Single-Photon Avalanche Diodes (SPADs) are highly sensitive detectors capable of registering individual photons. When operated in "Geiger mode"—biased above their breakdown voltage—a single photogenerated carrier can trigger a massive impact ionization cascade, creating a macroscopic, easily measurable current.
However, this violent avalanche process comes with a major drawback: afterpulsing. During an avalanche, millions of charge carriers flood the diode's multiplication region. Some of these carriers become trapped in deep-level material defects. When they are spontaneously released moments later, they can re-trigger the avalanche, creating a false count.
To mitigate afterpulsing, the total number of charge carriers passing through the diode per avalanche must be strictly limited. The most elegant way to achieve this is by integrating negative feedback directly into the device architecture to quench the avalanche prematurely.
The Problem with Full Capacitive Discharge
In a conventional SPAD circuit, the diode is connected to external quenching electronics. When an avalanche occurs, the internal capacitance of the diode, along with any parasitic capacitance from the surrounding interconnects and packaging, undergoes a full discharge.
The total charge $Q$ flowing through the device during this discharge is proportional to the total capacitance $C$ and the overvoltage $\Delta V$:
$$Q = C \cdot \Delta V$$
In standard configurations, a full capacitive discharge releases roughly $1 \times 10^6$ carriers or more per event. This massive volume of charge floods the lattice defects, guaranteeing high afterpulsing rates unless long, performance-limiting "hold-off" times are imposed before resetting the detector. To minimize afterpulsing, we must either minimize the capacitance $C$ or stop the discharge early by rapidly dropping $\Delta V$.
Quenching via Monolithic Negative Feedback
To stop the avalanche before full discharge occurs, researchers introduced the Negative Feedback Avalanche Diode (NFAD). In this architecture, a passive negative feedback element—specifically a thin-film resistor—is monolithically integrated directly onto the semiconductor chip next to the avalanche diode.
[ Bias Voltage ] ───[ Integrated Resistor ] ───[ Avalanche Diode ] ─── [ Ground ]
│
(Internal Node)
The mechanism works through a localized voltage-divider effect:
- Avalanche Initiation: A photon triggers a rapid rise in avalanche current ($I$).
- Negative Feedback: As this current flows through the integrated resistor ($R$), it creates an immediate voltage drop ($I \times R$) across the resistor.
- Self-Quenching: Because the total bias voltage is fixed, the voltage drop across the avalanche diode itself drops rapidly ($V_{\text{diode}} = V_{\text{bias}} - IR$). This lowers the internal electric field below the breakdown threshold, quenching the avalanche.
As detailed by Mark A. Itzler et al. (2010) and Mark A. Itzler et al. (2010), operating in a high negative feedback regime results in a partial capacitive discharge. Instead of emptying the entire capacitance, the avalanche is choked off so quickly that the measured charge flow drops to approximately $3 \times 10^5$ carriers—substantially lower than conventional designs. Fewer carriers mean fewer trapped charges, directly suppressing afterpulsing.
The Critical Role of Minimizing Parasitics
The success of this negative feedback loop hinges entirely on proximity. If a designer attempts to use an external, discrete resistor instead of a monolithically integrated one, the stray parasitic capacitance of the wire bonds and metal traces sits between the diode and the resistor. The diode will fully discharge this local parasitic capacitance before the resistor can exert any feedback, defeating the purpose.
To combat this, Tommaso Lunghi et al. (2012) emphasized that integrating feedback resistors directly onto the semiconductor surface minimizes parasitic capacitance to the absolute limit. This layout ensures that the negative feedback loop acts almost instantaneously, restricting the total charge flow and enabling clean, free-running single-photon detection with minimal noise.
Where monolithic integration is not fully utilized, packaging innovations can bridge the gap. Chong Hu et al. (2010) demonstrated that employing low-parasitic chip-to-chip wire bonding between a SPAD and its quenching circuit significantly lowers the total circuit capacitance. By minimizing this parasitic contribution, they achieved a remarkable reduction in avalanche charge, yielding an afterpulsing probability of just 0.3%.
Unlocking Photon Number Resolution
Beyond merely reducing noise, controlling the avalanche charge opens up an entirely new capability: photon number resolution.
In a standard SPAD, an avalanche is a binary event; the detector fully discharges regardless of whether one or three photons triggered it. However, because the integrated negative feedback in an NFAD regulates the avalanche process deterministically before it runs wild, the total charge output becomes sensitive to the initial number of triggering carriers.
As noted by Mark A. Itzler et al. (2009) and Mark A. Itzler et al. (2010), the quantized nature of the initial detection event can be preserved in the output charge pulse. This allows researchers to distinguish between single and multi-photon arrivals, transforming a traditionally binary "click" detector into a powerful tool for advanced quantum optics.
Key Foundational Work
Progress in self-quenching InP-based single photon detectors (2010) — contributed 6 concept elements to this direction.
Single-photon detectors based on InP avalanche diodes: status and prospects (2010) — contributed 5 concept elements to this direction.
Negative feedback avalanche diodes for near-infrared single-photon detection (2009) — contributed 5 concept elements to this direction.