ANINTRODUCTIONTOOPTOELECTRONICSBYDr. ASHOK JAYANTISOURCES OF LIGHTOptical radiation is electromagnetic radiation in the frequency range of approximately 10 to10Hertz. This corresponds to an energy range of approximately 0.01 to 1000 ev. Therefore if an electromagnetic radiation in the optical region is to be generated, it must be by an energy variation in the subatomic range.Hence a source of light radiationhasto be based on subatomic transitions. When electrons in the lower levels are excited by some means to higher level, they emit radiation when they fall back to the lower levels after some time.The energy of the emitted radiation will be equal to the energy difference between the two levels.One of the most common ways of generating optical radiation is by heating.As the temperatures reach about 2500k bodies become red hot. Red is the lowest frequency/energy in the visible spectrum.Then as the temperatures raise and the energy supplied increases other frequencies also add up till finally the body becomes white hot. The spectral radiant existence of a body is given by Plank's law.In incandescent lamps the heating is done by passing electric current through a filament in a sealed enclosure. However thermal light sources are bulky, fragile and are not energy efficient.They also have a short service life and have poor stability. All these reasons make heat radiators incompatible with Opto electronic and fiber technology.Nowwhat arethecharacteristicsthat arerequiredof asourceuseful for this technology?Thesourcemust besmall, ruggedandenergyefficient. It must have sufficiently high radiance over a narrow, well defined band of wavelengths.The out put should be capable of being easily modulated.The source must be cost efficient, reliable and have long life.The output power most be stable.In addition, if we are using fibre technology, the source must emit light in an angular cone, which matches the acceptance cone of the fibre as far as possible.Thus for Optoelectronic applications, we have to look for phenomena other than thermal radiation for generation of light.The energy transfer to the electrons must be in some other manner. The process by which a fraction of the energy absorbedby a substanceis remitted in the form of visible or near visible light radiationis knownas luminescence. If theexcitationis byphotons fallingonthe substance, we call it photoluminescence.If the excitation is by chemical reaction, we call it chemoluminescence. If the excitation is by electric field, we call it electroluminescenceof coursewehavethermoluminescence, about whichwealready know.There are two aspects of this luminescence, depending on the time lag between excitation and emission.If the time lag is negligible and emission happens more or less along with excitation, the phenomenon is called fluorescence.If the emission starts after the excitation ceases, it is called phosphorescence.Time lag of 10 seconds is generally taken to be the upper limit for fluorescence.Compatibility with integrated circuit technology suggests the use of semiconductors and the phenomenon of elctroluminescence. The energy band structure of semiconductors is suchthat centres for absorptionof electromagneticenergyare available in the form of luminescence centres.These are donor or acceptor atomswhich get ionised or vacant sites in crystal lattices.When sufficient energy is available, it can dislodge electrons from luminescent centres or release ions from the crystal lattice of the host material.Thus the semiconductor will have additional charge carriers in the form of electrons and positive ions. When these carriers encounter vacant lattice sites or the points of crystal defects they recombine and release the energy in the form of light in which case we call them radiative recombination's.These lattice sites or defect sites are known as recombination centers.Any luminescent phenomena therefore has two stages.The first one involves the creation of charge carriers.The second stage involves the recombination of these charge carriers. The driving of a semiconductor to an excited state is done by sending current through it and the phenomenon is injection electroluminescence.Before discussing the details of the transitions, which occur in the semiconductor, it will be appropriate to do a quick recap of semiconductor characteristics.Semiconductors are materials whose electrical conductivity is smaller than that of themetalsandhigherthanthatoftheinsulators.Theforbiddengapisnarrow.The material has a negative temperature coefficient of resistance. This means that the electricalconductivityincreases with temperature. This is because the forbidden gap being narrow, the energy from heating is sufficient for electrons in the valence band to be excited into the conduction band.

The energy band diagram of an intrinsic semiconductor appears as in the figure. Ev represents the highest energy level in the valence band, which is normally completely filled. Ec represents the lowest energy level in the conduction band.This band is usually empty at low temperatures.The energy difference Ec -Ev is known as the forbidden gap. The conduction band extends from Ec upto a level Eo known as the vacuum level.This representstheenergyof anelectronat rest, outsidethesemiconductor. Theenergy differenceEo - Ec isknown as the electron affinity of the material.In addition to these levels, EFrepresents the Fermi energy level.This is the highest filled energy levet.At O Kallthe levels belowthe Fermi level are filled and all the levels about it are empty. However at higher temperatures, it may so happen that some of the electrons below EF acquire energy and are excited to higher levels. Consequently some of the levels above E may be occupied and some of those below it may be vacant. In terms of the electron population statistics, the probability of the Fermi level being occupied is exactly one half. An electron energy level at EF, should one exist, would have a 50% probability of being occupied.This probability is obtained from the Fermi Dirac distribution functions.1F ( )= --------------------- [ 1 +exp(E-EF)/KT ]The difference in energy between the vacuum level and the Fermi level, Eo-E F is known as the work function of the semiconductor.Theenergydiagraminthefigurereferstotheconditionat lowtemperatures, whenthesemiconductor behavesasaninsulator.At roomtemperatureshowever, a numberoftheelectronsinthevalenceband, acquireenoughenergyandgointothe conduction band. This results in the creation of vacancies or holes in the valence band. The concentration of electrons, n in the conduction band is the same as the concentration of holes, P in the valence band.This is known as the intrinsic carrier concentration.Its value is dependant on the temperature and is given by n = p = ni= K exp(-Eg /2kt)where K is a constant of the material, k is the Boltzman constant and Eg is the energy gap.The electrical conductivity is dependent of the carrier concentration. When Eg/kT is more than 100, the material may be considered as an insulator.When Eg/kt is less than 10, the conductivity starts approaching that of a metal. The most commonly known 'semi conductor elements are germanium and silicon. These belong to group IV of the periodic table.The energy gap of germanium is 0.7ev and that of silicon is 1.1 ev.The crystal structure of these elements is tetrahedral and the inter atomic bonds are mainly covalent.Itisnotalwaysnecessarythatasemiconductingmaterialbeanelement.The requirement of four electrons in the outermost orbit can also be satisfied by compounds of elements from group III & V. These are known as III - V semiconductors.Thus we havesemiconductingmaterialslikeGallium Arsenide, GalliumPhosphide, Indium, Antimonideandsoon. Compounds of elements fromgroups II andVI suchas cadmiumsulfidearealsowidely known semi conductors. Using these compoundsas semi conductors gives us a greater range of energy structures. It also provides a wide rangeof latticeconstants, intrinsiccarrier concentration, carrier mobilities andother properties to choose from.There is yet another way in which to vary these values and obtain what can be called 'designer' semiconductors. All the III-V compounds referred to above are binary compounds.The elements from the groups III and V of the periodic table have a special property that they are completely miscible with each other.Thus, a part of the arsenic in Ga As can be replaced by phosphorus and we get the ternary compound GaASP.Again, a part of the gallium in this compound can be replaced by Indium and we get the quaternary compound InGaASP.Thus we can get any combination of group III elements and any other combination of group V elements each in any proportion we like to get the characteristics that are required for a particular design.The proportions of each material are designated conventionally by x and 1- x for group three elements and y and 1 - y for group five elements.Again, x and y are used for the lower atomic weight elements.Thus we have materials tike Ga (As- P), (Ga1-x Alx) As, (In1-xGax)(As1-yPy)andgo on. The advantage of the total miscibility of the materials is that we can choose x and y to be whatever we like.All the materials we so far discussed are intrinsic semiconductors. Their conductivity is a direct function of the temperature.Thus any component made using an intrinsic semi conductor will be heavily temperature dependent.It will be advantageous if this temperature dependence can be decreased as far as possible.This requires that the population of electrons in the conduction band must be for some reason other than raise in temperature.This brings us to the impurity or extrinsic semi conductors.The intrinsic semiconductor may be doped with a material with an excess of electrons or a depletion of electronswhichcanalsobetermedanexcessof holes.Asaresult of thisexternal addition of holes or electrons, the carrier concentration n and p are no longer equal.For normal levels of doping, the product of the carrier concentrations remains independent of the doping level.np = n = Kexp (- Eg/kT)Here the impurity concentration, rather than the temperature, is the main parameter, which determines the total number of free charge carriers. Hence the electrical conductivity of an extrinsic semi conductor is less dependant on temperature and reasonably stable components can be built.Group IV semi conductors can be made n type by doping with group V elements. They can be made p type by doping with group III elements.The III-V