Pyramids Produce Crack-free UV Lasers
UV lasers are used for various industrial and scientific tasks, including fluorescence spectroscopy, laser-ionization mass spectrometry, photolithography and material processing. These applications have driven the sales of nitrogen gas, HeCd, XeF excimer and triple harmonic generation Nd:YAG lasers, but these are bulky, inefficient, expensive to run and only cover a discrete range of wavelengths. In addition, HeCd and XeF lasers contain toxic substances.
Switching to UV-laser diodes could address all of these issues. However, it is not easy to extend the emission from 405 nm, the semiconductor laser wavelength of choice for Blu-ray and HD DVD players, into the UV. Far more aluminum is required in the cladding layers, along with aluminum rather than indium in the active region, and these changes lead to more defects in the epilayers, alongside greater strain, which can ultimately cause cracking.
This is not the only problem, as device development is also held back by substrate availability. AlN and AlGaN are the ideal choices, as they would allow crack-free growth of high-aluminum-content AlGaN layers, but these substrates are poor quality. The next best option is native GaN, but this is very expensive, so many devices are built on sapphire, SiC and silicon. However, these foreign platforms cause cracking and dislocations in the epilayers, due to significant differences in lattice constant and thermal expansion coefficients between the nitrides and the substrate.
These difficulties have hampered UV-laser diode progress and forced researchers to develop novel fabrication processes. These include the introduction of compliance layers for epitaxial lateral overgrowth (ELO) and methods described as a combination of low-temperature AlN interlayer technology and hetero-ELO. However, these techniques do not appear to produce epiwafers with crack-free regions, which is essential for high-yield laser diode manufacture.
It is possible to produce crack-free laser diode epiwafers on sapphire using a process called hetero-facet controlled (FAC) ELO, according to the work of a partnership between our team at the Central Research Laboratories of Hamamatsu Photonics KK and researchers at Meijo University, Japan (figure 1). This MOCVD-based technique involves the growth of a 2.5 µm thick GaN layer on sapphire. SiO2 stripes 3 µm wide, separated by 3 µm, are defined on this surface, so that subsequent GaN growth creates 6 µm high triangular-shaped "seeds" (figure 2). AlGaN is then grown laterally from the inclined facets of these seeds. Although dislocations are formed, they generally propagate in the horizontal direction and the dislocation density falls as more material is deposited.
The hetero-FACELO technique was initially applied to LEDs by researchers at Meijo University. However, we have extended this process to laser diodes by overcoming challenges associated with the growth of thicker and more complex structures that can provide optical and electrical confinement. This has led to the production of crack-free 2 inch material on sapphire with a total film thickness of 18 µm.
Laser stripes 5 µm wide were fabricated from these epiwafers, which span the wavelengths 355–362 nm. Threshold current densities for these emitters ranged from 7.3 to 19 kA/cm2. Room-temperature output under pulsed conditions peaked at 15 mW at a current density of 34 kA/cm2, and the typical polarization ratio (transverse-electric/transverse-magnetic) was 150.
The threshold current densities of these lasers are too high for commercialization. However, we have recently produced diodes with far lower threshold current densities that can deliver outputs of several tens of milliwatts under pulsed operation. These improvements resulted from optimization of the growth conditions for AlGaN, which increased the conductivity of the p-type AlGaN layers and decreased the number of defects acting as non-radiative centers.
We are continuing to improve our UV-laser diode characteristics and trying to produce novel devices operating at shorter wavelengths. In particular, we are aiming to produce a 337 nm emitter that offers an alternative source to the nitrogen gas laser.
S Kamiyama et al. 2002 Phys. Status Solidi A 192 296.
H Yoshida et al. 2007 Jpn. J. Appl. Phys. 46 5782.