optical detectors

Optical DetectorsAbdul Rehman

Optical Detector

Optical detector is an essential component of an opticalreceiver which converts received optical signal into anelectrical signal.

Improvement of detector characteristics and performanceimproves the link performance allowing fewer repeaters.

Optical detector has the same required characteristics asthat for the optical source

Abdul Rehman Optical Communication

RequirementsHigh sensitivity at operating wavelength

High fidelity

Large electrical response to received optical signal (high eff)

Fast response/ short response time

Low noise

High stability

Low cost

Low bias voltages

High reliability

Small size

lCGeneral ConceptIf the energy hνof the incident photonexceeds the band gapenergy (hν > Eg ) anelectron-hole pair isgenerated each time aphoton is absorbed by thesemiconductor

• Under the influence of an electric field set up byan applied voltage electrons and holes are sweptacross the semiconductor resulting in a flow ofelectric current

=

ResponsivityThe photo current Ip is directly proportional to the incident opticalpower Pin

Ip = RPin

Where R is resposivity of photodetector in units of A/W

R = output photo current I p

incident optical power Pin

Responsivity is the ratio of the electrical output to the optical input.Abdul Rehman Optical Communication

Three fundamental processes occurring between the two energystates of an atom: (a) absorption; (b) spontaneous emission; and(c) stimulated emission.


Quantum Efficieny• The responsivity can be expressed in terms of Quantum efficiency

electron generation rate=

photon incidence rateη

(Q E ) is a figure given for a photosensitive device (charge-coupleddevice (C C D ), for example) which is the percentage of photons hitting

the photo reactive surface that w ill produce an electron-hole pair. It is anaccurate measurement of the device's sensitivity.

Energy of a photon is ' hν ' so the photon incident rate may be written

in terms of incident optical power and the photon energy asAbdul Rehman Optical Communication

Piinhν

⋅

electron generation rate is given as I p

q

So Quantum efficiency can be written as

η = I p qPin hn

= hn I p

q Pin

=hν

qR [since R = I p

Pin]


hv h λ hc−6

=

⋅ = x−34 8 m=

R =ηq ηq= c =

ηqλ ηλ= hc

q

η λ ⋅ ⋅10- 6.624*10−34 Js 2.998*108 m s

1.602*10−19 C

η λ 10⋅ ⋅ −6 1.602*10⋅ −19 m ⋅ C C 16.624*10 r 2.998*10 Js ⋅ s s j

s

Abdul Rehman

=η λ⋅

1.24A W, where λ is in µm

Optical Communication

This happens for

R ∝ λ until hν < E g

because more photons are present for the same optical power

This linear dependence on λ is not continue forever since eventuallythe photon energy becomes too small to generate electrons.

hv < Eg , The quantum efficiencyη then drops to zero

If the facets of the semiconductor slab are assumed to have anantireflection coating, the power transmitted through the slab of

width W isptr = exp(-αW ) pin

Where α is absorption coefficientAbdul Rehman Optical Communication

The absorbed power is thus given by

pabs = pin - ptr = [1 - exp(- α W)] pin

Since each absorbed photon creates an electron-hole pair,the quantum efficiency η is given by

η= pabs

pin= 1 - exp ( -αW )

η becomes 'zero' when α = 0. On the other hand η approaches1 if αW >> 1


Photodiode Responsivities

Quantum efficiency

Optical absorption curve

The wavelength c at which α becomes zero is called cutoffwavelength

The material can be used as a photodetector only for < c

Indirect – bandgap semiconductors: Si, GeThe absorption is not as sharp as for direct band-bandgapmaterialsDirect – bandgap semiconductors : GaAs, InGaAs


Energy of incident photon must be greater than or equal to bandgap energy ofthe photodetector material. Therefore photon energy

h ν ≥ E g ⇒h c

λ≥ E g ⇒ λ ≤ h c

E g

Thus the threshold for detection commonly known as

the long wavelength cutoff point λ c is

λ c =h c

E g

This gives the longest wavelength of light for photodetection


Direct and Indirect Absorption(senior 425)Si and Ge absorbs light by both direct and indirect opticaltransitions .Indirect absorption requires the assistance of a photon sothat momentum as well as energy are conserved.For direct absorption no photon is involved so transitionprobability is more likely.Si is only weakly absorbing over the wavelength band ofinterest in optical fiber communications (800-900 nm).For Ge the threshold for direct absorption occurs above1530 nm and Ge may be used in the fabrication of thedetectors over the whole of the wavelength range ofinterest.Photodiode material should be chosen with bangap energyslightly less than the photon energy corresponding to thelongest operating wavelength of the system.

