressive and lateral tensile strain in the channel . This strain alters the silicon lattice's periodic potential and shifts the energy band structure ($E-k$ relationship) via the Schrödinger wave equation , which lifts the degeneracy of the conduction band valleys, lowers the electron effective mass, suppresses intervalley phonon scattering, and ultimately boosts electron mobility .
The choice of material and deposition method for the SMT oxide is driven by a critical trade-off between mechanical coupling efficiency, thermal budget, and etch selectivity . Unlike other oxide depositions in the flow—such as STI HARP Liner/Oxide Depositions which utilize high-ozone TEOS-based processes for high-aspect-ratio gap-fill, or CB and RV Silane Oxide Depositions which form permanent passivation and interlayer dielectrics —the SMT oxide must be temporary, thin, and highly conformal . It is typically deposited using Plasma-Enhanced Chemical Vapor Deposition (PECVD) or Low-Pressure Chemical Vapor Deposition (LPCVD) at low thermal budgets to prevent the premature diffusion of implanted S/D dopants . The deposition process parameters must be optimized because a thicker or less dense oxide layer exhibits greater viscoelastic relaxation during high-temperature annealing, which dampens the mechanical stress transfer from the SMT nitride to the channel . Conversely, an excessively thin oxide layer provides insufficient protection during the subsequent hot phosphoric acid or dry etch steps used for SMT Nitride Removal, risking active area pitting and spacer erosion .
At the 28nm planar node, SMT becomes highly critical as traditional geometrical scaling can no longer yield sufficient drive current performance on its own . SMT is exceptionally valuable at this node because its layout dependency is relatively low compared to permanent stress liners, making it compatible with tight 28nm gate pitches . Furthermore, by using SMT selectively, the nMOS performance can be enhanced by over 15% without incurring the severe pMOS mobility degradation that occurs when high-tensile stress liners are applied globally .
A free account opens the full rationale, risk analysis, and the paper and patent citations behind them.
Sign Up FreeOr
Need every step? Pay once for lifetime access.
Pay by card, PayPal, Apple Pay, or Google Pay — exact options shown at checkout · 14-day money-back guarantee