We report steady-state and time-resolved photoluminescence ͑TRPL͒ measurements on individual GaN nanowires ͑6-20 m in length, 30-940 nm in diameter͒ grown by a nitrogen-plasma-assisted, catalyst-free molecular-beam epitaxy on Si͑111͒ and dispersed onto fused quartz substrates. Induced tensile strain for nanowires bonded to fused silica and compressive strain for nanowires coated with atomic-layer-deposition alumina led to redshifts and blueshifts of the dominant steady-state PL emission peak, respectively. Unperturbed nanowires exhibited spectra associated with high-quality, strain-free material. The TRPL lifetimes, which were similar for both relaxed and strained nanowires of similar size, ranged from 200 ps to over 2 ns, compared well with those of low-defect bulk GaN, and depended linearly on nanowire diameter. The diameter-dependent lifetimes yielded a room-temperature surface recombination velocity S of 9 ϫ 10 3 cm/ s for our silicon-doped GaN nanowires.

Analysis of steady-state and transient photoconductivity measurements at room temperature performed on c-axis oriented GaN nanowires yielded estimates of free carrier concentration, drift mobility, surface band bending, and surface capture coefficient for electrons. Samples grown ͑unintentionally n-type͒ by nitrogen-plasma-assisted molecular beam epitaxy primarily from two separate growth runs were examined. The results revealed carrier concentration in the range of ͑3-6͒ ϫ 10 16 cm −3 for one growth run, roughly 5 ϫ 10 14 -1ϫ 10 15 cm −3 for the second, and drift mobility in the range of 500-700 cm 2 / ͑V s͒ for both. Nanowires were dispersed onto insulating substrates and contacted forming single-wire, two-terminal structures with typical electrode gaps of Ϸ3-5 m. When biased at 1 V bias and illuminated at 360 nm ͑3.6 mW/ cm 2 ͒ the thinner ͑Ϸ100 nm diameter͒ nanowires with the higher background doping showed an abrupt increase in photocurrent from 5 pA ͑noise level͒ to 0.1-1 A. Under the same conditions, thicker ͑151-320 nm͒ nanowires showed roughly ten times more photocurrent, with dark currents ranging from 2 nA to 1 A. With the light blocked, the dark current was restored in a few minutes for the thinner samples and an hour or more for the thicker ones. The samples with lower carrier concentration showed similar trends. Excitation in the 360-550 nm range produced substantially weaker photocurrent with comparable decay rates. Nanowire photoconductivity arises from a reduction in the depletion layer via photogenerated holes drifting to the surface and compensating ionized surface acceptors. Simulations yielded ͑dark͒ surface band bending in the vicinity of 0.2-0.3 V and capture coefficient in the range of 10 −23 -10 −19 cm 2 . Atomic layer deposition ͑ALD͒ was used to conformally deposit Ϸ10 nm of Al 2 O 3 on several devices. Photoconductivity, persistent photoconductivity, and subgap photoconductivity of the coated nanowires were increased in all cases. TaN ALD coatings showed a reduced effect compared to the Al 2 O 3 coated samples.

No abstract

We have demonstrated dramatic improvement in the quality of selective-area GaN nanowire growth by controlling the polarity of the underlying nucleation layers. In particular, we find that N-polarity is beneficial for the growth of large ordered nanowire arrays with arbitrary spacing. Herein, we present techniques for obtaining and characterizing polarity-controlled nucleation layers on Si (111) substrates. An initial AlN layer, which is demonstrated to adopt Al-(N-)polarity for N-(Al-)rich growth conditions, is utilized to configure the polarity of subsequently grown GaN layers as determined by piezoresponse force microscopy (PFM), polarity-dependent surface reconstructions, and polarity-sensitive etching. Polarity-dependent surface reconstructions observed in reflection high-energy electron diffraction (RHEED) patterns were found to be particularly useful for in situ verification of the nucleation layer polarity, prior to mask deposition, patterning, and selective-area regrowth of the GaN NW arrays. N-polar templates produced fast-growing nanowires with vertical m-plane side walls and flat c-plane tips, while Gapolar templates produced slow-growing pyramidal structures bounded by (11̅ 02) r-planes. The selective-area nanowire growth process window, bounded by nonselective and no-growth conditions, was found to be substantially more relaxed for NW arrays grown on N-polar templates, allowing for long-range selectivity where the NW pitch far exceeds the Ga diffusion length.