This will count as one of your downloads.
You will have access to both the presentation and article (if available).
In this paper, we discuss the changes in the electrical performance induced by operating time in hydrogen-terminated diamond MESFETs for high power and high frequency applications.
During the single stress step an increase of the current flowing in the sample is visible, possibly caused by the self-heating of the sample, as supported by temperature-dependent measurements, and by charge detrapping processes. In the full experiment the drain current was found to decrease, whereas the gate current remains below the detection limit.
In the characterization phase, we detected an increase in on-resistance, a decrease in the saturation current, a shift in the threshold voltage and a decrease in the transconductance peak. We found a time-dependent behavior for all these parameters, showing a further worsening up to 10 minutes after the end of the stress step. The time-dependent behavior is related to the creation of defects inside the structure and not to the self-heating, since the dynamic variation was found to increase as a consequence of stress, whereas the power dissipation decreases. The increase in the concentration of defects with activation energy of 0.30 eV was confirmed by ON-resistance and threshold voltage transient spectroscopy.
The variations in on-resistance and threshold voltage are not correlated in the full duration of the stress, suggesting that the generation of defects has (i) a different impact or (ii) a different generation rate in different parts of the device, with (iii) a possible role of the worsening of the contacts. Furthermore, a decrease in electroluminescence with higher magnitude than the decrease in drain current was found, compatible with an increased carrier-defect scattering.
Role of carbon in dynamic effects and reliability of 0.15 um AlGaN/GaN HEMTs for RF power amplifiers
The market for UV LEDs is experiencing a rapid growth, also driven by the need for effective and efficient disinfection systems. Before UV LEDs can be widely accepted by the market, they need to demonstrate a high reliability, with lifetimes of several thousands of hours. Several physical processes may limit the reliability of UVB and UVC LEDs, resulting in a loss in efficiency during long term operation.
This paper aims at discussing the most relevant processes that can lead to the degradation of UVB and UVC LEDs, with focus on: (i) instability of the electrical properties, which may result in gradual changes in the turn-on voltage of the devices during long-term operation. (ii) The generation of defects within the active region of the devices, with consequent increase in the Shockley-Read-Hall non-radiative recombination rate. Optical spectroscopy is found to be very effective for the identification of deep (midgap) traps during operation of the devices. (iii) trap states near the junction, with consequent impact on trap-assisted-tunneling of the current-voltage characteristics. (iv) the propagation of point defects, especially impurities, and accumulation of charges at heterointerfaces, that can reduce the carrier injection efficiency, thus leading to a decrease in the emitted optical power.
Deep defects have a fundamental role in determining the electro-optical characteristics and in the efficiency of InGaN light-emitting diodes (LEDs). However, modeling their effect on the electrical characteristics of the LED is not straightforward.
In this paper we analyze the impact of the defects on the electrical characteristics of LEDs: we analyze three single-quantum-well (SQW) InGaN/GaN LED wafers, which differ in the density of defects. Through steady-state photocapacitance (SSPC) and light-capacitance-voltage measurements, the energy levels of these deep defects and their concentrations have been estimated.
By means of a simulation campaign, we show that these defects have a fundamental impact on the current voltage characteristic of LEDs, especially in the sub turn-on region. The model adopted takes into consideration trap assisted tunneling as the main mechanism responsible for current leakage in forward bias.
For the first time, we use in simulations the defect parameters (concentration, energy) extracted from SSPC. In this way, we can reproduce with great accuracy the current-voltage characteristics of InGaN LEDs in a wide current range (from pA to mA).
In addition, based on SSPC measurements, we demonstrate that the defect density in the active region scales with the QW thickness. This supports the hypothesis that defects are incorporated in In-containing layers, consistently with recent publications.
In this paper we analyze the conduction properties, charge trapping and threshold voltage instability of normally-on β-Ga2O3 lateral MOSFETs for high power applications by means of threshold voltage transients. We found that a positive bias applied to the gate induces a rightward shift in the threshold voltage, caused by the trapping of electrons at border traps close to the semiconductor-dielectric interface.
The amount of trapped charge was investigated by an innovative fast-CV experimental setup and was found to follow a logarithmic kinetic in time, modeled by a generalization of the inhibition model that takes into account the effect of columbic repulsion in stress conditions.
