Ferroelectric neuron integrated circuits using SrBi2Ta2O9-gate FET's and CMOS Schmitt-triggeroscillators

A pulse frequency modulation (PFM) type ferroelectric neuron circuit composed of a metal-ferroelectric-semiconductor field effect transistor (MFSFET) and a CMOS Schmitt-trigger oscillator was fabricated on an SOI structure, in which SrBi₂Ta₂O₉ (SBT) was used as a ferroelectric gate material of the FET. It was found that the fabricated MFSFET showed a relatively good I_D-V_G (drain current versus gate voltage) characteristic with a hysteresis loop due to the ferroelectricity of the SBT film and that it acted as a synapse device with adaptive-learning function. It was also found that the output pulse height of the circuit was as high as the power supply voltage and that output pulse frequency was changed as the number of applied input pulses increased     

Ferroelectric characteristic of group IV elements added SrBi2Ta2O9 thin films

The SrBi₂Ta₂O₉ (SBT) thin film added IV group elements was fabricated on the Pt/Ti/SiO₂/Si substrate by the metal–organic decomposition (MOD) method, the Pt electrode was deposited on the SBT thin film by the DC spattering method. The electric properties of the ferroelectric capacitor were measured. The remanent polarization and the relative dielectric constant of the SBT thin film have decreased, according to the amount of IV group elements addition. The IV group elements added SBT thin film with low relative dielectric constant and low remanent polarization is suitable for the application of the ferroelectric gate FET type memory.     

Characteristics of metal-ferroelectric-insulator-semiconductor structure using Sr0.8Bi2.2Ta2O9 and Sr0.8Bi2.2Ta2O9-BaZrO3 for ferroelectric gates

Characteristics of BaZrO₃ (BZO) modified Sr_0.8Bi_2.2Ta₂O₉ (SBT) thin films fabricated by sol-gel method on HfO₂ coated Si substrates have been investigated in a metal-ferroelectric-insulator-semiconductor (MFIS) structure for potential use in a ferroelectric field effect transistor (FeFET) type memory. MFIS structures consisting of pure SBT and doped with 5 and 7 mol% BZO exhibited memory windows of 0.81, 0.82 and 0.95 V with gate voltage sweeps between −5 and +5 V, respectively. Leakage current density levels of 10⁻⁸ A/cm² for BZO doped SBT gate materials were observed and attributed to the metallic Bi on the surface as well as intrinsic defects and a porous film microstructure. The higher than expected leakage current is attributed to electron trapping/de-trapping, which reduces the data retention time and memory window. Further process improvements are expected to enhance the electronic properties of doped SBT for FeFET.     

High-performance Pt/SrBi/sub 2/Ta/sub 2/O/sub 9//HfO/sub 2//Si structure for nondestructive readout memory

Metal-ferroelectric-insulator-semiconductor (MFIS) capacitors with 390-nm-thick SrBi/sub 2/Ta/sub 2/O/sub 9/ (SBT) ferroelectric film and 8-nm-thick hafnium oxide (HfO/sub 2/) layer on silicon substrate have been fabricated and characterized. It is demonstrated for the first time that the MFIS stack exhibits a large memory window of around 1.08 V at an operation voltage of 3.5 V. Moreover, the MFIS memory structure suffers only 18% degradation in the memory window after 10/sup 9/ switching cycles. The excellent performance is attributed to the formation of well-crystallized SBT perovskite thin film on top of the HfO/sub 2/ buffer layer, as evidenced by the distinctive sharp peaks in X-ray diffraction (XRD) spectra. In addition to its relatively high /spl kappa/ value, HfO/sub 2/ also serves as a good seed layer for SBT crystallization, making the proposed Pt/SrBi/sub 2/Ta/sub 2/O/sub 9//HfO/sub 2//Si structure ideally suitable for low-voltage and high-performance ferroelectric memories.     

