What is responsivity in a photodetector?

Responsivity R (A/W) is the ratio of photocurrent to incident optical power. It is computed as R = QE × e × λ / (h × c), where QE is quantum efficiency, e is electron charge, λ is wavelength, h is Planck's constant, and c is the speed of light. A Si photodiode at 850 nm typically has R ≈ 0.6 A/W.

What is NEP (Noise Equivalent Power)?

NEP is the optical power that produces a signal-to-noise ratio of 1 in a 1 Hz bandwidth. It is given in units of W/√Hz. A smaller NEP means the detector can sense weaker light. NEP is calculated from the detector noise current divided by its responsivity. High-performance InGaAs detectors can achieve NEP below 10⁻¹⁴ W/√Hz.

How does an APD differ from a regular photodiode?

An APD (Avalanche Photodiode) uses a high reverse-bias field to trigger impact ionization, achieving internal multiplication of the photocurrent by a factor M (typically 10–200). This improves sensitivity but also adds excess noise characterized by the excess noise factor F = kA·M + (2–1/M)(1–kA). APDs are widely used in optical fiber receivers and LiDAR systems.

When should I use a PMT instead of a solid-state detector?

A PMT (Photomultiplier Tube) offers gain of 10⁶ or more via dynode multiplication, enabling single-photon counting. It is the preferred choice for extremely weak light such as in fluorescence spectroscopy, scintillation detectors, and PET scanners. However, PMTs are bulky, require high voltage (hundreds to thousands of volts), and are sensitive only in the UV–visible range (typically 300–650 nm).

Photodetector Design Calculator — SNR, NEP, Detectivity D*

Parameters

Wavelength λ (nm) 850

Optical power P (µW) 10.0

Bandwidth BW (MHz) 100

Temperature T (°C) 25

APD Gain M 10

—

R (A/W)

—

Ip (µA)

—

SNR (dB)

—

NEP (pW/√Hz)

—

D* (×10¹⁰ Jones)

—

i_shot (pA/√Hz)

Theory Reference

Responsivity: $R = \frac{QE \cdot e \cdot \lambda}{h c}$

Shot noise: $i_{shot}= \sqrt{2eI_p \cdot BW}$

NEP: $\displaystyle NEP = \frac{\sqrt{4kTBW/R_L + 2eI_d BW}}{R}$

Detectivity: $D^* = \frac{\sqrt{A \cdot BW}}{NEP}$

What is Photodetector Performance?

🧑‍🎓

What exactly is "responsivity" in this simulator? I see it changes when I slide the wavelength.

🎓

Basically, responsivity ($R$) tells you how good your detector is at converting light into electrical current. It's measured in Amps per Watt (A/W). In practice, a higher $R$ means more current for the same amount of light. Try moving the wavelength slider above from 400 nm to 1000 nm for a Silicon detector. You'll see $R$ increase because each photon has less energy, so you get more electrons per watt of optical power.

🧑‍🎓

Wait, really? So if I increase the optical power ($P$) in the simulator, the photocurrent just goes up proportionally? What limits this?

🎓

Exactly right—photocurrent $I_p = R \times P$. But the limit is noise. Even in total darkness, there's tiny random current called dark current. And the signal itself has "shot noise" because light arrives as discrete photons. A common case is in fiber-optic receivers: even with a strong signal, this inherent noise sets the minimum detectable power. Adjust the "Bandwidth" and "Temperature" sliders to see how they affect the noise values in the results.

🧑‍🎓

Okay, so NEP and D* in the results... they seem to get worse with more noise. Which one should I look at to choose the best detector for a low-light application?

🎓

Great question! NEP (Noise-Equivalent Power) is the signal power that gives a signal-to-noise ratio of 1. Lower NEP is better. D* (pronounced "D-star") normalizes NEP for the detector's area and bandwidth, so you can compare different sized sensors. For instance, choosing between a small, sensitive APD and a large-area PMT for a LIDAR system? Compare their D* values in this simulator by switching the detector type and keeping other parameters the same.

Physical Model & Key Equations

The core relationship is the responsivity, which converts incident optical power into photocurrent. It depends on the quantum efficiency (QE) and the photon energy.

$$R = \frac{I_p}{P}= \frac{QE \cdot e \cdot \lambda}{h c}$$

Where $R$ is responsivity (A/W), $QE$ is quantum efficiency (unitless, set per detector material), $e$ is electron charge ($1.602 \times 10^{-19}$ C), $\lambda$ is wavelength (m), $h$ is Planck's constant ($6.626 \times 10^{-34}$ J·s), and $c$ is the speed of light ($3 \times 10^8$ m/s).

