Академический Документы
Профессиональный Документы
Культура Документы
Astronomical Detectors
Learning Objectives
Astronomical detectors:
- major functions
- main characteristics
Human eye
Photographic film/plate
Photomultiplier tube
- photoelectric effect
- photomultiplier
Operation of Charge-Coupled Devices:
- band theory of solids
- semiconductor
- metal oxide semiconductor (MOS) capacitor
- charge coupling
Learning Objectives
Properties of Charge-Coupled Devices:
- plate scale
- quantum efficiency
- charge transfer efficiency
- output
- binning
Noise considerations for Charge-Coupled Devices:
- photon noise
- readout noise
- dark noise
Reducing CCD data:
- bias frame
- dark frame
- flatfield frame
Learning Objectives
Astronomical detectors:
- major functions
- main characteristics
Human eye
Photographic film/plate
Photomultiplier tube
- photoelectric effect
- photomultiplier
Operation of Charge-Coupled Devices:
- band theory of solids
- semiconductor
- metal oxide semiconductor (MOS) capacitor
- charge coupling
Functions of Astronomical Detectors
Astronomical detectors either initiate a chemical change in a compound (e.g., eye
retina, photographic film or plate) or transform energy from electromagnetic
radiation to electrical charge (e.g., photomultiplier tube, charged coupled device).
What are advantages of man-made astronomical detectors over human eyes?
Functions of Astronomical Detectors
Astronomical detectors either initiate a chemical change in a compound (e.g., eye
retina, photographic film or plate) or transform energy from electromagnetic
radiation to electrical charge (e.g., photomultiplier tube, charged coupled device).
What are advantages of man-made astronomical detectors over human eyes?
- higher intrinsic sensitivity. Human eye detects about 3 (daytime) to 10
(nighttime) out of every 100 incident optical photons. Similarly,
photographic film/plate also detect ~1-10 out of every 100 incident optical
photons. By contrast, CCDs can detect up to about 90 out of every 100
incident optical photons.
Functions of Astronomical Detectors
Astronomical detectors either initiate a chemical change in a compound (e.g., eye
retina, photographic film or plate) or transform energy from electromagnetic
radiation to electrical charge (e.g., photomultiplier tube, charged coupled device).
What are advantages of man-made astronomical detectors over human eyes?
- higher intrinsic sensitivity. Human eye detects about 3 (daytime) to 10
(nighttime) out of every 100 incident optical photons. Similarly,
photographic film/plate also detect ~1-10 out of every 100 incident optical
photons. By contrast, CCDs can detect up to about 90 out of every 100
incident optical photons.
- longer exposure (integration) time, thus collect more photons (signal).
Human eye integrates for only 10-15 ms (daylight) to 0.1-0.2 s (nighttime).
Functions of Astronomical Detectors
Astronomical detectors either initiate a chemical change in a compound (e.g., eye
retina, photographic film or plate) or transform energy from electromagnetic
radiation to electrical charge (e.g., photomultiplier tube, charged coupled device).
What are advantages of man-made astronomical detectors over human eyes?
- higher intrinsic sensitivity. Human eye detects about 3 (daytime) to 10
(nighttime) out of every 100 incident optical photons. Similarly,
photographic film/plate also detect ~1-10 out of every 100 incident optical
photons. By contrast, CCDs can detect up to about 90 out of every 100
incident optical photons.
- longer exposure (integration) time, thus collect more photons (signal).
Human eye integrates for only 10-15 ms (daylight) to 0.1-0.2 s (nighttime).
- PMTs/CCDs respond linearly to input photon signal (i.e., output input).
Response of human eye and photographic plate/film is non-linear.
Functions of Astronomical Detectors
Astronomical detectors either initiate a chemical change in a compound (e.g., eye
retina, photographic film or plate) or transform energy from electromagnetic
radiation to electrical charge (e.g., photomultiplier tube, charged coupled device).
What are advantages of man-made astronomical detectors over human eyes?
- higher intrinsic sensitivity. Human eye detects about 3 (daytime) to 10
(nighttime) out of every 100 incident optical photons. Similarly,
photographic film/plate also detect ~1-10 out of every 100 incident optical
photons. By contrast, CCDs can detect up to about 90 out of every 100
incident optical photons.
- longer exposure (integration) time, thus collect more photons (signal).
Human eye integrates for only 10-15 ms (daylight) to 0.1-0.2 s (nighttime).
- PMTs/CCDs respond linearly to input photon signal (i.e., output input).
Response of human eye and photographic plate/film is non-linear.
- provide a record that can be shared.
Functions of Astronomical Detectors
Astronomical detectors either initiate a chemical change in a compound (e.g., eye
retina, photographic film or plate) or transform energy from electromagnetic
radiation to electrical charge (e.g., photomultiplier tube, charged coupled device).
What are advantages of man-made astronomical detectors over human eyes?
- higher intrinsic sensitivity. Human eye detects about 3 (daytime) to 10
(nighttime) out of every 100 incident optical photons. Similarly,
photographic film/plate also detect ~1-10 out of every 100 incident optical
photons. By contrast, CCDs can detect up to about 90 out of every 100
incident optical photons.
- longer exposure (integration) time, thus collect more photons (signal).
Human eye integrates for only 10-15 ms (daylight) to 0.1-0.2 s (nighttime).
- PMTs/CCDs respond linearly to input photon signal (i.e., output input).
Response of human eye and photographic plate/film is non-linear.
- provide a record that can be shared.
- detect electromagnetic radiation at wavelengths not detectable by the human
eye.
In this course, we will only discuss detectors at optical (with occasional forays
into near-infrared and near-ultraviolet) wavelengths.
Characteristics of Astronomical Detectors
Quantum efficiency (QE):
- QE = number of detected photons/number of incident photons
- detections can be chemical changes (eye, photographic emulsion),
photoelectrons released (PMT), and charge-pairs created (CCD)
- generally a function of wavelength
Spectral bandwidth (or simply bandwidth) -
wavelength range over which photons can be detected
Characteristics of Astronomical Detectors
Linearity:
- response (detector output) linearly proportional to incident number of photons
(energy delivered to detector per unit time QE integration time)
- examples of non-linear detectors are human eyes and photographic emulsions
- examples of linear detectors are PMTs and CCDs, although in practice over
limited range of
input levels Starlight Xpress SXV-H9 CCD camera
Characteristics of Astronomical Detectors
Dynamic range (or contrast, in popular usage):
- ratio between largest and smallest values of detector output
- overall dynamic range may differ from dynamic range over which the
response is linear
QE of rods is ~10%,
QE of cones is ~3%.
Human Eye
Note that eye does not respond over
full adaptation range at a given time,
but over a limited range as indicated by
dashed line
Dynamic range: ~1,000 to ~10,000
depending on scene brightness and
complexity
Do you now understand why a lighted
room looks dark from the outside on a
sunny day, but looks bright from the
outside at night?
Human Eye
Note that eye does not respond over
full adaptation range at a given time,
but over a limited range as indicated by
dashed line
Dynamic range: ~1,000 to ~10,000
depending on scene brightness and
complexity
Do you now understand why a lighted
room looks dark from the outside on a
sunny day, but looks bright from the
outside at night? Outside on a sunny
day, the eye operates at a relatively
high level in its adaptation range. A
lighted room below the eye threshold
therefore appears to be dark.
