We develop and learn how to characterize non-classical light sources and single-photon detectors. To have an accurate model of a single-photon detector’s response, it is not enough to characterize detection efficiency and noise alone, there are many other essential parameters and carefully described recipes are necessary for measuring those parameters. Similarly, understanding non-classical sources requires characterizing multiple parameters, and we are learning which are the most useful, and why. These efforts give rise to new metrological methods, and in turn, those new measurements enable better designs for sources and detectors, as discussed below.
Detectors that can register individual photons are key to applications in quantum information, metrology, biology, and remote sensing, each having its own distinct detection requirements. In many ways, a single-photon detector is the device that spans the quantum-to-classical divide, giving classical researchers access to many quantum effects. We are developing and characterizing detectors with single-photon and photon-number-resolving capabilities. Research efforts include solid-state detectors such as avalanche photodiodes and the operational electronics that enables much of their performance, cryogenic detectors based on microbolometers with true photon-number-resolution, and the multiplexing of these detectors for significant advantages in maximum count rate and deadtime over individual detectors.
There are many physical processes that can be exploited to enable single-photon detection, including impact ionization, extremely temperature-dependent transitions to superconductivity in metal films, disruption of super-currents in nanowires, and others. There are advantages and disadvantages associated with each technology; there is no ideal detector for single photons. It is often important to treat single-photon detectors as devices that perform a quantum measurement, so characterization of a positive operator-valued measure (POVM) is also necessary. That is, single-photon detectors are treated as devices that make a projective (von Neumann-like) measurement on the quantum state of light. We use statistical methods to create and refine the measurements, and currently hold a world record for a verified calibration of a single-photon avalanche diode (SPAD).
A light source that emits photons by pairs: one in each of the conjugated modes can be used to calibrate and characterize a single-photon detector. Notably, such a source of photon pairs is all what’s needed – this method in theory requires no pre-calibrated components, i.e. the method is an absolute calibration technique.
We made a metrological verification of the correlated photon method for measuring photon counting detection efficiency to an uncertainty of ~0.15 % (k = 1). [From Polyakov and Migdall, Optics Express Vol. 15, Issue 4, pp. 1390-1407 (2007) https://doi.org/10.1364/OE.15.001390]
The overall response of a single-photon detector to an incident photon is complex, and this makes detector performance characterization similarly complex. Not only is the photon transduced, with a finite probability, to some form of material excitation, that excitation is just the first step in internal and/or system-level processes that ultimately produce a measurable output signal for the user. And then the detector must go through some recovery and reset process that returns the detection system to its initial, fully-armed state. There can also be internal processes, both dependent on, and independent of, the prior history of the detection system that generate noise or otherwise affect the output. All of these processes determine the performance of the detector and require numerous metrics to provide as complete a characterization of the detector as possible. For example, an actively quenched SPAD is characterized, in part, by its detection efficiency, dark count rate, total afterpulse probability, the afterpulsing profile’s exponential lifetime(s), dead time, reset time, and how it handles detections that occur during the reset process (sometimes called twilight events). Measuring each of these independently is difficult, especially in a manner that identifies the dependence of these metrics on the prior history of the detector.
We have developed a simple technique to accurately characterize these metrics, and their interdependence. In this method a detector is illuminated continuously by a Poisson source and all the detection events output by the detector are time-tagged and recorded. Central to this method is a time-tagging system that will not miss any of the detector’s outputs. From these recorded events, and with knowledge of the input photon statistics, the detector’s first-order parameters (such as efficiency, dark count rate, recovery times) can be extracted. And by autocorrelation second-order and higher effects, such as afterpulsing, and afterpulses of afterpulses can also be accurately characterized. With sufficient measurement times (e.g. tens of minutes for the second-order and tens of hours for higher order effects), the statistics from this simple measurement technique can be excellent. We have applied this method to a variety of Si-SPADs and identified heretofore unreported effects due to how some SPADs respond to twilight events. This method and setup is compact and can easily be disseminated to others. We are currently developing low-cost time-tagging systems suitable for this application.
By learning what processes affect the performance of single-photon detectors, we understand what would need to be changed or improved to achieve higher detector performance. One example is in our work to develop high count-rate SPADs. SPADs are convenient, room-temperature, low-cost, and high performing single-photon detectors. They operate with the diode reverse biased above its breakdown voltage, so that a single injected charge carrier (due to the absorption of a single-photon) is sufficient to trigger a runaway impact ionization process, or avalanche, thereby providing the gain to amplify a single carrier to a macroscope current. The amount of current that flows through the diode has a direct impact on afterpulsing, in which a secondary false detection event is triggered by the release of a carrier that remained trapped within the diode after recovery from a prior detection. The probability of a carrier being trapped, and thus the afterpulse probability, is directly proportional to the amount of charge that flows during an avalanche. Afterpulsing is typically suppressed by applying a hold-off time after an avalanche: holding the SPAD below its breakdown voltage for a period of time long enough to release the vast majority of trapped carriers. But the hold-off time obviously limits the maximum achievable count rate.
