光掃描式SWIR雷射雷達偵檢元件研製

Abstract

本論文探討SAGCM結構下之雪崩光電二極體，首先分析磊晶結構，接著討論不同保護環及漂浮保護環之擴散設計，從最簡單只有主動區擴散之元件開始研究，慢慢加入保護環和漂浮保護環結構，分析不同元件之擊穿電壓及崩潰電壓，最後討論不同元件之暗電流關係。 在TEM拍攝下，協助確認磊晶層的厚度，並藉由EDX line scan之結果，看出漸變層是由三層不同成分比例的四元合金所構成，並得知相對的元素比例。 藉由IV量測可以發現，擊穿電壓越大的元件，其崩潰電壓也越大。此外若元件的擴散方式若不當，便會受到drive-in的影響，造成擴散深度不穩定的現象。不同區域受到drive-in影響的程度不同，即使是同個擴散元件，受到drive-in後也會有不均勻的擴散深度，造成IV圖有多個擊穿電壓出現，drive-in效果嚴重的地方，其擴散深度甚至會比接觸到兩次擴散的區域還來得深。 此外收到drive-in影響的元件，暗電流會和主動區加上受drive-in影響之面積成比例。但少數元件之drive-in影響不明顯，其擊穿電壓及暗電流就會由主動區來決定。 因此有drive-in現象的擴散方法並不適合運用在光偵測元件上，可藉由改善絕緣層品質或改變擴散設計將保護環及漂浮保護環設計在第二次擴散時完成，如此便可增加元件的穩定性。

This thesis has an orientation towards avalanche photodiodes (APDs) under the SAGCM structure. The study analyzes the structure of epitaxy at first. Then, we discuss the design of the guard ring (GR) and floating guard ring (FGR). Start with the simplest doping structure that only includes active region and continue with more complicated one including GR. Finally, we introduce FGR to our research. The dark currents of the devices are also investigated in the end of our essay. The thickness of the epitaxy layer can be confirmed through the TEM images. Furthermore, by means of EDX line scanning, we can detect three layers with different compositions that construct the grading layer. The IV curves help us realizing the characteristics of the devices. The elements with larger punch-through voltages may also have larger breakdown voltages. However, if we diffuse the devices in an unfit process, they will suffer from drive-in effect and hence have unstable diffusion depths. The diffusion front will be rough although it is flat initially and hence form multiple punch-through voltages in the IV curve. What is more, areas that go through terrible drive-in effect can even have a deeper doping profile than the districts that contact twice to diffusion sources. The dark current of the device is directly proportional to the area of active region and region affected by drive-in. Nevertheless, some of the devices that only slightly influenced by the effect may have the properties merely according to the active region. So it’s improper to diffuse the APD in a method which will pass through the drive-in effect. Something we can do to improve the stability of our devices is depositing a better quality of insulating layer or designing the process to create the GR and FGR in the second step of diffusion.