The physical properties of InN crystals are known rather poorly, since the existing growth techniques have not produced epitaxial layers of good quality [1,2]. Even a key parameter of InN -the band gap E g -has not been firmly established so far. E g values of 1.8 eV to 2.1 eV have usually been estimated from the absorption spectra obtained on polycrystalline and nanocrystalline hexagonal InN [3][4][5][6]. No data on the band-to-band photoluminescence (PL) of InN are available in the literature. Recently an improved growth technique has made it possible to obtain single-crystalline InN layers [7]. Optical measurements on these InN layers have shown some strong differences from absorption data reported earlier [8]. In the present work the electronic structure of singlecrystalline InN layers was carefully studied by means of optical absorption, PL, and photoluminescence excitation (PLE) spectroscopy as well as by ab initio calculations. Our results revealed for hexagonal InN a band gap of about 0.9 eV, which is much smaller than the values of 1.8 eV to 2.1 eV reported previously.Single-crystalline InN epilayers were grown on (0001) sapphire substrates either by plasma-assisted molecular-beam epitaxy (PAMBE) [7] or metalorganic molecular-beam epitaxy (MOMBE) [9] and were characterized by many techniques. Only hexagonal symmetry, with no traces of other polymorphs, was established by X-ray analysis in all the samples. For characterization the symmetric (0002) and asymmetric ð11 2 24Þ Bragg reflexes were used. From these data the lattice constants in the InN layers were found to be c ¼ 5.7039 A and a ¼ 3.5365 A. The narrow profiles of q and q-2q scans at the (0002) reflex (250-300 arcsec and 50-60 arcsec, respectively) indicate a good crystalline quality. Polarized Raman spectra of InN show agreement with the selection rules for the hexagonal symmetry. The Raman phonon line widths correspond to a well-ordered crystal lattice [9,10]. Atomic force microscopy measurements did not reveal any columnar structure in the samples studied. According to the Auger data, the oxygen concentration did not exceed 0.1%. The Hall concentration of electrons n ranged from 9 Â 10 18 to 1.2 Â 10 19 cm -3 in the best samples, and their mobility was found to be as high as m $ 1900 cm 2 V -1 s -1 .The absorption coefficient a(w) for PAMBE-and MOMBE-grown InN samples at 300 K is shown in Fig. 1. The layer thickness was measured by means of scanning electron microscopy. The aðwÞ spectra were calculated from the transmission spectra with corrections for multiple reflections. It can be seen that the edge absorption rapidly reaches values of a(w) > 5 Â 10 4 cm À1 , which is typical of direct band-gap crystals. The inset in Fig. 1 shows that the absorption coefficient can be described by the relation a(w) $ ( hw -E g ) 1/2 usually applicable to allowed direct interband transitions. From the measurement of the absorption edges it can be concluded that the E g phys. stat. sol. (b) 229, No. 3, R1-R3 (2002)

The first- and second-order Raman scattering and IR reflection have been studied for hexagonal InN layers grown on (0001) and (11̄02) sapphire substrates. All six Raman-active optical phonons were observed and assigned: E2(low) at 87 cm−1, E2(high) at 488 cm−1, A1(TO) at 447 cm−1, E1(TO) at 476 cm−1, A1(LO) at 586 cm−1, and E1(LO) at 593 cm−1. The ratio between the InN static dielectric constants for the ordinary and extraordinary directions was found to be ε⊥0/ε∥0=0.91. The phonon dispersion curves, phonon density-of-state function, and lattice specific heat were calculated. The Debye temperature at 0 K for hexagonal InN was estimated to be 370 K.

On the page 117407-3, left column, the text in the last paragraph should be written as follows. ''Such a small L m value means that the respective clusters are elongated along the electric field vector of the incident light, like the interface indium inclusions. In this case, the depolarization is small, and the resonance energy is the lowest among those possible [16]. This type of clusters determines the effective absorption edge in InN. The resonances of the clusters with L m 0:33 are higher in energy than the absorption edge in InN, so the prolate intercolumn inclusions do not affect the effective band edge.'' This correction affects the speculation concerning the type of the In clusters involved, but changes none of our main results.We thank T.

We report on the thorough investigation of light emitting diodes (LEDs) made of core-shell nanorods (NRs) with InGaN/GaN quantum wells (QWs) in the outer shell, which are grown on patterned substrates by metal-organic vapor phase epitaxy. The multi-bands emission of the LEDs covers nearly the whole visible region, including UV, blue, green, and orange ranges. The intensity of each emission is strongly dependent on the current density, however the LEDs demonstrate a rather low color saturation. Based on transmission electron microscopy data and comparing them with electroluminescence and photoluminescence spectra measured at different excitation powers and temperatures, we could identify the spatial origination of each of the emission bands. We show that their wavelengths and intensities are governed by different thicknesses of the QWs grown on different crystal facets of the NRs as well as corresponding polarization-induced electric fields. Also the InGaN incorporation strongly varies along the NRs, increasing at their tips and corners, which provides the red shift of emission. With increasing the current, the different QW regions are activated successively from the NR tips to the side-walls, resulting in different LED colors. Our findings can be used as a guideline to design effectively emitting multi-color NR-LEDs.

No abstract