We developed a bias gating technique that seeks to operate at the minimum amount of charge per avalanche required to be discriminated from the noise floor of the detector. In this technique, the detector is biased above breakdown only for very short periods of time, typically a few hundred picoseconds, at a high rate (>1 GHz). Thus, the detector is only sensitive to single photons during the short bias gate. This also means that the avalanche current will only grow for the duration of the gate. In this manner, the total amount of avalanche charge is greatly curtailed by the duration of the gate, but this comes at the expense of the detector only being active for short periods of time, albeit at a high frequency. This is a highly effective way to suppress avalanche charge, and thus afterpulsing in SPADs. The challenge, however, is to discriminate the minute avalanche signal from the large systematic signals produced by the bias gate waveform. We use RF interferometry to both form short bias gates (by Fourier synthesis) and to suppress the systematic signal at the output (by destructive interference). This approach has been used to demonstrate InGaAs/InP SPADs with dramatic improvements in detection efficiency and maximum count rates, and high-efficiency (thick) Si SPADs capable of low-noise high-efficiency photon counting at rates above 100 million counts per second, a world record for detectors of this type.
As with detectors, sources at the single photon level are critical for many applications. We develop new sources of single photons and entangled light, particularly those based on spontaneous parametric down-conversion, and learn how to characterize these and other sources. Chip-scale (therefore easy to integrate) photon-pair sources that emit into a single spatial mode (therefore easy to collect) offer significant advantages in the goal for workable photonic computation systems. Among various contenders, waveguided spontaneous parametric down-conversion is a promising candidate for the above goal because it can generate correlated photons predominantly in a single spatial mode while having a very small footprint. We systematically study such a waveguide-based photon pair source by measuring important performance parameters such as the temperature dependence of photon-pair production and the amount of quantum entanglement produced, which is critical for the success of quantum information applications.
Some sources are complex and mixed. That is, their output is a mixture of one or more thermal, single-photon fields, and a laser field. We have developed a simple method to reconstruct the mode distribution of multimode classical and nonclassical optical fields using photon-number statistics. Knowing the underlying number and structure of occupied modes of a light field plays a crucial role in minimizing loss and decoherence of quantum information.
Conjugated quantum sources are very important for quantum information applications. Particularly, bright conjugated sources are sought to improve scalability of practical quantum information processing, yet characterization methods that are typically used for single-photon sources do not apply. We have extended our in-situ characterization method based on photon number resolved counting to the case of conjugated sources. Here, both conjugated outputs are measured with photon number resolving detectors and a joint photon number distribution is obtained. This joint photon number distribution is then analyzed to identify modes and their correlations. We also developed the complete software suite to aid this analysis, available for download.
When detecting photons one by one, the precise time of each photon's detection is available. We use statistical methods, such as correlation functions, to gain insight into the physics of different light sources without examining the sources themselves. That is, the only information we use is the statistics of the light produced by that source: i.e. we are only allowed to look, but not allowed to touch. In fact, in many cases light emitted by a light source is the only available asset to examine: a source can be buried deep in a solid-state crystal, be it a diamond, or a semiconductor, or something else.
We recently developed a method where we use time-resolved correlation functions to determine if the properties of a nonclassical field emitted by a source change during its excitation-emission cycle. Coupled with a physics model of a source, one can assess the underlying mechanisms and dynamics of light emission.
Another method that we recently introduced uses high-order photon number statistics to separate contributions from different modes of a mixed emitter. Particularly, we can reconstruct the contribution from a single-photon emitter (or a few emitters) over a (often much stronger) classical background, such as scattered laser light.
A motivating factor for source and detector development is our goal to tie our current radiometry standards, which are designed for relatively high power levels, to detectors and sources that operate at the few photon level. Making this connection offers the possibility of major improvements in the uncertainties over current methods. For example, we have verified a correlated-photon detector efficiency technique against conventional measurement to the best levels of uncertainty yet achieved. This resulted in the development of a number of transfer-standard detectors designed specifically for the low light levels encountered in photon-counting measurements. These detectors will be made available to outside users. We also use correlated photons as a tool to measure source spectral radiance. With this method we demonstrated fundamentally absolute radiometry measurements as far into the infrared as 8 μm while using only an uncalibrated visible detector.