Factors affecting Quantum Efficiency of High-Performance Cameras

Quantum Efficiency However, this calculation must factor in multiple

The quantum efficiency (QE) of a sensor is a measure of parameters including the size of the pixels in the detector
its ability to convert photons into a useful output signal. array, the noise sources associated with the camera
This is obviously a critical parameter to consider when system (e.g. read noise and dark noise), the number of
selecting which camera to use for a particular application, photons incident during the exposure time in addition
after all if a camera doesn’t respond to the light you are to the wavelength of the illumination. When using
trying to detect then there’s not much point in looking at QE plots to perform such calculations care must be
other specifications, such as array size, readout noise, taken to ensure the units of the quoted QE are
cooling performance, etc...! understood. For example, is the graph showing relative
For imaging devices, the QE is defined as the percentage of QE, detective QE or an absolute QE figure? Also is it
photons incident per pixel producing a charge a fractional figure or a percentage?
(usually electrons) which is successfully stored within the In some instances, customers can find themselves
pixel. For example, if a sensor has 75% QE and was looking at different cameras using the exact same image
illuminated with 100 photons per pixel during an exposure, sensor and yet still see differences in the QE plots
then in the absence of any noise sources, each pixel provided by the different vendors. How can this be the
would contain a charge of 75 electrons due to 75 photons case?
being absorbed and turned into charge (assuming each
detected photon produces one electron within the
device). The remaining 25 photons do not generate a
measurable charge within the photosensitive volume of
the device, possibly due to reflection losses,
recombination effects, absorption in a non-
photosensitive region or transmission through the device.
Determining whether or not a camera is going to be able
to capture images of the quality required by your
particular application, can be extremely difficult when
looking at the datasheet alone, as not all manufacturers
specify the cameras response to light in the same units
or under the same conditions. Vendors may quote a
detectors photosensitivity in radiometric units such as Figure 1: Absolute quantum efficiency for a back-illuminated, standard
silicon, CCD with a ‘Green’ optimized AR coating (courtesy of Teledyne e2v).
amps per watt while other may use photometric units,
such as volts per lux-second. Historically the
Examples of factors affecting Sensor
sensitivity of film and digital cameras were reported as
an ISO number however for low light scientific cameras Quantum Efficiency
Raptor Photonics (and many other vendors) now choose Sensor Material and Structure
to provide a graph of quantum efficiency as a function of QE is obviously going to be highly dependent upon
wavelength, as shown in Figure 1. the materials from which the image sensor is
This type of plot should make it possible to determine if constructed. In order to produce a detectable signal
a detectable signal will be produced within the sensor from any photons, firstly the photons must make it
for a given set of illumination conditions. to the photosensitive region of the sensor.

Technical Note
www.raptorphotonics.com | Factors that affect QE Tech Note Rev 1.0
Secondly the photons (ideally all of them) should be Anti Reflection Coatings
absorbed within this photosensitive region, each producing a
As described above, using back-illumination can minimize
charge stored within the pixel where they were absorbed,
absorption of photons before they reach the photosensitive
before being read out to produce an image. In practice the
region of the image sensor. However, silicon is a high
ideal situation is not achievable as there are many loss
refractive index material and therefore reflection at the
mechanisms which limit a sensors ability to convert incident
detector surface can prevent a large fraction of incident
photons into charge.
photons from being detected, as they never reach the
The majority of scientific sensors used for the detection
photosensitive region of the sensor. One way to mitigate
of visible wavelengths are made from doped silicon.
this effect is to apply an anti-reflection (AR) coating to the
The interaction of light with the silicon material
surface of the sensor thereby minimizing reflection losses
determines the limits of the useful detection range in such
and allowing transmission of more photons into the device.
devices. As the photon wavelength increases the
AR coatings can be tuned (by the sensor manufacturer) to
absorption length also increases, meaning that typically
optimize the sensor QE for different regions of the UV-
photons will travel further into the silicon before being
visible-NIR spectrum on silicon devices. Single layer
absorbed (and generating charge). This is OK until the
coatings can provide a dramatic increase in sensor QE
absorption length is so long that most of the absorption
across a wide range of wavelengths but generally have a
occurs outside of the photosensitive region of the device,
peak at a certain region of the spectrum. Multi-layer AR
i.e the detector is becoming increasingly transparent
coatings can also be applied to give a broader ‘top-hat’
to the longer wavelength photons. Under these
shape to the sensor QE curve.
conditions most absorbed photons no longer produce a
In some applications, such as direct detection of soft x-rays,
charge which is stored in the pixels of the device and as a
uncoated devices may be preferred / required, as the AR
result the QE plot for silicon detectors falls off at longer
coating can prevent some photon energies reaching the
wavelengths. Increasing the depth of the photosensitive
photosensitive region of the detector.
region (such as using deep depletion or red enhanced
Detectors fabricated from other materials, such as InGaAs,
silicon as the base material for the device) can
also benefit from the use of AR coatings to minimize
increase QE in the near-IR part of the spectrum for
reflections.
silicon-based sensors, however increasing the photon
Cover Glass / micro-lenses
wavelength beyond ~1150nm means that the photon
Many sensor packages and camera systems will protect the
energy becomes smaller than the band gap of silicon.
sensor from dust / foreign object debris with a transmissive
At these wavelengths absorption of a photon does not
glass cover. In the case of cooled sensors, the window
provide enough energy to produce charge within the pixel
material forms part of the sensor enclosure used to prevent
and so the QE falls towards zero. Detection of these
condensation forming on the sensor surface. The optical
wavelengths is possible using a different sensor
properties of the material used for the cover glass or
material, such as InGaAs which has an excellent
entrance window will obviously affect the QE of the overall
response for wavelengths up to approximately 1700nm.
camera system as all photons must pass through it before
At the blue end of the visible spectrum (and into the UV)
reaching the detector, so reflection and absorption losses at
the absorption length in silicon decreases and so most
this stage must be minimized. A range of materials are used
photons will be absorbed near the surface on which they
for this purpose including BK7 glass, fused silica and
are incident. Front illuminated CCDs have
sapphire, again AR coated components are typically used to
polysilicon gates on the illuminated surface and this
minimize reflection losses.
electrode structure prevents the photons from
Most scientific CCD sensors (excluding interline devices)
reaching the photosensitive region, resulting in the
have close to 100% fill factors. This means virtually 100% of
characteristic fall off in QE at shorter wavelengths.
the pixel area has a response to light intrinsic to the material
Back-illuminated sensors avoid this issue as the
from which the sensor is made. However, CMOS pixel
illumination does not have to pass through any
designs typically have a small photosensitive area in each
electrode structure before reaching the photosensitive
pixel surrounded by on-chip circuitry / components which
region, however the back thinning process must be
are not sensitive to light. This produces pixels which have a
done well to minimize any oxide layer and ‘dead-region’
small fill factor and a device with a poor response to light, as
near the surface of the device to provide high QE in
the number of photons detected per pixel area would be a
both the UV and soft x-ray regions.
small fraction of the total number incident per pixel.