Human Eye
Note that eye does not respond over
full adaptation range at a given time,
but over a limited range as indicated by
dashed line
Dynamic range: ~1,000 to ~10,000
depending on scene brightness and
complexity
Do you now understand why a lighted
room looks dark from the outside on a
sunny day, but looks bright from the
outside at night? Outside at night, the
eye operates at a relatively low level in
its adaptation range. A lighted room
above the eye threshold therefore
appears to be bright.
Learning Objectives
Astronomical detectors:
- major functions
- main characteristics
Human eye
Photographic film/plate
Photomultiplier tube
- photoelectric effect
- photomultiplier
Operation of Charge-Coupled Devices:
- band theory of solids
- semiconductor
- metal oxide semiconductor (MOS) capacitor
- charge coupling
Photographic Film/Plates
Photography invented in 1800s; use in astronomy became popular in 1900s.
Photographic film/plate comprises a thin coating of silver halide (e.g. AgBr,
micron-size crystals) suspended in a gelatin emulsion on a
plastic film/glass plate. Group 17 of
periodic table
When a photon strikes, (halogens)
Ag Br (crystal ) h (radiation) Ag Br e
The silver ion can then combine with the freed electron
Ag e Ag (atom)
Photographic Film/Plates
Chemicals used to remove the silver halide (fixing), leaving behind the
metallic silver. Metallic silver, which is opaque to optical light, makes up the
latent image (the negative).
QE of ~1-2%, reaching up to
~10% with special sensitizing
techniques.
Can integrate longer than
human eye to detect fainter
objects.
Photographic Film/Plates
Response of photographic film/plate is non-linear. At low light levels, response is
determined by physics of silver activation; as the film becomes progressively
more exposed, each incident photon is less likely to impact a still-unexposed grain.
Photographic Film/Plates
Silver Halide grains are manufactured by combining Silver Nitrate and Halide
Salts (Chloride, Bromide, and Iodide) in complex ways that result in a range of
crystal sizes, shapes, and compositions.
These primitive grains are then chemically modified on their surface to increase
their light sensitivity. The unmodified grains are only sensitive to the blue portion
of the spectrum, and they are not very useful in camera film.
Organic molecules known as spectral sensitizers are added to the surface of the
grains to make them more sensitive to blue, green and red light. These molecules
must adsorb (attach) to the grain surface and transfer the energy from a red, green,
or blue photon to the silver halide crystal as a photoelectron. Other chemicals are
added internally to the grain during its growth process, or on the surface of the
grain.
These chemicals affect the light sensitivity of the grain, also known as its
photographic speed (ISO, or ASA rating).
Photographic Film/Plates
When a photon of light is absorbed by the spectral sensitizer sitting on the surface
of a silver halide grain, the energy of an electron is raised into the conduction
band from the valence band, where it can be transferred to the conduction band of
the silver halide grain electronic structure. A conduction band electron can then
go on to combine with a positive hole in the silver halide lattice and form a single
atom of silver. This single atom of silver is unstable. However if enough
photoelectrons are present at the same time in the crystal lattice, they may
combine with enough positive holes to form a stable latent image site.
It is generally felt that a stable latent image site is at least 2 to 4 silver atoms per
grain. A silver halide grain contains billions of silver halide molecules, and it only
takes 2 to 4 atoms of uncombined silver to form the latent image site. Modern
color films generally take 20 to 60 photons per grain to produce a developable
latent image.
Learning Objectives
Astronomical detectors:
- major functions
- main characteristics
Human eye
Photographic film/plate
Photomultiplier tube
- photoelectric effect
- photomultiplier
Operation of Charge-Coupled Devices:
- band theory of solids
- semiconductor
- metal oxide semiconductor (MOS) capacitor
- charge coupling
Photoelectric Effect
Light incident on metallic surface release electrons provided light frequency
exceed threshold frequency. Ejected electrons are called photoelectrons. (More
generally, free electrons generated by the absorption of photons are called
photoelectrons.)
Increasing the light intensity increases the number of photoelectrons and hence
measured current. (The tube is evacuated to minimize collisions between
photoelectrons and air molecules.)
metal metal
plate plate
evacuated tube
Photoelectric Effect
By reversing the battery, measure the voltage required to prevent a current from
flowing. This voltage is a measure of the maximum kinetic energy of the
photoelectrons.
In contradiction to expectations if light have wave-like properties:
- increasing the light intensity (amplitude of the electromagnetic wave) does not
increase the maximum kinetic energy of the photoelectrons
- increasing the light frequency increases the maximum kinetic energy of the
photoelectrons
Photoelectric Effect
To explain the photoelectric effect, Einstein postulated that light interacts with the
metal (matter) not as a wave but as a particle (photon) with an energy given by
Plancks constant
frequency wavelength
A photon must have an energy exceeding the minimum energy necessary to eject
an electron from the metal surface before a photoelectron can be produced. This
minimum energy is known as the work function, W, of the metal.
Kinetic energy of photoelectron:
KEe =
Ephoton W = h W = h(-min)
threshold
frequency
Photoelectric Effect
To conserve linear momentum, should photoelectrons not be ejected on the other
side of the metal plate than drawn below? For this drawing to be correct, what
must have happened to photoelectrons released in the metal plate? Collided with
other atoms/electrons in the metal plate and bounced backwards.
Why does the current increase slowly with increasing light frequency?
metal metal
plate plate
evacuated tube
Photoelectric Effect
To conserve linear momentum, should photoelectrons not be ejected on the other
side of the metal plate than drawn below? For this drawing to be correct, what
must have happened to photoelectrons released in the metal plate? Collided with
other atoms/electrons in the metal plate and bounced backwards.
metal metal
plate plate
evacuated tube
Photoelectric Effect
Why does the current initially increase steeply with increasing light frequency?
metal metal
plate plate
evacuated tube
Photoelectric Effect
Why does the current initially increase steeply with increasing light frequency?
Sufficient kinetic energy to overcome collisions with other atoms/electrons in the
metal plate (work function) and escape. Work function is therefore a measure of
the minimum kinetic energy for photoelectrons to escape the plate.
metal metal
plate plate
evacuated tube
Photoelectric Effect
Why does the current then increase more slowly (up to a limit) with increasing
light frequency?
metal metal
plate plate
evacuated tube
Photoelectric Effect
Why does the current then increase more slowly (up to a limit) with increasing
light frequency? When majority of photoelectrons released in the metal plate
already has sufficient kinetic energy to escape, increasing the energy of incoming
photons only marginally increases the number of photoelectrons that escape the
metal plate.
metal metal
plate plate
evacuated tube
Photoelectric Effect
Light therefore has particle-like in addition to wave-like properties.
Light, in the form of electromagnetic waves, shows
its wave-like properties as it propagates through
space; e.g., refraction, diffraction, interference.
evacuated tube
Photomultiplier
Incident photons on photocathode (metal plate) releases photoelectrons.
Photoelectrons focused by electrode, then accelerated by first dynode (metal plate
biased to a voltage that is positive with respect to photocathode and electrode).
Impact of photoelectrons with first dynode releases more electrons, which are then
accelerated by second dynode (metal plate biased to a higher positive voltage with
respect to first dynode) in a process that multiplies the number of electrons
released before finally being collected at the anode (metal plate with highest
positive voltage).
Photomultiplier
Incident photons on photocathode (metal plate) releases photoelectrons.
