新设计实现雪崩光电二极管探测器性能优化

由于社交媒体、视频流、人工智能和物联网等新兴应用的出现,数据中心和高性能计算中的数据通信获得了巨大的发展。超大规模数据中心和百亿亿次超级计算机都需要高带宽和高能效的光学互连。在这种情况下,硅光互连技术由于高集成度和低成本优势变得非常有吸引力。高级调制码型由于可以在集成电路或者光学器件的有限带宽下进一步提高数据速率,而越来越受到学术界和工业界的青睐。但是,光链路的信噪比可能会遭受影响而降低。这时需要更高功率的激光器或更高灵敏度的接收器才能维持原有的误码率。

与使用高功率激光器相比,使用具有更高灵敏度的接收器可以产生更低的总链路功耗,从而提高能效。特别是高灵敏度探测器放宽了对有限输出功率的片上激光器的链路预算要求,因此具有内部增益的雪崩光电二极管探测器是提高接收器灵敏度的理想选择。

由于雪崩光电二极管探测器在带来倍增增益的同时也引入了额外的噪声,为了实现更高增益和更低噪声,研究人员一直致力于设计具有最佳硅锗层厚和掺杂分布的新型器件结构。通常来说,雪崩光电二极管探测器的设计指标需要在击穿电压、量子效率、倍增增益、器件带宽和额外噪声之间权衡取舍。打破这些权衡并优化整体性能对雪崩光电二极管探测器的设计是一个极大的挑战。

惠普实验室的王斌浩博士等在Photonics Research 2020年第7期(Binhao Wang, Zhihong Huang, Yuan Yuan, et al. 64 Gb/s low-voltage waveguide SiGe avalanche photodiodes with distributed Bragg reflectors[J]. Photonics Research, 2020, 8(7): 07001118)中提出了具有分布式布拉格反射器的雪崩光电二极管探测器,在无需额外的制造步骤情况下打破了量子效率和器件带宽之间的设计权衡,并仍具有高倍增增益、低击穿电压和低额外噪声的优点。

与大多数Ⅲ-Ⅴ族化合物探测器相比,硅具有较低的碰撞电离系数比,因此硅锗雪崩光电二极管探测器具有更低的噪声和更高的带宽。为了同时利用锗在近红外区的强吸收和硅的低倍增噪声优势,文中提出的硅锗雪崩光电二极管探测器采用了吸收层、电荷层和倍增层分离的结构。探测器采用波导设计而不是垂直入射,是因为其具有较小的寄生电容以及量子效率和载流子传输时间的解耦,从而实现更高的量子效率和带宽。

另外,波导雪崩光电二极管探测器可以集成在例如含有波分复用器的复杂光子集成电路中来满足各种应用需求。借助集成的分布式布拉格反射器,雪崩光电二极管探测器的量子效率在通信波段从60%提高到90%。具有分布式布拉格反射器的雪崩光电二极管探测器仍然可以达到25 GHz的带宽,这与没有分布式布拉格反射器的雪崩光电二极管探测器相当。这种设计还实现了10 V的低击穿电压和接近500 GHz的增益带宽积,并成功演示了在64 Gb/s速率下的无误码传输。

惠普实验室的黄志宏博士认为,创造性的雪崩光电二极管探测器设计在击穿电压、量子效率、倍增增益、器件带宽和额外噪声等方面具有出色的性能,将为大型数据中心和高性能计算中的下一代高带宽、高能效光学互连技术发挥重要作用。下一步,研究人员将演示包含硅锗雪崩光电二极管探测器的高速硅光波分复用收发器。

集成分布式布拉格反射器的波导硅锗雪崩光电二极管探测器示意图

64 Gb/s low-voltage waveguide SiGe avalanche photodiodes with distributed Bragg reflectors

Data communications in data centers and high-performance computing (HPC) have grown tremendously due to emerging applications in social media, video streaming, artificial intelligence (AI), and the internet of things (IoT). Hyperscale data centers and Exascale HPC require high-bandwidth and energy-efficient optical interconnects. In this context, silicon photonic interconnects are attractive, thanks to the high integration and low cost. To further increase the data rate, advanced modulation formats are preferred which can relax the bandwidth limitation of integrated circuitry or optical devices. However, the optical link may suffer degraded signal-to-noise ratio (SNR), which requires either a higher-power laser or a higher-sensitivity receiver to retain the bit error rate (BER). Using a receiver with better sensitivity can yield a lower total link power consumption compared to using a high-power laser and consequently improve the energy efficiency. Particularly, high-sensitivity detectors relax the link budget requirements for on-chip lasers with limited output power. An avalanche photodiode (APD) with internal gain is the ideal candidate to increase the receiver sensitivity. Since APDs introduce the excess noise while bringing the multiplication gain, a novel device structure with optimum layer thickness and doping profile is necessary for higher gain and lower noise. Typically, there are trade-offs among APD design metrics: breakdown voltage, quantum efficiency, multiplication gain, bandwidth, and excess noise. It is a big challenge to decouple these trade-offs and optimize the overall performance.

The APD with a DBR proposed by Dr. Binhao Wang from Hewlett Packard Labs in Photonics Research, Vol. 8, Issue 7, 2020 (Binhao Wang, Zhihong Huang, Yuan Yuan, et al. 64 Gb/s low-voltage waveguide SiGe avalanche photodiodes with distributed Bragg reflectors[J]. Photonics Research, 2020, 8(7): 07001118) breaks the trade-off between quantum efficiency and bandwidth while retaining high gain, low breakdown voltage and low excess noise without additional fabrication steps. Compared to most III-V compound devices, SiGe APDs have lower noise and higher bandwidth due to the low impact ionization coefficient ratio in silicon. The proposed SiGe APD is with a separate absorption and charge multiplication (SACM) structure to take advantages of the large absorption in Ge in the near-infrared region and the low multiplication noise in Si. The waveguide APD design is employed rather than normal incidence due to the smaller parasitic capacitance as well as the decoupling of quantum efficiency and carrier transit time, achieving higher quantum efficiency and bandwidth. In addition, waveguide APDs can be integrated in complex photonic integrated circuits (PICs) such as wavelength division multiplexers (WDMs) for many applications. With the help of the integrated distributed Bragg reflector (DBR), the APD quantum efficiency is improved from 60% to 90% in C band. APDs with DBRs can still achieve a 25 GHz bandwidth, which is comparable to APDs with no DBR. A low breakdown voltage of 10 V and a gain bandwidth product of near 500 GHz are obtained. Error-free transmission at 64 Gb/s is successfully demonstrated.

Dr. Zhihong Huang from Hewlett Packard Labs believes that the creative APD design with excellent performance in breakdown voltage, quantum efficiency, multiplication gain, bandwidth, and excess noise will play an important role for next generation high-bandwidth and energy-efficient optical interconnects in mega data centers and high-performance computing. Next step, we will demonstrate a high-speed silicon photonic WDM transceiver with the SiGe APD.

Schematic of a waveguide SiGe APD integrated with a DBR