compound can be made n - type by doping with group VI elements and p - type by doping with group II elements. There is an interesting variation here.If we consider the group IV elements, they contain four electrons in their outer most orbit. That is one less than the group V elements and one more than group III elements.Thus, if Si, Ge or Sn atoms are used to substitute sites of group Vatoms, the III - Vsemiconductor becomes a p type semiconductor. Similarly, if these group IV elements are made to substitute group III sitesthesemiconductorbecomesan-typesemiconductor. Thusthesamematerial, depending on which sites it is made to settle on, can be either a donor or an acceptor. They are known as amphoteric dopants. These occupy the sites of atoms which is the nearest in size to the impurity atom, because that requires the least energy.The case of Si inGa Asinspecial sinceit canoccupyeithersitedependingupontheconditionsof crystallisation. Using this, we can prepare successive layers of p &n type semiconductors.Theenergy band diagramofann type semiconductor appears as shown in the figure. Thentypeimpuritiesresult inlocalisedenergylevelsbeingformedjust belowthe conductionband. Theselevelsareoccupiedbythedonor atoms, whichareionised, giving up their electrons to the conduction band.Because of these additional levels, the Fermilevel in an n type material is nearer the conduction band.In the case of a p type semiconductor, acceptor levels are formed just above Ev. These levels are normally occupied by the electrons excited out of the valence band.Due to this increased population of holes, the Fermilevel in a p type semiconductor is lowered and is nearer Ev than Ec .p-n Junction:If a one part of a bulk semiconductor is doped with n type impurity and the other part with p type impurity with a practically abrupt junction between them, we call it a p-n junction. This is the most basic and most important of all semiconductors components. Whentherearenoappliedvoltages or thermal gradients, theFermi level is uniform throughout the material.In the n type material, the conduction band starts from close to the Fermilevel and in the p type material it is further off.So, as the energy band diagram of the p-n junction shows, the energy levels Ec and Ec are at different levels in the two halves. The same is the case with the levels Eo1, and Eo2, and Ev1 and Ev2. The difference in energies between Ec2, & Ec1, and the consequent difference in the free electron distribution results in a potential difference Va being set up across the junction. In the n type materials, the population of the electrons is more. They are the majority carriers. Their population is less in the p type region. So their tendency is to diffuse towardsthelowlypopulatedregion. However theelectricfieldduetoVdcausesa tendencyinthemtodrift exactlyintheoppositedirection. Thesetwoareexactly balanced and the majority population is maintained.The same thing happens to the holes also.The transition region contains a negligible number of carriers and is known as the depletion layer.The equilibrium concentrations of the majority and minority carriers on either side of the junction is governed by the equation. n Pn = np= nWhere n& np are the concentration of electrons in the n and p regions and p and p are the concentrations of holesNow if a voltage is applied across the junction, depending upon the polarity, it has the effect of either lowering or raising the potential barrier. If the p type region is mademorepositive, thenthepotentialbarrierislowered.Thisleadstosomeofthe majority carriers diffusing out of their regions. These means that the minority concentration increases on either side.This increase is by a factor exp (ev/kT )where v is theapplied voltage. Thepoint of interest as far as semiconductor light sources are concerned is this.These excess carriers injected into the region where they are minorities moveawayfromthedepletionregionbydiffusion. Someof theseexcess carriers, recombine with the majority carriers in that region and it is precisely this recombination process which results in the injection electroluminescence.The various transitions that are caused by recombination's in a semiconductor can be classified as follows.The first andthemostsimple one, shown in fig a is where an electron in the conduction band recombined with a hole, giving up its energy and coming down to the valence band. This is known as the direct band to band transition. The energy of the radiation which results is exactly equal to the energy of the forbidden gap.The impurity levels in the band gap can trap the electron as it loses energy.The energy taken by the trapping centre results in one of the impurity atoms combining with a hole or electron in the valence band.A similar thing can result from dislocations in the lattice structures. In this case there may not be any radiation of optical frequency since the transition occurs in two stages.This is illustrated in figure(b).The electron while losing energy can collide with phonons which are the particle form of mechanical vibrations and after some transfer of momentum, reach the valence band and recombine with a hole.This is known as the non-radiative Phonon interaction. Here the energy of the radiation is different from that of the band gap.This transition is illustrated in fig C.The transition may also be affected by the presence of shallow acceptor levels, as shown in figure D.These types of transitions may or may not be radioactive, depending on the position of the level.As against all these, the energy released by an electron in the process of recombination may be simply taken up by another electron to go to a higher level.This is known asAuger recombination.This is illustrated in figure e.To summarise, we can state that1) Light generation in semiconductors is by recombination of electrons and holes.2) Recombination may be brought about by many independent, parallel processes3) The transition may be directly between band and band or it may involve intermediate steps.4) The recombination may be radiative or non radiative.In a radiative process most of the energy is released as a single quantum of energy.When the recombination takes place in more than one stage, it is likely that it is non radiative.It may also result in more than one quantum of longer wavelength radiation being released.In any of the transitions, the principle of conservation must hold good. For radiative recombination, anelectroninahigherlevel releasesenergyandcometoalower level. While doing so, it releases a photon. If Pe, is the momentum of the electron before transition and Pe2 is the momentum of the electron after the transition,Pe1= Pe2+ Kp where kp is the momentum of the photon, which can be written as mp' C.Application of the deBroglie Principle leads to K= Eg / C. For gap energies of around ev,the value of the photon momentum is small and negligible compared to the electron momentum.In these circumstances, the electron momentummay be considered invariable. This transition is known as a direct transition. The momentum energy curves of semiconductors where these type of transitions occur are shown in figure.These type of semiconductors are known as direct band gap semiconductors.The curves in figure show that the valley of the conduction band is directly above the peak of the valence band.The natural transition always involves a minimum of change in energy and is from a valley to thenearestpeak. Thewavelength of the emitted radiation is given by = 1.23/ Eg. Some materials with these properties are GaAs, Ga Al As, Ga ASP etc.If the peak of the valence band is not exactly below the valley of the conduction band, as shown in figure,the semiconductor is known as an indirect band gap semiconductor.In this case, the transition has to involve a change in momentum.The photon momentum is small and can not account for this change.So for a transition to occur, a third particle has toparticipate. This maybeanimpuritycentre, aneutral trapor aphonon. The momentum relation for an indirect transition can be written asPe, = Pel Kp KGa P is an indirect band gap semiconductor.Here, donor - acceptor complex as (2n , O ) or neutral traps like a nitrogen atom instead of a phosphorus atom in the lattice act as centres of radiative recombination.In the case of a direct band gap semiconductor, there is no need for a third particle to participate in the transition.Therefore the probability of an optical transition is high in this case and these materials are likely to be efficient luminescent materials.On the other hand, for the same reason, the probability of the reverse transition, that of absorption of photons is also equally high. Thus self-absorption of emitted radiation is also strong in the direct band gap semiconductors.So the selection of material does not depend only on this aspect.The most important consideration is the required wavelength of light. The wavelength of theemittedphoton is given by 1.23/ Eg. From this expression,for the radiatortobeabletooperateinthe visibleregion,thesemiconductorshould have an energy gap of 1.5 to 3.0 ev.This rules out the use of materials like Si and Ge.We have to choose III-V compounds.Light Emitting diodes Injection Electro LuminescenceAs already discussed, any process of luminescence involves two stages.The first one is supply of energy to carriers. The second one is the release of this energy through theprocessof recombination. Electroluminescenceisthephenomenoninwhichthe energyisgivenbyelectricalmeans. Inthecaseofp-njunction, theexcitationisby creation of excess minority carriers. The process of making the majority carriers in one region flow across the junction, resulting in an increase in the concentration of minority carriers is known as injection electroluminescence. The energy stored by the creation of non-equilibrium carriers is released in the form of radiation.TheoperationofaninjectionLEDcanbebest understoodbyreferringtothe energy banddistributioninthep-n junction, as shown in figure. The flow of carriers fromtheregioninwhichtheyarethe minoritycarriers tothe other region,underthe influence of the internal fields in known as the drift flow of electrons or holes, as the case may be. The flow of carriers from the region in which they are the majority carriers to the other region is known as the diffusion flow. Under equilibrium conditions, the drift flow and the diffusion flow balance each other.When a forward bias is applied to the p-n junction, the potential barrier is lowered