https://mega.nz/file/scBBwCYS#o5SSZZehQEumxbGaOheUNXwgJSVX937kh8M3aGF_bTEHaving already achieved the objectives of efficiently converting electrons into photons and of being able to measure photons using other photons, I decided it was time to find a way to convert photons into electrons not only efficiently, but without forcing the photons to be absorbed by atomic orbitals and re-emitted, which is intrinsically inefficient. This proposal describes how to accomplish this. It is a boon not only for photovoltaic conversion efficiency, but its application, which took me a while to identify, is in enabling the efficient conversion of photons to electrons in the context of network interface cards. Large data centers, if they want to download data a very high rate must parse the job into a series of smaller jobs and purposefully fragment the downloaded data between multiple servers if they want to download at the fastest possible rate. This takes care of that problem as any single network interface would be able to make full use of the fiber-optic's potential with the remaining bottleneck being things like RAM speed and disk write speed.
This problem has been overlooked because solid-state drives usually form the ultimate bottleneck for individual nodes. However, creating servers with huge numbers of network interface cards for no reason makes no fundamental sense and drives up costs, as does the use of that amount CAT-5 cable in data centers.
Beyond the data conversion application, this opens the door to remarkably efficient photovoltaics. I wrote a paper in 2023 about how we could create efficient photovoltaics by mimicking natural photosynthesis, however that would only get us to about 88-92% efficiency. With this, we could get it to 99.5% It would also, naturally, have applications for photosensors, but I believe its best application is in making better network interface cards. I'm sure people will find a variety of applications for it.