qCMOS® camera vs. EM-CCD camera vol.2 – Qubit state detection

Quantum computing promises to overcome the limitations of Moore's Law, and significant research is focused on its industrialization. Various hardware approaches exist, with neutral atoms and ions (hereafter collectively referred to as “atoms”) being leading contenders.

These systems require high-speed, high-precision detection of atomic fluorescence to determine qubit states, making advanced cameras critical to overall performance. While EM-CCD cameras were traditionally used in this application, ultra-sensitive CMOS cameras, like the qCMOS camera (the ORCA-Quest series) from Hamamatsu Photonics, now offer comparable sensitivity with much superior speed and spatial resolution to EM-CCD cameras (See qCMOS vs EM-CCD Vol. 1).

To help users select the best camera, we simulated the fidelity of atomic fluorescence detection for quantum computing using our state-of-the-art qCMOS camera, a sCMOS Gen III camera (the generation preceding qCMOS), and a traditional EM-CCD cameras. This analysis aims to spark discussion on how advancing CMOS technology can accelerate breakthroughs in quantum computing.

 

Method for calculating the fidelity of atomic fluorescence detection

The fidelity of atomic fluorescence detection can be quantitatively evaluated based on the following three steps.

(i) Acquisition of integrated intensity via image simulation

For each of the 'On' and 'Off' states, 100,000 simulation frames are generated, and the integrated intensity value is determined for a 3×3 pixel array surrounding an atom in each frame.

Acquisition of integrated intensity via image simulation

(ii) Threshold determination via histogram analysis

A frequency histogram of the integrated intensity values is created for each state, and the position where the histograms intersect is set as the threshold for distinguishing between the 'On' and 'Off' states. (If the histograms do not intersect, the midpoint is used)

(iii) Calculation of fidelity metrics

The "True positive rate" (the proportion of 'On state' integrated intensity values that are above the threshold) and the "Error rate" (=1 – “True positive rate” / 100) are calculated.

  • True positive rate: The proportion of On-state frames with integrated intensity above the threshold.
  • Error rate: 1-True positive rate/100

Calculation of fidelity metrics

Simulation conditions for atomic fluorescence detection

In this simulation, we set the following conditions to quantitatively evaluate the fidelity of atomic fluorescence detection and compare the performance of different camera systems.

Wavelength

  • 493 nm (Ba+ ion)
  • 780 nm (Rb atom)

Optics

  • Spot size: Diffraction-limited spot with a spread of σspot=1.22 NA/λ
  • Objective lens NA: 0.8
  • Objective lens magnification: Set so that 2σspot equals to pixel size for each camera

Signals

  • Atom signal:  104 photons/sec/atom
  • Background signal: one-tenth of the fluorescence peak (referring to the actual experimental environment)

Simulation conditions for atomic fluorescence detection

Camera parameters

The camera parameters particularly relevant to this simulation are described below. (Since high-speed imaging is required for this application, the noise values are based on the fastest scan mode for each camera.)

Specifications 

ORCA-Quest 2

qCMOS camera

ImagEM® X2-1K*1

EM-CCD camera

ORCA-Fusion BT

sCMOS Gen III camera

Pixel size (μm) 4.6×4.6 13×13

6.5×6.5

QE (493 nm / 780 nm) 84 % / 49.1 % 90 % / 73 % 90 % / 63.3 %
Readout noise (e- rms) 0.43  < 1.0 (M =1200) 1.6
Data rate (Megapixel/sec) 1,130 10 470

*1 EM-CCD from Hamamatsu Photonics. It is now discontinued and no longer manufactured or sold.

While CMOS cameras have the high data rate (="the number of pixel" × "frame rate") required for quantum computers, a critical point in determining their viability for this application is how high a fidelity they can demonstrate compared to EM-CCD cameras. QE should particularly be noted, as EM-CCD cameras often have a relatively higher QE than qCMOS cameras, especially in the near-infrared region, which is 780 nm in this case.

Simulation results and discussion

The fidelity of atomic fluorescence detection strongly influences the performance of a quantum computer, and even the difference between 99 % and 99.9 % creates a significant impact. In this simulation, the comparison will focus on fidelities of 99 % or higher (an error rate of 1% or less).

Simulation result: 493 nm wavelength (Ba+ ion)

Simulation images

  • Exposure time: 10 ms
  • Atom signal: 100 photons/atom/frame
  • Background signal: 4.7 photons/pixel/frame

Frequency histogram comparison

  • Exposure time: 10 ms

Frequency histogram comparison

* For the EM-CCD camera, the horizontal axis is normalized to the number of detected photoelectrons by dividing by the gain.

  • Exposure time: 5 ms

Frequency histogram comparison

  • Exposure time: 2 ms

Frequency histogram comparison

Error rate comparison

Error rate comparison

* At the 10 ms exposure time, the ORCA-Quest 2 (Standard) and ORCA-Fusion BT (Fast) are not displayed on the logarithmic graph because their error rate is 0.

Simulation result: 780 nm wavelength (Rb atom)

Simulation images

  • Exposure time: 10 ms
  • Atom signal: 100 photons/atom/frame
  • Background signal: 4.7 photons/pixel/frame

Frequency histogram comparison

  • Exposure time: 10 ms

Frequency histogram comparison

  • Exposure time: 5 ms

Frequency histogram comparison

  • Exposure time: 2 ms

Frequency histogram comparison

Error rate comparison

Error rate comparison

Discussion

In this simulation, the qCMOS camera showed particularly superior results to the EM-CCD camera in terms of error rate at a wavelength of 493 nm, and slightly superior results even at 780 nm where there is a large difference in relative QE. Interestingly, the ORCA-Fusion BT, a sCMOS Gen III camera from the generation preceding the qCMOS, also showed superior results to the EM-CCD camera in the region of less than 1% error rate.

 

The reason for this is that while EM-CCD cameras achieve ultra-high sensitivity through electron multiplication, the fluctuation from this process (called “Excess noise”) is unavoidable and significantly affects quantitative measurement. In qubit state detection, which requires quantitative measurement to achieve high detection fidelity, this causes the histogram to widen, as seen in the figure provided in the document. As a result, the qCMOS camera showed better fidelity even with its relatively lower QE.

qCMOS camera and EM-CCD camera comparison

For reference, a simulation under conditions with absolutely no background signal is also attached. Although assuming zero background is unrealistic for quantum computers where lasers are used for atom manipulation and qubit state detection, this is included for comparison because the background significantly affects the sensitivity of EM-CCD cameras. Under these zero-background conditions, the qCMOS camera was still superior at 493 nm, but the EM-CCD camera was superior at 780 nm.

Conclusion

This simulation showed that for atomic fluorescence detection, the qCMOS camera demonstrates superior fidelity to the EM-CCD camera in most cases.

With the increasing sensitivity of CMOS cameras, we believe that qCMOS cameras, as ultra-sensitive imaging devices that combine high speed and high resolution, will contribute to the calculation speed enhancement and qubit scalability improvement on neutral atom and ion quantum computers.

Case study

Camera articles

Related products

Contact us for more information.

  • Literature
  • Price
  • Delivery
  • Custom order
  • Demo
  • Support
  • Other

Contact us