and more electrons flow into the p-region.This means that the minority concentration in the p region is increased. Simultaneously the minority concentration in the n region is also increased. These excess minority carriers combinewiththemajoritycarriersinthat regionandreleaseenergyintheformof electromagnetic radiation.It is preferable that the emission of the radiation is from either of the two regions. This is achieved by doping one of the regions more heavily than the other.Suppose the impurity concentration in the n region is heavier. We symbolise it as n+ or n++ as the case may be.Since n+ is more heavily doped than p, the flow of electrons from n+ to p is more than that of holes from p to n+ (figure) .As a result, the minority concentration in the p region becomes greater. The rate of recombination is proportional to the concentration and is more in the p region.Thus the light emission will be predominantly from the p region.In this case we call the n+ region the base and p the emitter.In general the more heavily doped region acts as the base and the other as the emitter.The efficiency of the light emission depends upon the total number of the excess minority carriers in the emitter region and the number of these that participate in radiative recombination. This in turn depends on the current that causes these transitions. The total current that is flowing through the diode depends on the effect of all the transitions occurring in the semiconductor, namely1) The component due to the electrons injected into the p region, In2) The component due to the holes injected into the n region, Ip3) Current due to non radiative recombination's at the p-n junction Irec4) Current due to the charge carriers piercing through the potential barrier. This is known as the tunneling current I tun.5) The leakage current at the junction ILThus I = In + Ip + Irec + I tun + ILThe ratio of the current providing the excess memory population in the emitter to the total current is knowas theinjectionefficiencyof thediode..Ne, theinternal quantum efficiency or simply the 'Yield' of a LED is defined as the ratio of the number of photons emitted to the number of the electrons injected into the p region.The parameter refers to the efficiency with which the charge carriers are injected intotheactiveregion. Theparameter Negives us theefficiencywithwhichthese injected charge carriers are converted into photons.Therefore, in order to determine the effectivenessoftheprocessofinjectionElectroluminescence,wehaveto look at the product Ne.Thereisanother wayof lookingat theefficiencyof aLED. Theefficiency dependsontherateofrecombination.Therateofrecombinationdependsuponthe number of excess carriers available. If the number of electrons in the p region before injection is Npo and their number after injection in N.The excess electrons are N - Npo. If the life time (average time before recombination of these electrons) is Tp, the rate of recombination can be written as popN NNt

A A TThis rate refers to both radiative and non radioactive recombination and we may writeRR NRdn dn dndt dt dt| ` | ` . , . ,Thus leads to the need for defining T RP and TNR as the life times of the electrons which would result if either of these two processes is operative and 1 1 1p RR NR +T T TThe internal quantum efficiencyeN may be written as 1111RR RRRRp NRdndtdndt| ` T. , T| `+ T T. ,Thus for an efficient source, it is necessary thatTRR/ TNR be as small as possible.The life time of charge carriers causing radiative recombination must be as small as possible and that of the charge carriers causing non radiative recombination should be as large as possible.In an indirect band gap material like silicon TRP is about 10ms and TNR is about 100ns, giving a value of Ne about 105.In a direct band gap material like GaAs, TRPis as low as 100ns, resulting in an efficiency of 0.5.ThevalueofNeisnormallyafunctionofthecurrent densityintheforward direction.At low current densities Ne more or less is proportional to the current density, showing a sharp rise with increasing J (figure.) At one stage, the diffusion component becomes more prominent in the total current. So Ne saturates at this value. After the luminescence centres are all saturated, there is no further possibility of recombination taking place.Thus any further increase in the current density is likely to show a decrease in Ne because the injected electrons become more in number and the number of photons emitted becomes less.However if the edge of the semiconductor are polishedto become mirrorsandform a resonant cavity,it is possible to initiate lasing action. This process will be discussed in detail later.The product r Ne give us the effectiveness of the injection electro luminescence phenomenon.But the usefulness of the LED also depends on how many of the photons that are created get out of the LED surface.Let us look at this in some detail.The photons that are created by the injection process may again be absorbed by the semiconductor itself, resulting in the excitation of electrons in the valence or conductionbandtohigherlevels. The photons also may be absorbed by the impurity levels or the crystal lattice.The photons may also causeexcitation of electrons from the valence band to the conduction band.This particular process is more probable in the case of direct band gap semiconductors.More importantly, as the photons try to come out from the semiconductor into air, they are travelling form a denser medium into a rarer medium.If the angle at which the photon meets the surface is greater than the critical angle, total internal reflection take place(fig). These photons are likely to be lost in self-absorption.Ingeneral, thelossbyabsorptionisinverselyproportional tothepenetration depthX and the surface area available A and directly proportional to the volume of the crystal. The loss by total internal reflection is most prominent of these losses because thesemiconductingmaterialshavehighrefractiveindciesmorethan3.5. thiscanbe minimisedbyshapingthesurfacesuchthat thepossibilityof theangleof incidence exceeding the critical angle is smaller.Tocover all theselosses, wedefineaparameter No, calledthelight out put efficiency of the LED No=Power emittedPower generatedThis parameter acts to modify the productNein expressing the efficiency of the LED. Thus we come to the parameter called the external quantum efficiency of the light emitting diode.Ultimately this value is equal to the ratio of the photons emitted by the LED in a given time interval to the total number of electrons passing through the diode in the same time.If wewant todesignanLEDwhichperformsbetter, wemust bethinkingof designs which improve either , Ne or No.Let usfirst lookat . If wewant toimprovetheinjectionefficiency, more electrons should go into the p region than holes into n region.We have already seen that a solution for this is to dope the n region more heavily.However, we can notsimply go on increasing the doping concentration.At levels higher that about 10m in III-V semiconductorstheimpuritylevels start to interact with oneanother andthe range of energy levels they occupy starts to merge with the band edge. The band edge it self is perturbed and a band tail is formed . The band gap therefore becomes narrower.This results in a change in the frequency of the light emitted.The Fermi level moves up into the conduction band in the n region and down into the valence band in the p region.A semiconductor where the impurity level is high enough for all these to happen is said to be degenerate.Thus the main effect of doping concentration increasing over a particular level is an unwanted change in the frequency of emitted radiation.Therefore there is a limit to the use of increase in doping concentration as a means of obtaining a higher minority concentration in the emitter p region.The main aim is to see that more electrons flow into the p region thanholes into the n region.If, somehow, we can make the potential barrier for the electrons lower than that for the holes, this can be achieved.This brings us totheconcept of heterojunctions, insemiconductors, Thepn junctionswesofarreferredtoarealwaysmadeupofasinglesemiconductoreither elemental or compound, which is doped with acceptors on one side and donors on the otherside. Thesejunctionareknown as Homo junctions - Junctions containing the same semiconductor.The complete miscibility of the elements in group III and those in group V gave the possibility of the tertiary and quaternary semi conducting compounds being formed. The same capability of forming perfect solid solutions also givesraise to the possibility of forming a junction with one compound on one side and another compound on the other side. This results in the band gap being different on each side (figure). Such junctions are called Hetero junctions.This single crystal grown intotwodifferent semiconductors shows different valuesofbandgapenergy, electronaffinity, permittivityandall theotherproperties. Each of the semi conductors may be doped with either donors or acceptors.Similarly the doping concentration may also be different.Thus there are four types of hetero junctions that may be formed n - N, n - P, P - P and N - p, where the upper case letters represent the higher concentration.The heterojunctionis the starting point in the design of a number of devices.The band diagram for a heterojunction in the Equilibrium State is shown in figure. In this case the fermi level is common to both materials.If Eg, is the energy gap in the material, and Eg2 is the energy gap in the material 2, an energy gap E = Eg-Eg2 exists at the junction. Since the energy gaps on either side are different, the potential barrier for electrons and holes are different.This results in the flow of the electrons from the N region to the P region being greater than that of the holes from the P region to the n region. Consequently the injection efficiency of the device is considerably greater than that ofhomojunctions. Ifwewant tofurtherimprovethedesign, thefractionofthe injected electrons, which participate in the recombination, must be increased. Thus we will be increasing the value of the internal quantum efficiency.For this we have to keep theinjectedelectronslocalisedinaspecificareanear theP-Njunction. Wehaveto prevent them from moving away from this region.One way to stop charge carriers from moving along the field direction is to create an additional potential barrier.This can be done by the provision of a third semiconductor