Shengchang Chen

Advantages of AlGaN-Based 310-nm UV Light-Emitting Diodes With Al Content Graded AlGaN Electron Blocking Layers

In order to improve the performance of deep ultraviolet light-emitting diodes (UV LEDs), the effects of different electron blocking layers (EBLs) on the performance of AlxGa1-xN-based deep UV LEDs at 310 nm have been studied through a numerical simulation. The simulation results show that the adoption of EBLs is critical to improve the device performance. In comparison with a conventional structure using EBL with constant Al composition (0.7), the device structure with an Al-content graded AlxGa1-xN (from 0.9 to 0.4 in the growth direction) EBL possesses numerous advantages such as lower working voltage, higher internal quantum efficiency, and less efficiency droop under high-current injection. By detailedly analyzing the profiles of energy band diagrams, distributions of carrier concentration, and electron current density, the advantages of Al-content graded AlxGa1-xN EBL are attributed to the resulting lower resistivity, higher barrier for electron leakage, and simultaneously reduced barrier for hole injection compared with the conventional EBL with constant Al composition.

Improved Ohmic contacts to plasma etched n-Al0.5Ga0.5N by annealing under nitrogen ambient before metal deposition

The contact characteristics of Ti/Al and Ti/Al/Ni/Au contacts to as-grown, plasma etched and plasma etched + annealed in N2 n-Al0.5Ga0.5N epilayers were compared. After a rapid thermal annealing, both Ti/Al and Ti/Al/Ni/Au contacts to as-grown and plasma etched + annealed in N2 n-Al0.5Ga0.5N became truly Ohmic, whereas the contacts to plasma etched samples still remained rectifying. Surface atomic concentration analysis indicates the N vacancies resulting from plasma treatment act more as deep-level states rather than shallow donors. However, these deep-level states could be effectively removed by annealing the plasma etched n-Al0.5Ga0.5N under N2 ambient, and as a result, the Fermi level was elevated toward conduction band edge, facilitating the formation of Ohmic contacts.

Defect Reduction in AlN Epilayers Grown by MOCVD via Intermediate-Temperature Interlayers

In this work, significant reduction of the density of threading dislocations (TDs) in AlN epilayers grown on sapphire substrates via metalorganic chemical vapor deposition has been obtained by insertion of thin intermediate-temperature interlayers (IT-ILs). The growth temperature of the IT-ILs ranged from 750°C to 950°C after the initial growth at high temperature of 1200°C. Detailed characterizations were performed to understand the mechanisms of the reduction in dislocation density. It is found that the relatively low growth temperature of the IT-ILs can modify the originally two-dimensional (2D) growth mode to a three-dimensional (3D) growth process and creates a high density of small islands at the interface. During the subsequent growth of the high-temperature AlN layer, the AlN islands initially coalesce to form larger grains as the growth proceeds, and some TDs present in the previous AlN epilayers are found to bend near the interlayer, leading to dislocation merging and annihilation.

Defect reduction in Si-doped Al0.45Ga0.55N films by SiNx interlayer method

The dislocation density in AlGaN epitaxial layers with Al content as high as 45% grown on sapphire substrates has been effectively reduced by introducing an in-situ deposited SiNx nanomask layer in this study. By closely monitoring the evolution of numerous material properties, such as surface morphology, dislocation density, photoluminescence, strain states, and electron mobility of the Si-Al0.45Ga0.55N layers as the functions of SiNx interlayer growth time, the surface coverage fraction of SiNx is found to be a crucial factor determining the strain states and dislocation density. The dependence of the strain states and the dislocation density on the surface coverage fraction of SiNx nanomask supports the very different growth models of Al-rich AlGaN on SiNx interlayer due to the reduced nucleation selectivity compared with the GaN counterpart. Compared with GaN, which can only nucleate at open pores of SiNx nanomask, Al-rich AlGaN can simultaneously nucleate at both open pores and SiNx covered areas. Di...

Efficiency Improvement Using Thickness-Chirped Barriers in Blue InGaN Multiple Quantum Wells Light Emitting Diodes

The advantages of blue InGaN light-emitting diodes (LEDs) with thickness-chirped barriers in the active region have been investigated by using the Crosslight APSYS programs. The results show that the output power of the proposed LED is increased 80% and the efficiency droop is decreased from 59% in conventional LED to 28% at the current of 250 mA. Based on the analysis of electrical and optical characteristics, these improvements are mainly attributed to the change of electrostatic field in the active region by using thickness-chirped barriers. In the even-numbered barriers, the fields are increased, which gets rid of more seriously bended valence band and results in decreased barrier heights for hole transport in the active region. Furthermore, the direction of electrostatic field in the last barrier is reversed to along the drift direction of holes, which not only can lead to upbended conduction band to rise the barrier height for electron escape but also can accelerate holes to increase the hole injection current. As a result, electrons blocking and holes injection are enhanced, and in turn, the performance of the proposed LED is improved.

Enhancement of magneto‐optic recorded domain image by mathematical morphology

To observe the magneto‐optic (MO) recording domains precisely it is necessary to erase noise and overcome poor contrast. By applying image averaging and subtracting nonmagnetic contrast in digital image processing system, the sharp domain image can be obtained. This is the first time the domain image has been combined with morphological signal processing. By using mathematical morphology in the processing algorithms, the nonmagnetical contrast (background image) can be generated from a real image directly. Subtracting it from the real image, the target domains can be segmented and enhanced.