Design optimization methodology for deep-submicrometer CMOS deviceat low-temperature operation

The design optimization for 0.3-μm channel CMOS technology at liquid-nitrogen temperature (77 K) is described. The tradeoff between circuit performance and reliability for deep-submicrometer CMOS devices at low-temperature operation is theoretically and experimentally examined. A simulator, which selects power-supply voltage and process/device parameters for low-temperature operation, has been developed. Based upon the simulated results, design optimization for low-temperature operation has been proposed to determine power-supply voltage and various process and device parameters. The optimized design has been demonstrated on a 0.3-μm CMOS device, by utilizing electron beam (EB) lithography· Excellent device characteristics and a functional ring oscillator circuit have been obtained at 77 K     

A low-resistance self-aligned T-shaped gate for high-performancesub-0.1-μm CMOS

This paper describes the high performance of T-shaped-gate CMOS devices with effective channel lengths in the sub-0.1-μm region. These devices were fabricated by using selective W growth, which allows low-resistance gates smaller than 0.1 μm to be made without requiring fine lithography alignment. We used counter-doping to scale down the threshold voltage while still maintaining acceptable short-channel effects. This approach allowed us to make ring oscillators with a gate-delay time as short as 21 ps at 2 V with a gate length of 0.15 μm. Furthermore, we experimentally show that the high circuit speed of a sub-0.1-μm gate length CMOS device is mainly due to the PMOS device performance, especially in terms of its drivability     

Merged BiCMOS logic to extend the CMOS/BiCMOS performance crossoverbelow 2.5-V supply

The authors discuss the merged BiCMOS (MBiCMOS) gate, a unique circuit configuration to improve BiCMOS gate performance at low supply voltages. MBiCMOS maintains a measured delay and power-delay advantage over CMOS into the 2-V supply range, in a simple four-device gate that does not require any change in the standard BiCMOS processing sequence. In a 2-μm technology, MBiCMOS outperforms CMOS down to a 2.6-V supply. Gates designed for fabrication in a 0.5-μm technology and simulated using measured device parameters indicate that MBiCMOS can be used to extend the performance crossover voltage to below 2 V in the submicrometer regime. A full-swing version of the MBiCMOS gate (FS-MBiCMOS) is introduced. Simulations of 2-μm gates show FS-MBiCMOS/CMOS performance crossover voltages of 2.2 V     

0.25 μm gate length CMOS devices for cryogenic operation

Under cryogenic operation, a low V_th realizes a high speed performance at a greatly reduced power-supply voltage, which is the most attractive feature of Cryo-CMOS. It is very important in sub-0.25 μm Cryo-CMOS devices to reconcile the miniaturization and the low V_th. Double implanted MOSFET's technology was employed to achieve the low V_th while maintaining the short channel effects immunity. We have investigated both the DC characteristics and the speed performance of 0.25 μm gate length CMOS devices for cryogenic operation. The measured transconductances in the saturation region were 600 mS/mm for 0.2 μm gate length n-MOSFET's and 310 mS/mm for 0.25 μm gate length p-MOSFET's at 80 K. The propagation delay time in the fastest CMOS ring oscillator was 22.8 ps at V_dd=1 V at 80 K. The high speed performance at extremely low power-supply voltages has been experimentally demonstrated. The speed analysis suggests that the sub-l0 ps switching of Cryo-CMOS devices will be realized by reducing the parasitic capacitances and through further miniaturization down to 0.1 μm gate length or below     

A 0.8-μm CMOS technology for high-performance ASIC memory andchannelless gate array

A 0.8-μm polycide-gate, double-layer-metal CMOS technology is described. Nominal device gate lengths down to 0.8 (±0.2) μm are used for both n- and p-channel transistors. Compact isolation, 175-A gate oxide grown in dry/wet/dry ambient, shallow-junction halo-implanted lightly doped drain n and p devices, TiN contact barrier, and a planarized double-layer-metal process are all integrated and demonstrated with a 0.8-μm full-CMOS 16K SRAM (static random-access memory) circuit. The device process integrity, design margins, performance, reliability, product yield and speed enhancement are all discussed in detail     

Threshold voltage-minimum gate length trade-off in buried channelPMOS devices for scaled supply voltage CMOS technologies

