As CMOS dimensions shrink, transistor performance becomes harder to improve through geometric scaling alone. Short-channel effects, mobility degradation, parasitic resistance and device variability all constrain further gains. Oxygen insertion technology is a materials and process approach investigated for addressing several of these challenges, from dopant-profile control to carrier transport and contact formation.
The research evaluates oxygen insertion (OI) in planar CMOS devices and advanced FinFET structures. Its findings are relevant to semiconductor process and device engineers assessing ways to extend CMOS scaling, although the results include both fabricated-device measurements and technology computer-aided design (TCAD) simulations; they should not be read as a single, universally validated process recipe.
Why CMOS scaling needs additional performance levers
Traditional transistor scaling historically delivered smaller, faster devices while reducing cost per function. That approach has become more difficult as device dimensions approach physical limits. Thinner gate dielectrics can increase leakage, stronger vertical electric fields can degrade channel mobility, and shrinking source/drain contact areas can increase parasitic resistance. At the same time, random dopant fluctuations and gate work-function variations can make device behavior less predictable.
These effects create competing design requirements. A transistor must provide adequate on-state current while limiting off-state leakage and maintaining electrostatic control of the channel. OI technology has been studied as a way to influence several parts of this problem rather than relying on dimension reduction alone.
How oxygen insertion technology works
In the approach described, multiple partial oxygen monolayers are incorporated into a crystalline silicon substrate using a low-temperature chemical vapor deposition epitaxy process. The oxygen is introduced between silicon layers while allowing subsequent silicon epitaxial growth. The inserted layers can then affect both carrier behavior and the movement of dopants during later thermal processing.
Those two effects underpin the main applications examined: carrier sub-band engineering and dopant-profile engineering. The technology is intended to complement other CMOS performance boosters, rather than replace every established device or process technique.
Carrier transport and leakage considerations
In a MOSFET channel, carriers occupy energy sub-bands and can scatter between states. Simulations and experimental results described in the research indicate that OI layers can alter the distribution of carrier wavefunctions, increasing separation between certain sub-bands. That separation can reduce inter-band scattering and contribute to higher electron and hole mobility.
The reported electron mobility benefit also involves changes to sub-band energies and electron transport effective mass. For bulk-silicon n-channel MOSFETs, measured gate leakage was lower with oxygen layers. The research attributes this result to a change in the electron tunneling effective mass. These mechanisms illustrate why OI is not simply a dopant-control method: the inserted layers can influence carrier transport and tunneling as well.
Dopant diffusion control for shallow junctions
Ultra-shallow junctions are important in scaled planar CMOS, but making junctions shallower can raise sheet resistance (Rsh). Subsequent annealing can also broaden dopant profiles. Transient-enhanced diffusion (TED), driven by silicon interstitials generated during processing, is one contributor to this movement.
Experimental measurements and secondary ion mass spectrometry (SIMS) analyses in the research indicate that OI can reduce dopant diffusion and promote dopant retention during post-implantation annealing. The reported result is a reduction in both Rsh and junction depth (XJ) for ultra-shallow junctions. OI also reduced the unfavorable increase in sheet resistance associated with a low-temperature-oxide capping layer, which the research links to lower dopant activation.
SIMS observations of boron and phosphorus profiles further support the proposed mechanism. Dopants accumulated, or “piled up,” around the oxygen-insertion region, consistent with interstitial trapping. In oxidation experiments, the oxygen layers helped limit oxidation-enhanced diffusion (OED), preserving buried boron profiles that broadened in control samples.
FinFET channel engineering and SRAM implications
The work also considers a super-steep retrograde (SSR) fin-channel doping profile enabled by OI. In an SSR profile, doping varies sharply with depth rather than remaining uniform. This can support electrostatic integrity while avoiding the performance and variability trade-offs associated with heavily doping the entire fin channel.
Three-dimensional TCAD simulations for a 7/8-nm low-power target found that the bulk-silicon SSR FinFET design could achieve higher on-current-to-off-current ratios and lower sensitivity to device variation than a conventional bulk-silicon control design. In six-transistor SRAM yield estimates, SSR and silicon-on-insulator (SOI) FinFET designs achieved similarly low minimum supply voltage, VDD,min. Both were estimated to reduce VDD,min by as much as 100 mV compared with the control design. The study presents the OI-enabled SSR design as a potentially lower-cost alternative to SOI, not as a demonstrated universal replacement.
Contact engineering through barrier-height modification
A separate part of the research examines a Pt/Ti/p-type silicon metal-semiconductor contact. For p-type silicon contacts, reducing the Schottky barrier height (ΦBp) can help lower specific contact resistivity. Electrical measurements, SIMS and X-ray photoelectron spectroscopy (XPS) associated OI with a reduction in ΦBp and changes in titanium and silicon core energy levels.
The proposed explanation is that OI suppresses grain-boundary-enhanced diffusion of platinum into crystalline silicon during forming-gas annealing. This favors formation of low-barrier platinum monosilicide. The result adds contact behavior to the set of process interactions that may be influenced by oxygen insertion.
Practical implications for process development
For semiconductor engineers, the key takeaway is that OI may provide multiple, linked process levers: changing carrier confinement, limiting interstitial-mediated dopant diffusion, shaping channel profiles and influencing contact formation. The evidence spans fabrication and characterization as well as simulation, so application-specific validation remains essential.
In practical evaluation, teams should distinguish the mechanisms relevant to their device target. Junction optimization concerns dopant activation, diffusion and sheet resistance; FinFET development concerns channel electrostatics, variability and SRAM operating voltage; contact work concerns barrier formation and interface chemistry. Oxygen insertion technology is most useful to consider within that broader process context—as a potential means of extending CMOS performance where conventional scaling faces physical and manufacturing constraints.