layer, whose energy gap is such that an additional potential well is provided (figure). This type of a structure is known as a double heterostructure.There is an additional advantage we gainfromthissecondpotentialbarrier.A narrower energygap meansalonger wave length absorption band and the phenomenon of self absorption of the generated photons will be considerably smaller Again, the different energy gaps mean different refractive indices and the transmission of photons will be through three media. The critical angle for the two individual stages will be higher than what it would have been, if there were only two media, thus the total internal reflection loss is also reduced.Another reason for improvedperformanceisthatthewiderbandgapregionsaremoretransparenttothe recombination radiation of the narrow band gap materials.In order to further reduce the absorption and TIR losses, it has be come common to usea five layerstructure as shown in figure. In addition to the three layers which make up the double hererojunction there are two contact layers, one of them being the substrate on which the rest of the layers are grown epitaxially. The middle layer is the narrow band gap active layer.The layers on either side of it are N and P.If the middle layer is n we get a N n P double Hetero structure and if the muddle layer is p we get the N p P structure.The outer most layers are usually narrow band gap materials.These permit the fabrication of good low resistance ohmic contacts at the device terminals.Iftheactiveemittingregionismadenarrowenoughthephenomenonoftotal internalreflectioncanbeutilisedtocreateawaveguideeffect.Thisfactorbecomes particularlyimportant in the design of edge emitting LEDS.We can have an idea of the performance of DHLEDS by trying to figure out its internal quantum efficiency let us assume the recombination time constant of the active player is Tp1.This is a function of the radiative and non-radiative time constants. 11 11 1 1p rr nr +T T TOneither sideof theactiveregion, theheterojunctionsgiveraisetorecombination centres.These recombinations are largely non-radiative.It the recombination are largely non-radiative.If the recombination velocity at either heterojunction iss (assumed to be the same, for simplicity), the thickness of the active region is 2d and the electron concentration is 1n, we can write the net rate of recombination per unit cross sectional area as1 1 111 11 12 2 221 1 1rr nrrr nrn d n d n dn ssandd+ + T T T + +T T THence the over all internal quantum efficiency is int1 rrnTTThusthesmaller thevalue of s the better the device will be GaAS-GaALAS junctions givesofthe order of 10m/s and In GaAsp-In P system give even smaller valuesPractical LED structures :Light emitting diodes are generally classified as surface emitting (SLED)or edge emitting ( ELED ).One typical design of a surface emitting conventional LED is shown in figure.Far visible range, Ga As P or Ga P doped with N or Zn O may be used.This type of structure has a problem in that there is a limit to the current densities that can be used. Thus there is a limit to the radiation intensity that can be obtained. If we try to increasethecurrent densityaboveacertaincritical value, thetemperaturestarts to increase, thewave lengthdistribution of the emitted radiation changes. Non radiative recombinationsstart toincreaseandtheinternal quantumefficiencyfalls. Increasing junction temperature also results in the decrease of the life of the LED. These problems are solved to a large extent in what is widely known as the Burrus design (figure). In this the temperature rise is controlled by keeping the active region near a heat sink layer, which is also the positive contact layer.The light emission is from thesideof thesubstrate.Aportionof thesubstrateisetchedawaytominimisethe distance between the active layer and the emitting layer. The active layer is separated fromtherearcontact byaninsulatingoxidelayer, leavingasmall areafor thelight emitting region.This concentrates the flow of current into a well-defined lateral region. These designs is known as aBurrus LEDTheBurrus principleextendedtodoubleheterostructures results inwhat are called the Burrus DHSLED s.In these devices, the etched well in the substrate is taken toasfarasthen-GaAlAsconfininglayer. Theisolationofthecontact area, when achieved by the oxide layer as in figure, leaves a large area of inactive Pn junction which contributes to the depletion layer capacitance. In order to overcome this problem, the electrical resistivity of the semiconductor layer is increased by bombarding selectively, with high-energy protons.This results in lattice damage and increase in insulation.The thicknessoftheinsulatinglayeris a function of the penetration depth of the protons which in turn can be controlled by the bombarding energy.The same effect can also be achieved by implanting oxygen ions. This type of DH SLED can with stand active area current densities around 50 A/mm and an optical power of around 60mw.A number of modifications of the design are possible, mostly involving the formation of micro lenses in the material.These micro lenses make it possible to use a smaller active area and still getting a large beam aperture.This results in higher radiance SLEDS.Edge Emitting LEDS:The conventional double heterojunction SLED mainly consists of an active layer and two confining layers.The narrow band gap active layer is the region where most of theradiativerecombination's takeplace. Thismeansthat thephotongenerationis predominately in the active layer.The two confining layers have a comparatively wider band gap and confine the charge carriers to the active region which is between the two junctions.Thewiderbandgapalso resultsina lower refractive index. Henceifthe photons are to cross into the confining layers they have to move from a region of higher refractive index to a region of lower refractive index-i.e. from a denser medium to a rarer medium. Thisbringsinthepossibility of total internal reflection. If the active layer thickness is sufficiently decreased, it is possible to initiate a wave guiding action, where the photon stream is confined to the active layer, without moving into the confining layers and finally emerges from the sides rather than the surface.This principle results in the design of edge emitting LEDs or ELEDs as theyarecalled.Here,thetwowide band gap layers are providing not only electrical confinement but optical confinement also.Since the active layer is made fairly thin, the possibility of self absorption is also decreased.The light beam emerging from the ultra thin active region from the sides rather thanfromthesurfaceresultsinafairlythinoutput beam. Thisisof great helpin communication set ups where the beam is likely to be coupled into an optical fiber.If a surface emitting structure is used it may be necessary to use a converging lens system to launchthetotalopticalfluxinto the fiber.In ELEDsystemsthisneed is eliminated. Also, since the total optical flux is concentrated over a small cross section, the radiance of the component is greater.The number of radiative recombinations is a function of the current density. In order to get a given current density with lower drive currents, the contact layer at the top can be laid in the form of a narrow stripe, the rest of the surface being crated with an insulating layer. This is the bases for the structure commonly referred to as the stripe geometryDHLED. (fig.). Thestrip geometry results in greater efficiencies since the input drive current can be reduced.Ifcanbedesignedsuchthatmost ofthelightemissionisatoneendonlyby providing a reflecting crating at one and face and an antireflection creating at the front end.Thestripegeometryalsocauses the total optical flex to be emitted from a near circular or elliptical cross section rather than a fan. Thus the effective radiance can be very high.The size of the output beam gives an increased coupling efficiency into small NAopticalfibers.Itmust, however,bekeptin mindthat surfaceemittersgenerally radiate more power into air.Super Luminiscent LEDs :Communication applications call for light sources with (a) high out put power (b) directionality and (c) a narrow spectral line width.Devices known as super Luminescent LEDsoffer morephotonoutput but stopshort of providingoptical resonancewhich would result in a Lasing output. (figure)The photon population is increased by giving sufficient drive currents to initiate stimulated emission.However conditions for optical feed back are notprovided, this not allowinglasingaction. Thisdevisethusgiveshighpower incoherent output. The stimulatedemissionalsocausesanarrowingdownof thespectral linewidth. The structure has its drawbacks in the requirement of high drive currents, non linear output characteristics and increased temperature sensitivity. ---------

Any device that converts optical energy into electrical energy is a photodetector and the phenomenon of conversion is the photo effect.The photo effect is the result of excitation of charge carrier inside a material by the optical radiation.There are some materials whose work function is such that the energy contained in the optical radiation is sufficient to release the electrons from the material.This is known as photoemission and the process is the EXTERNAL PHOTOEFFECT.This effect is used in a number of photo detecting devices such as photo tubes, photo multipliers etc., These are of limited interest as far as semiconductor Optoelectronic are concerned though there are areas where they are highly useful.TheINTERNALPHOTOEFFECTdealswiththecreationofadditional charge carriers inside a semiconductor.This generation causes an increase in the conductivity on exposure to optical radiation and is known as the PHOTOCONDUCTIVE EFFECT. If the semiconductor is in the form of a p-n junction, the potential barrier seperates the two types of charge carriers.The spatially separated electrons and holes produce a potential difference known as the photo e.m.f. This is the PHOTOVOLTAIC EFFECT.The photo voltaic effect is the basis for the design of photo diode, photo transistors, photothyristers etc., whilethephotoconductiveeffect is usedinphoto resistors.The creation of charge carriers in a semiconductor is due to the photons giving up energy to electrons in the valence band and lifting them to the conduction band.We can also visualise this as the optical radiation ionising the semiconductor atoms.If the photo energy is used in lifting an electron from the