Logic

We present a novel calibration methodology that (i) integrates dopant diffusion, mechanical strain and bandgap narrowing for accurate device short channel effect modeling and (ii) deploys stress dependent mobility model for robust device performance projection especially on effective drive current (Ideff). Good agreement is obtained between the model calibration and experimental measurements over the full gate length range examined. Moreover, a general mobility gain with respect to uniaxial stress is presented.

This paper presents a BSIM-based method for source/drain series resistance and mobility extraction in nanoscale strained-silicon metal oxide semiconductor field effect transistors (MOSFETs) with halo implants. This method is more accurate than the conventional channel-resistance and shift and ratio method because it considers the gate-length dependence of mobility caused by local uniaxial stress and laterally nonuniform channel doping. We have verified this method using samples with different stressor/doping conditions and good agreement with experimental data has been obtained. The accuracy of the Berkeley Short-channel IGFET model (BSIM) extraction method is also proven by simulated current–voltage characteristics with different external resistant values. Significant mobility degradation in the short-channel regime has been observed for various uniaxial stressors. This method may serve as a suitable process monitor tool for ultrashallow junction and strained process development.

We demonstrate, for the first time, an integration-friendly selective PMOSFET fully silicided (FUSI) gate process. In this process, a millisecond-anneal (MSA) technique is utilized for the nickel silicide phase transformation. A highly tensile FUSI gate electrode is created and hence exerts compressive stress in the underlying channel. The highly flexible integration scheme successfully, and exclusively, implements uniform P + FUSI gates for PMOSFETs while preserving a FUSI-free N + poly-Si gate for NMOSFETs with the feature size down to 30 nm. A 20% improvement in FUSI- gated PMOSFET I on - I off is measured, which can be attributed to the enhanced hole mobility and the elimination of P + poly-gate depletion.

A 32 nm gate-first high-k/metal-gate technology is demonstrated with the strongest performance reported to date to the best of our knowledge. Drive currents of 1340/940 muA/mum (n/p) are achieved at I off =100 nA/mum, V dd =1 V, 30 nm physical gate length and 130 nm gate pitch. This technology also provides a high-Vt solution for high-performance low-power applications with its high drive currents of 1020/700 muA/mum (n/p) at total I off ~1 nA/mum @ V dd = 1V. Low sub-threshold leakage was achieved while successfully containing I boff and I goff well below 1 nA/um. Ultra high density 0.15 um 2 SRAM cell is fabricated by high NA 193 nm immersion lithography. Functional 2 Mb SRAM test-chip in 32 nm design rule has been demonstrated with a controllable manufacturing window.

A continuum model of phosphorus diffusion with germanium and carbon coimplant has been proposed and calibrated based on secondary ion mass spectroscopy (SIMS) profiles aiming at ultra shallow junction (USJ) formation in advanced CMOS technologies. The phosphorus diffusion behaviors are well captured by our model under various implant and annealing conditions, representing a significant step towards advanced n-type USJ formation technique using phosphorus and carbon coimplant for aggressively scaled CMOS technologies.

Highest planar HK/MG PFET performance (I ON = 790 muA at I off = 100 nA, Vdd= 1 V and Lg= 33 nm) has been demonstrated with a gate-first dual-metal CMOS integrated process and proven by functional SRAM cell. Integrating modern stressors without IL re-growth and achieving band edge work function without increasing T INV are two major challenges for gate-first HK/MG processes. In this work, band-edge effective work function has been achieved without increasing T INV . Furthermore, with successful integration of stress techniques like SiGe-S/D, SMT and CESL, not only performance was improved by 30% but also no reliability degradation was observed. Finally, no degradation from decreasing poly-pitch also suggests its good scalability to next generations.

For the first time, we present a state-of-the-art 32 nm low power foundry technology integrated with 0.15um 2 6-T high density SRAM, low standby transistors, analog/RF functions and Cu/low-k interconnect for mobile SoC applications. To our knowledge, this is the smallest fully functional 2Mb SRAM test-chip for 32nm node. Low power transistors with Lg of 30nm achieve current drive of 700/380 uA/um at 1.1V and off-leakage current of 1 nA/um for NMOS and PMOS, respectively. An NPoly/NWell MOS varactor shows capacitance ratio of >5.0. The MOM unit capacitance of 3.5 fF/um 2 is achieved with only 4 metal layers.

A highly scaled, high performance 45 nm CMOS technology utilizing extensive immersion lithography to achieve the industry's highest scaling factor with ELK (k=2.55) BEOL is presented. A record gate density 2.4X higher than that of 65 nm is achieved. Refined strained-CMOS demonstrated 1200/750 μA/μm Idsat at 100 nA/μm Ioff, Vdd=1 V, which has the best Ion-Lg performance reported for bulk CMOS device. The proposed 45 nm technology is not only manufacturing friendly but also has well-controlled leakage and mismatch evidenced by a functional 32 Mb 0.242 μm 2 SRAM.

In this paper, we present an advanced integration approach using milli-second anneal technique to enhance device performance. In addition to enhanced poly-silicon activation, the device gain resulted from channel stress modulation, and retarded dopant diffusion can be obtained through process optimization including rapid-thermal anneal (RTA), capping layer, and milli-second anneal. More than 15% NMOS performance gain is demonstrated without undergoing milli-second-anneal-induced pattern loading effect and re-crystallization defect. No obvious stress relaxation and driving current degradation are observed in epi-SiGe PMOS. Moreover, the performance gain is increased while lowering the RTA temperature, suggesting that our proposed approach may open an alternative pathway for 45nm technology node and beyond

Mobility enhancement techniques have become pervasive in advanced CMOS technologies. Non-scalable and scalable approaches to mobility enhancement are widely used in various application segments (high-speed, low-operating power, and low- standby power). Non-scalable techniques rely on preferential channel directions having fundamentally higher carrier mobility than the conventional <110> directions of (100) substrates. On the other hand, scalable techniques rely on process built-in mechanical strain to boost mobility as a result of band- structure response impacting both carrier-effective mass and scattering rates. Scalable techniques are based on the channel strain induced by contact-etch stop layers or Si(1-x)Ge(x) structures primarily. This paper will review both non-scalable and scalable mobility enhancement approaches from two angles, namely fundamental device physics and overall device integration schemes. The paper will also review and discuss issues related to superposition of process built-in strain for performance enhancement.