氮化鋁鎵�氮化鎵異質結構的電性傳輸

Electrical transport in AlGaN/GaN heterostructures

吳坤達(Kun-Ta Wu)

指導教授 ： 梁啟德

共同指導教授 ： 張本秀(Pen-Hsiu Chang)

國立臺灣大學/理學院/物理研究所/碩士(2005年)

https://doi.org/10.6342/NTU.2005.01223

摘要

在本論文中，我將描敘氮化鋁鎵�氮化鎵高電子遷移率電晶體的兩項研究。論文總共分為兩大部份。

1.氮化鋁鎵�氮化鎵高電子遷移率電晶體在不同鋁成份下傳輸特性

　　我們探討三塊不同鋁成份的氮化鋁鎵�氮化鎵高電子遷移率電晶體的傳輸性質。鋁的成份分別為１１％、１５％和２５％。這三塊樣品都成長在藍寶石基板上，而遷移率的量測顯示１５％鋁成份的氮化鋁鎵�氮化鎵高電子遷移率電晶體有最高的遷移率。其遷移率在溫度１０Ｋ時，為６６００ｃｍ２�Ｖｓ。如果鋁的成份大過１５％，那樣品的遷移率將大幅下降。
　　此外，我們發現三塊樣品的遷移率在室溫下幾乎一樣。這是因為在溫室下，聲子散射主宰電子的全部散射，而由樣品雜質所產生的散射相較之下微不足道。
　　再者，我們發現二維電子濃度隨著在氮化鋁鎵中鋁成份的增加而增加。我們也對二維電子濃度做定量計算也發現同樣的趨勢。這個理由是因為當鋁成份增加，在氮化鋁鎵中的極化也增加。而這個增加的極化現象，在氮化鋁鎵和氮化鎵的界面產生更多片電荷形成。因此，更多電子將被吸引到這裡來補償在氮化鋁鎵�氮化鎵界面附近的片電荷。所以對於一個有較高鋁成份的氮化鋁鎵�氮化鎵的樣品，其二維電子濃度也會比較高。
　　我們同時測量氮化鋁鎵�氮化鎵高電子遷移率電晶體的活化能，而發現活化能隨著鋁成份增加而增加。當鋁成份增加，在氮化鋁鎵和氮化鎵之間的晶格不匹配愈來愈大。而這個增加的晶格不匹配將導致較大的缺陷密度。因此需要更高的活化能來熱激發載子到二維電子系統中。
　　總而言之，我們研究結果顯示出１５％鋁成份的氮化鋁鎵�氮化鎵高電子遷移率電晶體有最高的遷移率。而這個事實告訴我們可以用１５％鋁成份來成長其它氮化鋁鎵�氮化鎵高電子遷移率電晶體，以獲得最高的遷移率。

2.成長在正型矽基板上關於氮化鋁鎵�氮化鎵高電子遷移率電晶體的測量

　　在前一章節所使用的基板是藍寶石。然而，藍寶石在工業上並沒有矽來得普遍。因此，我們企圖在矽基板上成長我們的樣品。然而，在矽基板上成長高品質的氮化鎵是有困難的，因為在矽與氮化鎵之間有約１７％的晶格不匹配和約５４％的熱膨脹係數不匹配的缺點。
在本章節中，我們在樣品中分別沈積５秒和１０秒的氮化矽薄膜來增加氮化鎵的品質，而再和沒有沈積氮化矽薄膜的樣品做比較來觀察這樣一個薄膜如何影響在氮化鋁鎵和氮化鎵界面處的二維電子遷移率。在溫度１０Ｋ時，沒有沈積氮化矽薄膜的樣品，其遷移率為７４４ｃｍ２�Ｖｓ。相反地，有沈積５秒和１０秒氮化矽薄膜的樣品，其遷移率分別為２３３３ｃｍ２�Ｖｓ和２３８７ｃｍ２�Ｖｓ。這結果顯示出在樣品中沈積氮化矽薄膜後，樣品的遷移率大幅提升三倍之多。
　　再來把這章的研究結果和上一章比較，我們發現在藍寶石基板上，鋁成份１５％的氮化鋁鎵�氮化鎵有更高的遷移率。其遷移率為６６００ｃｍ２�Ｖｓ。因此我們可以推斷出，在矽基板上成長氮化鋁鎵�氮化鎵高電子遷移率電晶體，其遷移率要和在藍寶石上成長的一樣高是有困難的。然後，樣品遷移率大帳提昇的這項事實展示了我們在樣品裡沈積氮化矽薄膜這項技術的實用性。如果我們對成長樣品的技術做更進一步的研究，比如成長溫度，樣品結構，還有最重要的沈積氮化矽的技術，也許在矽基板上成長的高電子遷移率電晶體可以和在傳統藍寶石上成長的一樣好。

關鍵字

霍爾效應 ； 氮化鋁鎵 ； 氮化鎵 ； 藍寶石基板 ； 矽基板

並列摘要

In this thesis, I will report on two measurements on AlGaN/GaN high electron mobility transistors (HEMTs). This thesis consists of the following two parts.