The trade-off between threshold voltage (V_th) and the minimum gate length (L_min) is discussed for optimizing the performance of buried channel PMOS transistors for low voltage/low power high-speed digital CMOS circuits. In a low supply voltage CMOS technology it is desirable to scale V_th and L_min for improved circuit performance. However, these two parameters cannot be scaled independently due to the channel punch-through effect. Statistical process/device modeling, split lot experiments, circuit simulations, and measurements are performed to optimize the PMOS transistor current drive and CMOS circuit speed. We show that trading PMOS transistor V_th for a smaller L_min results in faster circuits for low supply voltage (3.3 to 1.8 V) n⁺-polysilicon gate CMOS technology, Circuit simulation and measurements are performed in this study. Approximate empirical expressions are given for the optimum buried channel PMOS transistor V _th for minimizing CMOS circuit speed for cases involving: (1) constant capacitive load and (2) load capacitance proportional to MOS gate capacitance. The results of the numerical exercise are applied to the centering of device parameters of a 0.5 μm 3.3 V CMOS technology that (a) matches the speed of our 0.5 μm 5 V CMOS technology, and (b) achieves good performance down to 1.8 V power supply. For this process the optimum PMOS transistor V_th (absolute value) is approximately 0.85-0.90 V     

Deep-submicrometer CMOS technology with reoxidized or annealednitrided-oxide gate dielectrics prepared by rapid thermalprocessing

Deep submicrometer CMOSFETs with re-annealed nitride-oxide gate dielectrics have been demonstrated to satisfy 3.3-V operation, unlike conventional oxide FETs. The 1/4-μm re-annealed nitrided-oxide CMOS devices achieve (1) an improved saturation transconductance g _m of ~250 μS/μm for n-FETs together with acceptably small degradation in p-FET g_m resulting in a CMOS gate delay time of 55 ps/stage comparable or superior to the device/circuit performance of oxide FETs, and (2) device lifetimes improved by ~100 times to exceed 10 years with respect to both ON- and OFF-state hot-carrier reliability for n-FETs as well as gate-dielectric integrity together with unchanged p-FET hot-carrier reliability, all at 3.3-V operation. To achieve these CMOS performance/reliability improvements, both a light nitridation and subsequent re-annealing in O ₂ (reoxidation) or in N₂ (inert-annealing) are found to be crucial     

Power-supply voltage impact on circuit performance for half andlower submicrometer CMOS LSI

Based on theoretical understanding, the concept that the lower power supply voltage limit can be simply expressed by 1.1E_cL_eff, where E_c is the critical electric field necessary to cause carrier velocity saturation and L_eff is the effective channel length, is introduced. Experimental results confirmed that 1.1E_cL_eff predicts a good guideline for power-supply voltage for CMOS devices over a wide range of gate oxide thickness (7-45 nm) and design rule (0.3-2.0 μm). On the basis of theoretical models and experimental results, trends for power-supply voltage with MOS device scaling are demonstrated. It is shown that 1.1E_cL_eff can be regarded as the lower power-supply voltage limit in order to maintain the improvement in delay time for below 0.6-μm channel length at reduced power supply. The transconductance behavior for a MOSFET under high electric fields was investigated in order to explain the physical meaning of 1.1E_cL_eff     

Extremely scaled silicon nano-CMOS devices

Silicon-based CMOS technology can be scaled well into the nanometer regime. High-performance, planar, ultrathin-body devices fabricated on silicon-on-insulator substrates have been demonstrated down to 15-nm gate lengths. We have also introduced the FinFET, a double-gate device structure that is relatively simple to fabricate and can be scaled to gate lengths below 10 nm. In this paper, some of the key elements of these technologies are described, including sublithographic patterning, the effects of crystal orientation and roughness on carrier mobility, gate work function engineering, circuit performance, and sensitivity to process-induced variations.     

Device design for the submicrometer p-channel FET with n+polysilicon gate

CMOS has become one of the most important technologies for VLSI applications. If the conventional n⁺polysilicon gate approach is to be maintained for VLSI CMOS, the p-channel transistor will cause problems in scaling down to submicrometers due to the counter-doping that is necessary to adjust the threshold voltage to a reasonable value. The depth of the p⁺source-drain junctions will also cause short-channel effects. This paper presents in-depth analysis of the submicrometer p-channel transistor structure. The effects of the counter-doping junction depth and the source-drain junction depth on the device subthreshold characteristics are discussed. Criteria for the submicrometer p-channel transistor structure with good subthreshold characteristics are presented. A new technique for minimizing the counter-doping junction depth is also presented. Submicrometer p-channel transistors with n⁺polysilicon gate were fabricated using this new technique as well as techniques for forming very shallow p⁺-junctions. Devices with submicrometer channel lengths showed very good subthreshold characteristics, as predicted by simulations.     