This gives a sufficiently high absorption coefficient to ensure agood response and at the same time limits the thermallygenerated carriers in order to achieve a low dark currentGe diodes have relatively large dark currents - a disadvantage ofGeTo overcome this problem certain alloys of Ge are made wherethe bandgap is adjusted as per the requirements like InGaAs,GaAlSb


Bandwidth(agrawallll 135)The bandwidth of photodetector is determined by the speed withwhich it responds to variation in the incident optical power.The rise time is defined as time over which the current builds upfrom 10 % to 90 % of its final value when the incident opticalpower is changed abruptly.The rise time is written as;

Tr = (ln 9)( tr + RC )

Abdul Rehman

transit time


time constant

The transit time is added to time constant of equivalent RC circuitbecause it takes some time before the carriers are collected aftertheir generation through absorption of photons.The maximum collection time just equal to the time an electrontakes to traverse the absorption region.The transit time can be reduced by decreasing W but the quantumefficiency begins to decrease significantly for

αW < 3

Thus, there is a trade-off between the bandwidth and the responsivity(speed versus sensitivity) of a photodetector.Abdul Rehman Optical Communication

Thus there is trade-off between the bandwidth and theresponsivity (speed versus sensitivity) of a photodetector.The bandwidth is given as

Abdul Rehman

f = 12(tr + RC )


Rise and fall times


•

•

Dark current IdDark current is the current generated in absence of optical signal

It originates from:

stray light

thermally generated electron – hole pairs

Id should be negligible for a good photodetector

(Id<10nA)


•

–

–

•

–

–

–

–

•

–

4.2. Photodetector designPhotodetectors can be broadly classified into two categories

photoconductive

photovoltaic

Photoconductive detector

homogeneous semiconductor slab as shown in fig before

Little current flows when no light is incident

Incident light increases conductivity through electron-hole generationand current flow is proportional to the optical power

reverse biased p-n junction

Photovoltaic detectors

solar cells, produce voltage in the presence of lightAbdul Rehman Optical Communication

p-n photodiodesA reverse bias p-n junction consists of a region, known as depletion regionElectron-hole pairs are created through absorption when such p-n junction isilluminated with light on one sideBecause of the large built-in electric field, electrons and holes generated insidethe depletion region accelerate opposite directions and drift to n and p sidesrespectively.The resulting flow of current is proportional to the incident optical power

A reverse bias pn junction acts as aphotodetector and is referred as pnphotodiodeFigure showing structure of a p-nphotodiodeLight is falling on one side of thephotodiode (p-side)The depletion region has width W


As shown in the figure incident lightis absorbed mostly inside thedepletion regionThe responsivity of photodiode isquite high (R~1 A/W) because ofhigh quantum efficiency

The electron-hole pairs generatedexperiences a large electric field anddrift rapidly towards the p or n side,depending on the electric chargeThe resulting current flow becauseof incident optical power

I p = RPin

ms

5 m

Bandwidth often limited by transit time .If W is the width of the depletionregion and vd is the drift velocity, the transit time is given by

tr =W

vd

W = 10m ,Vd =0 5

tr =10 ⋅ 10 −6 m

10 s= 10 −10 s = 100ps

Good enough to 1 Gbit/sAbdul Rehman Optical Communication

Both W and Vd can be optimized to minimize tr

The depletion-layer width depends on theacceptor and donor concentrations and can becontrolled through them.The velocity Vd depends on the applied voltagebut attains a maximum value called saturationvelocity ~ 105 m/s that depends on the materialused.


Limiting factorLimiting factor for the bandwidth of p-n photodiode is thepresence of a diffusive component in the photocurrent.The physical origin of diffusive component is related to theabsorption of incident light outside the depletion region.Electrons generated in the p-region have to diffuse to thedepletion-region boundary before they can drift to the n-side,similarly holes generated in the n-region must diffuse to thedepletion-region boundary.Diffusion is an inherently slow process; carriers take ananosecond or longer to diffuse over a distance of about 1 µm.


