Fully-integrated Isolators for Silicon Photonics using WAMO (Wrap Around Magneto-Optics)

摘要：非技术：想象一个集成电路，其中光携带信号而不是电子。这有望使许多信号传输更快，同时，没有热量。实现这种具有高密度的“光子集成电路”的有效应用的主要缺失环节是集成隔离器。隔离器使用磁光石榴石来控制光的方向，类似于电子器件中的二极管。虽然已经提出了几种集成隔离器，但它们并不能适用于所有的光偏振，事实上，大多数隔离器只适用于与所有硅集成激光器相反的偏振。本文提出了一种新颖的隔离器，从一种独特的简单设计开始，这种设计在当今的激光器中具有很高的成功概率，并逐步升级到一种最终设计，该设计包括一个集成磁铁并提供极化多样性（这意味着所有的极化都可以用一个设备隔离）。这个项目的技术新颖性有三个方面。首先，将磁光石榴石集成到硅上的材料挑战将被克服。其次，新的硅光子设计将首先实现对激光匹配偏振的隔离，然后实现对所有偏振的隔离。第三，磁性设计将采用封闭磁通结构来磁化石榴石，同时最大限度地减少器件之间的磁串扰。技术方面：光子系统通过光纤连接社会，通过光互连连接计算机内部。光子学也影响着医学、化学和许多其他领域。集成隔离器是在完全集成激光源的光子系统成为现实之前需要解决的缺失环节，这意味着这里的结果将产生非常广泛的影响。制造非互易器件（如隔离器）的唯一无源（零功率）方法是使用非互易（磁光）材料。磁光石榴石（例如，掺铈钇铁石榴石）的性能比任何其他材料好几个数量级。然而，这些石榴石很难以单相薄膜的形式与硅集成，并且多相会造成损耗。此外，迄今为止提出的隔离器设计太大，无法实际实施，而且它们大多使用非互易相移的几何形状，仅适用于横向磁极化，而集成激光器仅输出横向电光。在这里，经过验证的高旋转熵，低损耗和硅集成石榴石薄膜（Stadler）与新型硅光子设计（Li）的结合将确保具有极化多样化功能的全无源硅隔离器。将探索三个层面的创新。首先，提出了一种横向电隔离器，其中石榴石沉积在干涉仪分支之间，以提供简单的不对称，这样，在一个外加磁场下，将发生“推挽”非互易横向电相移。下一层创新将应用上面的基本光子结果，并将扩展设计，包括集成的静磁工程磁偏。具体来说，将使用封闭磁通环结构（类似于封闭马蹄形磁铁）来磁化石榴石包层，而不需要外部场，也不会在电路中产生边缘场。偏振分集的最后一步是接近封闭的磁通设计，这将产生第一个在全无源、完全集成的磁光学隔离器中提供偏振分集的设备。

Title: Fully-integrated Isolators for Silicon Photonics using WAMO (Wrap Around Magneto-Optics)Abstract: Nontechnical: Imagine an integrated circuit where light carries the signal rather than electrons. This promises to allow many signals to transmit faster, simultaneously, and without heat. The major missing link in achieving impactful applications of such "photonic integrated circuits" with high densities is an integrated isolator. Isolators use magneto-optical garnets to control the direction of light, similar to diodes in electronics. Although several integrated isolators have been proposed, they do not work for all light polarizations, and in fact the large majority only work for the polarization that is the opposite of all silicon-integrated lasers. Here, novel isolators are proposed starting with a uniquely simple design that has a high probability of success with today's lasers, and escalating to a final design that includes an integrated magnet and offers polarization diversity (meaning all polarizations can be isolated with one device). The technical novelty of this project are three-fold. First, materials challenges will be overcome to integrate magneto-optical garnets onto silicon. Second, new silicon photonic designs will initially enable the isolation of the laser-matched polarization and later the isolation of all polarizations. Third, magnetic design will be used to incorporate closed flux structures to magnetize the garnets while minimizing the magnetic cross-talk between devices.Technical: Photonic systems keep society connected with optical fibers and inside computers with optical interconnects. Photonics also impacts medicine, chemistry, and many other fields. An integrated isolator is the missing link that needs to be solved before photonic systems with fully-integrated laser sources can be a reality, and this means the results here will have very broad impact. The only passive (zero-power) way to produce a nonreciprocal device, such as an isolator, is to include a nonreciprocal (magneto-optic) material. Magneto-optic garnets (e.g., Cerium-doped Yttrium Iron Garnet) yield orders of magnitude better performance than any other material. However, these garnets are difficult to integrate with silicon as single phase films, and multiple phases cause loss. Also, isolator designs proposed to date are too large for practical implementation, and they mostly use non-reciprocal phase shift in a geometry that only applies to transverse magnetic polarizations while integrated lasers output only transverse electric light. Here, the combination of proven high-gyrotropy, low-loss, and silicon-integrated garnet films (Stadler) with novel silicon photonic designs (Li) will ensure all-passive silicon isolators with polarization diverse functionality. Three layers of innovation will be explored. First, a transverse electric isolator is proposed where garnet is deposited between the branches of an interferometer to provide a simple asymmetry such that, with one applied magnetic field, a "push-pull" non-reciprocal transverse electric phase shift will occur. The next layer of innovation will apply the fundamental photonic results from above and will extend the design to include an integrated magnetostatically-engineered magnetic bias. Specifically, a closed flux loop structure (similar to a closed horseshoe magnet) will be used to magnetize the garnet claddings without requiring an external field and without producing fringing fields in the circuit. The final step towards polarization diversity is a nearly closed flux design that will yield the first device to provide polarization diversity in an all-passive, fully-integrated isolator with wrap around magneto-optics.