Spectral power distribution (SPD) is the radiation power intensity at different wavelengths, containing the most basic photometric and colorimetric performance of the illuminant, which is able to predict the lifetime of LEDs. This paper proposes an SPD model assisted by machine learning algorithms to detect the early failure of white LEDs. The SPD features of 3W high-power white LEDs were firstly extracted by the statistical models of Gaussian, Lorentz, and Asym2sig functions. An unsupervised learning method, principal component analysis (PCA), was then used to reduce the extracted features parameters’ dimensions. Next a K-nearest neighbor (KNN)-based method was used to detect LEDs’ anomalies by dividing the main cluster into groups, and estimating the distance from the center of mass of each cluster to the test point. The results showed the following: (1) for selected white LEDs, the Asym2sig function has a better fitting result than Gaussian and Lorentz functions; (2) machine learning methods can significantly assist in LED anomaly detection and can decrease the amount of anomaly detection time to 789.6 h, compared to the 1311 h when lumen maintenance degradation reaches 70% as required by IES TM21.
A distributed Bragg reflector (DBR) was designed and fabricated to reflect the excess blue light to both increase the utilization of the blue light from micro-LED and improve the color purity of the red and green light from QDs. Red and green QDs arrays on glass substrate were presented by transfer printing via polydimethylsiloxane (PDMS) stamp. The red QDs transferred on the DBR as color-converting layer was then attached with the blue micro-LED, and the red QDs down-converted the blue emission from the micro-LED to the red emission. These results provided the promising potential of transfer printing QDs as color-converting layer to realize full-color micro-LED display.
The well-known "Green gap" issue related to LEDs efficiency significantly limits the full color applications in both display and lighting. In order to solve this problem, in this study, a new green LED structure is proposed, which includes a regular short wavelength MQWs for the electroluminescence (EL) emission, and a green photoluminescence (PL) color conversion region, so called EP-LED. The shorter wavelength EL region is responsible for carrier injection and recombination, which suffers from a much weaker QCSE, and the longer wavelength PL region is responsible for high quantum efficiency color conversion, whose strain field has no effect on carrier injection. With this novel hybrid active-region design, the green EP-LED shows more than 30% improved WPE over the conventional one at current density of 50 A/cm2, and 1/3 suppressed blueshift in wavelength in the current range of 1 mA to 100 mA, as well as the much-reduced efficiency Droop. It is very encouraging that the green or long wavelength EP-LEDs will meet many new needs in future display and lighting applications.
Micro LED (light-emitting diode) is maturely used in the fields of space display, medical detection and visible light communication due to its unique features of miniaturization, low power consumption, fast response speed and high resolution. However, since the driver circuits are usually composed of Si devices, a large number of Micro LED pixels must be transferred from their GaN substrates to bond with Si field-effect transistors (FETs) by mass transfer techniques. Since there are certain technical bottlenecks in the mass transfer technology of Micro LED, relevant scholars at home and abroad have successively put forward the idea of integrating LED and field effect transistor driving circuit on the same substrate, so as to avoid the issues of low yield and poor accuracy. This paper summarizes the research on monolithic integrated active driving of LEDs and Micro LEDs at home and abroad in recent years. These researches include: monolithic integration of high mobility field effect transistors (HEMTs) and LEDs, metal oxide field effect transistors (MOSFET) and LED monolithic integration and thin film field effect transistor (TFT) monolithic integration with Micro LED.
With the development of chip scale package (CSP) light-emitting diode (LED) technology, this technology is more and more widely concerned and applied in semiconductor lighting industry. When LED is used in the lighting on education, its characteristics such as color rendering index (CRI) and color temperature are particularly important. This paper introduces one design of CSP LED with 630 nm K 2 SiF 6 (KSF) fluoride phosphors in the mixed phosphors layer. The spectral characteristics, color parameter drift and thermal characteristics of the CSP LED are studied. The CSP technology achieve high CRI (>95) along with high luminous efficacy (158.81 lm/W). The results show that: 1) The driving current and temperature have different effects on the optical, color parameter drift and thermal properties of the device; 2) The added KSF fluoride phosphors expand the luminescence spectrum distribution and improves the luminescence intensity, especially in the red band (600 nm~650 nm); 3) The CRI is improved without significant reduction in luminous efficiency, and it achieved the trade-off between CRI and luminous efficiency.
