在不同種類的光偵測器中，單光子雪崩二極體（single-photon avalanche diode, SPAD）因其能在接近常溫的條件下操作，具有較低的暗計數率，且在光纖通訊的1550 nm波段中可作為量子密鑰分發（quantum key distribution, QKD）的光子接收偵測器而備受關注。因此，本文選擇研究基於III-V族材料的單光子雪崩二極體。 實驗室過往研究中，是對輸出電壓振幅進行次數機率統計以實現光子數解析。然而，當欲增加偏壓以提升增益時，雜訊隨機分布的可能性增加，導致光子解析的準確度受限；相反地，當偏壓下降致增益過低時，電壓分離度（voltage separation）會降低，進而影響光子數準確辨識的能力。因此，為提升元件的光子數解析性能，實驗室過去研究多著力於在增益與非理想效應之間取得平衡。 本文採用並設計了時間多工的量測方式，以減少非理想效應對光子數解析的影響。這一方法使量測結果符合泊松分布，並能針對3顆光子態進行分析。此外，本文分析並解釋了偵測效率與光子數解析之間的關係，實現了對入射光脈衝平均光子數近乎準確的重建。展望未來，我們提出了如何進一步提高光子數解析最大數目以及提升光子數辨識準確率的具體方法。 ;Among various types of photodetectors, single-photon avalanche diodes (SPADs) are highly regarded for their ability to operate near room temperature, low dark count rates, and suitability as single photon detectors in quantum key distribution (QKD) within the 1550 nm wavelength range used in fiber-optic communication. Therefore, this study focuses on investigating SPADs based on III-V materials. In previous research conducted in our laboratory, photon-number resolution (PNR) was typically achieved by distinguishing output voltage amplitudes. However, increasing the bias voltage to enhance gain also increases the random noise distribution, thereby limiting the accuracy of PNR. Conversely, reducing the bias voltage lowers gain and reduces voltage separation, compromising the ability to accurately distinguish photon numbers. To address this challenge, prior studies had to carefully balance gain and non-ideal effects to optimize PNR performance. In this study, we designed and employed a time-multiplexing measurement approach to reduce the impact of non-ideal effects on PNR. This approach resulted in measurement outcomes consistent with a Poisson distribution and enabled the analysis of three photon number states. Furthermore, we comprehensively analyzed the relationship between detection efficiency and PNR distribution, achieving a near-accurate reconstruction of the mean photon number for incident light pulses. Looking ahead, we proposed specific methods to further enhance the maximum resolvable photon number and improve the accuracy of photon-number discrimination.