Photoelectrons focused by electrode, then accelerated by first dynode (metal plate
biased to a voltage that is positive with respect to photocathode and electrode).
Impact of photoelectrons with first dynode releases more electrons, which are then
accelerated by second dynode (metal plate biased to a higher positive voltage with
respect to first dynode) in a process that multiplies the number of electrons
released before finally being collected at the anode (metal plate with highest
positive voltage).
Photomultiplier
Multiplicative factor can be designed
to be very high, thus producing strong
current even for weak incident light.
(Total electrons per photoelectron)
Photomultiplier Tube
Photomultiplier therefore converts weak incident light into strong electrical signal
that can be accurately measured.
Advantages of photomultiplier tube (PMT) over other astronomical detectors:
- linear over wide range of incident light intensity
- ultrafast time response (~0.0001 ms)
- low noise
A supernova explosion of a massive star
can leave behind a neutron star (i.e., the
remnant core of the star, having a mass
of ~1 M and radius ~10 km). A
rapidly-rotating neutron star can produce
beams of radiation along its magnetic
poles (a good analogy is a lighthouse).
If the beams of radiation sweep past the
Earth, the neutron star is seen as a pulsar.
A neutron star can rotate as far as once
every millisecond.
Photomultiplier Tube
Optical light pulses from the Crab pulsar (rotation period of 33 ms) sampled with
microsecond (10-6 s) temporal resolution. The pulse profile changed over the
period of the observations.
The Crab pulsar resulted from a supernova explosion in 1054 that was recorded
by Chinese astronomers.
Photomultiplier Tube
Disadvantages of PMT over other astronomical detectors:
- does not immediately produce an image
- moderate quantum efficiency (up to ~30%); why so low?
- lifetime limited by buildup of charge at anode
Photomultiplier Tube
Disadvantages of PMT over other astronomical detectors:
- does not immediately produce an image
- moderate quantum efficiency (up to ~30%); why so low? What can happen to
photoelectrons released in the metal?
Photomultiplier Tube
Disadvantages of PMT over other astronomical detectors:
- does not immediately produce an image
- moderate quantum efficiency (up to ~30%); why so low? What can happen to
photoelectrons released in the metal? Collide with other atoms/electrons and
either do not escape the photocathode or escape on the opposite side.
Photomultiplier Tube
Disadvantages of PMT over other astronomical detectors:
- does not immediately produce an image
- moderate quantum efficiency (up to ~30%)
- lifetime limited by buildup of charge at anode
Why does QE
drop away at long ?
Why does QE
drop away at short ?
Photomultiplier Tube
Disadvantages of PMT over other astronomical detectors:
- does not immediately produce an image
- moderate quantum efficiency (up to ~30%)
- lifetime limited by buildup of charge at anode
Why does QE
drop away at long ?
Threshold frequency.
Why does QE
drop away at short ?
Will return to this later.
Photomultiplier Tube
Even when the PMT is shielded from light, it generates a weak current known as
dark current. That is, electrons continue to be released from the photocathode.
What processes are responsible for the dark current?
Photomultiplier Tube
Even when the PMT is shielded from light, it generates a weak current known as
dark current. Processed responsible for the dark current are: -
electrons liberated from the photocathode and dynodes because they have
kinetic energies due to thermal motion sufficiently large to overcome the
work function of the respective metals
Photomultiplier Tube
Even when the PMT is shielded from light, it generates a weak current known as
dark current. Processed responsible for the dark current are:
- cosmic rays and energetic particles produced by radioactive decay that
liberate electrons through ionization or transfer of kinetic energy
electron
nucleus
Atoms
Energy level diagram of hydrogen, showing atomic orbitals having different
energies.
(third shell)
(second shell)
(first shell)
Covalent Bonds
Electrons in the outermost shell are called valence electrons, which can participate
in the formation of chemical (covalent) bonds with other atoms. Atoms/molecules
with unfilled outermost shell are especially reactive, whereas those with filled
outmost shells are especially inert.
Number of molecular orbitals is equal to the number of atomic orbitals in the
atoms being combined to form the molecule. A hydrogen molecule has a total of
two molecular orbitals.
Covalent Bonds
In a multi-electron atom, the situation is more complicated. The sharing of
electrons in the outer shell can have an effect on electrons in the inner shell(s).
A single water molecule has eight molecular orbitals, only five of which are
occupied at a given time. Electrons can be excited from a lower to higher
molecular orbital, and vice versa.
Band Theory of Solids
In a solid, atoms pack closely together by sharing their electrons to form what
resembles a single large molecule.
How many molecular orbitals are there in a solid like pure silicon?
(2nd excited
state)
(1st excited
state)
(Ground
state)
Band Theory of Solids
Any solid has a large number of bands; in theory, a solid can have infinitely many
bands (just as an atom has infinitely many energy levels).
Bands have different widths, based upon the
properties of the atomic orbitals from which
they arise. Also, allowed bands may overlap,
producing (for practical purposes) a single large
band.
All but a few of these bands lie at energies so
high that any electron that attains those energies
will escape from the solid (i.e., classic
photoelectric effect). These bands are usually
disregarded within a solid.
Band Theory of Solids
Below is a simplified diagram where we consider only two electronic bands in a
solid (the rest are ignored) that allows the three major types of materials to be
identified: metals, semiconductors, and insulators.
Valence band = ground states that are almost fully occupied in an insulator or
semiconductor.
Conduction band = excited states, corresponding to states whereby electrons
can move freely in a solid, and which are partially occupied in a metal, weakly
occupied in a semiconductor, and virtually unoccupied in an insulator.
Band Theory of Solids
Below is a simplified diagram where we consider only two electronic bands in a
solid (the rest are ignored) that allows the three major types of materials to be
identified: metals, semiconductors, and insulators.
Valence band = ground states that are almost fully occupied in an insulator or
semiconductor. In this state, electrons are in molecular orbits such that they are
tightly bound to atoms.
Conduction band = excited states, corresponding to states whereby electrons
can move freely in a solid, and which are partially occupied in a metal, weakly
occupied in a semiconductor, and virtually unoccupied in an insulator. In this
state, electrons are in molecular orbits such that they are loosely bound to atoms
(alternatively, they are almost equally shared among neighboring atoms).
Band Theory of Solids
Energy bandgap (Eg) is the minimum energy required to excite electrons from the
valence to the conduction band.
What are two ways in which electrons can be excited into the conduction band?
Band Theory of Solids
Energy bandgap (Eg) is the minimum energy required to excite electrons from the
valence to the conduction band.
What are two ways in which electrons can be excited into the conduction band?
- thermal energy (heat), causing the kinetic energy of atoms in the valence band
to increase with temperature; as a consequence, electrical conductivity of
semiconductor increases with increasing temperature
- absorption of photons (creation of photoelectrons); sometimes also referred to
as the non-photoelectric effect
Silicon Silicon
Band Theory of Solids
If we can store and measure the number of photoelectrons in the conduction band
(electrons excited from the valence to conduction band by the absorption of
photons) of a semiconductor, we can use a semiconductor to detect and measure
the intensity of incident light. This is the underlying concept of a CCD.
Semiconductors
Elemental semiconductors: column IVa , most popular Si, Ge.
Compound semiconductors: elements in columns Ib, IIb, IIIa, Va, VIa, VIIa
symmetrically spanning column IVa to form diatomic or triatomic molecules such
as GaAs, InSb, HgCdTe.