In this paper, a new structure named plasma spreading layer (PSL) is introduced into planar SiC MOSFETs to improve the short-circuit robustness. The short-circuit performance under 400V, 600V, and 800V bus voltage is tested. And two failure modes are discovered. The 3D TCAD simulations, Emission Microscope (EMMI), and Focused ion beam (FIB) are used to investigate the short-circuit failure mechanisms. Under 400V and 600V bus voltage, the interlayer dielectric between the source and the gate is damaged due to the high temperature generated in the short-circuit process, and the melted source metal Al is connected to the gate polysilicon, causing short circuit of the gate and source electrodes. Compared with commercial devices, the short-circuit capability of the device with PSL is improved by 3 to 7μs under 400V bus voltage. Under 800V bus voltage, the thermal power is almost twice that of 400V, which reduces the short circuit withstanding time (SCWT) of the device rapidly to just over 5µs. Due to such high instantaneous power, the device failure is characterized by the short circuit between the three electrodes, and thermal runaway is the cause of failure in this situation.
As the device size continues to shrink, the gate oxygen layer in MOS devices continues to become thinner, which leads to problems such as increased leakage current in conventional Si MOS devices. SiC, an important semiconductor material used in high-power devices, is not immune to the problem of premature breakdown due to the gate oxygen layer. Based on this, this paper adopts the innovative approach of replacing the conventional SiO 2 as the gate oxygen layer with the high-k material, hafnium dioxide (HfO 2 ), and continues the in-depth study on the electrical characteristics of the gate oxygen structure. In this paper, hafnium dioxide (HfO 2 ) thin films were deposited on SiC and Si substrates using the atomic layer deposition(ALD) method. The metal-insulated semiconductor (MIS) structures with Al as the upper and lower electrodes were prepared, and the leakage currents and breakdown voltage characteristics of the MOS capacitors on the two different substrates were investigated. The leakage current density of the MOS capacitor structure with SiC substrate is 10 -11 A/cm 2 ,and the breakdown voltage is about 32 V. Based on the capacitance-voltage (C-V) measurements, the frequency dependence of the movable charge at the interface of the MOS capacitor structure is investigated. The lowest interfacial defect density (D it ) of the MOS capacitive structure with SiC substrate is only of the order of 10 10 eV -1 cm -2 , while the movable charge N eff of the oxide layer of this sample is also controlled to the order of 10 13 cm -2 . It is worth mentioning that the HfO 2 /SiC structure performs better than the HfO 2 /Si structure in these electrical parameters, and the sample designed in this paper also performs better in electrical parameters than previous studies. These results undoubtedly demonstrate that the combination of High-k material (HfO 2 ) and SiC in the MOS gate oxygen layer is a promising research topic.
Silicon Carbide (SiC) power module has the advantages in high-frequency and high-power applications due to the superior properties of fast switching and low loss under condition of high-current output. However, as the switching speed increases, the abnormal transient waveforms are developed, so that the fast switching performance is restricted eventually. In this paper, two kinds of SiC power modules are fabricated, and the variation trend in turn-on transient waveforms of SiC power modules with increasing turn-on speed is investigated. A correlation existing between the variation trends of drain-source voltage and gate-source voltage waveforms with rising turn-on speed is disclosed by experiments. A criterion for the allowable maximum turn-on speed of SiC power module is also proposed based on this correlation. The findings in this paper provide a new reference for enabling the fast-switching performance of SiC power module in high-current applications.