A robust 8F2 ferroelectric RAM cell with depletiondevice (DeFeRAM)

This paper describes an area-penalty-free, leakage-compensated, and noise-immune 8F² cell design suitable for high-density, low-power ferroelectric RAM (FeRAM) generations. The new concept features a 1T1C ferroelectric memory cell containing an additional depletion device (DeFeRAM) controlled by the passing word line in a folded bit-line architecture. The depletion device permits the use of a common cell plate at intermediate voltage level. A highly reliable three-level word-line driver circuit design is discussed     

Ferroelectric DRAM (FEDRAM) FET with metal/SrBi2Ta2O9/SiN/Si gate structure

N-channel ferroelectric dynamic random access memory (FEDRAM) FETs with SrBi₂Ta₂O₉/SiN/Si structure were fabricated and characterized. The estimated switching time (t_sw ) of the fabricated FET, measured at applied electric field of 376 kV/cm, was less than 50 ns, which could be significantly reduced upon scaling. Its remnant polarization (2P_r) was measured to be about 1.5 μC/cm², which is more than one order of magnitude higher than that required for FEDRAM operation. The stored information retains more than three orders of magnitude of on/off ratio up to three days at room temperature, with little fatigue after 10¹¹ switching cycles     

Stack gate PZT/Al2O3 one transistorferroelectric memory

We have developed a single transistor ferroelectric memory using stack gate PZT/Al₂O₃ structure. For the same ~40 Å dielectric thickness, the PZT/Al₂O₃/Si gate dielectric has much better C-V characteristics and larger threshold voltage shift than those of PZT/SiO₂/Si. Besides, the ferroelectric MOSFET also shows a large output current difference between programmed on state and erased off state. The <100 us erase time is much faster than that of flash memory where the switching time is limited by erase time     

Superfluorescence dynamics in 130Te2

High-resolution analysis results are presented for Te₂ superfluorescence induced by short pulses from a YAG-pumped dye laser pulses, and a comparison is made with CW-laser-induced fluorescence observed by Fourier transform spectroscopy. A complete identification of the transitions involved, based on intensity measurements and supported by calculated Franck-Condon factors, has been achieved. The cooperative emission in Te₂ has been studied by time resolved spectroscopy as a function of density. With increasing density or temperature of the sample, the emission changes continuously from characteristic spontaneous emission to superfluorescence. The evolution of the time characteristics of the signal with the number of partners are presented and compared with the predictions of current theories     

Q-switching of the CO2laser

The laser pulses obtainable from aQ-switched CO2laser are calculated and compared with the results of a number of different techniques of performing theQ-switching. The continuously operating laser is considered first. The transition rates between the molecular vibrational states and their occupations are derived from the measured CW power. The laser tube was 1.9 meters long, had a diameter of 2.4 cm, and used flowing CO2-N2-He gas. For rapidQ-switching, maximum pulses of 4.5 mJ energy and 85 ns half width are predicted. Such pulses were observed with a rotating mirrorQ-switch. However, that technique has a limited pulse repetition rate and experiments on closely spaced pulses are difficult to interpret. A more flexible technique, which allows a much greater variation in the experimental parameters, is the use of a fast shutter to interrupt the laser beam in the cavity. While this switch is somewhat slower than the rotating mirror it does produce pulses of the same energy at repetition rates up to 5000 per second, and smaller pulses at any desired higher rate. From these measurements the upper and lower laser level lifetimes are deduced. They are found to agree well with the values obtained from the CW measurements.     

Fast response of PLZT pyroelectric detectors to megawatt CO2laser pulses

PLZT ceramic was used to make fast pyroelectric detectors. These detectors were able to: 1) respond to 10 ns events; 2) directly drive a 50 Ω CRO input with load resistors as small as 10 Ω; and 3) withstand irradiation with CO2laser pulses of 2 MW/cm²without cracking or need for repoling. The performance characteristics are described and are found to be in general agreement with a previously developed analysis.     

