| 摘要: | 砷化銦鎵/磷化銦近紅外光單光子雪崩型偵測器偵測的光波長可適用於光纖通訊，但三五族材料在磊晶過程中易產生缺陷，因此相較於矽製程的單光子雪崩型偵測器有較差的暗計數特性且有明顯的二次崩潰 (afterpulsing)。因此本論文會在不同溫度下進行暗計數分析，並提出各溫度區間產生暗計數的機制，除了已知的熱產生載子、SRH載子、穿隧載子與二次崩潰外，我們另外觀察到電荷堆積效應，並藉由照光量測實驗深入探討此效應。
在本研究中為了探討暗計數，我們以閘控電路操作元件，利用液態氮將元件降溫至77K。各溫度區間造成暗計數的主要機制可劃分為：高溫度區間(225-300K)，此區為熱產生載子主導暗計數來源，由活化能大小可加以驗證；以及低溫度區間(77-125K)，在此區間内，若是沒有足夠的推遲時間 (hold-off time) 便會由二次崩潰主導暗計數產生機制，使得暗計數率大幅且迅速上升；中間溫度區間介於高低兩溫度區間之間，其有極低的暗計數率，在此區由穿隧電流主導。在中間溫度區間我們觀察到了暗計數率的變化存在有局部極值，為分析此現象，我們改變了不同的直流偏壓與短脈衝電壓，並觀察到調整直流偏壓會使得暗計數隨溫度變化的曲線發生水平移動，我們將此結果歸因於電荷堆積效應。
為了驗證電荷堆積效應我們進行照光量測，將雷射光在turn on前的不同時間點入射於元件，並改變光量、溫度及直流偏壓來觀察光子計數率的變化。由量測可以發現計數率會隨延遲時間越長呈指數衰減，我們可萃取出兩段指數衰減的時間常數並依其快慢區分為兩種機制，由文獻指出較短的時間常數為暮光效應 (twilight effect)，而較長的時間常數則推論為電荷堆積效應。藉由改變溫度，可以觀察到高溫300K時有電荷堆積效應的影響，此時電荷堆積效應的電荷來源除了光產生載子，亦有熱產生載子及SRH載子；而在200K時，因暗載子來源被有效的抑制，電荷堆積效應並不明顯；在150K時，即是局部極值發生之處，觀察到非常嚴重的電荷堆積效應。將暗量測結果與照光量測的結果比對可以發現，在暗計數率高的溫度下，其電荷堆積效應亦是相當嚴重，因此可驗證我們的推論。
;The InGaAs/InP single photon avalanche detectors (SPADs) detecting light wavelength of near infrared are suitable for the fiber-based communication. However, the III-V material system has experienced serious issues of defects due to the difficulty of epitaxy, so III-V based SPADs could bear poor dark count characteristics and significant afterpulsing. In this thesis, we present a study of the dark count rate for InGaAs/InP SPADs at different temperature ranges. Besides the thermal generation, SRH carriers, tunneling carriers and afterpulsing, we additionally address the charge persistence effect by the temperature-dependent dark count probability. This effect is further discussed and evidenced by the experiment of light injection.
In this study, for investigating dark count characteristics, the SPAD is operated using the gated mode. The temperature-dependent experiments are implemented by cooling the device down to 77K using liquid nitrogen. The dominant mechanisms at different temperature regions can be categorized as follows. First is the high temperature region, between 225K and 300K, in this region thermal generation and SRH carriers are the dominant sources of dark counts, which can be evidenced by extracting the value of activation energy. Second is the low temperature region, between 77K and 125K, in such region serious afterpulsing effect is observed if there is no sufficient hold-off time for releasing the trapped carriers. The final is the intermediate temperature region between high and low temperature region. There are extremely low dark counts for this region. It is known that the main source of dark counts originates from the tunneling current in such region. In the temperature-dependent dark count rate measurements, an abnormal peak occurs in the intermediate temperature region. In order to clarify the cause of abnormal peak, we further measure the temperature-dependent dark count rate under different dc bias and gate pulse height. It is observed that the curve of dark count rate shifts to the lower temperature with increasing dc bias. We attribute this observation to the charge persistence effect, which can be further evidenced by the following measurements under light injection.
To evidence the charge persistence effect observed under dark conditions, we illuminate the SPAD prior to the gate pulse. With varying the delay time between light pulse and gate pulse, the charge persistence effect can be investigated. The count rate is found to exponentially decay with the delay time. By fitting the curve with exponential decay formula, we can extract two time constants which are involved by two mechanisms. The literatures have addressed that a faster time constant is caused by the twilight effect and the slower time constant is due to the charge persistence effect. The count rate as a function of delay time shows that there is slight charge persistence effect at 300K, where the persistent charges are mainly contributed by thermal generation and SRH carriers besides the photo-generated carriers. At 200K, the dark carriers are effectively suppressed, therefore no obvious charge persistence is observed. At 150K, at which an abnormal peak occurs, serious charge persistence effect is observed. Comparing the results under dark and light conditions, it is found that at the temperature where higher dark count occurs the charge persistence effect is also distinct, which evidences our argument. |
