Wednesday, October 09, 2024

Artilux paper on room temperature quantum computing using Ge-Si SPADs

Neil Na et al from Artilux and UMass Boston have published a paper titled "Room-temperature photonic quantum computing in integrated silicon photonics with germanium–silicon single-photon avalanche diodes" in APL Quantum.

Abstract: Most, if not all, photonic quantum computing (PQC) relies upon superconducting nanowire single-photon detectors (SNSPDs) typically based on niobium nitride (NbN) operated at a temperature <4 K. This paper proposes and analyzes 300 K waveguide-integrated germanium–silicon (GeSi) single-photon avalanche diodes (SPADs) based on the recently demonstrated normal-incidence GeSi SPADs operated at room temperature, and shows that their performance is competitive against that of NbN SNSPDs in a series of metrics for PQC with a reasonable time-gating window. These GeSi SPADs become photon-number-resolving avalanche diodes (PNRADs) by deploying a spatially-multiplexed M-fold-waveguide array of M GeSi SPADs. Using on-chip waveguided spontaneous four-wave mixing sources and waveguided field-programmable interferometer mesh circuits, together with the high-metric SPADs and PNRADs, high-performance quantum computing at room temperature is predicted for this PQC architecture.

Link: https://doi.org/10.1063/5.0219035

Schematic plot of the proposed room-temperature PQC paradigm with integrated SiPh using the path degree of freedom of single photons: single photons are generated through SFWM (green pulses converted to blue and red pulses) in SOI rings (orange circles), followed by active temporal multiplexers (orange boxes that block the blue pulses), and active spatial multiplexers (orange boxes that convert serial pulses to parallel pulses) (quantum sources), manipulated by a FPIM using cascaded MZIs (quantum circuits), and measured by the proposed waveguide GeSi SPADs as SPDs and/or NPDs (quantum detectors). An application-specific integrated circuit (ASIC) layer is assumed to be flipped and bonded on the PIC layer with copper (Cu)–Cu pillars (yellow lines) connected wafer-level hybrid bond, or with metal bumps (yellow lines) connected chip-on-wafer-on-substrate (CoWoS) packaging. The off-chip fiber couplings are either for the pump lasers or the optical delay lines.

 


 (a) Top view of the proposed waveguide GeSi SPAD, in which the materials assumed are listed. (b) Cross-sectional view of the proposed waveguide GeSi SPAD, in which the variables for optimizing QE are illustrated.

 

 

(a) QE of the proposed waveguide GeSi SPAD without the Al back mirror, simulated at 1550 nm as a function of coupler length and Ge length. (b) QE of the proposed waveguide GeSi SPAD with the Al back mirror, simulated at 1550 nm as a function of gap length and Ge length. (c) QE of the proposed waveguide GeSi SPAD with the Al back mirror, simulated as a function of wavelength centered at 1550 ± 50 nm (around the C band) and 1310 ± 50 nm (around the O band), given the optimal conditions, that is, coupler length equal to 1.4 μm, gap length equal to 0.36 μm, and Ge length equal to 14.2 μm. While the above data are obtained by 2D FDTD simulations, we also verify that for Ge width >1 μm and mesa design rule <200 nm, there is little difference between the data obtained by 2D and 3D FDTD simulations.


Dark current of GeSi PD at −1 V reverse bias, normalized by its active region circumference, plotted as a function of active region diameter. The experimental data (blue dots) consist of the average dark current between two device repeats (the ratio of the standard deviation to the average is <2%) for five different active region diameters. The linear fitting (red line) shows the bulk dark current density and the surface dark current density with its slope and intercept, respectively.



For the scheme of photon-based PQC: (a) The probability of successfully detecting N photon state and (b) the fidelity of detecting N photon state, using M spatially-multiplexed waveguide GeSi SPADs at 300 K as an NPD. (c) The difference in the probabilities of successfully detecting N photon state, and (b) the difference in the fidelities of detecting N photon state, using M spatially-multiplied waveguide GeSi SPADs at 300 K and NbN SNSPDs at 4 K as NPDs. Note that no approximation is used in the formula for plotting these figures.



For the scheme of qubit-based PQC: (a) The probability of successfully detecting N qubit state, and the fidelity of detecting N qubit state, using single waveguide GeSi SPADs at 300 K as SPDs. (b) The difference in the probabilities of successfully detecting N qubit state, and the difference in the fidelities of detecting N qubit state, using single waveguide GeSi SPADs at 300 K and NbN SNSPDs at 4 K as SPDs. Note that no approximation is used in the formula for plotting these figures.




4 comments:

  1. Is GeSi SPAD still promising for autonomous driving LiDARs?

    ReplyDelete
  2. When have GeSi SPADs ever been promising for Lidars?

    I strongly doubt that. Firstly, I question in general whether the wavelength range of 1-1.5um is really beneficial for automotive. There is still a lot of scattering (Mie scattering isn't strongly wavelength dependent). Sure you can shoot out a bit more photons, but you also have to spend the power on that. Most lasers in that wavelength range are for telecommunication and not for shooting out high peak powers. In daylight conditions you may want to argue that at certain wavelengths your background light is lower than at e.g. 940nm, but therefore your dark current is many orders of magnitude higher than at Silicon which very likely outweighs the advantage of reduced background light. Especially for SPADs dark count rate should be challenging as you'd either have to create avalanching in the Germanium layer itself, which should have much higher defect densities than Silicon potentially amplifying the tremendous dark carrier generation even more. Or you have to use Germanium as an absorber and try to realize amplification in Silicon - but then how do you create a defect free interface between Silicon and Germanium to enable good charge transfer without trapping and dark carrier generation?

    I believe Germanium-SPADs are limited to applications for which you have some more fundamental reason forcing you to not operate at 940nm... There is no fundamental reason why you have to operate a Lidar at 1-1.5um over 940nm. But if you try to build a microscope looking at certain absorption spectra or if you want to transmit light through some medium that absorbs ~940nm but not 1-1.5um, then maybe you have a use-case...

    ReplyDelete
  3. Would eye safety be a good reason to operate at 1.5um?

    ReplyDelete
    Replies
    1. That is what I implied by "Sure, you can shoot out a few more photons". But you can't build a system based on a single aspect. There are pros and cons and in my experience the cons outweigh the pros. That, of course, is different if you focus on applications where you have to operate above 1um wavelength. E.g. if you want to use the Silicon Photonics platforms operating at SWIR/telecom wavelengths, you can't use Silicon (which is the premise of the paper being highlighted here). But then again, for some (not all) use-cases you could use Silicon-Nitride based Photonics and then you could very well use Silicon SPADs...

      Delete

All comments are moderated to avoid spam and personal attacks.