Response to “Comment on ‘AlN/GaN double-barrier resonant tunneling diodes grown by rf-plasma-assisted molecular-beam epitaxy’ ” [Appl. Phys. Lett. 83, 3626 (2003)

Abstract

the bottom electrode’’ by recent additional tests with some other samples fabricated in those days, we hereby correct it. We reported in the paper 2 prior to the APL article of concern that the negative differential resistance of nitride resonance tunneling diodes ~RTDs! shows the asymmetric I ‐V curve and there the relationship with piezoelectric effect was suggested. After that, the other group has confirmed theoretically the asymmetry by the self-consistent simulation based on the ‘‘real-time tight-binding Green’s function.’’ 3 This is really an effect resulted mainly from a large piezoelectric field. These results definitely show that the RTD behavior in the GaN/AlN system could not be explained by a simple RTD scheme. In the comment, Belyaev et al. 4 has been insisting that we tried to explain the behavior of nitride RTD with a normal simple resonance concept in the APL letter, but this is misunderstood. In the letter reporting an early demonstration of GaN based RTD, we only gave the trial calculation of resonant levels as an approximation. We also predicted that the GaN/AlN RTD could not be explained completely with a simple model in the system with a large piezoelectric field. Also, Belyaev et al. infer that current‐voltage characteristics are sensitive to the way of voltage sweep, and this would be a phenomenon caused by a large piezoelectric field and electron trapping. We agree that this phenomenon deeply relates to the dislocation density in crystals and insufficient heteroboundaries. In this sense, regretfully, it is unclear that Belyae et al. and us have observed the completely same phenomenon, because between both samples, the RTD device structure is different and moreover the dislocation density included could be different. In the Belyaes sample, an asymmetric thickness combination was utilized for the bottom and top AlN barriers without undoped GaN spacers, which were placed on both sides of the double resonant region in our structure. The great difference is the absence of AlN multiple intermediate layers ~MILs! in their structure. We placed the AlN‐MILs underneath the RTD structure, i.e., just on metalorganic chemical vapor deposition-grown GaN templates to suppress propagation of screw-character threading dislocations to the upper RTD region. According to the previous papers of ours, 5,6 the insertion of AlN MILs functions to suppress the screw-character dislocations by three orders of magnitude. The suppressed screw character dislocation contributes to produce uniform heteroboundaries for molecularbeam epitaxy ~MBE!-grown AlN/GaN heterostructures. Belyaev wrote that we might also observe the current‐ voltage instability or the hysteresis characteristics, but it is not real. As described in the letter, we made the experiment with a single voltage scan mode based on the equipment character we used. During a long period after that, we did not have a chance to fabricate GaN-based RTD devices again,