valence band to the conduction band it is knownastheINTRINSICphotoeffect. If the energy is spent in the excitation of an impurity trap centre, it is known as EXTRINSIC photo effect.These are comparable to the radiative andnon - radiative transitions we described in the previous chapter. However, since the population of the impurity trap centres is very small, it is the intrinsic photo effect that is the dominant phenomenon and causes almost all of the photo e. m. f. In intrinsic photo effect, the electrons may make direct transitions or indirect transitions dependingonwhetherthematerial is a direct band gap semiconductor or an indirect band gap semiconductor. One fundamental condition that has to be satisfied is that the energy of the photon must be equal to or greater than the band gap energy.EpEc - EvThis can be expressed in terms of the frequency of the radiation.f ( Ec - Ev ) / h.This give raise to the concept of a critical frequency.If the frequency is less than this, excitation of the electrons across the bandgap is not possible.In terms of the wave length, we definea longwavelength cut off of the spectral response c is given as 1.23/Eg.The number of electrons that are excited is dependent on the number of photons, which cause the transitions. This in turn, depends on how deep the radiation can penetrate into the material. Once an electro magnetic radiation enters a material, the oscillations begin to get damped gradually. The depth at which the radiation reduces to 1/e of its original value is known as the penetration depth, o, of the material.If o, is large, the photonscancausetransitions over a larger distance and the number of charge carriers created will be more.The reciprocal of the penetration depth is known as the absorption coefficient.PHOTODIODES:In the two types of detectors, which employ the internal photo effect, the photo voltaic detectors are used more widely, because of better responsivity and greater flexibility of design.So we would leco discuss these devices indetail.The simplest of all photo voltaic devices is the p-n junction diode, whose structure is shown in fig 2.1.Let us assumethat theradiationis incident inadirectionnormal tothep-n junctionplane, onthensurfaceo, isthepenetrationdepthandastheradiationis incident on the surface, electron-hole pairs are created up to this depth from the surface. These carriers diffuse intothe inner regions andgivensufficient time, theywould recombine.This time period depends on the life time of the charge carriers.Keeping the lifetimeandthemobilityof thechargecarriersinmindit ispossibletodesignthe distance between the surface and the junction plane such that the possibility of recombination is minimised.In other words the charge carrier have to reach the junction before they havetime to recombine.Once they reach the junction, the potential barrier separatesthem. Theholesmoveintothepregionandtheelectrons, whichcannot negotiatethejunctionfield, remaininthenregion, pilingupat theboundary. The magnitude of the photo current caused by the movement of the charge carriers, depends uponthenumberofphotonswhich charted the charge carriers. Thus it is possibleto control the current by changing the level of illumination. In other words, we can exert OPTICALCONTROLontheelectricparametersanditisthiscontrolthatmakesthe Optoelectronic devices so important for instrumentation. It mustbenotedthat thephotocurrentisentirelyduetothedrift flowofthe minoritycarriersacrossthejunction. Let us compare this with the case in which the junction is forward biased.In that case, the potential barrier is lowered and current flows because of the diffusion of the majority carriers across the junction.This diffusion flow overwhelms the drift component.However, it is only the drift component, which can be controlled optically. Thus the diffusion current is parasitic in photo diodes.It becomes impossible to exert optical control if the diode is forward biased.Thus it is necessary that aphoto diode should always be operated with a reverse bias or no bias at all. The potential difference caused by the accumulation of charges on either side of the junction is called the photo EMF.The current caused by the drift flow of the carriers is the photo current.As the drift flow increases the charge accumulated on either side of the junction and consequently the photo EMF increase.The photo EMF is opposite in sign to thejunctionfield. Therefore, theincreaseinphotoEMFcausestheloweringof the potential barrier.This decreases the ability of the potential barrier to separate the charge carriers.Ultimately there comes a point where the potential barrier is no longer capable of separating the holes and electrons.This is the point of saturation of the photo diode.If is the optical flux incident on the photodiode, we can write the number of photonsfallingonthesurfaceas/hf, hfbeingtheenergyofeachphoton. Let us assume eachphotoncreates charge carriers. This is known as the INTERNAL QUANTUM EFFICIENCY.The total charge generated is /hf .Now we have to see howmanyofthesecarriersescape recombination on their way to the junction region. This is given by the base transport factor rB. Altogether the photocurrent can be written as Ip=q rB. /hf Where q is the charge on the electronThis equation suggests that the Ip- curve must be perfectly linear.However all the parameters in the above expression are dynamic parameters and vary with conditions. We have already seen those at large values of , the curve tends to level off.Similarly at lower levels of also the base transport factor is slightly lower and upto a certain min value, the raise in Ip is slow. (figure 2.2)The spectral response curve between and is also not linear (fig 2.3). At the short wave length region, the energy absorption in the surface layer of depth Yo is much higher. This also causes greater levels of recombination. Due to this increased recombination, there is a decrease in the responsivity at the short wave length region and there is a practical short wavelength length cut off, in addition to the long wavelength cut offc.Thedynamicnaturesoftheparametersmaketheresponsecurvesevenmore complicated.For instance, the penetration depth Xo depends on the wave length and the fraction of photons absorbed by the semiconductor depends on Xo so,if the variation of Xo or its reciprocal, the absorption coefficient with wavelength is sharp, then the peak response is a function of the cut-off wave length. It the rate of change of Yo with is small, thenthepeakresponsedependsonthe rate of recombination r.This,wecan adjust by varying the base width. Silicon is an example of the second type of material and germanium is an example of the first type.We have already mentioned the internal quantum efficiency of the photodiode as the number of electrons created by each photon absorbed by the semiconductor.The external quantum efficiency or simply quantum efficiency of a photodiode is defined as thenumber of electronscollectedat thedetector terminal tothenumber of photons incident on the photo diode.sinsecsin secnumberofelectronscollectednumberofphoton cidentrateofcollectionofelectronsper ondrateofphoton cidentper ondn The value of is influenced by (a) the absorption coefficient of the material, (b) the base transport factor and most importantly (c) the frequency of incident light.Amorepracticalparameter of a photo detector is the actual photo current that results fromunit optical power incident on the detector. This is known as the RESPONSIVITY, R./poIR amperes wattP where Ip is the output photo current and Po is the input optical power in watts. If the frequency of the incident radiation is f, then the energy of each photo ishf.From thiswecanwriterpthenumberofphotonsincidentpersecondasPo/hf.Thephoto current can be written as the number of electrons collected per second multiplied by the charge on each electron, re* ethuso pp eP r hfandI r eThis gives us epr e eRr hf hfn This equation gives us a way of calculating the amount of optical power required to give a certain photo current which can be very important in designing sensor systems.Noise in Detectors:The performance of anoptoelctronic instrumentation device depends on the faithfulness with which the detector output follows the optical input variations. Unfortunately, this faithfulness is not this.The basic one is what is generally called shot noise. This stems from the fact that the current flow is not continuous but quantum in nature. The amount of charge flowing has to be the charge on so many electrons. The smallest step of increment is the charge on each electron. This on a microscopic scale thereisalwaysafluctuationaroundtheaveragevalue.Ther.m.svalueofthisshot current is given by( ) ( )1 122 22si eBI where B is the Band width.The other contributor to the noise current is the dark current in the device, which can be minimised by suitable designs and materials. A parameter, which is used to describe the performance of a detector, is the Noise Equivalent power or NEP. The NEP is defined as the incident optical power, at a particular wavelength required to produce a photocurrent equal to the r.m.s.noise current at unit Bandwidth.We have the responsively defined as00ppP hfRI eI hcPenni Now we use the photo current value equal to the shot noise current( )122pI eIB The photo diode average current would include the dark current also ( )( )122p dp p dI I II e I I B + ] + ] If the dark current is negligible 2pI eB 02hcNEP Pni (since B has to be put equal to 1 )Ontheother hand ifthere is a considerable dark current, as in the case of the photo transistor, if over whelms the shot noise and( )( )121222p ddI eI Bhc eINEPe ni The reciprocal of NEP is defined as the detectivityD =1/ NEP.The Specific Detectivity is a parameter which takes into account the area of the detector collecting light, and is represented by D*1*2D DA SPEED OF RESPONSE:Theusefulness of aphotodetectingdeviceininstrumentationsystemgreatly depends on the speed with which it responds to the input variations.This is not only due to the time lapse between the output signal and the input signal.It is also likely to effect the output pulse shape, as shown in fig (2.4). The responsetime must alwaysbewithin specific critical values depending on the process parameters. Ingeneral, itcanbesaidthatthespeedofresponsemust beashighas possible.Thespeedofresponseofaphotodiodedependsonseveral factors. Thefirst among them is the efficiency with which the electric field at the junction separates the charge carriers. The