Figure shows how the presenceof a diffusive component candistort the temporal response ofa photodiode.The diffusion contribution can bereduced by decreasing thewidths of the p- and n-regionsand increasing the depletion-region width so that most of theincident optical power isabsorbed inside it.This is the approach adopted forp–i–n photodiodes.


p-i-n energy-band diagram


p-i-n photodiodesA simple way to increase the depletion region width is to

insert a layer of undoped or lightly doped semiconductor

material between the p-n junction

Since the middle layer consists of nearly intrinsic material

so such a structure is referred as p-i-n photodiode

Because of its intrinsic nature the middle i-layer a large

electric field exists in the i-layer.

The depletion region extends throughout the i-region and

its width W can be controlled by changing the i-layer

thickness

Most power is absorbed in i-region so

drift dominates over diffusion

The width W depends on a compromise

between speed and sensitivity.

The responsivity can be increased by

increasing W so that quantum efficiency

η approaches 100% but at the same

time the response time also increases

as it takes longer time for carriers to

drift across the depletion region.

Optimum W is a compromise between

responsivity and response time.

(PAGE 139 AGRAWAL)For indirect Si and Ge W = 20 - 50m, for reasonable quantum efficiency

The bandwidth of such photodiode is limited by relatively

long transit time, tr > 200 ps

By contrast in InGaAs,W = 3 - 5m that uses in direct bandgap semiconductor, the transit time is reduced

So 10ps and the detector bandwidth f 1

2πτ tr 10GHz, τ tr >> τ RC

20 GHz possible, even 30 GHz with reduced ηAbdul Rehman Optical Communication

λ

5-6

Characteristics of common p-i-n diodesParameter

Wavelength

Symbol Unit

µ

Si

0.4-1.1

Ge

0,8-1,8

InGaAs

1,0-1,7

Responsivity

Quantum efficiency

Dark current

R

η

Id

A/W

%

nA

0,4-0,6

75-90

1-10

0,5-0,7

50-55

50-500

0,6-0,9

60-70

1-20

Rise time τr ns 0,5-1 0,1-0,5 0,05-0,5

Bandwidth

Bias voltage

∆f

Vb

GHz

V

0,3-0,6

50-100

0,5-3

6-10

1-5

56

Intrinsic and extrinsic semiconductorsAn intrinsic semiconductor is one which is pure enough thatimpurities do not appreciably affect its electrical behavior. Inthis case, all carriers are created by thermally or opticallyexcited electrons from the full valence band into the emptyconduction band.

Thus equal numbers of electrons and holes are present in anintrinsic semiconductor. Electrons and holes flow in oppositedirections in an electric field, though they contribute to currentin the same direction since they are oppositely charged.


N-type dopingThe purpose of n-type doping is to produce an abundance of mobile or"carrier" electrons in the material. To help understand how n-typedoping is accomplished, consider the case of silicon (Si).Si atoms have four valence electrons, each of which is covalentlybonded with one of four adjacent Si atoms. If an atom with five valenceelectrons, such as those from group VA of the periodic table (eg.phosphorus (P), arsenic (As), or antimony (Sb)), is incorporated into thecrystal lattice in place of a Si atom, then that atom will have fourcovalent bonds and one unbonded electron. This extra electron is onlyweakly bound to the atom and can easily be excited into the conductionband.At normal temperatures, virtually all such electrons are excited into theconduction band. Since excitation of these electrons does not result inthe formation of a hole, the number of electrons in such a material farexceeds the number of holes. In this case the electrons are the majoritycarriers and the holes are the minority carriers. Because the five-electron atoms have an extra electron to "donate", they are called donoratoms.