The total noise current determines the minimum detectable signal. It combines shot noise from the signal and dark current, plus thermal (Johnson) noise from the load resistor.

$$i_{noise}= \sqrt{i_{shot}^2 + i_{dark}^2 + i_{thermal}^2}= \sqrt{2e(I_p + I_d) \cdot BW + \frac{4kT \cdot BW}{R_L}}$$

Where $I_d$ is dark current (A), $BW$ is electrical bandwidth (Hz), $k$ is Boltzmann's constant, $T$ is temperature (K), and $R_L$ is load resistance (Ω). From this, we derive NEP = $i_{noise}/ R$ and D* = $\sqrt{A \cdot BW} / $NEP, where $A$ is detector area.

Real-World Applications

Fiber-Optic Communications: InGaAs photodetectors are standard in telecom receivers for 1310 nm and 1550 nm light. Engineers use calculations like these to balance responsivity and bandwidth to achieve the required data rate and bit-error rate for your internet backbone.

LIDAR and 3D Sensing: Avalanche Photodiodes (APDs) are favored here for their internal gain (M). Designing a LIDAR for a self-driving car requires optimizing the gain to amplify weak return signals without introducing excessive excess noise, which you can explore with the "APD Gain M" slider.

Scientific Spectroscopy: For analyzing faint starlight or fluorescent samples, Photomultiplier Tubes (PMTs) offer extremely high gain and low dark current. Their superior NEP and D* make them ideal for photon-counting applications where every single photon matters.

Consumer Electronics: Silicon photodiodes are everywhere, from the ambient light sensor in your phone that adjusts screen brightness to barcode scanners. The design trade-off is between sensitivity (high R) and speed (high BW), which are inversely related—a key relationship visible in the simulator outputs.

Common Misconceptions and Points to Note

First, while it's easy to think "a sensor with high sensitivity is a good sensor," this is a pitfall. For example, an InGaAs photodiode with high sensitivity R excels in the near-infrared but is inferior to silicon in the visible range. By using the tool to observe sensitivity while changing the wavelength, you can appreciate that "high sensitivity" is specific to a particular wavelength band. Next, understand the relationship between bandwidth BW and "response speed." A 10MHz bandwidth corresponds to a response time of approximately 35ns. For high-speed applications like communications, you would set a larger bandwidth, but you can use this simulator to confirm the trade-off of increased noise that comes with it. Finally, dark current Id is highly sensitive to temperature. Try lowering the tool's temperature T parameter from room temperature (300K) to a cooled state (e.g., 250K). You should see a significant improvement in NEP. In actual design, pursuing higher performance means considering the balance between cost and performance, as cooling mechanisms become necessary.

Related Engineering Fields

The calculation logic of this tool forms the foundation for various advanced engineering fields. One is LiDAR. The LiDAR in autonomous vehicles detects extremely weak light pulses that are reflected back. The SNR and NEP calculated here are core parameters that directly determine the maximum detection range and point cloud quality. Also, in biomedical imaging, such as brain function measurement using near-infrared light (fNIRS), very weak light transmitted through biological tissue is detected. Here, whether the detector's specific detectivity D* is sufficiently high determines the measurable depth and signal quality. Furthermore, it connects to quantum information technology. Receivers in quantum communication handling single-photon levels require "photon counters" that suppress dark current and thermal noise to the extreme, and the foundation for evaluating their performance is the noise concepts you learn here. Thus, photodetection is a "common language" across a wide range of measurement, communication, and sensing applications.

For Further Learning

As a first next step, we recommend delving deeper into "Types of Noise and Their Physical Origins". Beyond the shot noise and thermal noise handled by this tool, explore 1/f noise (flicker noise) and the multiplication noise (excess noise factor F) specific to APDs. Mathematically, understanding that shot noise follows a Poisson process and thermal noise follows a Gaussian distribution will help you grasp the reasoning behind noise combination methods (root sum of squares). Next, practice interpreting actual sensor datasheets using the tool's outputs, NEP and D*. For example, trying to replicate the NEP listed in a commercial photodiode's datasheet using the tool is excellent practice. Finally, to advance to a system-level design perspective, study the topic of "Receiver Design". Here, you'll consider how the amplifier's noise figure (NF) affects the overall system SNR—essentially, the "next" step in the calculation process of this tool.