speed with which the charge carriers cross the junction determines the time taken for separation.Ifis the junction width andVmaxthe drift velocity the transittime is/Vmax.The junction thickness depends on the impurity concentration. The drift velocity depends on the time constant of the junction capacitance. These are someofthefactorstobetakeninto account while designing the diodes. The normal range for Si and Ge photodiodes is about 0.1 ns.As already pointed out, what is important in the performance of a photodiode is not the instantaneous speed but the way in which the current through the load varieswith time, when the input is in the shape of a pulse, the output pulse shape is dependant on this factor. The time taken for the output current to raise to a particular value and the time takenfor thecurrent tofall tozerofromthat valuearetheimportant parametersin decidingtheoutput pulseshape. Sowewant therateofchangeofexcessminority carriers at a particular point with time.Theilluminationofthephotodioderesultsinanexcessminorityconcentration being created in the base or n region (fig 2.5) Let this excessholeconcentrationbep.Thedistributionoftheseexcesscarriersat different values of x can be written as d (p )/d x. Now as the diode continues to be illuminated, therewill beavariationinthisdistribution.Thisvariationindependant onthehole distribution constant and can be written as Dp. d( P) / d x.However, this number has to be reduced by the number of charge carriers which recombine.If the life time of the holes in the n region is Tp, then the fraction of holes which recombine is given by P / Tp.So putting these together, we get the time rate of change of minority carriers to be

d (p)/dt=Dp d( P) / d x- P / Tp ------- (2.3)Whenthereisnoillumination, theexcessholeconcentrationthroughout then region is zero.Suppose the illumination starts at t=0.Instantaneously a hole population starts to build up at the surface, where x = 0.The current caused by the hole population is Ip, and the boundary condition at the surface can be written at d(p)/dx |=Ip/(q*Dp)x=0At the junction, where x=w, there is no current at t=0 and ( P)x=w=0.Applying these two conditions, equn.2.3.has the solution ip=Ip[1-exp(-t/Ttr)]-------- (2.4)WhereTr = W/2Dp is the transit time.Once the charge carriers reach the junction it is the efficiency with which they are separated which determines the growth of the photo emf and the slope of the photovoltage pulse.Let us once again look at the whole process. Suppose a rectangular light pulse arrives at the n side of the diode at time t = 0. This triggers the flow of the minority carriers in the n region.The concentration of the minoritycarriersincreases witht.Asthenumber of carriersincreases sodoes the probabilityof collisionsandconsequent rateof recombination.Alsoastheflowof minority carriers increases the hole current through the junction begins to grow and the p regioncontinuestochargepositively with respect to the n region. This results in the photo emf increasing.After sometime tp, the rate of recombination will be balanced by the rate of carrier generation and the photo emf reaches a steady value.The time TRis theraisetimeof thepulse.After thepulseends, concentrationof minoritycarriers decreases and the current and photo emf also decay.Thus the build up of photo emf takes a specific time period dependant upon the rateof generationof excess carriers andtherateof recombination. Eachof these processes is governed by the life time of holes tp.The speed of response also depends on the base width WB and the hole diffusion constant Dp.Improving the speed of response of a photo diode therefore has to look into either decreasing the transit time Ttr or increasing the hole life time. It will be the electron life time if we started by assuming that light is incident on the p side Tp.Ttr depends on WBandDp.The minimum value for WB normally is 10 microns. If we decrease WBfurther the response lowers and the response peak also shifts towards the short wavelength region.Dp for normal doping levels is around 0.01 m s . Which gives Ttr around100ns. Soit becomes necessarytolookoutside Si andGe as fabricatingmaterials. III-Vsemiconductors, whichhavebeendiscussedearlier, have been widely used.Thus in order to increase the response speed over a larger wavelength region, 1) we need a greater depth through which the charge carriers are generated 2) we need to increase the speed of diffusion while at the same time keeping WB at around 10 microns. The transit time can also be written as W/V and we have to look to increasing V. The first requirement means we have to decrease the rate of recombination.Most of these can be achieved by using 1) a narrow p region where the generation of charge carriers takes place 2) following this by a sufficiently wide region of semiconductor so thinly doped that we can call it intrinsic and 3)a final n region which is comparatively doped heavily whichwillprovidegoodexternal contacts.This is known asthePINDiode.See fig (2.6).The advantages of this structure are many.The junctions P-I and I-N can be made very sharp, thusdecreasingthetimeconstant toaround10 sec.Thusthedrift timeis decreased. The initial p region can be brought to the minimum possible thickness but need not be made too thin.Since the resistivity of the Iregion is pretty high, almost all the applied voltage drops across this region.This creates a strong and homogenous field whichpervadesthewholeI region. Thisfieldsweepsthefreecarrierstowardsthe junction. Since the impurity concentration is very small, the collisions are also comparatively less.The velocity with which the charge carriers are swept across the intrinsic region is given by E, where E is the electric field.So the transit time can be written as W/ E =(W/ ) (W/VSR) = W / VSR.In case of the PN diode, this was W/2Dp 2220.5( )2TRpinsrTRpn sr srpKTTV qapproxT V VDuu Given a reverse voltage of 0.1v, gives double the speed of the PN Diode. For a given speedPINDiodescanoperateat longer wavelengths. However theintrinsicregion requires high purity and may be difficult to obtain.Schottky Barrier diode : The Schottkyjunction characteristics alsocanbe utilisedtoobtaina photo detector with a wider response band and greater speed.The Schottky diode is a metal semiconductor junction.Figure(2.6) shows a metal p type semiconductor junction. The fermi level in the metal is higher than that in the semiconductor. Thismeansthat if weconsider anyoneparticular energylevel, the probability of its being occupied is greater in the case of the metal than in the case of the p type semiconductor. Hence, the electron occupancy of the conduction band of semiconductor is lower than that in the metal.When there is an ohmic contact between the two, a fraction of the electrons move from the metal into the p type semiconductor and will neutralise some of the holes present there.As a result a negative space charge appears at the interface (Similarly, if the semiconductor is the n-type, a positive space charge develops).When a forward bias is applied the barrier lowers and when a reverse bias is applied it raises.Used as a photo detector, the light energy is made to be incident on the metal side. Thus it becomes possible that photons of energy less then the semiconductor band gap energyalsoareabsorbed. For theflowof photocurrent it will besufficient if the quantum energy is greater than the schottky barrier.Thus the long wavelength cut off is shifted towards longer wavelengths. Similarly, since the absorption region extends till thespace chargelayer, photonsof higher energy can penetrate up to the space charge layer. This extends the short wavelength cut off also. Since the base resistance is also very small the base being a metal, the time lag is also considerably reduced. Thus the schottkybarrier photodiodeoperatesover awider spectral regionwithconsiderable speed. However the formation of the metal semiconductor junction is not easy and the characteristics are also not reasonably stable.Heterojunction detectorsManyof the shortcomings of junction detectors canbe overcome byusing Heterojunctions.For instance let us consider a hetero structure of GaAs and GaAlAs as shown in figure (2.7). The wide band gap Ga Al As does not absorb the photons and acts as a window. Almost all the radiant energy is absorbed in the active layer, the n type Ga As, and gives raise to charge carriers.The holes travel freely to the pregion while the electrons travel to the n region. In this case the difference of band gaps on either side of the heterojunction is about o.4 ev.If the width of the active region is chosen to absorb all the photons and the purity of the region is high, the loss of photo carriers due to recombination can be minimised and a highefficiencyverynear 100%canbe obtained. Since the entrance windowis transparent, the photons reach the active region rapidly and the speed of the photo diode is also very high.The wave length of operation depends on the difference between the band gaps on eithersideof the heterojunction. The penetration depth for the transparent window is anyway very large.Thus, in order to design a detector for a specific wavelength region, we have to properly select the pairs of semiconductors. The use of a heterojuction results in an increase in the breakdown voltage of the diode. This means that the normal operating reverse bias given to the detector is much less thanthebreakdownvoltage.As aresult thereverseleakagecurrent is kept a minimum.Which improves the responsivity of the diode.However the fabrication of heterodiodes is a much more complicated process than that of homojunctions and this stands in the way of their being used more widely. The technological bottlenecks are very rapidly being solved.All the detectors we discussed so far, the photo emf or the photo current depends onthenumberofchargecarrierswhichescaperecombination.Thus the efficiencyis always less than 100%.When it becomes necessary to detect feeble light signals, this is not always sufficient.In order to get an output pulse of sufficient amplitude we have to amplifytheoutput of thedetector. Insteadofusinganexternal amplifiersit isalso possible to obtain an internal gain in the detector itself.PHOTODETECTORS WITH INTERNAL GAINThe first type of detector we consider under this family is the photo transistor.In principle, we use the photo current here as the base current and the output is drawn from the collector. The structure of the device is shown in figure (2.8). Since