Semiconductors
Key to usefulness of semiconductors as detectors at optical and infrared
wavelengths is that their bandgap energies match those of visible/IR photons.
For each semiconductor, long wavelength cutoff long,cut = hc/Eg.
InSb 0.18 IR
6.89 m
Ge 0.67 IR
1.85 m
Si 1.11 1.11 m NIR-Optical
Conduction
band
Valence
band
Semiconductors
Consider how an electric current can be conducted in a semiconductor.
Electrons excited to the conduction band leave behind electron holes, i.e.
unoccupied states (i.e., unoccupied molecular orbitals) in the valence band.
In this situation, what charge carriers can conduct an electrical current? Both the
conduction-band and valence-band electrons.
Conduction
band
Valence
band
Semiconductors
When a neighboring electron moves to fill an unoccupied molecular orbital (hole),
the electron leaves an unoccupied molecular orbital (hole) where it came from,
which is filled by a neighboring electron, and so on. A hole therefore appears to
move in the opposite direction to the motion of a series of electrons, and behave
as if it was an actual positively charged particle.
Semiconductors
It is easier to imagine the conduction of electricity by electrons in the valence
band by considering the motion of holes rather than valence electrons.
Although holes are discussed as if they are real positively-charged particles,
remember that they are not real particles and all the effects we see are actually
caused by the movement of valence electrons.
Semiconductors
Electrical conductivity of a semiconductor can be radically altered by adding an
impurity (different element), a process called doping.
Consider adding an element with 5 valence electrons (e.g., P or As) to Si to
produce a compound semiconductor. 4 of these 5 electrons bond with adjacent Si
atoms as before, but the 5th electron cannot form a bond. This electron can easily
be excited into conduction band.
Resulting crystal (which is neutral) has an excess of current-carrying electrons
and is designated "N-type" (negative type).
Note: N-type
semiconductor
is electrically
neutral.
Semiconductors
Alternatively, consider adding an element with 3 valence electrons (e.g., B or Al)
to Si to produce a compound semiconductor. These 3 electrons bond with
adjacent Si atoms as before, but the expected fourth bond cannot form thus
leaving a hole (unoccupied molecular orbital).
Electrons from the valence band can be easily excited into the hole, leaving
behind a hole in the valence band.
Resulting crystal (which is neutral) has an excess of current-carrying holes and is
designated P-type" (positive type).
Note: P-type
semiconductor
is electrically
neutral.
Semiconductors
Of course, what is really happening is that valence electrons can easily move to
fill holes, leaving behind holes that can be easily filled by other valence electrons.
PN Junctions
Consider what happens when we join two crystals, one N-type and the other P-
type, together, as is the case in a CCD.
What do you think might happen at the junction of these two crystals?
depletion region
+
PN Junctions
Plot of the charge density,
electric field, and voltage
across a p-n junction at
equilibrium (no applied
voltage).
In practice, the N-type and
P-type materials may be
manufactured with
different concentrations of
dopants, resulting in a
depletion region having
different widths in the N-
type and P-type materials.
PN Junctions
Energy bands of the N-type and P- N-type P-type
type materials before they are conduction
joined together. The loosely-bound band
donor electrons in the N-type
material is at a higher energy level
than the acceptor holes in the P-
valence
type material. band
Energy bands of the N-type and P-
type in the depletion region (i.e.,
N-type P-type
after the two materials are joined
together and the donor electrons
have diffused into the acceptor
holes). Notice that the energy conduction
levels of the valence/conduction band
bands in the N-type material (which
has a net positive charge) is lower
than those in the P-type material valence
(which has a net negative charge). band
PN Junctions
If an electron in the depletion region is excited (e.g., by the absorption of a photon,
or by thermal collisions) to the conduction band of this material, will this electron
move to the conduction band of the P-type or N-type material?
N-type P-type
E
+ conduction
band
depletion region
valence
band
PN Junctions
If an electron in the depletion region is excited (e.g., by the absorption of a photon,
or by thermal collisions) to the conduction band of this material, will this electron
move to the conduction band of the P-type or N-type material? Move to a lower
energy state in the conduction band of the N-type material, a concept that will
become useful when we discuss how a MOS capacitor in a buried-channel CCD
works. Electrically, this is because the N-type material in the depletion region has
a positive charge and that of the P-type material a negative charge.
+
+
depletion region
+
PN Junctions
Suppose we apply a voltage to the outside ends of our PN crystal, such that the
positive voltage is applied to the N-type material (reverse external voltage).
What do you think might happen near the junction of these two crystals? The +ve
voltage applied to the N-type material attracts loosely-bound electrons in this
material towards the +ve end away from the junction, while
the ve voltage applied to the P-type material attracts
holes in this material towards the ve end away
from the junction.
+
+
depletion region
+
PN Junctions
The result is that all available current carriers (electrons and holes) are attracted
away from the junction, and the depletion region grows correspondingly larger.
The positive charge in the N-type material increases, and the negative charge in
the P-type material also increases.
+
+
depletion region
+ +
PN Junctions
In reverse external voltage, the N-type P-type
difference in energy levels of
the valence/conduction bands in
the N-type and P-type materials conduction
increases. Electrically, the band
depletion region in the N-type
material acquires a larger
positive charge and that in the
P-type material acquires a valence
band
larger negative charge.
conduction
band
+
valence
band
PN Junctions
In reverse external voltage, would you expect a current to flow?
+
+
depletion region
+ +
PN Junctions
In reverse external voltage, would you expect a current to flow? There is no
current flow if the applied voltage is not too large so that the positive charge in the
depletion region attract the excess electrons and the negative charge in the
depletion region attract the excess holes to maintain an electrical balance.
+
+
depletion region
+ +
PN Junctions
As the voltage is increased, the depletion region gets larger. If either end of the
depletion region approaches its electrical contact too closely, the applied
positive/negative voltage begins to strongly attract/repel electrons in the molecular
bonds. If the applied voltage is too large, an electrical arc (i.e., breakdown of
bonds between atoms, thus releasing
electrons and destroying
the crystal) can be
generated between
the electrodes. +
+
depletion region
+ +
MOS Capacitor
Metal-oxide semiconductor (MOS) capacitor is the basic unit of a charge-coupled
device (CCD).
A MOS capacitor comprises a layer of electrically-insulating oxide deposited on a
semiconductor, with the two materials sandwiched between metal electrodes.
There are two types of MOS capacitors used in CCDs,
surface channel and buried channel.
MOS Capacitor
In a surface channel MOS capacitor, applying a positive voltage V G in the case of
a P-type semiconductor (-VG in the case of a N-type semiconductor) repels holes
away from (attracts valence-band electrons towards) the oxide surface, creating a
depletion region in the upper layer of the P-type material.
Depletion region
MOS Capacitor
If, in the depletion region, electrons in the valence band of the semiconductor are
excited (by the absorption of photons) to the conduction band, where do the
electrons move towards?
Depletion region
MOS Capacitor
If, in the depletion region, electrons in the valence band of the semiconductor are
excited (by the absorption of photons) to the conduction band, where do the
electrons move towards? Conduction band electrons are attracted towards VG,
and the holes left behind are repelled away from VG.
Depletion region
MOS Capacitor
If, in the depletion region, electrons in the valence band of the semiconductor are
excited (by the absorption of photons) to the conduction band, where do the
electrons move towards? Conduction band electrons are attracted towards VG,
and the holes left behind are repelled away from VG. As a consequence, the
depletion region reduces in size.
