Improvement of luminescence properties of n-GaN using TEGa precursor

Highlights

• Growth from TEGa precursor improved luminescence and structural properties of n-GaN.

• Type of carrier gas significantly influences properties of n-GaN.

• Nitrogen carrier gas supports creation of big complexes of V_Ga and V_N.

Abstract

The aim of this work is to compare and improve optical and structural properties of GaN layers prepared using TMGa or TEGa precursors. MOVPE grown GaN buffer layers on sapphire substrates are usually grown from TMGa precursor at the temperatures above 1000 °C. These layers contain deep and shallow acceptor levels which are responsible for blue and yellow defect bands in luminescent spectra. Both defect bands are detrimental for all major nitride device applications. Especially n-doped GaN layers suffer from strong yellow defect bands. In this work, it is shown that yellow band photoluminescence intensity can be suppressed by using TEGa precursor during the growth of n–doped GaN layers. Different kinds of growth parameters, such as growth temperature or growth rate, have been studied. It is also shown that the change of carrier gas (H₂ or N₂) has very strong influence on the layer quality. H₂ carrier gas increased intensity of yellow band in sample grown from TEGa precursor while N₂ carrier gas had the same effect for sample grown from TMGa precursor. Variable energy positron annihilation spectroscopy showed creation of single V_Ga in H₂ atmosphere and clustering of V_Ga to big complexes ((V_Ga)₃(V_N)_n) in N₂ atmosphere.

Introduction

Epitaxial GaN layers were first grown by Metal Organic Vapor Phase Epitaxy (MOVPE) technology in the late 1960 s by Manasevit and co-workers [1]. Layers were grown on c-plane sapphire substrates from trimethylgallium (TMGa), ammonia and a H₂ carrier gas. A significant breakthrough in nitrides growth was done in the late 1980s, when Amano and Akasaki invented a nucleation layer for the growth of nitrides on sapphire substrate [2]. This invention enabled to grow an active layer containing InGaN/GaN heterostructure on GaN buffer layer with sufficient quality. Since 1990s, GaN buffer layers grown on sapphire substrates have been extensively studied and the quality was improved. However, there are still a lot of open questions and unsolved problems today.

MOVPE grown GaN layers can be prepared from TMGa or TEGa precursors. TMGa is usually used for higher temperature growth while TEGa can be used at temperatures below 1000 °C. However, we have found that there are not many works comparing properties of GaN layers prepared using these two precursors. Therefore, one aim of this work is to contribute in this field and compare optical properties of GaN layers prepared from both precursors at different technological conditions.

GaN buffer layers are typically prepared from TMGa and ammonia precursors with a H₂ carrier gas at a growth temperature above 1000 °C, with V/III ratio between 1000 and 2000 and at growth pressure around 200 mbar. Silane is used for n-type doped GaN (similar growth conditions). GaN layers grown on sapphire substrate (the most common substrate for optoelectronic devices) have crystal structure with dislocation density around 10⁸ cm⁻². Undoped GaN layers are always unintentionally n-doped, due to impurities like oxygen or silicon; native defect formation cannot be excluded. All these defects and impurities in GaN are the cause of various bands in luminescent spectra. Many kinds of luminescent bands were described up to now, for example yellow band (YB) with the maximum around 2.2 eV or blue band (BB) with maximum around 2.9 eV. Most of them are described in review article [3].

The source of YB was attributed to some deep acceptor level [4]. Suppression of deep levels in GaN layers is a crucial point for all nitride applications, such as light emitting diodes (LEDs), high electron mobility transistors (HEMTs) or others. Deep levels (caused for example by carbon) can reduce electron mobility, which is detrimental for HEMTs [5]. In the active region of LEDs, they can be responsible for non-radiative recombination and decrease of the efficiency. For some other applications, such as fast scintillation detectors, non-radiative recombination is less detrimental than a radiative recombination, which has a long decay time [6], [7]. In unintentionally n-doped GaN layers, the formation of acceptor levels is the most probable. Suppression of acceptor levels' formation in undoped GaN is difficult and in intentionally n-doped GaN layers is even more challenging (due to the lower formation energy of acceptor defects). Therefore, we have studied n-doped GaN layers in this work.

Undoped GaN layers grown from TMGa and triethylgallium (TEGa) were studied and compared in few works [8], [9]. It was shown that GaN layers grown from TEGa have in some way better properties and layers contain less carbon. TEGa precursor is usually used for GaN growth when the growth temperature is lower than 1000 °C (typically during the growth of a barrier between InGaN quantum wells), because TEGa decomposition pathway involving β-hydrogen elimination can reduce carbon incorporation into GaN layers. Carbon usually sits on N lattice site (C_N) and acts as a deep acceptor 0.9 eV above the valence band [10]. Carbon atoms can be incorporated as an interstitial defect (C_i) or on Ga lattice site (C_Ga) as well, but the formation energies of these two defects are higher in usual growth conditions compared to C_N. So in GaN samples which are unintentionally n-doped, the C_N defect will be the most probable [11]. YB was attributed to C_N defect [10], C_N-O_N complex [12], which is energetically favourable, or C_N-Si_Ga [13] in n-type doped GaN especially. Another sources of YB, which were proposed in literature, were V_Ga-3H and V_Ga-O_N-2H complexes [14] or V_Ga-O_N [15]. The source of the YB is still not clear and more experiments are needed.

In our work we will compare MOVPE grown n-type doped GaN layers grown from TMGa and TEGa precursor. Source of the YB will be discussed; suggestions for YB luminescence intensity suppression will be given and confirmed by the experiment. The supposed source of YB in n-type doped GaN layers will be also discussed and compared with our experimental results.