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.     

Moore's law lives on [CMOS transistors]

We discuss several device structures suitable for scaling CMOS devices well into the nano-CMOS era, perhaps down below 10 nm physical gate length. The ultra-thin body MOSFET device structure has many features in common with today's bulk MOSFET, which makes it easier for industry to introduce into manufacturing. On the other hand, the double-gate structure as represented by the FinFET appears to offer greater scalability down to 10 nm gate length or perhaps even below. While a number of significant challenges remain to be overcome, including device parasitics, interfaces, and threshold voltage control techniques, it appears that the continued evolution of CMOS integrated circuit technology into this regime will not be impeded by basic limitations of the underlying transistor technology. The implication of this is that "Moore's law" may continue for yet another 15-20 years before the ultimate device limits for CMOS are reached.     

Supply voltage design tradeoffs between speed and NMOSFETreliability of half-micrometer BiCMOS gates

Voltage supply scaling for a BiCMOS gate was investigated experimentally and analytically. For half-micrometer technology, the supply voltage design tradeoffs between propagation delay and NMOSFET reliability were studied. It was found that the minimum BiCMOS operating voltage ranges from 2.5 to 3.0 V. This minimum can be explained by the sum of twice the base-emitter potential and the NMOSFET threshold voltage, due to the emitter grounded Bi-MOS structure of the BiCMOS gate. As a result, 3.3-V operation is inherently marginal for BiCMOS gates. On the other hand, NMOS reliability in the BiCMOS gate is drastically improved because effective drain voltage is reduced by the base-emitter potential. Thus, NMOSFETs with a channel length shorter by more than 0.2 μm can be used, and this ensures reliable operation of BiCMOS gates at 5 V. In terms of the tradeoffs between gate speed and NMOS reliability, it is verified that BiCMOS gates offer the highest speed in half-micrometer design     

Comparison of characteristics of n-channel and p-channel MOSFET's for VLSI's

A comparison of device characteristics of n-channel and p-channel MOSFET's is made from the overall viewpoint of VLSI construction. Hot-carrier-related device degradation of device reliability, as well as effective mobility, is elaborately measured for devices having effective channel lengths of 0.5-5 µm. From these experiments, it is found that hot-electron injection due to impact ionization at the drain, rather than "lucky hot holes," imposes a new constraint on submicrometer p-channel device design, though p-channel devices have been reported to have much less trouble with hot-carrier effects than n-channel devices do. Additionally, p-channel devices are found to surpass n-channel devices in device reliability in that they have a highest applicable voltage BVDCthat is more than two times as high as for n-channel devices. It is also experimentally confirmed that the effective hole mobility approaches the effective electron mobility when effective channel lengthL_{eff} < 0.5µm. These significant characteristics of p-channel devices imply that p-channel devices have important advantages over n-channel devices for realization of sophisitcated VLSI's with submicrometer dimensions. It is also shown that hot holes, which may create surface states or trap centers, play an important role in such hot-carrier-induced device degradation as transconductance degradation.     

Submicrometer device design for hot-electron reliability and performance

Hot-electron stressing effect on different lightly doped drain device (LDD), As/P, and conventional As source/drain device structures are investigated. Increasing the overlap between the gate and drain is found to reduce hot-electron degradation significantly when stressed under the same substrate current level. By increasing the gate-to-drain overlap, it is possible to design LDD and As/P devices with a shorter n-region and still have good hot-electron reliability. These devices have better current drive and are scalable down to the submicrometer region. The As/P device with a short n-region is a good candidate for a submicrometer VLSI device because of the simplicity in processing, the good device performance, and the low susceptibility to hot-electron degradation.     