Depletion region
MOS Capacitor
The conduction band electrons collect in a channel below the oxide surface, hence
a surface channel device.
The maximum number of electrons that can collect below the oxide surface
depends on the applied voltage VG.
MOS Capacitor
Applying VG reduces the energy levels of the valence/conductions bands to lower
values the closer to the oxide surface, creating a potential well below the oxide
surface where conduction-band electrons in the depletion region can collect.
As conduction-band electrons fill the potential well below the oxide surface, the
potential well becomes shallower and eventually becomes filled.
before
charge
stored
MOS Capacitor
In a buried channel MOS, the N-type material is more heavily doped and much
thinner than the P-type material. A depletion region develops at the PN-junction,
and is especially wide in the P-type material.
Applying a positive voltage to the N-type material (reverse external voltage)
widens the depletion region at the PN junction; loosely-bound electrons in the N-
type material are attracted toward and collect
+VG
below the oxide layer, whereas holes are
repelled to the base of the P-type material.
+
+
depletion region
MOS Capacitor
If, in the depletion region at the PN-junction, electrons in the valence band of the
semiconductor are excited to the conduction band, where do the electrons collect?
+VG
+
+
depletion region
MOS Capacitor
If, in the depletion region at the PN-junction, electrons in the valence band of the
semiconductor are excited to the conduction band, where do the electrons collect?
In a channel within the N-type material, hence a buried channel. The holes left
behind by the valence-band electrons excited to the conduction band move away
from the depletion region to the base of the P-type material.
+VG
+
+
depletion region
MOS Capacitor
If the photon is absorbed in the N-type material rather than the depletion region,
what can happen to the resulting electron excited from the valence to the
conduction band?
+VG
+
+
depletion region
MOS Capacitor
If the photon is absorbed in the N-type material rather than the depletion region,
what can happen to the resulting electron excited from the valence to the
conduction band? The electron, or one of the other electrons in the conduction
band, can recombine with the hole left behind. For use in a CCD, you do not
want this to happen.
+VG
+
+
depletion region
MOS Capacitor
If the photon is absorbed in the P-type material rather than the depletion region,
what can happen to the resulting electron excited from the valence to the
conduction band?
+VG
+
+
depletion region
MOS Capacitor
If the photon is absorbed in the P-type material rather than the depletion region,
what can happen to the resulting electron excited from the valence to the
conduction band? The electron does not move across the depletion region (and
eventually recombines with the hole left behind). The CCD therefore does not
record photons that pass beyond the depletion region into the P-type material.
+VG
+
+
depletion region
MOS Capacitor
In a buried-channel MOS, the minimum in the potential well is located within the
N-type material, away from the oxide layer.
Disadvantage of surface-channel MOS is lower charge transfer efficiency (due to
surface effects) than buried-channel MOS when used in a CCD.
Disadvantage of buried-channel MOS is shallower potential well and so inability
to store as much charge as surface-channel MOS.
Surface channel Buried channel
Charge-Coupled Devices
Layout of MOS capacitors used in CCDs. Note that the rows are drawn as if they
are physically separated; in practice, they are physically connected.
M=metal
O=oxide
S=semiconductor
electrode
Charge-Coupled Devices
A single CCD pixel comprises three MOS capacitors, one used to detect photons
and the other two acting as electrical barriers to contain photoelectrons within the
middle MOS capacitor.
M=metal
O=oxide
S=semiconductor
electrode
Charge-Coupled Devices
A single CCD pixel comprises three MOS capacitors, one used to detect photons
and the other two acting as electrical barriers to contain photoelectrons within the
middle MOS capacitor.
photoelectrons photoelectrons
M=metal
O=oxide
S=semiconductor
electrode
Charge-Coupled Devices
A single CCD pixel comprises three MOS capacitors, one used to detect photons
and the other two acting as electrical barriers to contain photoelectrons within the
middle MOS capacitor.
electrical barrier
photoelectrons photoelectrons
M=metal
O=oxide
S=semiconductor
electrode
Charge-Coupled Devices
A single CCD pixel comprises three MOS capacitors, one used to detect photons
and the other two acting as electrical barriers to contain photoelectrons within the
middle MOS capacitor.
photoelectrons photoelectrons
photoelectrons photoelectrons
M=metal
O=oxide
S=semiconductor
electrode
Charge-Coupled Devices
To prevent photoelectrons from moving to adjacent MOS capacitors, the latter are
biased to a lower positive (or negative) voltage.
photoelectrons photoelectrons
+
electrical barrier electrical barrier
+ photoelectrons photoelectrons
photoelectrons photoelectrons
+
electrical barrier electrical barrier
+ photoelectrons photoelectrons
photoelectrons photoelectrons
+
electrical barrier electrical barrier
+ photoelectrons photoelectrons
+
electrical barrier electrical barrier
+ photoelectrons photoelectrons
photoelectrons photoelectrons
+
electrical barrier electrical barrier
+
electrical barrier electrical barrier
photoelectrons
VERTICAL
RAIN (PHOTONS) CONVEYOR
BELTS
(CCD COLUMNS)
BUCKETS (PIXELS)
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After the exposure is finished, buckets contain (different amounts of) rain.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
Conveyor belt starts turning and transfers buckets (along CCD columns). Rain
collected on the vertical conveyor is tipped into buckets on the horizontal
conveyor.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
Vertical conveyor stops. Horizontal conveyor starts up and tips each bucket in
turn into the measuring cylinder.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After each bucket has been measured, the measuring cylinder is emptied, ready
for the next bucket load.
`
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After each bucket has been measured, the measuring cylinder is emptied, ready
for the next bucket load.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After each bucket has been measured, the measuring cylinder is emptied, ready
for the next bucket load.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After each bucket has been measured, the measuring cylinder is emptied, ready
for the next bucket load.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After each bucket has been measured, the measuring cylinder is emptied, ready
for the next bucket load.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After each bucket has been measured, the measuring cylinder is emptied, ready
for the next bucket load.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
After each bucket has been measured, the measuring cylinder is emptied, ready
for the next bucket load.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
A new set of empty buckets is set up on the horizontal conveyor and the process is
repeated.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
Eventually all the buckets have been measured, the CCD has been read out.
MEASURING
CYLINDER
HORIZONTAL (OUTPUT
CONVEYOR BELT AMPLIFIER)
(SERIAL REGISTER)
Charge-Coupled Devices
The last row of a CCD is called the serial register (also called multiplexer)
Line address readout:
- Shift all columns by one pixel into multiplexer
- Readout all the multiplexer pixel electrons to an amplifier by shifting charges
- When multiplexer is empty, repeat
If CCD continues to be exposed during readout,
image will be smeared.
During readout, cover CCD with shutter. This
may not be necessary if the amount of light
collected during the readout process is relatively
small (e.g., object relatively dim).
Charge-Coupled Devices
Pictograph of a corner of a CCD. The serial register is bent double to move the
output amplifier away from the edge of the chip. This useful if the CCD is to be
used as part of a mosaic. The arrows indicate how charge is transferred through
the device.
160mm
Image Area
Serial Register
Bus wires
Edge of
Silicon
Read Out Amplifier
Charge-Coupled Devices
In wide-field cameras, a CCD mosaics are used to increase the field of view that
can be imaged.
