The SOI Papers at VLSI ’14 (Part 2):


Last week we posted Part 1 of our round-up of SOI papers at the VLSI Symposia – which included the paper showing that 14nm FD-SOI should match the performance of 14nm bulk FinFETs. (If you missed Part 1, covering the three big 14nm FD-SOI and 10nm FinFET papers, click here to read it now.)

This post here gives you the abstracts of all the other papers we couldn’t fit into Part 1.  (Note that as of this posting date, the papers are not yet available on the IEEE Xplore site – but they should be shortly.)

There are in fact two symposia under the VLSI umbrella: one on technology and one on circuits. We’ll cover both here. Read on!

 

(More!) SOI Highlights from the Symposium on VLSI Technology

4.2: III-V Single Structure CMOS by Using Ultrathin Body InAs/GaSb-OI Channels on Si, M. Yokoyama et al. (U. Tokyo, NTT)

The authors propose and demonstrate the operation of single structure III-V CMOS transistors by using metal S/D ultrathin body (UTB) InAs/GaSb-on-insulator (-OI) channels on Si wafers. It is found that the CMOS operation of the InAs/GaSb-OI channel is realized by using ultrathin InAs layers, because of the quantum confinement of the InAs channel and the tight gate control. The quantum well (QW) InAs/GaSb-OI on Si structures are fabricated by using direct wafer bonding (DWB). They experimentally demonstrate both n-and p-MOSFET operation for an identical InAs/GaSb-OI transistor by choosing the appropriate thickness of InAs and GaSb channel layers. The channel mobilities of both InAs n- and GaSb p-MOSFET are found to exceed those of Si MOSFETs.

 

4.4:  High Performance InGaAs-On-Insulator MOSFETs on Si by Novel Direct Wafer Bonding Technology Applicable to Large Wafer Size Si, S. Kim et al. (U. Tokyo, IntelliEPI)

The authors present the first demonstration of InGaAs-on-insulator (-OI) MOSFETs with wafer size scalability up to Si wafer size of 300 mm and larger by a direct wafer bonding (DWB) process using InGaAs channels grown on Si donor substrates with III-V buffer layers instead of InP donor substrates. It is found that this DWB process can provide the high quality InGaAs thin films on Si. The fabricated InGaAs-OI MOSFETs have exhibited the high electron mobility of 1700 cm2/Vs and large mobility enhancement factor of 3× against Si MOSFETs.

 

6.1: Simple Gate Metal Anneal (SIGMA) Stack for FinFET Replacement Metal Gate Toward 14nm and Beyond, T. Ando et al. (IBM)

The authors demonstrate a Simple Gate Metal Anneal (SIGMA) stack for FinFET Replacement Metal Gate technology with a 14nm design rule. The SIGMA stack uses only thin TiN layers as workfunction (WF)-setting metals for CMOS integration. The SIGMA stack provides 100x PBTI lifetime improvement via band alignment engineering. Moreover, the SIGMA stack enables 9nm more gate length (Lg) scaling compared to the conventional stack with matched gate resistance due to absence of high resistivity WF-setting metal and more room for W in the gate trench. This gate stack solution opens up pathways for aggressive Lg scaling toward the 14nm node and beyond.

 

8.1: First Demonstration of Strained SiGe Nanowires TFETs with ION Beyond 700μA/μm, A. Villalon et al. (CEA-LETI, U.Udine, IMEP-LAHC, Soitec)

The authors presented for the first time high performance Nanowire (NW) Tunnel FETs (TFET) obtained with a CMOS-compatible process flow featuring compressively strained Si1-xGex (x=0, 0.2, 0.25) nanowires, Si0.7Ge0.3 Source and Drain and High-K/Metal gate. Nanowire architecture strongly improves electrostatics, while low bandgap channel (SiGe) provides increased band-to-band tunnel (BTBT) current to tackle low ON current challenges. They analyzed the impact of these improvements on TFETs and compare them to MOSFET ones. Nanowire width scaling effects on TFET devices were also investigated, showing a 1/W3 dependence of ON current ION per wire. The fabricated devices exhibit higher Ion than any previously reported TFET, with values up to 760μA/μm and average subthreshold slopes (SS) of less than 80mV/dec.