The vast majority of sEMG acquisition systems use wired transmission, which makes the signal highly susceptible to power-frequency interference. In addition, the sEMG signal is relatively weak, and preprocessing such as amplification, filtering, and denoising is crucial. In this paper, a low-noise and high-precision multi-channel active surface electromyography acquisition device that supports up to 16 channels of acquisition is designed and developed, which is used to collect muscle signals of various upper limb movements. The system uses active electrodes to achieve a higher signal-to-noise ratio (SNR), which can be improved by 64.04% compared to passive electrodes. And the integrated functions of active electrode array drive, weak signal acquisition and right leg drive create hardware conditions for the subsequent research on surface EMG signal processing.
As a compelling part of the next-generation wireless communication technologies, visible light communication (VLC) enables the use of light bulbs as transmitters while addressing the shortage of spectrum resources. To achieve both lighting and communication applications, a large field of view (FoV) is desirable. Laser-based VLC technology has attracted increasing attention due to its high data rate but visible laser light communication (VLLC) system with a large FoV has yet been well studied. In this work, we design a VLLC system using a white laser transmitter and a silicon photomultiplier (SiPM) receiver with ultra-wide FoV reaching 180°. The preliminary experimental results show that the system enables transmission data rates over 500Mbps. This work suggests that the demonstrated VLLC system is promising for wide-coverage LiFi applications.
Electric-thermo-mechanical interaction has significant effect on the lifetime of a semiconductor power module. It is crucial to conduct collaborative simulations during the design phase to optimize the electrical and thermal performance of the power module and to minimize the thermomechanical stress from lifetime consideration. This study highlights the challenges of power module design in electronic, thermal, and mechanical co-simulation and presents a methodology in characterizing power module’s transient thermal response through transient CFD simulation and introduces a method of conducting mold flow analysis of power module by general CFD software. A concept for conducting multiphysics simulation is also proposed for power modules design using a single integrated general computer-aided engineering (CAE) software.
Micro-LED is a promising candidate of the next generation display technology. The size effect caused by sidewall surface traps is a serious issue that restricts its industrialization progress. However, the current research on the size effect of micro-LED is limited to the reduce of quantum efficiency and luminous intensity with decreased pixel size, but there is few research on the mechanism of the transient switching characteristics of Micro-LED affected by sidewall surface traps. In this paper, the size effect of GaN based micro-LEDs is studied by TCAD simulation. The decrease of quantum efficiency is attributed to the depletion effect of the sidewall traps induced by dry etching, the enhancement of SRH non-radiative recombination, and the lateral current component caused by sidewall traps capturing carriers. In addition, the sidewall trap density and trap energy level of micro-LEDs with 10 μm pixel size were varied to study their influence on the transient switching characteristics of micro-LEDs. This work provides a design guideline for the process control of micro-LEDs.
Epitaxial lateral overgrowth (ELOG) is one of the effective means to improve crystal quality. In this work, gallium nitride (GaN) was grown using a hexagonal patterned graphene as mask via ELOG by metal organic chemical vapor deposition (MOCVD). Two kinds of multi-layer graphene were adopted on GaN/Sapphire compound substrate respectively, one was graphene by plasma enhanced chemical vapor deposition (PECVD), another was wet transferred graphene by chemical vapor deposition (CVD) growth on copper foil. We compared the ELOG behavior of epitaxial GaN and found that the integrity and homogeneity of PECVD graphene made GaN preferentially nucleate and completely cover in the window region. However, because of the large number of defects and winkles introduced during the transfer process, GaN nucleated both in the window region and mask region on the transferred graphene substrate, consequently, epitaxial GaN on transferred graphene was not a ELOG mode. At the same time, we found that both PECVD graphene and transferred graphene was gradually decomposed during the growth process of GaN, but PECVD graphene could still act the mask role. Finally, ELOG GaN on PECVD graphene was obtained and the threading dislocations of GaN on PECVD graphene were greatly reduced and the dislocation distribution was similar to the window pattern. In contrast, the transferred graphene did not act as a mask, and the threading dislocation distribution of epitaxial GaN did not show any regularity.