A modified lightly doped drain structure for VLSI MOSFET's

A new n-MOS LDD-like device structure (the J-MOS transistor) is proposed. Its design, simulation, and fabrication are studied in this paper, n-channel MOSFET's with Leffbelow 2 µm suffer from high-field effects that must be overcome to secure reliable 5-V operation. LDD structures alleviate these effects, but their reliability is better than that of conventional MOSFET's only if the n^-regions have a peak doping density above 1 × 10¹⁸cm^-3. To overcome this limitation and to allow constant voltage scaling for devices into the submicrometer regime, the J-MOS structure uses a series drain JFET to drop part of the supply voltage. Both 2-D device simulations and experimental results are presented to demonstrate the operation of this device and its potential for applications requiring reliable submicrometer device operation under maximum supply voltage. The major experimental findings are that the J-MOS structure can sustain 5-V operation even for submicrometer effective channel lengths. As has been the case with all LDD-like structures, improved device reliability has been achieved at the expense of some performance. However, the advantages of keeping 5-V operation in micrometer-sized devices may outweigh this performance loss.     

Influence of Extreme Thinning on 130-nm Standard CMOS Devices for 3-D Integration

The influence of thinning standard 130-nm CMOS technology device wafers to residual silicon thicknesses of 20 and 5 mum has been studied. Electrical performance was evaluated at wafer level by characterizing various basic device parameters before and after thinning. An increase in the well sheet resistance and a reduction in the gate leakage current were observed. However, both at 25degC and 100degC, no performance degradation was found that could be correlated to the applied thinning techniques, including extreme grinding down to 5 mum. These electrical results are consistent with the experimentally observed submicrometer thinning-induced subsurface damage. Hence, the feasibility of extreme thinning in 3-D integration schemes for standard bulk-Si CMOS was demonstrated.     

TFSOI complementary BiCMOS technology for low power applications

A Thin-Film-Silicon-On-Insulator Complementary BiCMOS (TFSOI CBiCMOS) technology has been developed for low power applications. The technology is based on a manufacturable, near-fully-depleted 0.5 μm CMOS process with the lateral bipolar devices integrated as drop-in modules for CBiCMOS circuits. The near-fully-depleted CMOS device design minimizes sensitivity to silicon thickness variation while maintaining the benefits of SOI devices. The bipolar device structure emphasizes use of a silicided polysilicon base contact to reduce base resistance and minimize current crowding effects. A split-oxide spacer integration allows independent control of the bipolar base width and emitter contact spacing. Excellent low power performance is demonstrated through low current ECL and low voltage, low power CMOS circuits. A 70 ps ECL gate delay at a gate current of 20 μA is achieved. This represents a factor of 3 improvement over bulk trench-isolated double-polysilicon self-aligned bipolar circuits. Similarly, CMOS gate delay shows a factor of 2 improvement over bulk silicon at a power supply voltage of 3.3 V. Finally, a 460 μW 1 GHz prescaler circuit is demonstrated using this technology     

具有应变沟道及EOT 1.2nm高性能栅长22nm CMOS器件

深入研究了亚30nm CMOS关键工艺技术，特别是提出了一种新的低成本的提高空穴迁移率的技术--Ge预非晶化S/D延伸区诱生沟道应变技术，它使栅长90nm pMOS空穴有效迁移率在0.6MV/cm电场下提高32%. 而且空穴有效迁移率的改善，随器件特征尺寸缩小而增强. 利用零阶劳厄线衍射的大角度会聚束电子衍射分析表明，在沟道区相应的压应变为-3.6%. 在集成技术优化的基础上，研制成功了高性能栅长22nm应变沟道CMOS器件及栅长27nm CMOS 32分频器电路（其中分别嵌入了57级/201级环形振荡器）, EOT为1.2nm,具有Ni自对准硅化物.     

Disposable polysilicon LDD spacer technology

An advanced 0.5-μm CMOS disposable lightly doped drain (LDD) spacer technology has been developed. This 0.5-μm CMOS technology features surface-channel LDD NMOS and PMOS devices, n⁺/p⁺ poly gates, 125-A-thick gate oxide, and Ti-salicided source/drain/gate regions. Using only two masking steps, the NMOS and PMOS LDD spacers are defined separately to provide deep arsenic n⁺ regions for lower salicided junction leakage, while simultaneously providing shallow phosphorus n^- and boron p^- regions for improved device short-channel effects. Additionally, the process allows independent adjustment of the LDD and salicide spacers to optimize the LDD design while avoiding salicide bridging of source/drain to gate regions. The results indicate extrapolated DC hot-carrier lifetimes in excess of 10 years for a 0.3-μm electrical channel-length NMOS device operated at a power-supply voltage of 3.3 V