the input to the base is the photo current, base does not require a separate terminal. So usually a photo transistor has only two terminals.Absorptionoflight fluxbythebaseresultsinthecreationofchargecarriers. These are separated by the collector junction.The holes cross into the p region while the electrons remain in the n region.This build up of electrons causes a space charge layer at the emitter junction. As a result the potential barrier at the emitter junction is lowered and holes cross in to the n region and the total current is increased.The out put characteristics of a photo transistor are therefore similar to those of an ordinary bipolar transistor .. The current responsivity of the photo transistor will bep times the responsivity if only the collector junction was operating, p being the common emitter current gain.As long as there is a potential difference between the collector and emitter, there will be some current flow even if there is no base current (see curve for -O in fig ) Therefore the dark current in a phototransistor has an appreciable value and is given by Id=Ieo(1+) The temperature stability of the device is also low because the base operates with a d.c. bias on it.In order to improve the gain and decrease the switching time the base width has to be decreased. But if we decrease the base width, the area available for flux absorption will be less.Thus the optimum switching time obtainable is about 10s.To improve both the gain and switching time it is preferable to use a diode for detectionandtransistorforamplification. Ifbothofthemarefabricatedonasingle monolithic chip; the device is known as a composite photo transistor. The structure is shown is figure (2.9).The diode can be designed for high speed and the transistor can be designed for high gain, neither one affecting the other switching time are around 10 s.A more commonly used structure is the photo darlington , where two transistors areconnectedinadarlingtonconfiguration, thusachievingatwostageamplification. The structure and its schematic are shown in fig (2.10).When light falls on the first base, a photo current is generated. This is the base current of the 1st transistor IB1 .The current through the first collector Ic, is given by IB1.The current through the first emitter is ( 1 + ) IB1. The current through the first emitter is fed to the second base and can be redesigned as IB2.The current through the second collector is I B2 2 or (1 + ) 2 I B1 .The total output current is Ic=Ic+Ic2= IB1+ 2 I B2 = I B1( + 2 + 2)The photo darlington can give on output current that is about 1000 times that of the ordinary photo diode. The switching time of a photo darlington is, however, high, being around 10s.Avalanche Photodiodes:Devices using the avalanche breakdown phenomenon in p-n junctions give better out put voltages than any of these.The APD or the Avalanche photodiode is very widely used in optical fiber communication systems, inspite of some draw backs.Inprinciple, anavalanchephotodiodeis apnjunctionoperatedunder heavy reverse bias causing breakdown of the diode (fig. 2.11 a). Photon generated carriers acquire high energies while crossing the break down field.It becomes possible for the carriers with sufficient energy to collide with the lattice sites and create new electron hole pairs.This is known as impact ionisation (fig.2.11b).It is likely that thenewlygeneratedpairwill, in turncreateanother pair andthisaction can cascade. Current gains in excess of 100 are readily obtainable. However there are two specific drawbacks the first is that the current gain varies with the value of the bias voltage.The second one is that the multiplication process is not necessarily limited to photo generated carriers. So any noise is also automatically multiplied. The third is that the avalanche process takes some time to build up and this increases the switching time.The avalanche process itself is erratic and random.There are fluctuations in the actual distancebetweensuccessivecollisions.Againtheprobabilityof avalancheis greater for electrons than for holes.All of these give raise to noise and increase the delay in switching.Normally the bias voltages required for impact ionisation are about 50 to 400v dependinguponthejunction. Newer designshowever madeit possibletousebias voltages around 15-25 volts.As avalanche photo diode usually produces asymmetric pulses.This is due to the pulse rise time depends on the fast electron transit and fall depends on the slower hole transit, silicon, germaniumandIII-Vheterostructureshavewidelybeenusedin APD design.We will discuss these structures in some detail in chapter 2.B.Special types of DetectorsWe now come to two special types of structures. Even though these can not be said to be radiation measuring systems; they very certainly are radiation sensing systems and very useful in instrumentation and / or control.The photo thyristor is a four layer semiconductor structure as shown as figure (2.12a). The schematic representation is as in (2.12b).When radiant energy is incident on the devise as shown, photo currents start to flow at the three junctions. Let Ip, be the photo current generated at p1 n1 junction, Ipa at the n1p2junction and Ip3at the p2n2 junction.Each of these currents will have some effect on the current at the other junctions. Let the total current through the p1n1junction be Ij1, n1p2junction be Ij2and p2n2 junction be I j3.

The structure can be considered to consist of two transistors, one p-n-p and the othern-p-n.Letus say theintrinsic current through the collector junction of the first transistor is Ico, the total current through that junction consists of three parts. The first one is the current generated at that junction.This is the sum of the intrinsic current and thephotocurrent, (Ico+Ip2). Thesecondpart istheeffect ofthecurrent at thefirst junction.This current is Ij= I + Ip .If , is the current gain then the effect of Ijat J2will be (I+Ip1 ) . The third part is the effect of the current at the third junction. This is given byIj3=I+Ip3.If 2, is the current gain of the second transistor, the effect of Ij3 at J2 will be ( I + Ip3 ) 2Thus we can write Ij2=(Ico+Ip2)+ (I+Ip1 ) +(I+Ip2 ) 2 .= I because of the current continuumThis leads toI (1- - 2) = Ico+ Ip1 + Ip2+ Ip3 2When there is no light incident on the device

Ip1 = Ip2=Ip3 =0 AndI(1- - 2)=IcoIf we draw the V-I characteristics for = o, it will appear as the curve (a) of fig (2. 13)Whenthephotothyristor is illuminated, the R.H.S.of eqn (2.) will have the additional terms.This results in the required current appearing at lesser voltages. As the incident flux, increases to 1, 2, 3etc. the characteristics will appear as shown in fig 2.13.Thus, the switching voltage of the thyristor decreases with increasing illumination.The quantityIp1 + Ip2+ Ip3 2plays the role of the control current in the ordinary thyristor.Photo thyristors are very useful in high power switching applications.----------FIBEROPTIC SENSORS FOR PROCESS INSTRUMENTATIONAninstrumentationsystemingeneral consistsofthreemajorparts.Aninput derived from a sensor, an electronic device which performs some processing on this input and an output device.Inthepresentcontextweareconcerned with sensors whichuse optical fibers. The essential function of a sensor is to convert one form of input energy, which we call the measurand into another form of energy. The output can also be the same form of energy but in another mode,for instance amplified.The process of energy conversion in a transducer can be a single stages processor a two-stage process.In a fiber optic sensor the measurand is made to change the features of a light beam transmitted along the fiber andthesechangesareusedtocreate or modify an electronic signal in the receiver or actuator.Thistwostageprocessinabsolutelybasictotheoperationofafiberoptic sensor.Thereforeitisnecessary that both the stages of interaction are very carefully designed.Fiber optic sensors are extremely useful in several applications. In areas where high and variable electromagnetic fields are present, the immunity of fiber optic systems makes them very useful.In a large member of avionics and space systems where weight and radar or radiation signatures from the instrumentation are important considerations, fiberopticsystemshave found extensive applications. The ever increasing activityin medical instrumentationwheresafety, particularlyinintravenous useof paramount importance, we canuse fiber optic sensors withadvantage. Similarlythe security industry is mainly attracted by the noise immunity, remote operation and safety considerations.Anopticalfibersensoris defined as a means by which light guided through a fiber can be modified in response to external.Physical, chemical, biological biomedical or similar influence. The light from a source is guided through a fiber to the point at which the measurement is made.At this point there are two options available.The light beam can be made to come out of the fiber, modulated in a separate zone and re launched into either a second fiber or the same fiber. This type of sensor is called an Extrinsic sensor.Here the fiber is simply used as a carrier of information and plays no part in the modulation. Ontheotherhand, we can modulate the light while passing through the fiberinresponsetothemeasured. Inthisthecharacteristicsofthefiberareusedin modulation.This type of a sensor is called an Intrinsic sensor. Thespecificmeritsofafiber optic sensor system depend on thepossibilityof energytransmissionwithout anyelectrical connection. Inadditionthe information carrying capacity of a fiber is enormous and thus gives us the capability of multiplexing several systems and form distributed sensing networks.The basic structure of a fiber optic sensor can therefore be shown as in the figure. The modulator M can be either intrinsic or extrinsic.If the signal from the source is S and the transmission factor of F, is T, the energy that reaches the modulator is ST. The modulator superimposes a modulation function M on this signal, dependant on Q, the measurand . If the transmission factor of Fis T, the signal that reaches the detector system isST, TQM. IfR isthe responsive its of the detector. The output electrical signal is given by ST,T QMR.The heart of a fiber optic sensor is obviously the modulator where the light signal ismodulatedinresponsetothephysical measurand. Thismodulationcanbeeither digital or analogue. Theforemost of theanaloguequantities is theintensity. The