Note that this is a difficult paper to understand for the non-techies (as myself)

Neverthelss, I'm still happy to find out a recent paper acknowledging the viability of GaAs as a tech platform.

I wonder what those professors/researchers/lecturers would think of POET GaAs-based platform.

Abstracts from the paper:

This paper reviews the recent progress in the development of the key building blocks for GaAs quantum photonics and the perspectives for their full integration in a fully-functional and densely integrated quantum photonic circuit

A comparison between GaAs and other quantum photonic technologies is useful. The LiNbO3, Si and SiN platforms can also in principle provide the key functionalities of photon production, passive control and detection, while the integration of efficient sources on silica-based QPICs appears very challenging. In terms of integration density, the small index contrast in LiNbO3 represents a major fundamental limitation, making its application to large-scale circuits involving tens of photons and thousands of componentsdoubtful. Si and SiN therefore represent the strongest competitors to GaAs for integrated quantum photonics.

These materials benefit from decades of investments and technology developments in the electronics industry, allow fabricating devices with high quality and reproducibility, and are in principle compatible with production in CMOS fabs.

The index contrast in both systems is high, enabling high integration levels. Indeed, fast progress, particularly in quantum silicon photonics, has recently led to the demonstration of QPICs with tens to hundreds of components [21, 262]. However, fundamental challenges remain in the large scale integration of sources, detectors and low-power reconfigurable circuits on the Si and SiN platforms. On the one hand, single-photon sources in these materials can only be based on two-photon production via spontaneous four-wavemixing from a pump laser and heralding of a non-vacuum state by detection of one of the photons. This approach is intrinsically limited in terms of efficiency (probability of generating a single photon) since the average photon number must be kept low to avoid multiphoton events. A potential solution is the multiplexing of several sources [263], which implies a very significant increase in the number of components and puts stringent requirements on the loss of switches [20, 264]. On the other hand, the integration of superconducting nanowire detectors implies additional technological challenges on the Si and SiN platform, related to the extreme filter performance required to suppress the pump used for photon production, the impossibility to use thermal phase tuning at low temperature, and the incompatibility of most superconducting materials with CMOS fabs.

A solution to these problems might lie in the use of hybrid platforms such as III-V materials on silicon by direct heteroepitaxial growth on top of each other or by wafer bonding [265].

While it is unclear at this point which technology will be suited for large-scale QPICs, the physical properties of GaAs and other III-Vs, including the direct bandgap and the non-centrosymmetric crystal structure (enabling a linear electro-optic effect), provide them a very fundamental advantage which could become crucial in the long term.

Source: http://128.84.21.199/pdf/1601.06956v1.pdf