CCD Mosaic used in MEGACam on the CFHT (FOV=1 x 1)
Learning Objectives
Properties of Charge-Coupled Devices:
- plate scale
- quantum efficiency
- charge transfer efficiency
- output
- binning
Noise considerations for Charge-Coupled Devices:
- photon noise
- readout noise
- dark noise
Reducing CCD data:
- bias frame
- dark frame
- flatfield frame
Plate Scale
Recall that the linear size of an image at the focal plane is given by
y = f tan
f
Plate Scale
Recall that the linear size of an image at the focal plane is given by
y = f tan
f
Plate scale is defined as
/y = 1/f
and is usually expressed in units of arcseconds/millimeter for
photographic plates, and arcseconds/pixel for CCDs (thus requiring knowledge of
the linear size of the pixel).
Field of view depends on focal length of telescope and size of detector. For a
detector with linear dimensions w, field of view = tan-1(w/f) w/f when fw as
is usually the case.
Plate Scale
How should the plate scale of a CCD be matched to the telescope, which has an
angular resolution determined by the telescope diffraction limit or seeing?
Is this CCD well matched to the telescope?
Plate Scale
How should the plate scale of a CCD be matched to the telescope, which has an
angular resolution determined by the telescope diffraction limit or seeing?
Is this CCD well matched to the telescope? No, does not utilize the available
angular resolution.
Here the image is undersampled, which
- does not utilize full angular resolution
- adds uncertainty to positional
measurements
Plate Scale
How should the plate scale of a CCD be matched to the telescope, which has an
angular resolution determined by the telescope diffraction limit or seeing?
Is this CCD well matched to the telescope?
Plate Scale
How should the plate scale of a CCD be matched to the telescope, which has an
angular resolution determined by the telescope diffraction limit or seeing?
Is this CCD well matched to the telescope? Utilizes the available angular
resolution, but does not maximize the field of view.
Here the image is oversampled, which
- reduces field of view
- reduces sensitivity because light
spread over more pixels
So, what is optimum match?
Nyquist Sampling
Suppose we are sampling a signal comprising a sinusoidal wave. How often do
we need to sample the signal to figure out its frequency?
Suppose the sinusoidal wave has a period of 1 s. Is sampling the signal once
every 1 s sufficient to figure out its frequency?
Nyquist Sampling
Suppose we are sampling a signal comprising a sinusoidal wave. How often do
we need to sample the signal to figure out its frequency?
Suppose the sinusoidal wave has a period of 1 s. Is sampling the signal once
every 1 s sufficient to figure out its frequency?
Nyquist Sampling
Suppose we are sampling a signal comprising a sinusoidal wave. How often do
we need to sample the signal to figure out its frequency?
Suppose the sinusoidal wave has a period of 1 s. Is sampling the signal once
every 1 s sufficient to figure out its frequency? No, we then think the signal is
constant in time, and so we have to sample the signal more than once per cycle.
Nyquist Sampling
Is sampling the signal 1.5 times per cycle (once every 2/3 s in the present example)
sufficient?
Nyquist Sampling
Is sampling the signal 1.5 times per cycle (once every 2/3 s in the present example)
sufficient? No, we then think this signal is a lower-frequency sinusoidal wave.
Nyquist Sampling
In the example below, the purple sinusoidal wave is sampled twice per cycle (once
every 0.5 s in the present example) at the locations of the green points. Is this
sampling rate sufficient?
Nyquist Sampling
In the example below, the purple sinusoidal wave is sampled twice per cycle (once
every 0.5 s in the present example) at the locations of the green points. Is this
sampling rate sufficient? Appears to be as drawn below.
Nyquist Sampling
In the example below, the purple sinusoidal wave is sampled twice per cycle (once
every 0.5 s in the present example) at the locations of the green points. What if
the sampling points are shifted by one quarter of a period so that the sinusoidal
wave is sampled at the locations of the red points?
Nyquist Sampling
In the example below, the purple sinusoidal wave is sampled twice per cycle (once
every 0.5 s in the present example) at the locations of the green points. What if
the sampling points are shifted by one quarter of a period so that the sinusoidal
wave is sampled at the locations of the red points? No, we then think the signal is
constant in time. Sampling twice per cycle is sufficient under most but not all
circumstances.
Nyquist Sampling
In general need to sample at more than twice the frequency of the signal, a
condition first recognized by the physicist Harry Nyquist.
Nyquist Sampling
In the previous examples we only considered sinusoidal waves. Any waveform,
however, can be assembled from a suitable collection of sinusoidal waves of
different amplitudes and frequency.
Nyquist Sampling
How often do we need to sample this particular waveform to figure out the
frequencies of all the sinusoidal waves that make up this waveform?
intensity
time
Nyquist Sampling
How often do we need to sample this particular waveform to figure out the
frequencies of all the sinusoidal waves that make up this waveform? Slightly
more than twice the frequency of the highest frequency sinusoidal wave.
intensity
time
Nyquist Sampling
Images are simply waveforms that vary in intensity over space.
How often do we need to sample this 1-dimensional image to figure out the
frequencies of all the sinusoidal waves that make up this image?
intensity
spatial dimension
Nyquist Sampling
Images are simply waveforms that vary in intensity over space.
How often do we need to sample this 1-dimensional image to figure out the
frequencies of all the sinusoidal waves that make up this image? Slightly more
than twice the spatial frequency of the highest spatial-frequency sinusoidal wave.
intensity
spatial dimension
Nyquist Sampling
The astronomical image shown below contains both stars and galaxies.
Which object or objects varies most quickly in spatial dimension (i.e, from one
pixel to an adjacent pixel), thus having the highest spatial frequencies?
Nyquist Sampling
The astronomical image shown below contains both stars and galaxies.
Which object or objects varies most quickly in spatial dimension (i.e, from one
pixel to an adjacent pixel), thus having the highest spatial frequencies? Stars
(Airy pattern + diffraction spikes + seeing).
To record all the available spatial
information and maximize the field
of view, how many pixels do we
need to devote to a star image?
Nyquist Sampling
The astronomical image shown below contains both stars and galaxies.
Which object or objects varies most quickly in spatial dimension (i.e, from one
pixel to an adjacent pixel), thus having the highest spatial frequencies? Stars
(Airy pattern + diffraction spikes + seeing).
To record all the available spatial
information and maximize the field
of view, how many pixels do we
need to devote to a star image? A
precise answer would require us to
figure out the frequency of the
highest-frequency sinusoidal
waveform from which the star image
can be assembled. A good rule of
thumb is at least two pixels over the
linear size of the Airy or seeing disk
(as measured at FWHM),
corresponding to the angular
resolution.
Quantum Efficiency
You might expect images taken with CCDs will have many blank spots where
photons impact the MOS capacitors acting as electrical barriers rather than light
detectors.
photoelectrons photoelectrons
photoelectrons photoelectrons
M=metal
O=oxide
S=semiconductor
electrode
Quantum Efficiency
By placing a microlens above each pixel, light incident on a given pixel can be
focused onto the middle MOS capacitor of that pixel.
Quantum Efficiency
CCDs have the highest quantum efficiency (QE), up to ~90%, of any astronomical
detectors yet invented.
Higher detector QE utilizes more of the collecting power of telescopes, and
reduces the integration time to reach a given sensitivity.