8.2: Band-to-Band Tunneling Current Enhancement Utilizing Isoelectronic Trap and its Application to TFETs, T. Mori et al. (AIST)

The authors proposed a new ON current boosting technology for TFETs utilizing an isoelectronic trap (IET), which is formed by introducing electrically inactive impurities. They  demonstrated tunneling current enhancement by 735 times in Si-based diodes and 11 times enhancement in SOI-TFETs owing to non-thermal tunneling component by the Al-N isoelectronic impurity complex. The IET technology would be a breakthrough for ON current enhancement by a few orders in magnitude in indirect-transition semiconductors such as Si and SiGe.

 

9.1: Ge CMOS: Breakthroughs of nFETs (I max=714 mA/mm, gmax=590 mS/mm) by Recessed Channel and S/D, H. Wu et al. (Purdue U.)

The authors report on a new approach to realize the Ge CMOS technology based on the recessed channel and source/drain (S/D). Both junctionless (JL) nFETs and pFETs are integrated on a common GeOI substrate. The recessed S/D process greatly improves the Ge n-contacts. A record high maximum drain current (Imax) of 714 mA/mm and trans-conductance (gmax) of 590 mS/mm, high Ion/Ioff ratio of 1×105 are archived at channel length (Lch) of 60 nm on the nFETs. Scalability studies on Ge nFETs are conducted sub-100 nm region down to 25 nm for the first time. Considering the Fermi level pining near the valence band edge of Ge, a novel hybrid CMOS structure with the inversion-mode (IM) Ge pFET and the accumulation-mode (JAM) Ge nFET is proposed.

 

13.4: Lowest Variability SOI FinFETs Having Multiple Vt by Back-Biasing, T. Matsukawa et al. (AIST)

FinFETs with an amorphous metal gate (MG) are fabricated on silicon-on-thin-buried-oxide (SOTB) wafers for realizing both low variability and tunable threshold voltage (Vt) necessary for multiple Vt solution. The FinFETs with an amorphous TaSiN MG record the lowest on-state drain cur-rent (Ion) variability (0.37 %μm) in comparison to bulk and SOI planar MOSFETs thanks to the suppressed variability of Vt (AVt=1.32 mVμm), drain induced barrier lowering (DIBL) and trans-conductance (Gm). Back-biasing through the SOTB provides excellent Vt controllability keeping the low Vt variability in contrast to Vt tuning by fin channel doping.

 

13.6: Demonstration of Ultimate CMOS based on 3D Stacked InGaAs-OI/SGOI Wire Channel MOSFETs with Independent Back Gate (Late News), T. Irisawa et al. (GNC-AIST)

An ultimate CMOS structure composed of high mobility wire channel InGaAs-OI nMOSFETs and SGOI pMOSFETs has been successfully fabricated by means of sequential 3D integration. Well behaved CMOS inverters and first demonstration of InGaAs/SiGe (Ge) dual channel CMOS ring oscillators are reported. The 21-stage CMOS ring oscillator operation was achieved at Vdd as low as 0.37 V with the help of adaptive back gate bias, VBG control.

 

17.3: Ultralow-Voltage Design and Technology of Silicon-on-Thin-Buried-Oxide (SOTB) CMOS for Highly Energy Efficient Electronics in IoT Era (Invited), S. Kamohara et al. (Low-power Electronics Association & Project, U. Electro-Communications, Keio U, Shibaura IT, Kyoto IT, U.Tokyo)

Ultralow-voltage (ULV) operation of CMOS circuits is effective for significantly reducing the power consumption of the circuits. Although operation at the minimum energy point (MEP) is effective, its slow operating speed has been an obstacle. The silicon-on-thin-buried-oxide (SOTB) CMOS is a strong candidate for ultralow-power (ULP) electronics because of its small variability and back-bias control. These advantages of SOTB CMOS enable power and performance optimization with adaptive Vth control at ULV and can achieve ULP operation with acceptably high speed and low leakage. In this paper, the authors describe their recent results on the ULV operation of the CPU, SRAM, ring oscillator, and, other lcircuits. Their 32-bit RISC CPU chip, named “Perpetuum Mobile,” has a record low energy consumption of 13.4 pJ when operating at 0.35 V and 14 MHz. Perpetuum-Mobile micro-controllers are expected to be a core building block in a huge number of electronic devices in the internet-of-things (IoT) era.

