Brian S. Doyle

Benchmarking nanotechnology for high-performance and low-power logic transistor applications

Recently there has been tremendous progress made in the research of novel nanotechnology for future nanoelectronic applications. In particular, several emerging nanoelectronic devices such as carbon-nanotube field-effect transistors (FETs), Si nanowire FETs, and planar III-V compound semiconductor (e.g., InSb, InAs) FETs, all hold promise as potential device candidates to be integrated onto the silicon platform for enhancing circuit functionality and also for extending Moores Law. For high-performance and low-power logic transistor applications, it is important that these research devices are frequently benchmarked against the existing Si logic transistor data in order to gauge the progress of research. In this paper, we use four key device metrics to compare these emerging nanoelectronic devices to the state-of-the-art planar and nonplanar Si logic transistors. These four metrics include: 1) CV/I or intrinsic gate delay versus physical gate length L/sub g/; 2) energy-delay product versus L/sub g/; 3) subthreshold slope versus L/sub g/; and 4) CV/I versus on-to-off-state current ratio I/sub ON//I/sub OFF/. The results of this benchmarking exercise indicate that while these novel nanoelectronic devices show promise and opportunities for future logic applications, there still remain shortcomings in the device characteristics and electrostatics that need to be overcome. We believe that benchmarking is a key element in accelerating the progress of nanotechnology research for logic transistor applications.

High performance fully-depleted tri-gate CMOS transistors

Fully-depleted (FD) tri-gate CMOS transistors with 60 nm physical gate lengths on SOI substrates have been fabricated. These devices consist of a top and two side gates on an insulating layer. The transistors show near-ideal subthreshold gradient and excellent DIBL behavior, and have drive current characteristics greater than any non-planar devices reported so far, for correctly-targeted threshold voltages. The tri-gate devices also demonstrate full depletion at silicon body dimensions approximately 1.5 - 2 times greater than either single gate SOI or non-planar double-gate SOI for similar gate lengths, indicating that these devices are easier to fabricate using the conventional fabrication tools. Comparing tri-gate transistors to conventional bulk CMOS device at the same technology node, these non-planar devices are found to be competitive with similarly-sized bulk CMOS transistors. Furthermore, three-dimensional (3-D) simulations of tri-gate transistors with transistor gate lengths down to 30 nm show that the 30 nm tri-gate device remains fully depleted, with near-ideal subthreshold swing and excellent short channel characteristics, suggesting that the tri-gate transistor could pose a viable alternative to bulk transistors in the near future.

Tri-Gate Transistor Architecture with High-k Gate Dielectrics, Metal Gates and Strain Engineering

We have combined the benefits of the fully depleted tri-gate transistor architecture with high-k gate dielectrics, metal gate electrodes and strain engineering. High performance NMOS and PMOS trigate transistors are demonstrated with IDSAT=1.4 mA/mum and 1.1 mA/mum respectively (IOFF=100nA/mum, VCC =1.1V and LG=40nm) with excellent short channel effects (SCE)-DIBL and subthreshold swing, DeltaS. The contributions of strain, the lang100rang vs. lang110rang substrate orientations, high-k gate dielectrics, and low channel doping are investigated for a variety of channel dimensions and FIN profiles. We observe no evidence of early parasitic corner transistor turn-on in the current devices which can potentially degrade ION-IOFF and DeltaS

A 50 nm depleted-substrate CMOS transistor (DST)

In this paper we show a Depleted-Substrate Transistor (DST) technology which demonstrates significant performance gain over bulk Si transistors without the floating body effect (FBE). We have fabricated depleted-substrate CMOS transistors on thin silicon body (/spl les/30 nm) with physical gate lengths down to 50 nm which show much steeper subthreshold slopes (/spl les/75 mV/decade) and improved DIBL (/spl les/50 mV/V) over both partially-depleted (P-D) SOI and bulk Si, for both PMOS and NMOS transistors. The salicide formation and high parasitic resistance problems associated with the use of thin Si body can be overcome by using raised source/drain. Depleted-substrate PMOS transistors with 50 nm physical gate length and raised source/drain were fabricated and achieved I/sub on/=0.65 mA/um and I/sub off/=9 nA/um at V/sub cc/=1.3 V. This PMOS drive current is the highest ever reported, and is about 30% higher than any previously published PMOS I/sub on/ value for both PD-SOI and bulk Si at a given I/sub off/. The use of raised source/drain improved the I/sub on/ of the depleted-substrate NMOS transistors by /spl sim/20%. Depleted-substrate NMOS transistors with 65 nm physical gate length and raised source/drain achieved DIBL=45 mV/V, subthreshold slope=75 mV/decade, I/sub on/=1.18 mA/um and I/sub off/ =60 nA/um at V/sub cc/=1.3 V, as well as significant improvement in Id-Vd characteristics due to a 60% reduction in DIBL and >25% improvement in subthreshold slope over the bulk Si.