Why does the QE of CCDs
(made from Silicon) drop to
virtually 0 beyond ~ 1000 nm?
Quantum Efficiency
CCDs have the highest quantum efficiency (QE), up to ~90%, of any astronomical
detectors yet invented.
Higher detector QE utilizes more of the collecting power of telescopes, and
reduces the integration time to reach a given sensitivity.
Why does the QE of CCDs
(made from Silicon) drop to
virtually 0 beyond ~ 1000 nm?Semiconductor Egap(eV) Detection
Threshold for exciting photons
from valence to conduction InSb 0.18 IR-Optical
6.89 m
band in Si is at a wavelength
of 1100 nm. Ge 0.67 IR-Optical
1.85 m
Si 1.11 1.11 m NIR-Optical
Crystalline Si
Amorphous Si
Absorption of Light in Solids
Why does increase with increasing temperature and decreasing wavelength?
Absorption of Light in Solids
Why does increase with increasing temperature and decreasing wavelength? In
a semiconductor, more electrons are excited to the conduction band with
increasing temperature. In addition to electrons in the valence band at sufficiently
higher energy levels, electrons in the conduction band are able to absorb photons.
Absorption of Light in Solids
Why does increase with increasing temperature and decreasing wavelength?
Photons at shorter wavelengths are able to excite a larger number of electrons (at
progressively lower energy levels) in the valence band to the conduction band.
Photon Penetration in Si
The distance 1/ is called a scale height (or optical depth).
Most photons have been absorbed by a few scale heights
e.g., at z = 3/, f0 exp(-z) = f0 exp(-3) = 0.05f0.
For blue photons of =0.4 m, scale height in Si at room temperature is 1/ 0.2
mm; most blue photons absorbed by ~0.6 mm.
For red photons of =0.8 m, scale height in Si at room temperature is 1/ 10
mm; require thickness of ~30 mm to absorb nearly all red photons.
In front illuminated CCDs, photons also absorbed by electrode and oxide.
Photon Penetration in Si
Let us briefly return to the quantum efficiency of the photomultiplier tube. What
causes the falloff in quantum efficiency at long and short wavelengths?
Photon Penetration in Si
Let us briefly return to the quantum efficiency of the photomultiplier tube. What
causes the falloff in quantum efficiency at long and short wavelengths? At long
wavelengths, incident photons do not have sufficient energy to excite electrons
from an energy band in which the electron remains in the metal to that of an
energy band in which the electron has escaped from the metal.
Photon Penetration in Si
Let us briefly return to the quantum efficiency of the photomultiplier tube. What
causes the falloff in quantum efficiency at long and short wavelengths? At short
wavelengths, absorption by the glass (Silicon) wall of the photomultiplier tube.
Quantum Efficiency
Photons at shorter wavelengths more easily absorbed in electrode, oxide, or at
shallower depths in semiconductor, whereas photons at longer wavelengths are
able to penetrate more deeply.
For short-wavelength photons absorbed before the depletion layer, what might
happen to the photoelectrons generated?
Quantum Efficiency
Photons at shorter wavelengths more easily absorbed in electrode, oxide, or at
shallower depths in semiconductor, whereas photons at longer wavelengths are
able to penetrate more deeply.
For short-wavelength photons absorbed before the depletion layer, what might
happen to the photoelectrons generated? Recombine back with the hole created.
Quantum Efficiency
Photons at shorter wavelengths more easily absorbed in electrode, oxide, or at
shallower depths in semiconductor, whereas photons at longer wavelengths are
able to penetrate more deeply.
For long-wavelength photons absorbed beyond the depletion layer, what might
happen to the photoelectrons generated?
Quantum Efficiency
Photons at shorter wavelengths more easily absorbed in electrode, oxide, or at
shallower depths in semiconductor, whereas photons at longer wavelengths are
able to penetrate more deeply.
For long-wavelength photons absorbed beyond the depletion layer, what might
happen to the photoelectrons generated? Recombine back with the hole created.
Quantum Efficiency
To improve QE and also response to shorter wavelengths, illuminate CCD from
the back: back illuminated CCDs.
To ensure that short wavelength photons penetrate into depletion region, etch the
P-type substrate into a thin layer: thinned CCDs.
Thinned CCDs are more fragile, and also more expensive because of extra
processing together with lower yields.
Quantum Efficiency
Compare the quantum efficiencies of standard front-illuminated and back-
illuminated thinned CCDs.
Quantum Efficiency
Ongoing research to make CCDs with ever higher QE over broader spectral
ranges. An example: coat front surface of (front illuminated) CCD with a
phosphor coating that emits light in the 500-580 nm region when excited by 120-
450 nm ultraviolet and blue light (Blue Plus CCD).
Charge Transfer Efficiency
Some electrons may not be transferred (i.e., left behind) from one pixel to another
during the readout process, including those comprising the serial register.
Charge Transfer Efficiency
The charge transfer efficiency CTE = Number of electrons transferred/Number of
electrons collected.
Consider a 1024x1024 CCD with a 3-phase charge transfer and CTE of 0.999.
Average of 3x512x2 transfers (why?), so that average ratio of charges left behind
is 1 (0.999)3x512x2= 0.95.
What are the two consequences of an imperfect CTE?
Charge Transfer Efficiency
The charge transfer efficiency CTE = Number of electrons transferred/Number of
electrons collected.
Consider a 1024x1024 CCD with a 3-phase charge transfer and CTE of 0.999.
Average of 3x512x2 transfers (why?), so that average ratio of charges left behind
is 1 (0.999)3x512x2= 0.95.
What are the two consequences of an imperfect CTE? Non-linearity as the CTE
can change dramatically from one pixel to another in the CCD.
Charge Transfer Efficiency
The charge transfer efficiency CTE = Number of electrons transferred/Number of
electrons collected.
Consider a 1024x1024 CCD with a 3-phase charge transfer and CTE of 0.999.
Average of 3x512x2 transfers (why?), so that average ratio of charges left behind
is 1 (0.999)3x512x2= 0.95.
What are the two consequences of an imperfect CTE? An image trail.
Charge Transfer Efficiency
The charge transfer efficiency CTE = Number of electrons transferred/Number of
electrons collected.
Consider a 1024x1024 CCD with a 3-phase charge transfer and CTE of 0.999.
Average of 3x512x2 transfers (why?), so that average ratio of charges left behind
is 1 (0.999)3x512x2= 0.95.
What are the two consequences of an imperfect CTE? An image trail. In the
image shown below, CTE = 0.99995.
CCD Output
At the end of the multiplexer, an Analog-to-Digital Converter converts the
electrons to a digital signal.
The Gain (G) is the number of electrons combined to generate one signal count,
called a Analog-to-Digital Unit, or ADU. So ADU = Ne/G, where Ne = number of
electrons arriving at the Analog-to-Digital Converter.
Normally, G is set as a positive number larger than 1; e.g., for KAF-1602E,
G = 2.3 e-/ADU.
The dynamic range of CCD is limited (in practice) by the number of digital bits of
the output, e.g. KAF-1602E produces a maximum 16-bit output (allowing 216 =
65536 values).
How many photoelectrons can the KAF-1602E CCD accumulate (assuming 100%
QE and CTE) before its output saturates?
CCD Output
At the end of the multiplexer, an Analog-to-Digital Converter converts the
electrons to a digital signal.