P-type dopingThe purpose of p-type doping is to create an abundance of holes. In thecase of silicon, a trivalent atom (such as boron) is substituted into thecrystal lattice. The result is that one electron is missing from one of thefour covalent bonds normal for the silicon lattice.Thus the dopant atom can accept an electron from a neighboring atoms'covalent bond to complete the fourth bond. Such dopants are calledacceptors. The dopant atom accepts an electron, causing the loss of onebond from the neighboring atom and resulting in the formation of a"hole." Each hole is associated with a nearby negative-charged dopantion, and the semiconductor remains electrically neutral as a whole.However, once each hole has wandered away into the lattice, one protonin the atom at the hole's location will be "exposed" and no longercancelled by an electron. For this reason a hole behaves as a quantity ofpositive charge. When a sufficiently large number of acceptor atoms areadded, the holes greatly outnumber the thermally-excited electrons. Thus,the holes are the majority carriers, while electrons are the minoritycarriers in P-type materials. Blue diamonds (Type IIb), which containboron (B) impurities, are an example of a naturally occurring P-typesemiconductor.


Double-hetereostructureThe performance of p-i-nphotodiode can be improvedconsiderably by using double-heterostructure designIn this the middle i-type layeris sandwiched between the p-type and n-type layers of adifferent semiconductorwhose bandgap is chosensuch that light is absorbed inthe middle i-layerA p-i-n photodiode commonlyused for lightwaveapplications uses InGaAs forthe middle layer and InP forthe surrounding p-type and n-type layers.


= −19

Double – heterostructure design

λ c =hc 6.626x10−34 x2.998x108

Eg 1.35x1.602x10= 0.92µm

InP: Eg=1.35 eV C=0.92 µ(transparent for wavelengths greater than 0.92 µ)

InGaAs: Eg=0.75 eV λC=1.65 µ

lattice matched

The middle InGaAs layer absorbs strongly in thewavelength region 1.3-1.6 µThe diffusive component of the detector current iseliminated completely because light is only absorbedinside the depletion region

The quantum efficiency η can be made almost 100%by using an InGaAs layer 4-5 µ thickBandwidth as high as 70 GHz were realized using thinabsorption layer (W< 1 µ) but only at the expenseof a lower quantum efficiency η and responsivity

By 1995 photodiodes exhibit bandwidth f≈110 ∆ GHzand τRC ≈1s

How to improve efficiency ??Fabry-Perot cavity (laser like structure) can be formedaround p-i-n structure to enhances quantum efficiency to∼94%. FP cavity has a set of longitudinal modes at whichthe internal optical field is resonantly enhanced throughconstructive interference

The result of such structure is that when the incidentwavelength is close to longitudinal mode such photodiodeexhibit high sensitivity

Another approach for efficient high-speed photodiodes is use of opticalwaveguide into which the optical signal is edge-coupled

It enhances quantum efficiency to ∼100% as absorption takes place along thelength of the optical waveguide ( 10 ∼ µ )50 Ghz bandwidth was realized in 1992 for waveguide photodiode.The bandwidth could be increased to 110 GHz by adopting mushroom-mesawaveguide structure.In this structure the width of the i-type absorbing layer was reduced to 1.5 µmwhile the p- and n- type cladding layers were made 6 µm wideIn this way both parasitic capacitance and the internal series resistance were

minimized, reducing TRC to about 1 ps

∆f ≈ 172 GHzη ≈ 45%

Speed of responseThe main factors limit the response of photodiode are

Drift time of carriers through the depletion region

Diffusion time of carriers generated outside the

depletion region

Time constant incurred by the capacitance of photodiode

with its load


hv

Avalanche photodiodesAll detectors require a certain minimum current to operate reliably

The current requirement translate into minimum powerrequirements through

Pin =I p

R

• Detectors with large responsivity R are preferred sincethey require less optical power

• The responsivity of p-i-n photodiode can takes itsmaximum value

R = q for η = 1Abdul Rehman Optical Communication

hν η= ---

q

R

APDs can have much largervalues of R as they aredesigned to provide aninternal current gainThey are used when theamount of optical power thatcan be spared for the receiveris limitedThe physical phenomenonbehind the internal gain isknown as impact ionizationUnder certain conditions, anaccelerating electron canacquire sufficient energy togenerate a new electron-holepair


The energetic electrongives a part of its kineticenergy to another electronin the valence band thatends up in the conductionband, leaving behind aholeThe net result of impactionization a single primaryelectron, generatedthrough absorption of aphoton, creates manysecondary electrons andholes, all of whichcontribute to photodiodecurrent