30 nm physical gate length CMOS transistors with 1.0 ps n-MOS and 1.7 ps p-MOS gate delays

Planar CMOS transistors have been fabricated to evaluate the 70 nm technology node using conventional transistor design methodologies. Conventional CMOS transistors with 30 nm physical gate length were fabricated using aggressively scaled junctions, polysilicon gate electrode, gate oxide and Ni silicide. These devices have inversion Cox exceeding 1.9 /spl mu/F/cm2, n-MOS gate delay (CV/I) of 0.94 ps and p-MOS gate delay of 1.7 ps at V/sub cc/=0.85 V. These are the smallest CV/I values ever reported for Si CMOS devices. The transistors also show good short channel control and subthreshold swings. The n-MOS and p-MOS have drive currents equal to 514 /spl mu/A//spl mu/m and 285 /spl mu/A//spl mu/m respectively with I/sub off/ at or below 100 nA//spl mu/m at Vcc=0.85 V. The saturation gm is equal to 1200 mS/mm for n-MOS and 640 mS/mm for p-MOS. These are among the highest gm values ever reported. The junction edge leakage is reasonably low with less than 1 nA//spl mu/m at 1.0 V and 100 C for both n-MOS and p-MOS. These encouraging results suggest that the 70 nm technology node is achievable using conventional planar transistor design and process flow.

Hydrogen induced silicon surface layer cleavage

Physical mechanisms of hydrogen induced silicon surface layer cleavage were investigated using a combination of microscopy and spectroscopy techniques. The evolution of the silicon cleavage phenomenon is recorded by a series of microscopic images. The underlying hydrogen profiles under (between 250 and 500 °C) annealing are characterized by secondary-ion-mass spectroscopy and hydrogen forward scattering experiments. An idea gas law model calculation suggests that internal pressure of molecular hydrogen filled microcavities is in the range of Giga-Pascal, high enough to break silicon crystal bond. A dose threshold, which prevents cleavage, is observed at 1.6×1017 cm−2 for 40 kV hydrogen implantation.

AC versus DC hot-carrier degradation in n-channel MOSFETs

The origins of the enhanced AC hot-carrier stress damage are examined. The enhancement in hot-carrier stress damage under AC stress conditions observed with respect to damage under DC stress conditions can fully be explained by the presence of three damage mechanisms occurring during both DC and AC operation: interface states created at low and mid-gate voltages, oxide electron traps created under conditions of hole injection into the oxide, and oxide electron traps created under conditions of hot-electron injection. It is shown that the quasi-static contributions of these mechanisms fully account for hot-carrier degradation under AC stress. The AC stress model is applied to devices from several different technologies and to several different AC stress waveforms. Excellent agreement is obtained in each case. The results demonstrate the validity of the model for frequencies up to 1 MHz. The absence of any transient effect indicates that the model could be applicable at much higher frequencies. >

Recovery of hot-carrier damage in reoxidized nitrided oxide MOSFET's

Recovery of channel hot-carrier damage in reoxidized nitrided oxide (RNO) n- and p-MOSFETs is examined. It is found that recovery is substantially greater in RNO versus conventional oxide (CO) devices, particularly for p-MOSFETs. The authors believe this recovery is due to the detrapping of electrons trapped in the nitridation-induced traps near the substrate interface. The more rapid recovery of the hot-carrier damage in RNO devices will produce greater circuit lifetime improvements over conventional oxides than predicted by accelerated static high-voltage tests.<<ETX>>

Simultaneous growth of different thickness gate oxides in silicon CMOS processing

A method is proposed that allows the growth of gate oxides of different thicknesses on a single wafer. The method does not require masking the gate oxide during oxidation with its inherent risk to the oxide quality, but rather relies on the implant of nitrogen into the silicon wafer before both oxide growth and preoxidation cleans. This implant is performed at the same step as the normal threshold voltage implants, avoiding possible contamination. Using nitrogen implant doses of the order of 3/spl times/10/sup 14/-3/spl times/10/sup 15/ cm/sup -2/, it is shown that it is possible to grow oxides of 30-70 /spl Aring/, for a process with a nominal oxide thickness of 90 /spl Aring/.<<ETX>>

High-frequency response of 100 nm integrated CMOS transistors with high-K gate dielectrics

This paper reports, for the first time, the high-frequency response of NMOS and PMOS transistors in an integrated CMOS technology with 100 nm physical gate length and alternative gate dielectrics such as ZrO/sub 2/ and HfO/sub 2/ with TiN/PolySi gate electrode. It is shown that the dielectric constants of ZrO/sub 2/, HfO/sub 2/ and SiO/sub 2/ are invariant with respect to operating frequency at least up to 20 GHz. In addition, the cutoff frequency f/sub t/ of the 100 nm CMOS transistor test structures with ZrO/sub 2/ gate dielectric was measured to be equal to 46 GHz for NMOS and 47 GHz for PMOS. The corresponding f/sub t/ values for HfO/sub 2/ were 45 GHz for NMOS and 35 GHz for PMOS. High-K film transistors with 80 nm physical gate lengths, 7 /spl mu/m gate width and layout optimized for high frequency testing were also fabricated. The NMOS devices achieved an f/sub t/ of 83 GHz and an f/sub max/ of 35 GHz, while the PMOS yielded 41 GHz and 25 GHz respectively. These results are very similar to those of CMOS transistors with SiO/sub 2/ gate dielectric at similar physical gate lengths and widths. These results are very encouraging and suggest that high-K gate dielectrics can be used for high-frequency logic applications.