The Gain (G) is the number of electrons combined to generate one signal count,
called a Analog-to-Digital Unit, or ADU. So ADU = Ne/G, where Ne = number of
electrons arriving at the Analog-to-Digital Converter.
Normally, G is set as a positive number larger than 1; e.g., for KAF-1602E,
G = 2.3 e-/ADU.
The dynamic range of CCD is limited (in practice) by the number of digital bits of
the output, e.g. KAF-1602E produces a maximum 16-bit output (allowing 216 =
65536 values).
How many photoelectrons can the KAF-1602E CCD accumulate (assuming 100%
QE and CTE) before its output saturates? 65535 x 2.3 = 151,000 e- (use 1 bit for
0 counts).
CCD Output Read Noise
Since digital signals can only have integer values, fractional ADU counts are
rounded off, leading to error of 0.5 ADU (differential nonlinearity).
E.g., fractional counts of 20.1, 20.2, , 20.49999 ADU count = 20
fractional counts of 20.5, 20.6, , 21.49999 ADU count = 21
E.g., If the Gain is set at 5 e-/ADU, presence of 13 17 photoelectrons will
generate the same ADU output.
CCD Output Read Noise
A bigger problem can be integral nonlinearity, whereby the ADU output departs
from a linear relationship. This leads to an error of up to N ADU, where N is the
maximum deviation of the output from the linear relationship.
Error associated with the readout process (both differential and integral
nonlinearity) is called read noise (details associated with the detector electronics).
CCD Output Binning
An effective method to reduce read noise is pixel binning.
Combine signals from adjacent pixels before arriving at the Analog-to-Digital
Converter.
Additional advantages are improvement in sensitivity as well as reduction in total
readout time for the whole CCD chip, e.g. pixel digitization rate (pixel readout
rate) for KAF-1602E = 30,000 Hz.
Space
Photon Noise
Photons (discrete particles) produced by any physical body are emitted with a
random distribution in time; i.e., a Poisson distribution. In a Poisson distribution,
the resulting probability, P, of a events during a particular time interval is
where m is
the average number of random occurrences per time interval.
Photon Noise
A Poisson distribution therefore expresses the probability of a number of events
occurring in a fixed time if these events occur with a known average rate, and is
independent of the time since the last event.
If you measure an average of number of events having a Poisson distribution,
the standard deviation of your measurements is . In the plot below, notice that
the width, and hence standard deviation, of the Poisson distribution increases with
the average number of events .
Photon Noise
A Poisson distribution therefore expresses the probability of a number of events
occurring in a fixed time if these events occur with a known average rate, and is
independent of the time since the last event.
If you measure an average of number of events having a Poisson distribution,
the standard deviation of your measurements is . In the plot below, notice that
the width, and hence standard deviation, of the Poisson distribution increases with
the average number of events .
In the case of light, average rate is the light intensity. Why then do we not
perceive the light intensity from a light source (e.g., light bulb) to change with
time depending on the number of photons received over a given time interval (0.1-
0.2 s)?
Photon Noise
A Poisson distribution therefore expresses the probability of a number of events
occurring in a fixed time if these events occur with a known average rate, and are
independent of the time since the last event.
If you measure an average of number of events having a Poisson distribution,
the standard deviation of your measurements is . In the plot below, notice that
the width, and hence standard deviation, of the Poisson distribution increases with
the average number of events .
In the case of light, average rate is the light intensity. Why then do we not
perceive the light intensity from a light source (e.g., light bulb) to change with
time depending on the number of photons received over a given time interval (0.1-
0.2 s)? A 100 W light bulb emits ~1020 photons each second. If we were to make
measurements of the number of photons emitted by the light bulb each second, we
will find that our measurements fluctuate randomly with a standard deviation of
1020 = 1010 photons about a mean of 1020 photons. The signal-to-noise is
therefore 1020/1010 = 1010. Our eye cannot perceive a fluctuation of 1010/1020 =
10-10!
Photon Noise
For a large number of events, a Poisson distribution approaches the (normal or)
Gaussian distribution.
Photon Noise
Say you make repeated measurements with a photomultiplier or repeated short
exposures with a CCD of an object having a constant brightness. If the detector
itself contributed no noise, then if you count an average of Ne for the number of
photoelectrons collected over time an integration time t, the standard deviation of
these counted numbers will be Ne.
Thus, if only photon noise exists, the signal-to-noise ratio S/N = Ne /Ne = Ne,
where Ne is the average number of photoelectrons recorded. This represents the
highest S/N we can ever achieve (imposed by Nature) even if the detector
contributes no noise!
Dark Noise
Even when a CCD is not exposed to light, some electrons are excited from the
valence to conduction band by thermal collisions. These electrons are
indistinguishable from photoelectrons (i.e, those excited by the absorption of
photons).
The electrons collected by a CCD even in the absence of incident light is called
dark current. (Of course, when the CCD is exposed, it continues to build up a
dark current.)
Dark Noise
Dark current is expressed in terms of the number of e- per pixel per second (i.e.,
average number of electrons excited by collisions from conduction to valence
band per unit time in a pixel). E.g., for the Kodak CCD type KAF-6303E, dark
current at 25oC = 15 e-/pixel/second (quite high).
Dark current can be reduced by operating CCD at lower temperature.
Suppose you expose a CCD having a dark current NDC for a time interval t. The
dark current accumulated over this time interval is NDC t. Suppose you make
repeated exposures, each having a time interval t. The corresponding standard
deviation of the dark current is DC = (NDC t)1/2. This fluctuation of the dark
current is known as dark noise.
The total signal collected in each exposure is Ntot = Ne + (NDC t).
The total noise, that is the standard deviation in the total signal collected during
each exposure, from photon and dark noise is (Total)2 = (Photon) 2 + (DC) 2
Read Noise
Recall that the error associated with the readout process (both differential and
integral nonlinearity) is called read noise (details associated with the detector
electronics).
For a given CCD circuitry, the read noise is a constant independent of the signal
received, and expressed in terms of root-mean-square number of electrons; e.g.,
read noise of KAF-602E has a root-mean-square of 15 e-. (Root-mean-square =
standard deviation when the mean is 0, as it is in read noise).
Total Noise
The total noise, that is the standard deviation in the total signal collected during
each exposure, of a CCD is therefore (Total)2 = (Photon) 2 + (DC) 2 + (RN) 2.
If Photon >>DC + RN , photon noise dominates and so the data is sky limited
(here sky refers to the astronomical object under study).
If DC + RN >> Photon, non-photon noise dominates and so the data is noise
limited (here, noise refers to dark noise and read noise).
Examples of CCD Properties
Learning Objectives
Properties of Charge-Coupled Devices:
- plate scale
- quantum efficiency
- charge transfer efficiency
- output
- binning
Noise considerations for Charge-Coupled Devices:
- photon noise
- readout noise
- dark noise
Reducing CCD data:
- bias frame
- dark frame
- flatfield frame
Reducing CCD Data
The process of detecting light introduces systematic effects into the collected data
(e.g., dark current, different QE among different pixels).
The process also may include random effects (e.g., cosmic ray hits, defective or
unstable pixels), which we shall ignore in this introductory course.
In reducing CCD data, we want to correct for systematic errors present in the data
without introducing additional random errors. The method involves using
separate calibration frames to isolate each individual aspect of the systematic
effects.
Median-combined Bias
frame
Cosmic ray hits
Reducing CCD Data