αe:

The primary hole can also generate secondary electron-hole pairsthat contribute to the current

The generation rate is governed by two parameters,

impact-ionization coefficient of electrons ,averagenumber of electrons created per length

αh: impact-ionization coefficient ofaverage number of holes created per length

holes,

The numerical value of αe and αh depends on the semiconductormaterial and on the electric field that accelerates electrons andholes

5 ⎪⎭

• In the last figure

E = 2 - 4 ×10 V cm ⇒α e ⎫

⎬α h ⎪

≈1⋅104 cm-1

These values can be realized by applying a high voltage ( 100 V)∼to the APDAPDs differ in their design from that of p-i-n photodiodes is thatan additional layer is added in which secondary electron-hole pairsare generated through impact ionization


Reversebias

An APD with electrical field distribution inside various layersunder reverse bias


Ref to last figure: under reverse bias a high electric field exists in the p-type layer sandwiched between i-type and and n+ type layersThis p-type layer is referred to as multiplication layer, since secondaryelectron-hole pairs are generated here through impact ionizationThe i-layer still acts as the depletion region in which most of the incidentphotons are absorbed and primary electron-hole pairs are generatedThe APD drawbacks include

Fabrication difficulties due to their more complex structure and henceincreased costThe random nature of the gain mechanism which gives an additionalnoise contributionThe high bias voltage required (50 to 400 V) which are wavelengthdependentThe variation of gain (multiplication factor) with temperature, thus thetemperature compensation is necessary to stabilize the operation ofthe deviceThe bandwidth of APD depends on M- multiplication factor (decreasewith increasing M)Avalanche process takes time to build up

λ

-

-

Characteristics of common APDsParameter Symbol Unit Si Ge InGaAs

Wavelength

Responsivity RAPD

µ

A/W

0,4-1,1

80-130

0,8-1,8

3-30

1,0-1,7

2-20

APD gain M 100-500 50-200 10-40

k-factor kA= αh/ αe 0,02-0,05 0,7-1,0 0,5-0,7

Dark current Id nA 0,1-1 50-500 1-5

Rise time

Bandwidth

τr

∆fns

GHz

0,1-2

0,2-1,0

0,5-0,8

0,4-0,7

0,1-0,5

1-3

Bias voltage Vb V 200-500 20-40 20-30

Si: very strong for short wavelength

Performance of InGaAs APDs can be improved through designmodificationsThe main reason of poor performance of APDs is comparablenumerical values of the impact-ionization coefficient αe and αh. Theresult is that bandwidth is considerably reduced and the noise isrelatively highAt the same time InGaAs undergoes tunneling breakdown at electricfields of about 1x105 V/cm, the value that is below threshold foravalanche multiplicationThis problem can be solved in heterostructure APDs by using an InPlayer for the gain region because quite high electric fields (> 5x105

V/cm) can exists in InP without tunneling breakdownThe absorption region (i-type InGaAs layer) and the multiplicationregion (n-type InP layer) are separate in such device so this structureis known as SAM (separate absorption and multiplication regions)

αh > αe for InP so the holes initiate the avalanche processAbdul Rehman Optical Communication

E g= 0,75eV ⎪⎪

SAM APD (separate absorption and multiplication regions)

–InP :InGaAs :

InGaAs

E g = 1,35eV ⎫ ⇒ ⎬ valenceband step 0,4eV

⎭InP

Holes trapped at interface

εv

A problem with SAM APD is related to large bandgap differencebetween InP (Eg = 1.35 eV) and InGaAs (Eg = 0.75 eV). Becauseof this valence band step of 0.4 eV, holes generated in the InGaAslayer are trapped at the heterojunction interface and are slowedbefore they reach the InP multiplication regionSuch APD has extremely slow response time and relatively smallbandwidthThis problem can be solved using SAGM (separate absorption,grading, and multiplication regions)APDs.In SAGM (separate absorption, grading and multiplication) APD,we use another layer (InGaAsP) between absorption andmultiplication regions whose bandgap is intermediate to those ofInP and InGaAs layersThis additional layer improves the bandwidth considerablyGain-bandwidth product (M ∆f) of 100 GHz was demonstrated in1991 by using a charge region between the grading andmultiplication regions

