Abstract Currently there is a high level of interest in the development of ultraviolet (UV) light sources for solid-state lighting, optical sensors, surface decontamination and water purification. III-V semiconductor UV LEDs are now successfully manufactured using the AlGaN material system; however, their efficiency is still low. The majority of UV LEDs require AlxGa1-xN layers with compositions in the mid-range between AlN and GaN. Because there is a significant difference in the lattice parameters of GaN and AlN, AlxGa1-xN substrates would be preferable to those of either GaN or AlN for many ultraviolet device applications. However, the growth of AlxGa1-xN bulk crystals by any standard bulk growth techniques has not been developed so far. There are very strong electric polarization fields inside the wurtzite (hexagonal) group III-nitride structures. The charge separation within quantum wells leads to a significant reduction in the efficiency of optoelectronic device structures. Therefore, the growth of non-polar and semi-polar group III-nitride structures has been the subject of considerable interest recently. A direct way to eliminate polarization effects is to use non-polar (001) zinc-blende (cubic) III-nitride layers. However, attempts to grow zinc-blende GaN bulk crystals by any standard bulk growth techniques were not successful. Molecular beam epitaxy (MBE) is normally regarded as an epitaxial technique for the growth of very thin layers with monolayer control of their thickness. In this study we have used plasma-assisted molecular beam epitaxy (PA-MBE) and have produced for the first time free-standing layers of zinc-blende GaN up to 100 μm in thickness and up to 3-inch in diameter. We have shown that our newly developed PA-MBE process for the growth of zinc-blende GaN layers can also be used to achieve free-standing wurtzite AlxGa1-xN wafers. Zinc-blende and wurtzite AlxGa1-xN polytypes can be grown on different orientations of GaAs substrates - (001) and (111)B respectively. We have subsequently removed the GaAs using a chemical etch in order to produce free-standing GaN and AlxGa1-xN wafers. At a thickness of ∼30 µm, free-standing GaN and AlxGa1-xN wafers can easily be handled without cracking. Therefore, free-standing GaN and AlxGa1-xN wafers with thicknesses in the 30–100 μm range may be used as substrates for further growth of GaN and AlxGa1-xN-based structures and devices. We have compared different RF nitrogen plasma sources for the growth of thick nitride AlxGa1-xN films including a standard HD25 source from Oxford Applied Research and a novel high efficiency source from Riber. We have investigated a wide range of the growth rates from 0.2 to 3 µm/h. The use of highly efficient nitrogen RF plasma sources makes PA-MBE a potentially viable commercial process, since free-standing films can be achieved in a single day. Our results have demonstrated that MBE may be competitive with the other group III-nitrides bulk growth techniques in several important areas including production of free-standing zinc-blende (cubic) (Al)GaN and of free-standing wurtzite (hexagonal) AlGaN.

中文翻译：

---

分子束外延作为实现自支撑闪锌矿 GaN 和纤锌矿 Al x Ga 1-x N 的生长技术

摘要 目前，用于固态照明、光学传感器、表面净化和水净化的紫外 (UV) 光源的开发受到高度关注。III-V半导体UV LED现在使用AlGaN材料系统成功制造；然而，它们的效率仍然很低。大多数 UV LED 需要 AlxGa1-xN 层，其成分介于 AlN 和 GaN 之间。由于 GaN 和 AlN 的晶格参数存在显着差异，因此对于许多紫外器件应用，AlxGa1-xN 衬底比 GaN 或 AlN 衬底更可取。然而，迄今为止，尚未开发出通过任何标准体生长技术来生长 AlxGa1-xN 体晶体。纤锌矿（六方）III族氮化物结构内部存在非常强的电极化场。量子阱内的电荷分离导致光电器件结构效率的显着降低。因此，非极性和半极性 III 族氮化物结构的生长最近已成为人们相当感兴趣的主题。消除极化效应的直接方法是使用非极性 (001) 闪锌矿（立方）III 族氮化物层。然而，通过任何标准体生长技术来生长闪锌矿 GaN 体晶体的尝试都没有成功。分子束外延 (MBE) 通常被认为是一种外延技术，用于生长非常薄的层，单层控制其厚度。在这项研究中，我们使用了等离子体辅助分子束外延 (PA-MBE)，并首次生产了厚度达 100 微米、直径达 3 英寸的闪锌矿 GaN 独立层。我们已经证明，我们新开发的用于生长闪锌矿 GaN 层的 PA-MBE 工艺也可用于实现自立式纤锌矿 AlxGa1-xN 晶片。闪锌矿和纤锌矿 AlxGa1-xN 多型体可以在不同方向的 GaAs 衬底上生长 - 分别为 (001) 和 (111)B。我们随后使用化学蚀刻去除了 GaAs，以生产自立式 GaN 和 AlxGa1-xN 晶片。在约 30 µm 的厚度下，独立式 GaN 和 AlxGa1-xN 晶片可以轻松处理而不会开裂。所以，厚度在 30-100 μm 范围内的独立式 GaN 和 AlxGa1-xN 晶片可用作衬底，以进一步生长基于 GaN 和 AlxGa1-xN 的结构和器件。我们比较了用于生长厚氮化物 AlxGa1-xN 薄膜的不同射频氮等离子体源，包括来自 Oxford Applied Research 的标准 HD25 源和来自 Riber 的新型高效源。我们研究了从 0.2 到 3 µm/h 的各种生长速率。使用高效的氮射频等离子体源使 PA-MBE 成为一种潜在可行的商业工艺，因为可以在一天内获得独立的薄膜。