ISO 9001
Registered
Quality
Management
Thankfully micro-lens arrays can be used to steer a high
fraction of the light incident on the pixel into the photosensitive
region and in some cases achieve peak QE values of ~80%.
Typically micro-lens materials have poor transmission in the UV
region and must be designed carefully to maximize the
wavelength range over which they effectively steer light into
the photosensitive area. Attempting to couple another optical
component directly onto a micro-lens array, eg. by using
coupling gel or adhesive will negate the effectiveness of
the micro-lenses and revert the sensor to a poor response / low
fill factor state.
Micro-lens arrays are also used on interline CCD devices, again
to steer light away from the on-chip structures which are
not sensitive to light, in this case the vertical shift registers,
and achieve high fill factor pixel performance.
Temperature
Temperature is another parameter that will alter the QE of an Figure 2: Temperature variation of the absolute quantum
imaging device. Many cameras are designed to operate with efficiency for a back-illuminated, deep depletion CCD with a
the image sensors cooled to relatively low temperatures ‘NIR’ optimized AR coating (courtesy of Teledyne e2v).
<-80°C. Such cooling can be necessary to reduce the
Conclusion
thermally generated dark signal produced within the
It can be very tricky to evaluate and compare different
pixels of the imaging device in the absence of any light.
camera sensitivities based solely on the manufacturers
This dark signal could otherwise limit the usable dynamic
datasheets. When attempting such calculations it is
range of the sensor and introduce a large dark noise
important to ensure like for like comparisons are being
component resulting in reduced image quality.
made, this requires the units for the various camera
The downside to cooling the sensor is that the QE of the
performance parameters to be identical and unambiguous.
device actually decreases the more the image sensor is
In addition care should be taken to check that the quoted
cooled, this is particularly evident in the NIR response of
parameters, such as QE, are actually applicable to the
silicon-based detectors as shown in Figure 2. There are many
device under the operating conditions intended to be
factors contributing to this QE change, however a primary
used. It may be worthwhile also evaluating
cause is that the band gap increases slightly at lower
which operating conditions are best suited to your
temperatures, reducing the probability of a photon being
application, for example in some NIR detection
absorbed and creating an electron-hole pair.
applications, if short exposures are to be used (i.e.
Many camera vendors provide QE plots for the image sensors
dark signal will have a minimal impact on the
used within their cameras which represent the device
acquired images) it may be advantageous to
performance at room temperature, however they neglect to
maintain the sensor at a slightly warmer
mention that this is not the response the customer will obtain
temperature than the headline cooling figure on the
when the sensor is cooled to the operating temperature within
camera datasheet, so that you can utilize the higher
the camera. When comparing QE graphs from different
QE values and still not suffer from dark
datasheets it is very important to confirm that the plot
signal effects.
illustrates the performance of the device at the actual
Determining the best selection of camera for your
operating temperature used.
application can be a challenging and complex decision to
make, our experienced engineering team at Raptor
Photonics would welcome the opportunity to assist
camera users in making this choice, please contact us
directly at [email protected]

Raptor UK (Headquarters) Raptor USA

T: +44(0)2828 270 141 T: 1(877) 240-4836
E: [email protected] E: [email protected]
www.raptorphotonics.com www.raptorphotonics.com

ISO 9001
Registered
Quality
Management