Microelectronic Engineering最新文献

Hafnia or hafnium oxide is a high-$\kappa$ dielectric material with paramount importance in the realm of semiconductor devices. Recent advancements in 3D device structures require a few nanometer-thick conformal films on non-planar substrates. During the fabrication stage, the annealing process of thin films has been discovered to mitigate delamination issues at the film-substrate interface. However, it has been observed that the residual stress, which emerges as the film cools to room temperature, may lead to delamination. In this study, we propose an idealized atomistic model to mimic the critical region of a 3D-NAND structure, to get insights into the effect of thermal stress and delamination during the annealing of hafnia-made thin film. We employ molecular dynamics simulation using charge-optimized many-body potential (COMB) to perform heating and cooling simulations for different thicknesses of the hafnia layer. Our results suggest that, during heating, as the annealing temperature increases, the severity of delamination decreases. At extremely low thickness of the hafnia layer, delamination does not occur. However, significant delamination is observed during the cooling process, especially when the high temperature gradient is high.

In this work, a capacitor multiplier based on a Multiple Output-Voltage Difference Transconductance Amplifier (MO-VDTA) is built by using Arbel-Goldminz cells with extensive performance analysis. Considering the large chip area occupation of capacitors, capacitor multipliers are one of the most required analog building blocks in most of low frequency applications. In this respect, the obtained capacitor multiplier is tested in a 2nd order low-pass filter by changing the cut-off frequency from 2 kHz to around 12.4 kHz. The multiplication factor (denoted as “k”) of the proposed architecture can be adjusted electronically from 120 to 750 for approximately two decades, while the structure contains only a single active element with a base capacitance. Additionally, the multiplication factor can be safely increased by using additional transconductance stages in the MO-VDTA active block. In the performance analysis, post-layout results are provided in conjunction with process corners, Monte-Carlo analyses and experimental verifications on the basis of commercial off-the-shelf elements such as AD844 and LM13700s.

Medical images contain rich individual health information, making the protection of their privacy and security crucial. This study first proposes a novel multi-segment memristor based on a multi-segment linear function. Then, building upon the Sprott-B chaotic system, a mirror-symmetric memristor multi-scroll chaotic attractor (MMSCAs) is introduced by incorporating logic pulse signals and the novel multi-segment memristor. Dynamic analysis of MMSCAs is conducted from four aspects: equilibrium points, Lyapunov exponents and bifurcations, coexisting attractors, and complexity. Lyapunov exponents and bifurcation diagram analysis reveal rich dynamical behaviors in MMSCAs, including inverse period-doubling bifurcations, burst chaotic, transient chaotic, and offset boosting. MMSCAs exhibit periodic and chaotic attractors co-existing under different initial conditions, along with multi-stability and super multi-stability. Complexity analysis results indicate that MMSCAs possess higher complexity and better randomness compared to other memristor chaotic systems. The accuracy of the MMSCAs mathematical model is verified through circuit design and simulation, and the implementation of MMSCAs in the embedded domain is extended using the STM32 microcontroller. Finally, a new cryptographic system is designed by integrating MMSCAs with RNA computation and applied to medical image encryption. The security of the cryptographic system is evaluated through key space and sensitivity, histogram, and correlation, while its robustness is evaluated through resistance to cropping and noise. The analysis results demonstrate high security and strong robustness of the cryptographic system, offering a novel solution for the protection of medical image information.

The characterization of the self-heating effect (SHE) has been an important research topic in advanced technology, but the existing characterizations are few and the characterization process is relatively complex. In this research, a SHE characterization model is established based on the relationship between output transconductance variation ($∆g_m$), gate source voltage (V_GS) and temperature variation ($∆T$) caused by SHE through machine learning, and then the model is validated by theoretical analyses and experimental simulation. The characterization model is capable of directly calculating the $∆T$ caused by SHE during I - V testing and simplifying the SHE characterization steps while ensuring characterization accuracy ($∆T$ difference < 1 °C), thus saving costs. It is also found that the model can expand the characterization range (V_GS: 0.3–0.7 V) of SHE and conducts quantitative characterization with model calculation under different V_GS, realizing a high characterization resolution of V_GS: 0.01 V. The circuit level application proves that the method can be effectively applied to the characterization of the SHE and solves the problem of the characterization of the circuit level SHE.

We introduce a detailed design and fabrication process of Silicon microcantilever arrays for biomolecular detection in liquid environment, utilized with laser readout. We present typical fabrication problems and provide related solutions to obtain high quality resonators via a robust, reproducible and high-yield process. Sensors in these arrays are individually functionalized with self-assembled chemical monolayers exposing various pH-active end-groups into solution. Dynamic-mode controlled frequency measurements in varying pH solutions result in stress-induced change of the sensor spring constant. pH changes in the solution lead to deprotonation of exposed functional chemical groups at high pH and the repulsive charges induced strain is proportional to the quantity and confinement of charges at the sensor interface. These built-up strains that affect the mechanical stiffness can be reversibly relaxed when exposed again to low pH environments.