1.Transport in AlxGa1-xN/GaN HEMTs with different Al compositions

We performed measurements on three AlxGa1-xN/GaN HEMTs with different Al contents (11%, 15%, and 25% respectively). All three samples are grown on sapphire substrates, and the mobility measurements indicate that AlxGa1-xN/GaN HEMTs with 15% Al content have the highest mobility (6600 cm2/Vs at 10K). If the Al content of AlxGa1-xN/GaN HEMTs exceeds 15%, the mobility will drop drastically.
In addition, we found that the mobilities of the three samples are almost identical at room temperature. This is because electron-phonon scattering dominates the electrical scattering at room temperature, and the electron-imperfection scattering due to impurities in the sample is small compared with electron-phonon scattering.
Furthermore, we found that the 2DEG concentration increases with increasing Al content in AlxGa1-xN. Quantitative calculation of the 2DEG concentration was performed and the same trend was obtained. The reason for this is that when the Al fraction increases, the polarization in AlxGa1-xN increases and induces more sheet charges at the interface of AlxGa1-xN and GaN. Therefore, for an AlxGa1-xN/GaN heterostructure with a higher Al composition, more electrons would be attracted to compensate for the sheet charges near the AlxGa1-xN/GaN interface. Hence the 2DEG concentration would be higher for an AlxGa1-xN/GaN sample with a higher Al composition.
We also measured the activation energy of AlxGa1-xN/GaN HEMTs, and found that the activation energy increases with increasing Al content. As the Al content increases, the lattice mismatch between AlxGa1-xN and GaN becomes more prominent. This increased lattice mismatch could induce a larger defect density. Therefore a higher activation energy is required to thermally activate the carriers into the 2DEG region.
Finally, the fact that AlxGa1-xN/GaN HEMTs with 15% Al content has the highest mobility suggests that we can grow other AlGaN/GaN HEMTs with 15% Al content for the greatest mobility.

2.Measurements of AlGaN/GaN HEMTs grown on p-type silicon substrates

The substrate used in the previous section is sapphire. However, sapphire is not as popular as Si in the industry. We therefore attempt to grow our samples on Si substrates. However, growing high-quality GaN on silicon substrates proves difficult because of the large lattice mismatch (about 17%) and large thermal mismatch (about 54%) between Si and GaN.
In this section, we inserted a thin Si5N4 film with deposition time of 5 seconds and 10 seconds to improve the GaN quality, and observed how such a thin film could affect the 2DEG mobility at the interface of AlGaN and GaN in comparison with the case without a Si5N4 thin film. At 10 K, the sample without the Si5N4 thin film had mobility 744 cm2/Vs, whereas for the samples with 5 seconds and 10 seconds deposition time of Si5N4 it was 2323 cm2/Vs and 2387 cm2/Vs respectively. The result showed that the mobility was greatly enhanced by as much as three times after the Si5N4 thin film was inserted.
Upon comparing the mobility with the results of the previous chapter, we see that Al0.15Ga0.85N/GaN HEMTs on sapphire substrates had a much higher mobility of 6600 cm2/Vs. We conclude that it is difficult to grow AlGaN/GaN HEMTs on silicon substrates with mobilities as high as those grown on sapphire substrates. Nevertheless, this great enhancement of the mobility demonstrates the usefulness of our technique. If we perform further investigations on the optimization of our growth temperature, HEMT structure, and most importantly, Si5N4 treatment technique, it is expected that the quality of HEMTs grown on Si substrates may well be as high as those grown on conventional sapphire substrates.

並列關鍵字

Hall effect ； AlGaN ； GaN ； sapphire substrate ； silicon substrate

參考文獻

Chapter 1

[2] Daniel W. Koon, Rev. Sci. Instrum. 60, 271 (1989).

[2] U. K. Mishra and Y.-F. Wu, IEEE Trans. Microwave Theory Tech. 46, 756 (1998).

[11] A. Krost and A. Dadgar, Phys. Stat. Sol. (a) 194, 361 (2002).

[12] L. J. Van der Pauw, Philips Research Report No. 13, 1 (1958).