Then, we developed a model for the gate conduction based on temperature dependent IG-VG characteristics. We detected that the gate current characterized in temperature and bias conditions similar to the ones used for the stress is dominated by Poole-Frenkel conduction assisted by a deep level at EC - 0.12 eV.
In this paper, we analyze the electrical behavior and the deep levels present in nitrogen-implanted gallium oxide Schottky barrier diodes annealed at increasing temperature from 800 °C to 1200 °C. In gallium oxide, nitrogen implantation is used in order to achieve controlled isolation of parts of the final device, and its stability and performance is therefore of high importance.
The high temperature annealing carried out after implantation causes a reduction in the leakage current flowing in the structure, confirming the feasibility of nitrogen implantation as isolation procedure and the annealing of the defects caused by the implantation process.
Repeated current-voltage measurements show the presence of an electron trapping process in the structure. The involved deep levels were investigated by means of isothermal transient spectroscopy tests, and both current and capacitance were used to monitor the trapping level in the devices. A model was developed to explain the full set of collected data on all the annealing temperatures, based on thermionic injection of electrons into an intermediate deep level and on charge injection into the space charge region.
By means of deep level transient spectroscopy experiments we analyzed the various defects present in the samples. Their concentration correlates with the annealing, decreasing at high temperature. All the detected deep levels are consistent with previous reports in the literature, and are attributed to gallium vacancies, native point defects and extrinsic defects.
We analyze the effects of stress in off-state condition at increasing drain bias. The main variations are an increase in on resistance (Ron) and in threshold voltage (Vth) and a decrease in the peak transconductance value, whereas the gate diode remains stable during the test. The Ron and Vth variations are correlated, suggesting a common degradation mechanism for both effects.
By means of sampled filling and recovery measurements, it is possible to highlight a trapping process occurring in off-state condition and dependent on the drain filling bias. This process is related to deep levels located both in the access regions and in the region under the gate, since it results in both Ron and Vth shifts. By means of temperature-dependent recovery measurements, a dominant thermal activation energy of 0.30 eV was found. The good fit quality according to the “stretched exponential” model indicates that the deep levels are extended defects or energy mini-bands.
The amplitude of the recovery transient collected after the same filling condition increases after stress at higher drain voltage, confirming that the deep levels causing the detected dynamic variation are increasing in concentration as a consequence of the stress. The good correlation between the amplitude and the Ron variation suggests that the detected variation in concentration is the root cause for the degradation of the device.
We submitted devices to 80 A/cm2 constant current stress, monitoring optical power and voltage by I-V and L-I characterization at each step. All the structures showed an increase in reverse leakage and low forward bias current, possibly due to trap-assisted tunneling ascribed to an increase in trap concentration. Reverse current was found to increase with the square root of stress time, indicating the presence of a diffusion process. The intensity of both QWs decreased during stress time; remarkably, degradation rate of reference QW (495 nm) was found to be much stronger for device B, where the 495 nm QW is closer to the p-side.
The defects responsible for degradation were characterized by Steady-State Photocapacitance measurements, indicating the presence of a ~2 eV level, whose signal changes during stress time. Shallower defects were detected by C-DLTS, that identified a level with 0.284 eV activation energy, possibly related to VN, whose concentration decreases during stress, due to defect annealing.
The results collected within this paper are explained by considering that stress promotes the diffusion of defects towards the active region of the devices. This mechanism results in a decrease in the SRH recombination lifetime, and in the subsequent increase in threshold current and drop in sub-threshold emission. An increase in the SRH rate next to the quantum dots can also reduce the injection efficiency into the QDs, thus inducing a drop in the slope efficiency of the lasers.
The study showed three main effects: (i) the decrease in the sub-threshold optical power, which shows two different slopes, that we ascribe to the regions where A, Shockley-Read-Hall (SRH) recombination coefficient, and B, radiative coefficient, dominate. (ii) a logarithmic decrease during the stress time of the characteristic temperature T0. (iii) the presence of a parasitic peak, with energy close to the main emission peak. This peak is ascribed to recombination in a second quantum well with slightly different energy, due to the different internal field. The intensity of this excited emission decreases during stress time, possibly due to a change in the injection efficiency.
We have also found an initial increase in the optical power at very low current levels, followed by a decrease with increasing stress time. This behavior is ascribed to an initial annealing, that favors the activation of magnesium, followed by an increment of the density of defects in the material caused by the stress.
View contact details
No SPIE Account? Create one