 

18.1: Direct Measurement of the Dynamic Variability of 0.120μm2 SRAM Cells in 28nm FD-SOI Technology, J. El Husseini et al. (CEA-Leti, STMicroelectronics)

The authors presented a new characterization technique successfully used to measure the dynamic variability of SRAMs at the bitcell level. This effective method easily replaces heavy simulations based on measures at transistors level. (It’s worth noting that this could save characterization/modeling costs and improve the accuracy of modeling.)  Moreover, an analytical model was proposed to explain the SRAM cell variability results. Using this model, the read failure probability after 10 years of working at operating conditions is estimated and is shown to be barely impacted by this BTI-induced variability in this FD-SOI technology.

 

18.2: Ultra-Low Voltage (0.1V) Operation of Vth Self-Adjusting MOSFET and SRAM Cell, A. Ueda et al. (U. Tokyo)

A Vth self-adjusting MOSFET consisting of floating gate is proposed and the ultra-low voltage operation of the Vth self-adjustment and SRAM cell at as low as 0.1V is successfully demonstrated.  In this device, Vth automatically decreases at on-state and increases at off-state, resulting in high Ion/Ioff ratio as well as stable SRAM operation at low Vdd. The minimum operation voltage at 0.1V is experimentally demonstrated in 6T SRAM cell with Vth self-adjusting nFETs and pFETs.

 

18.3: Systematic Study of RTN in Nanowire Transistor and Enhanced RTN by Hot Carrier Injection and Negative Bias Temperature Instability, K. Ota et al. (Toshiba)

The authors experimentally study the random telegraph noise (RTN) in nanowire transistor (NW Tr.) with various NW widths (W), lengths (L), and heights (H). Time components of RTN such as time to capture and emission are independent of NW size, while threshold voltage fluctuation by RTN was inversely proportional to the one-half power of circumference corresponding to the conventional carrier number fluctuations regardless of the side surface orientation. Hot carrier injection (HCI) and negative bias temperature instability (NBTI) induced additional carrier traps leading to the increase in the number of observed RTN. Moreover, threshold voltage fluctuation is enhanced by HCI and NBTI and increase of threshold voltage fluctuation becomes severer in narrower W.

 

SOI Highlights from the Symposium on VLSI Circuits

C19.4: A 110mW, 0.04mm2, 11GS/s 9-bit interleaved DAC in 28nm FDSOI with >50dB SFDR across Nyquist. E. Olieman et al. (U.Twente)

The authors presented an innovative nine-bit interleaved DAC (digital-to-analog converter) implemented in a 28nm FD-SOI technology. It uses two-time interleaving to suppress the effects of the main error mechanism of current-steering DACs. In addition, its clock timing can be tuned by back gate bias voltage. The DAC features an 11 GS/s sampling rate while occupying only 0.04mm2 and consuming only 110mW at a 1.0V supply voltage.

 

UTwenteC194VLSI14lowres
(Courtesy: VLSI Symposia)

A nine-bit interleaved digital-to-analog converter (DAC) from the University of Twente uses two-time interleaving to suppress the effects of the main error mechanism of current-steering DACs. The low-power device features an 11 GS/s sampling rate and occupies only 0.04mm2. From A 110mW, 0.04mm2, 11GS/s 9-bit interleaved DAC in 28nm FDSOI with >50dB SFDR across Nyquist, E. Olieman et al. (University of Twente)

 

 

C6.4: A Monolithically-Integrated Optical Transmitter and Receiver in a Zero-Change 45nm SOI Process, M. Georgas et al . (MIT, U.Colorado/Boulder)

An optical transmitter and receiver with monolithically-integrated photonic devices and circuits are demonstrated together for the first time in a commercial 45nm SOI process, without any process changes. The transmitter features an interleaved-junction carrier-depletion ring modulator and operates at 3.5Gb/s with an 8dB extinction ratio and combined circuit and device energy cost of 70fJ/bit. The optical receiver connects to an integrated SiGe detector designed for 1180nm wavelength and performs at 2.5Gb/s with 15μA sensitivity and energy cost of 220fJ/bit.

Leave a Reply

Your email address will not be published. Required fields are marked *