In this work, a scandium-doped aluminum nitride (ScAlN)-based piezoelectric breathing mode dual-ring resonator with high temperature stability is presented. The designed resonator consists of two identical rings and a coupling straight beam in between. A combination of highly doped silicon and composite structure using silicon oxide is implemented to improve the frequency-temperature stability of the resonator. The dual-ring resonator is fabricated based on a ScAlN-based thin-film piezoelectric-on‑silicon (TPoS) platform. The measurement results show that the fabricated dual-ring resonator has a loaded quality factor ($Q_l$) of 6889 and an insertion loss of 13.898 dB at its resonant frequency of 16.766 MHz, corresponding to a motional resistance of 395 $\Omega$, and an unloaded quality factor ($Q_{\mathrm{un}}$) of 8681. The resonator's $Q_{\mathrm{un}}$ is almost constant within the pressure range of less than 300 Pa, indicating a good process tolerance in the vacuum packaging process. With the aid of the passive temperature compensation, the reported resonator exhibits an overall frequency variation of less than ±70 ppm over the entire temperature range of 20 °C to 105 °C, which agrees well with the predicted value obtained by finite element method (FEM) analysis. Moreover, Allan deviations of the resonator-based oscillator frequency are collected.

Laser Induced Reverse Transfer (LIRT) is a versatile technique as a single-step deposition method allowing the localized transfer of a variety of different metals and polymers on transparent, ultra-thin and stretchable substrates. Also referred to as laser induced backward transfer (LIBT), the process can be manipulated to transfer material from bulk materials to transparent targets, providing a direct method potentially sustainable to generate microelectronic circuitry. In this work, a fs-pulsed UV laser (343 nm) was employed for the first time to transfer electrically conductive copper tracks and layers from bulk Cu in the form of sheet metal onto ultra-clear soda lime glass slides with sub-micrometric thickness. The process development started from the selection of the materials for adequate energy transfer between the beam source and the donor/receiver combination. In the single-track study, the effect of donor/receiver gap was analyzed while tracks ranges with 5 to 233 nm thickness and 7 to 41 μm average width were produced. Based on the results, multi-track layer deposition was assessed by varying the overlap between the tracks. Functional demonstrator cases were produced. The work confirms the suitability of LIRT as a direct approach to create microelectric circuitry by using readily available and sustainable bulk Cu material.

Al_xSc_1-x thin films have been studied with compositions around Al₃Sc (x = 0.75) for potential interconnect metallization applications. As-deposited 25 nm thick films were x-ray amorphous but crystallized at 190 °C, followed by recrystallization at 440 °C. After annealing at 500 °C, 24 nm thick stoichiometric Al₃Sc showed a resistivity of 12.6 μΩcm, limited by a combination of grain boundary and point defect (disorder) scattering. Together with ab initio calculations that found a mean free path of the charge carriers of 7 nm for stoichiometric Al₃Sc, these results indicate that Al₃Sc bears promise for future interconnect metallization schemes. Challenges remain in minimizing the formation of secondary phases as well as in the control of the non-stoichiometric surface oxidation and interfacial reactions with underlying dielectrics.

Device stability is one of the key parameters for transistor applications. To improve the stability of Indium oxide (In₂O₃) after a long-time gate bias, a synthetic solution of hafnium chloride (HfCl₄) and indium nitrate (In(NO₃)₃∙xH₂O) reagents were used to obtain 0% to 5-at.% Hf doped In₂O₃ thin-film transistors. With the increase of Hf doping concentration, oxygen vacancies and residual hydroxyl groups continue to decrease, suppressing the carrier concentration and influencing the trap state density of In₂O₃. The sub-threshold slope (SS) 0.78 V·dec⁻¹ for the undoped In₂O₃ transistor in this work is a typical value. When the dopant dose is up to 5-at.%, SS decreases to 0.32 V·dec⁻¹. According to the proportional relationship between SS and the density of trap states, it shows that the density of trap states in the dielectric layer and the semiconductor/dielectric interface SS is greatly reduced after 5-at.% Hf doping. The probability of the charge being trapped is dropped as well. At the same time, under the doping of Hf, the transistor exhibits a very small threshold voltage shift. Especially at the dopant dose of 5-at.%, the transfer characteristic curve hardly shifts. This work demonstrates an In₂O₃ transistor with high bias stability by doping method.

This work demonstrates the potential use of Cu-Sn-In metallurgy for wafer-level low-temperature solid-liquid interdiffusion (LT-SLID) bonding process for microelectromechanical system (MEMS) packaging. Test structures containing seal-ring shaped SLID bonds were employed to bond silicon and glass wafers at temperatures as low as 170 °C. Scanning acoustic microscopy (SAM) was utilized to inspect the quality of as-bonded wafers. The package hermeticity was characterized by cap-deflection measurements and evaluated through finite element modelling. The results indicate the bonds are hermetic, but residual stresses limit the quantitative analysis of the hermeticity. The microstructural studies confirm the bonds contain a single-phase intermetallic Cu₆(Sn,In)₅ that remains thermally stable up to 500 °C. This work shows Cu-Sn-In based low-temperature bonding method as a viable packaging option for optical MEMS or other temperature-sensitive components.