Highly efficient UV emission with semipolar quantum wells
Sterilization techniques are commonly used to maintain the safety of air, water, food, and surgical instruments. Mercury lamps, for instance, are currently used to realize UV luminescence at a wavelength of 253.7nm, which is effective for germicidal irradiation.1 Although deep-UV emitters have practical potential for various fields beyond sterilization (e.g., medical research and high-density optical data storage beyond Blu-ray discs), current deep-UV sources have very low efficiencies and they contain toxic elements such as fluorine and mercury. To achieve benign and highly efficient deep-UV emitters, solid-state lighting devices—based on semiconductors—are therefore required.
Aluminum gallium nitride (AlGaN) has recently attracted attention as a deep-UV light emitter because of its large energy bandgap (3.4–6.0eV). Many AlGaN-based LEDs have been reported,2–6 but the external quantum efficiency (EQE) of these devices remains very low (about 10%). In addition, lasing oscillations at wavelengths less than 300nm have yet to be reported for laser diodes (LDs). For highly efficient deep-UV LEDs and LDs, the internal quantum efficiency (IQE)—a component of the EQE—must be drastically improved (especially at short emission wavelengths).7, 8 Two different methods have been proposed to improve the IQE of AlGaN quantum wells (QWs). In the first method, the non-radiative recombination centers (e.g., lattice defects) are reduced. In the second option, the radiative recombination probabilities are enhanced. The effectiveness of the first method depends strongly on the type of lattice defect, because not all lattice defects are active non-radiative recombination centers. Moreover, it is difficult, in practice, to drastically reduce the number of threading dislocations and native point defects (e.g., cation vacancies) through epitaxial growth techniques alone. On the other hand, the second method can be used to improve the IQE in a more direct manner, even if the number of non-radiative recombination centers is unchanged.
Conventional AlGaN QWs, which are grown on the (0001) plane, suffer from spontaneous and piezoelectric polarization fields that drastically decrease the radiative recombination probabilities. Non-polar or semipolar plane QWs, inclined from the (0001) plane, are a promising way to achieve shorter radiative recombination lifetimes of the carriers, because they can suppress the internal electric field. Challenges in growth conditions have restricted the realization of high-quality semipolar QWs thus far. Recently, however, we optimized metalorganic vapor phase epitaxy growth conditions to fabricate high-quality semipolar QWs on aluminum nitride (AlN) bulk substrates.9 In this work,10 we also demonstrate the growth of AlGaN QWs on other semipolar planes and demonstrate their optical superiority compared with conventional (0001) QWs.
The growth planes and photoluminescence (PL) spectra of our fabricated (0001) and semipolar AlGaN/AlN QWs (on AlN substrates) are depicted schematically in Figure 1. The semipolar ,
, and
planes are tilted at angles of 43, 58, and 75° from the (0001) plane, respectively. We have also confirmed that our fabricated QWs have atomically smooth surfaces and very abrupt QW interfaces.11 All our samples clearly show strong emissions close to the band-edge in the deep-UV region. It is also important to note that our semipolar QWs have rather narrow line widths.




The well-width dependences of the emission line widths are quantified in Figure 2 for our (0001) and semipolar QWs. We find that the widths of the emission lines for the semipolar QWs are much smaller than for that of the (0001) QW, over the entire well-width region. This is because weaker internal electric fields cause smaller energy fluctuations, even if (in principle), the Al composition and well-width fluctuations are the same.9 In fact, the estimated internal electric fields in the semipolar Al0.8Ga0.2N/AlN QWs are less than a third of that in the (0001) QW. Furthermore, the line-width difference is more pronounced for wider well widths because of the larger effect of the internal electric fields. Semipolar AlGaN QWs are therefore more attractive structures for deep-UV LDs than (0001) QWs because narrow emission line widths are desirable for such devices.

The relationship between temperature and the integrated PL intensity, under very weak excitation conditions (initial carrier density of about 1014cm−3), for our fabricated (0001) and semipolar QWs is illustrated in Figure 3. These results show that the thermal quenching of semipolar QWs is much weaker than for that of the (0001) QW. This is because of higher radiative recombination probabilities that arise from the smaller internal electric fields. Consequently, the emissions of our semipolar QWs are more than 100 times stronger than those of conventional (0001) QWs.

In summary, we have fabricated high-quality semipolar AlGaN/AlN QWs. These have small internal electric fields—less than one third of a conventional (0001) QW—which allows very narrow emission line widths and very strong emissions to be realized. Our study demonstrates that semipolar AlGaN QWs are attractive structures for highly efficient deep-UV LEDs and LDs. We are currently working to further improve the IQE of our devices. We are thus investigating methods to reduce the dominant non-radiative recombination centers in AlGaN QWs, and to enhance the radiative recombination probability.
The authors thank JFE Mineral Company, Ltd. for providing the AlN substrates.