SAGM: Separate absorption, grading, multiplication

Improved bandwidthM ∆f≈70 GHz for M>12⋅

SAGCM n-doped charge layerM ⋅ f≈100 ∆ GHz


A high performance APDs uses superlattice structure

The major limitation of InGaAs APD results from comparable values of αe andαh.In this structure

k A =α hα e

can be reduced

• In one structure absorption and multiplication regionalternate and consists of thin layers (~ 10 nm) ofsemiconductor material with different bandgaps


Its use is less successful for InGaAs/InP material system and the socalled staircase APDs were developedIn this InGaAsP layer is graded to form a sawtooth kind of structure inthe energy-band diagram that looks like staircase under reverse biasAnother scheme uses alternate layer of InP and InGaAs for the gradingregionSuperlattice structure can be used for multiplication region so that∆f=15GHz for M=10 , 10 times more sensitive than p-i-n diode.In MSM (metal semiconductor metal) photodetectors a semiconductorabsorbing layer is sandwiched between two metals forming a schottkybarrier at each metal-semiconductor interface that prevents flow ofelectrons from metal to semiconductorThis structure result in planner structure with inherently low parasiticcapacitance that allows high speed operation (up to 300 GHz)1.3 µ MSM photodetectors exhibit 92% quantum efficiency and lowdark currentThe planner MSM structure is most suitable for monolithic integration

MSM PhotodiodesMetal

Schottky

BarrierSemiconductor

MetalPrevents flow of electrons from metal tosemiconductor

Integrated1µm

Low dark currentRise time of about 16 ps

BW Up to 300 GHz

υd

Bandwidth of p-n photodiode often limited by transit time

as ∆f (bandwidth) = 12π (τ tr +τ RC)

If W is is the width of the depletion region andtransit time is given by

is the drift velocity then the

τtr =Wυd

Abdul Rehman

Typically W = 10µm, υd ≈ 10


5ms

V(E)

GaAs

Abdul Rehman

Si


vd ≈ 105 m

Es

by

Band gapthe band gap is the energy difference between the top of thevalence band and the bottom of the conduction band ininsulators and semiconductors.An intrinsic (pure) semiconductor's conductivity is stronglydependent on the band gap. The only available carriers forconduction are the electrons which have enough thermalenergy to be excited across the band gap, which is defined asthe energy level difference between the conduction band andthe valence band.Band gap engineering is the process of controlling or alteringthe band gap of a materiallb controlling the composition ofcertain semiconductor alloys, such as GaAlAs, InGaAs, andInAlAs


Band gapsCommon materials at room temperature

Abdul Rehman

Ge

InN

InGaN

Si

InP

GaAs

AlGaAs

AlAs

SiC 6H

SiC 4H

GaN

Diamond

0.67 eV

0.7 eV

0.7 - 3.4eV

1.14 eV

1.34 eV

1.43 eV

1.42 - 2.16 eV

2.16 eV

3.03 eV

3.28 eV

3.37 eV

5.46 - 6.4 eVOptical Communication

PhotodiodeA photodiode is an electronic component and a type of photodetector. It is a p-n junction designed to be responsive to optical input. Photodiodes are providedwith either a window or optical fibre connection, in order to let in the light tothe sensitive part of the device.

Photodiodes can be used in either zero bias or reverse bias. In zero bias, lightfalling on the diode causes a voltage to develop across the device, leading to acurrent in the forward bias direction. This is called the photovoltaic effect, andis the basis for solar cells - in fact a solar cell is just a large number of big,cheap photodiodes.

Diodes usually have extremely high resistance when reverse biased. Thisresistance is reduced when light of an appropriate frequency shines on thejunction. Hence, a reverse biased diode can be used as a detector bymonitoring the current running through it. Circuits based on this effect aremore sensitive to light than ones based on the photovoltaic effect.

A phototransistor is in essence nothing more than a normal bipolar transistorthat is encased in a transparent case so that light can reach the Base-Collectordiode. The phototransistor works like a photodiode, but with a much highersensitivity for light, because the electrons that tunnel through the Base-Collector diode are amplified by the transistor function. A phototransistor has aslower response time than a photodiode however


optical detectors

Documents