含有銻元素的半導體成長與其高電子遷移率電晶體之研究

Abstract

本論文的目的是為了全面性地了解由分子束磊晶所成長的含銻半導體材料系統和在砷化鎵基板上的第一型式的砷化銦/銻砷化鋁的高電子遷移率電晶體的電特性。 為了抑制由銻化鋁緩衝層與砷化鎵基板之間的8% 晶格不匹配所引起的晶格失配和錯位的形成，我們已經使用了兩種不同的方法。首先，我們開發了一種低缺陷緩衝層的生長技術，它提供了具有光滑表面的高電阻率的緩衝層。這種技術使用了幾對薄銻化鋁所組成的週期性溫度變化的緩衝層。透過利用此方法，在砷化鎵基板上的砷化銦通道的高電子遷移率電晶體是可以被製造出來。 對於高品質的砷化銦量子井通道，我們已經發展出了另一種方法是透過改善在砷化銦通道與含銻的位壘層之間的界面。在含銻/砷化合物的界面，因為砷化銦與銻砷化鋁之間的陰離子和陽離子的變化，它可能靠著分子束磊晶的方式生長出兩個不同的界面。一種是銻化銦的界面和另一種可能是砷化鋁的界面。在這項研究中，我們發現砷化銦通道的晶體品質對所使用的界面類型有著很強的依賴性。在有著銻化銦界面的處理，砷化銦的晶體品質將大大被改善。這是因為砷化銦的晶格與緩衝層的晶格是對齊的而沒有任何的晶格鬆弛。與像砷化鋁的類型相比，砷化銦的晶格是鬆弛的且砷化銦通道的晶體品質變的較差。優越的砷化銦量子井的晶體品質證明了表現出高的電子遷移率和良好的表面形態。 除了材料的改進之外，我們還開發了新的元件結構，它在碰撞電離的存在有著優越的電洞侷限。它導致了元件大大改善了輸出特性。在這項研究中所使用的新結構包括第一型式的砷化銦/銻砷化鋁的異質結構。採用這種結構所製成的場效電晶體不具有來自由碰撞電離所產生的電洞的不期望的反饋效果。用加入了上和下部的銻砷化鋁的屏障層，阻斷了由碰撞游離所產生的電洞從通道移動到閘極和下方的緩衝層並且將其限制在通道中。因此，顯示出了沒有任何隆起的正常閘極電流特性和大大地改善了輸出的電流電壓特性與可用的汲極電壓範圍。 為了更清楚地了解含有著不同的五族元素像磷，砷，銻的三五族的三元化合物的生長特性，我們做了這些元素在成長過程中的摻入和對生長層的應變與組成之間的關係進行了詳細的研究。我們首先研究了銻砷化鎵在砷化鎵基板上的生長。我們發現儘管固定流量比例在該層中有著自然形成的銻梯度現象。我們還研究了銻分佈是如何取決於生長層的應變量與其類型。接著在磷化銦基板上生長了應變的銻砷化鎵。在這種方式中，我們調查了在各種應變條件下的銻元素的摻入行為。我們發現在拉伸應變下銻的分佈隨著遠離界面有著逐漸遞減的趨勢和在壓縮應變下其分佈則是朝向表面而持續增加。 在本文介紹的工作，不僅為我們提供了對銻化合物生長的更好的了解，而且也提供了一個對許多重要的元件生長的指引。

The purpose of this dissertation is to understand comprehensively the growth of the Sb-contained semiconductor material systems by MBE and the electronic characteristics of the type-I InAs/AlAsSb high electron mobility transistor on GaAs substrates. To suppress the formation of misfit and threading dislocations caused by the 8% lattice mismatch between the AlSb buffer layer and GaAs substrate, we have used two different methods. First, we developed a low defect buffer layer growth technique, which provides a high resistivity buffer with a smooth surface. This technique uses a periodically temperature-varying buffer layer composed of several pairs of thin AlSb layers. Through exploiting this method, an InAs-channel high electron mobility transistor on GaAs substrate was fabricated. Another method we have developed for high quality InAs quantum well channels is by improving the interfaces between the InAs channel and the Sb containing barrier layers. At antimonide/arsenide interfaces, it is possible to grow two different interfaces by molecular beam epitaxy because of the change of both the anion and cation between InAs and AlAsSb. One is an InSb-like interface and the other possibility is an AlAs-like interface. In this study, we found that the crystal quality of the InAs channel has a strong dependence on the type of the interface used. With the use of the InSb-like interface, the InAs quality is greatly improved. This is because the InAs lattice is aligned with the lattice of the buffer layer without any lattice relaxation. Compared with the AlAs-like type, the InAs lattice is relaxed and the crystal quality of the InAs channel is poor. The superior InAs quantum wells was demonstrated by showing high electron mobility and good surface morphology. Besides material improvement, we have also developed a new device structure, which had superior hole confinement in the presence of impact ionization. It results in much improved output characteristics for the devices. The new structure used in this study includes a type-I InAs/AlAsSb heterostructure. The FETs fabricated using this structure does not have the undesirable feedback effect from the holes generated by impact ionization. With the addition of the upper and lower AlAsSb barriers, impact-ionized holes are blocked from moving to the gate and buffer layer underneath and are confined in the channel. Thus, the gate current shows the regular leakage characteristic without any hump and the output I-V characteristics and the usable drain voltage range are greatly improved. To more clearly understand the growth behavior of III-V ternary compounds with different group-V elements like P, As and Sb, we did a detailed study on the incorporation of these elements during growth and its dependence on the strain and composition of the grown layer. We first study the growth of a GaAsSb layer on GaAs. We found that there is a naturally formed Sb gradient in the layer despite a fixed beam flux ratio. We also studied how the Sb distribution depends on the amount and type of strain in the grown layer. The strained GaAsSb layers on InP substrates were then grown. In this way, we investigated the incorporation of Sb with various strain conditions. We found the Sb distribution under tensile strain decreases as we move away from the interface and the Sb composition increases under compressive strain toward the surface. The work presented in this thesis not only gives us a better understanding on the growth of antimonide compounds but also provides a guideline for the growth of many important devices.