What is quantum efficiency?
Quantum efficiency (QE) is the percentage of incident photons that an imaging device can convert into electrons. For example, if an image sensor has 75% QE and is exposed to 100 photons, it will be able to convert to 75 electronic signals.
The QE is different for each sensor technology, with high-end image sensors achieving 95% QE. However, it is determined by the wavelength of the detected light and the semiconductor material. For CCD, EMCCD, (em)ICCD, and sCMOS technologies, 95% QE may be achieved in some wavelength ranges, but photons in the near-red and ultraviolet regions of the visible spectrum have lower QEs, so the efficiency of the sensor is lower. To improve QE in these areas, Deep-depleted silicon sensors and coated sensors have been developed, increasing QE.
Silicon sensors
Most image sensors are made of silicon. Since QE depends on the material, the properties of the element and how it interacts with light are very important.
In high-purity crystal forms, adjacent silicon atoms are covalently bonded to each other. Breaking these keys to produce an electron/hole pair (~1.1 eV) requires energy greater than the band gap energy. The wavelength of incident light is directly related to the depth of absorption of photons; the shorter the wavelength, the shorter the depth into the silicon.
Deep-depleted silicon sensors are thicker than conventional silicon sensors and are therefore capable of detecting light with longer wavelengths (i.e. > 700 nm, NIR). NIR light penetrates deeper in silicon than a typical silicon sensor, so the silicon sensor is effectively transparent to incident NIR light without depth depletion. As shown in the figure below, Deep-depleted silicon sensors provide >90% QE in the range of 700 – 850 nm, while conventional silicon sensors provide >60% QE.
To further improve QE, the orientation of the sensors in the device can be changed by means of a front-illuminated or back-illuminated device. Incident light from a front-illuminated device usually enters the sensor through a door in parallel registers. These gates consist of very thin polysilicon that are fairly transparent at long wavelengths but become opaque at wavelengths shorter than 400 nm. Therefore, at short wavelengths, the gate structure attenuates incident light.
If the silicon sensor is uniformly thinned, the image can be focused on the rear end of the sensor without a gate structure. Because the gate structure has no light limitation, the back-illuminated device exhibits high sensitivity to light, making 95% QE possible. Image sensors using front-illuminated technology, where incident light must pass through microlenses and metal wires before hitting the sensor, reducing maximum quantum efficiency. The incident light of the back-illuminated image sensor first illuminates the sensor, so the QE of the device does not decrease.
InGaAs sensors
Semiconductors detect photons only when they have higher or shorter wavelengths than the band gap energy of the material. InGaAs sensors are semiconductors made from alloys of InAs and GaAs, and traditional InGaAs sensors have an x:1-x InAs:GaAs ratio. Since InGaAs are not naturally occurring materials, single crystals must be grown on the InP substrate.
InGaAs sensors typically have less bandgap energy than silicon, which means they are capable of detecting longer wavelengths, such as the short-wave infrared (SWIR) region (900-1700 nm). As a result, the QE of the InGaAs camera in the 950-1600 nm region > 80%. Shows the QE curve of a typical InGaAs sensor. By increasing the concentration of InAs within a single crystal, the cutoff wavelength can be extended to 2600 nm.
Although InGaAs cameras have high QE in the 900 – 1700 nm range, the remote wavelength cutoff decreases as the device cools. For every 10 degrees Celsius of cooling, this is typically offset by 8 nanometers. This means that maximizing the throughput of the photons entering the device is important, but this shift in remote cutoffs can be advantageous because it allows the sensor to act as an "tunable" low-pass filter.
summary
QE is a measure of the effectiveness of a device in converting incident photons into electrons. Not only does the QE wavelength depend, it also depends on the sensor material.
If the energy is higher than the semiconductor band gap energy, the sensor will detect incident photons. This is why silicon has a 95% QE between 500-600 nm, but a lower QE for longer infrared/shorter violet wavelengths, while InGaAs has a high QE in the SWIR range (900-1700 nm) instead of the visible region or mid-infrared wavelength range (> 1700 nm).