Developing a conductive oxygen barrier for ferroelectric integration

For high-density FeRAM memories, the 1T–1C architecture has been proposed. In this scheme, the ferrocapacitor (FeCAP), a metal\ferroelectric film\metal-like sandwich, is placed directly on top of polysilicon or tungsten plugs contacted to the underlying CMOS technology. Our ferroelectric material of choice, SrBi₂Ta₂O₉ (SBT), requires crystallization anneals ranging from 650 to 800 °C in full oxygen. The bottom electrode (BE) needs to be conductive, as well as an oxygen barrier to protect the underlying plug from oxidation. We have developed a BE stack combination consisting of Pt/IrO₂/Ir/Ti(Al)N. Each layer plays a critical role in the performance of the barrier. Issues such as thermal expansion, stress relaxation, grain growth, and oxidation can be critical in order to have a working bottom electrode. For capacitor formation, the complete stack is etched and covered by SBT. Since all layers are in contact with SBT, it is important to understand how the ferroelectric interacts with both the individual layers and the combined structure. In this paper, we will present our understanding of the chemical reactivity between SBT and the BE stack, which has been successfully engineered to prevent oxidation of the underlying plug.     

A sealed high-repetition-rate TEA CO2laser

A compact atmospheric pressure CO2laser utilizing a double-discharge technique has been constructed and operated at repetition rates to 100 pulses/s. With the addition of small amounts of hydrogen and carbon monoxide to give a gas mixture of He:N2: CO2:CO:H2= 69.3:11:15:4:0.7, sealed operational lifetimes exceeding2 times 10^{6}pulses have been obtained. Operating in this mode, the output energy density is about 8-9 J/l at repetition frequencies of 30-40 pulses/s for input energy densities of 60-70 J/l. The operation of the sealed laser has been studied by means of mass spectroscopic measurements of the gas mixture. It has been determined that sealed operation is possible as long as the oxygen concentration is kept below 1-2 percent. It has also been found that the addition of small amounts of H2and CO will keep the oxygen concentration below 2 percent by reducing the CO2decomposition, allowing sealed operation. The experimental results are compared to the predictions of a theoretical model in which neutral and negative-ion processes have been included. The calculations indicate that when small amounts of oxygen or water are present in the discharge the negative-ion population is significantly increased and the ratio of negative-ions to electronsN_{n}/N_{e}can approach values near unity. These are the conditions under which discharge arcing was found to occur. The model also predicts that the dissociation equilibrium of the CO2can be controlled by the addition of the above concentrations of hydrogen and CO.     

Switching in Nb—Nb2O6—Nb devices with doped Nb2O5amorphous films

Switching properties of Nb-Nb2O5-Nb devices with doped Nb2O5amorphous films were studied as a function of applied voltage and temperature. A continuous recording of the switching process made possible the determination of switching parameters as a function of small voltage increments. Using this technique, it is found that switching occurs within 2.5 mV with a delay time > 1 ms. A switching device of this type can support an input power of ∼1 W.     

An electrically modifiable synapse array composed ofmetal-ferroelectric-semiconductor (MFS) FET's using SrBi2Ta2O9 thin films

An array of metal-ferroelectric-semiconductor field effect transistors (MFSFETs) is fabricated on a silicon-on-insulator (SOI) structure using SrBi₂Ta₂O₉ as the gate insulator. It is demonstrated that each FET shows good characteristics as a nonvolatile analog memory due to partial polarization of the SrBi ₂Ta₂O₉ film and that the array operates as an electrically modifiable synapse circuit for carrying out the weighted sum operation in an artificial neural network     

Pulsed CO2laser with double modulation

For the pulse pumped CO2laser we used Q-switching to investigate inversion kinetics. The pump was synchronized with the Q-switching mirror. It was possible to obtain the Q-switch pulses at any phase of the pump pulse. The repetition rate was 50 Hz and the pump pulse duration was 4 or 10 ms. It was found that there is an optimum delay in switching the cavity Q-factor after the discharge started. The giant pulse intensity increased several times after the discharge was switched from CW to pulses. The inversion rise time was 1-2 ms and its lifetime was strongly dependent on the discharge current, due to plasma heating during the current pulse. Increasing the helium percentage in the discharge prolonged the inversion lifetime because of high thermal conductivity of helium gas. Gaseous BCl3was chosen to obtain the CO2laser giant pulses with bleachable filter. The vibronic frequency V3of the B¹¹Cl3molecules coincides with thePlines of the CO2laser. Rarer B¹⁰Cl3corresponds to less activeRlines. Pure BCl3did not give the giant pulses. Adding helium to the BCl3cell gave the pulses. The best results came from adding very small amounts of ammonia NH3because of the effective vibrational energy transfer between colliding BCl3and NH3molecules.     