With the advancement of semiconductor lighting technology and the popularization of human-centred lighting concepts, colour preference and colour discrimination of lighting are now at the forefront of colour quality research of light source. However, individual case in psychophysical research have the inherent drawback of being inapplicable in multiple scenarios, and thus the two dimensions of colour quality of light source have not yet been effectively quantified. To address this issue, the team has successfully constructed the most accurate colour preference model MCPI, and the colour discrimination model CDM at this stage, based on a meta-analysis of multiple sets of psychophysical research data, which have also yielded convincing results in terms of visual mechanism. In addition, the CDM-E prediction model has been constructed by extending the CDM model with illumination factors.
The degradation behavior under repetitive unclamped inductive switching stresses on the subsequent time-dependent dielectric breakdown (TDDB) of commercial SiC power MOSFETs are investigated. It is reported that repeated avalanche stresses can cause changes, including Vth (the threshold voltage) and Ron (on-state resistance) but not Igss (gate leakage current). However, the results show a significant decrease in the TDDB lifetime after 180k avalanche cycles, which suggests the degradation of the gate oxide. We attribute these results to defects at the SiC/SiO2 interface, which are caused by hot hole injection and trapping into the gate oxide above the channel and JFET region. The subthreshold curve analysis reveals that the density of the interface states changes during aging experiments, which further confirms our view. Meanwhile, interface states should be responsible for the increase in Ron, which was different from the previously research.
With the increase of power consumption and running speed, the junction temperature of three-dimensional stacked memory devices has increased. It eventually leads to chip overheating and may cause device failures. It is essential to analyze and optimize the thermal performance of the package. In this paper, the finite element analysis method is used to study the influence of the package module on the chip junction temperature and thermal resistance for the typical six-layer stacked memory chip packaging structure. The main influencing factors are investigated. In addition, the thermal characteristics of the stacked packaging structure are optimized by using the orthogonal experimental design method. After further optimization, the junction temperature decreases by 7.17% and the thermal resistance decreases by 11.59%.
Abnormal phenomenon that on-resistance of JBS diode increases with forward bias has been studied in this work. It is found that the forward bias is mainly applied to the p-type Schottky junction, resulting in a slower electric potential increasing rate of p-GaN than n-type channel between adjacent p-grid. Therefore, the lateral PN junction at channel is reversely biased and the effective conduction area of JBS diode reduces, which results in the increase of on-resistance. JBS diodes with different Schottky barrier height at p-type Schottky contact are also simulated to verify our analysis. The decrease of barrier height at p-type Schottky junction results in that the anode voltage is mainly applied to the PN junction, which can shorten the lateral depletion region and decrease the on-resistance of JBS diode. The simulation results in this paper are useful for providing reference advice for following experiments.
This paper investigates high-voltage light emitting diodes (LEDs) driven linearly by three-phase AC power, which features high reliability, cost effectiveness, all-solid-state components, and a minimum number of driving elements. To verify the optical and electrical properties of high-voltage LED, a high-voltage LED lighting system driven linearly by three-phase AC power was built as a prototype in this paper. At the rated voltage of 380V, the LED lighting system has input power of 145W, total luminance flux of 18500lm, luminous efficacy of 127lm/W, correlated color temperature (CCT) of 5617K, and color rendering index (Ra) of 70.6, power factor (PF) of 0.95. Further, the experimental results show that stable output luminance, high luminous efficacy, high PF, and flicker-free are achieved within 380V±10% voltage fluctuation.
In this paper the SiC MOSFET with different channel length was fabricated and were irradiated with Cobalt-60 γ-ray under different bias stress. Synergetic effects of structure and gate bias-stress design is experimentally studied. The turn-on voltage in 3 rd quadrant could determine the channel leakage and indicates the state of the device. Long channel length could inhibit the conduction under large dose of radiation. The device failure mechanism under high radiation dose is also analyzed.