Picosecond CO2 laser-pulse generation and amplification

The generation of 30-ps 10-μm laser pulses from an optical-free-induction-decay (OFID) 10-μm CO₂ laser system and the first realization of a UV-preionized high-pressure CO₂-laser-pulse amplifier with Ernst-electrode profiles for these ultrashort CO₂ laser pulses are described. For a given transverse electric discharge width, the Ernst profile makes it possible to position the UV-radiation sources nearer to the discharge than with the Rogowski or Chang profile. A more homogeneous preionization of the discharge area is possible. Thus, the amplifier works up to 15 bar with a laser-gas mixture of 5:5:90 (CO₂:N₂:He). The authors amplified 30-ps OFID 10-μm CO₂-laser pulses up to ten times by passing through the amplifier twice. Thus, 10-μm OFID pulse energies of 1 mJ were achieved. With the amplifier used as an oscillator, a maximum output energy of 600 mJ was obtained in longitudinal-multimode operation for 50-ns pulses     

Novel capacitor process using diffusion barrier rounded by Si3N4 spacer for high density FRAM device

A novel capacitor process was successfully implemented in 4 Mb FRAM device by developing a barrier layer rounded by Si₃N₄ spacer (BRS) scheme. Using this process, it is possible to eliminate an undesired barrier etching damage, which is a major role in degrading ferroelectric properties. The novel capacitor process was generated by etching an Ir barrier layer and rounding the barrier by a Si₃N₄ spacer before preparing Pb(Zr _1-xTi_x)O₃ (PZT) films. It was observed that uniform sol-gel derived PZT films were prepared on the patterned Ir substrate by using Si₃N₄ spacer, which provides a smooth edge of the patterned cell. The contact resistance between bottom electrode and polysilicon plug after full integration was monitored below 700 Ω per contact with contact size 0.6×0.6 (μm²). Compared to the ferroelectric capacitor damaged by barrier etching, the novel Pb(Zr_1-xTi_x)O₃ (PZT) capacitor exhibited a well-saturated Q-V curve. The fully processed novel capacitor having 1.2×1.2 (μm²) effective area displayed remnant polarization of 14 (μC/cm²) at an operating voltage of 3.0 V. The BRS ferroelectric capacitor showed a reliable retention property until 100 h at 125°C. Same state retention (Qss) was stable with time up to 100 h while opposite state retention (Qos) showed a log-linear decay rate at 125°C thermal stress     

Considerations on high peak powers from CO2and CO2- mixture lasers

A pulsed excitation was used to study high peak power generation in CO2and CO2-mixture flowing gas lasers. A rotating mirror was used as aQswitch with a variable delay between excitation pulse and mirror alignment. A variation of the time delay and multiple-exposure photographs can permit a measure of the population inversion versus time to be photographed. High-voltage pulses were used with higher than usual gas pressures to generate peak powers of about 30 kW. Peak power was shown to be primarily dependent on CO2partial pressure in CO2-He and CO2-N2mixtures.     

Subnanosecond (Er,Yb) Glass $Q$ -Switched Microlasers: 3-D Transient Modeling and Experiments

We present a new 3-D transient model and new experimental data for Q-switched eyesafe microlasers based on (Er,Yb) glass with a Co²⁺ :spinel saturable absorber. Subnanosecond pulses have been achieved for the first time using shorter, more highly doped laser glass. The model gives reasonable agreement with data for different Er ³⁺ concentrations and cavity lengths, using only published or measured input parameters     

Pulse train generation by soliton fission in highly nonlinear chalcogenide (As2S3) waveguide Bragg grating

The conversion of a 3.9 ps optical pulse into a train of six 450 fs pulses within a 37 mm length of a highly nonlinear chalcogenide (As₂S₃) waveguide Bragg grating is reported. Here, the initial pulse develops into a sixth-order soliton and is split into its six fundamental solitons through soliton fission. The very large optical nonlinearity and strong photosensitivity of As₂S₃ enables the use of pulses that are 25 x shorter than in previous experiments, and have 500 x less energy. The results are compared to numerical modelling using the nonlinear coupled mode equations and find satisfactory agreement between experiment and theory after accounting for imperfections observed in the grating.