intensityistheparametertowhichall optical detectorsfinallyrespond.Thesensors typically use the position of a moving mirror back into a fiber or a mask, which varies the aperture, and in turn the transmission through a fiber.Shot noise in the source or detector and variations in the response of the detector system are likely to cause false triggering and must be corrected.Another optical parameter which lends itself to analogue modulation is the optical phase, where even changes of the order a microradian can be detected. Phase measurement is typically the measurement of time delay and the principles of interferometry have to be used.Polarisation modulation is very much similar to the interferometric technique but here we use the phase variation between two orthogonal polarisation vectors. Intrinsic fiber sensors using brefringent fibers use this principle excellently.A fourth parameter that can be modulated is the color or the frequency of the light beam. Thislendsitselftounambiguousdetectionbytheuseof filters. Thewidest applicationofthistechniqueiswhereweusetemperaturedependant luminescenceof several materialstomeasuretemperature. Thedigital modulationtechniquescanbe implementedbyon-offintensitymodulationuseful inlimit switches. Similarlyifwe make the input light signal to fall on a vibrating mirror the out put light has a modulation frequency same as that of the mirror. This frequency can be made to be dependant on the measurand.Flowmeasurement sensors commonlyuse the Doppler shift as a means of unambiguous determinationof apparent speed. This techniquecanalsobeusedto measure particle velocities inside a liquid. Hybrid fiber optic systems are those where conventional transducers are interfacedwithopticallypoweredlowpower systems. Inthesetheoutput froma conventional transducer is digitally intensity modulated on to a fiber optic system.In this foregoing, we will look at the structural details of a number of fiber optic sensor systems which are useful in measuring and monitoring process variables. Displacement sensors :Fiber optic displacement sensors are generally intensity modulated. These find application, in addition, as proximity sensors and in hybrid sensors for measurement of pressure, temperature and other parameters.The simplest configurations of displacement sensors are shown in figure. (a) This is a sensor for measuring transverse displacement.The sensitivity is dependent on the NA of the fibers. (b) The same configuration can also be used as a long itudinal displacement sensor, the numerical aperture again playing the decisive role.( c )This is a sensor for measurement of angular displacement.Theconfigurationcanbemodifiedtoobtainadifferential displacement. The output current giving the difference between the two detectors outputs. This configuration is useful in maintain the position of a component.The accuracy of these sensors can be substantially improved by using beamexpanders as shown in the accompanying figure. (d)(e) In the place of the shutter if we use a moveable grating in conjunction with another fixed grating, the sensitivity can be greatly enhanced.Theprincipleoftheshuttermodulatedon-offsensorsisappliedincomponent known as the optical microswitch, as shown in the figure The optical microswitch can be operated neither the normally off configuration or the 'normally ON' configuration. With this component however it becomes difficult to say whether the switch is off or if there is any fault in the fiber link. Therefore usually the shutter is arranged to close the optical path only partially.Due to this, even if the switch is off there is still some energy coupling between the two fibers.If this energy does not appear it means that there is a discontinuity in the fiber.The fiber optic on-off switches find application in components called optical fiber interrupters.The principle function of the interrupters is to detect moving objects.These canbeeither reflectiontypeor transmissiontype. Inthereflectionsensor thelight becomeemittedfromthefiber isreflectedbackintothesamefiber. Incaseof the transmission type the objects act like a shutter, resulting in no light being coupled into the output fiber.In general, their interrupters employ incoherent light, use large core plastic fibers and the minimum delectable size of the object is.These can Operatefrom -40C to 70 C. These can be utilised for the following applications among others.1) Mark detection---- Date stamping and character pattern recognition.2) Color differencerecognition.--- Electricwires. Plasticsheet paints, Resistance Color codes.3) Defect detection 4) Label detection5) Counting of discrete products.It is possible to design complex position and mark sensors like bar code readers or signature verification using arranged fiber bands.These again can be reflective or transmittive types as shown in the figure.Optical Microswitch:Thetypical differential attenuation, whichdeterminestheON-OFFcontrast, is about 20dB.When a greater level of fault detection sensitivity is required use is made of positionsensitivemicroswitches inarobust design. Inthis weusearetroreflector material for reflecting a light signal back into the input fiber.The retroreflector is mounted on flexible spring blades.The spring blades in turn are connected to the mechanical interface, which takes in the parameter under test. When the retroreflector is shifted or inclined form its normal position, the light flux is no longer reflectedintothe fiberand theoutput signal is zero. This type of microswitches find special application in the oil industry for detecting value positions, indicating whether the value is open or not. When it becomes necessary to test several positions, it becomes necessary to code the test signals. In these circumstance we use what are called delay lines.These are additional lengths of fiber added in any particular path.This works on the simple principle that the greater the distance to be traveled, the more is the time taken by the light pulse to return.Thus one pulse can be made to return after another.Let us examine one commercial design.The 3dB coupler devides the pulses from the LED into the inputs of the two paths oneofthepathshasadelayline incorporated. After the delay line the pulses go the microswitch, come back, and are detected by the PIN Diode.Signals of the type shown are figure are receive o.Dependinguponthe microswitch configurations, these signals may mean valve closed, valveopen, intermediateetc. Thissystemmakesit possibletosensevalue position over a distance of 600m with a PCS fiber having an 8dB/km-transmission loss. Longerdistancescanbeachieved usingstronger lightsources and fiberswithsmaller transmission losses.Temperature sensors:Practical FOSfor temperaturesensinguse1) thetemperaturedependenceof energy band gap in a semiconductor. 2) The photo luminescent spectrumof a semiconductor which again is dependent on the band gap and 3) the temperature dependence of fluorescent emission from phosphors.Systems based on these phenomena can operate in temperature ranges of -100 C to 400 C.High temperature sensors with ranges of 500 C to 2000 C are usually based on black body radiation and principles of optical pyrometry. Thefirst typeof FOTSareusuallypoint contact typeandfind application in microwave processes in the industrial plants like drying methods for tablets (pharmaceutical), andwood, curingprocesses for glues, resins andplastics, heating processes for food, rubber and oil, joint welding. More than all of these, the strongest attention is now centred on semiconductor device fabrication which may use processes likeplasmaetching, ionimphanlationetc whicharedone inelectricallyhostile environments. Similarly since the OFTS are chemically inert, they have advantages for temperature sensing in electrochemical processes.Temperature sensing in electrical power machines such as transformers generators bus bars etc. can be conveniently done by using OFT's.Hybrid temperature sensors use the temperature sensing elements like a platinum resistance and combine them with fiber optic transmission.We will now describe some common Fiber Optic Sensors for temperature measurement.Semiconductor absorption sensor:The operation of this sensor depends on the fact that the energy band gap of most semiconductorsdecreasesalmost linearlywithtemperature. Iftheincident energyis greater than the band gap energy, that energy is absorbed by the semiconductor and used up in upward transitions of the electrons.If the incident energy is less than the band gap energy. It will not be absorbed by the semiconductor. Therefore for energies less than the band gap energy the light intensity transmitted by the semiconductor will be considerable.As the energy crosses the band gap energy, the intensity will show a sharp fall and practically no transmitted light will be visible. Thisis shown in curve a)in figure frequency- transmission curve will be similar to the energy Transmission curve. On the other hand, if we try to see the wave length transmission relation, at wavelengths below a particular value corresponding to the energy band grap there willbe no transmittedintensity and above that wavelength value the transmitted intensity is considerable.This is shown in figure. (b) The critical value of the wavelength is known as the absorption band edge. If a light signal near the absorption band edge falls on the semiconductor, as shown in the figure, the transmitted, intensity will depend on the point where the absorption band edge intercepts the signal pulse.Now with increasing temperature, the energy gap of the semiconductor decreases. As a consequence, the absorption edge shifts towards longer wavelengths.This result in decreased transmission of the signal pulse through the semiconductor, as shown in figureTherefore this variation in intensity will be a measure of the temperature. The basic requirement is that the signal pulse must be in the region of the absorption edge. Therefore it is common to use Ga As as the sensor and a Al Ga As LED which gives out light at 0.88 in conjunction. The band edge shift for Ga As is about 3A per degree kelvin.A polished Ga As chip is attached to the fiber and mounted in a stainless steel capillary tube of 2mm diameter.The front surface of Ga As is coated for anti-reflection.The back surface of the GaAs chip is coated with gold to return the light into the fiber.Whendesigninganinstrumentationsystembasedonthisprinciple, weshould take care that loss either in the fiber or connector will not cause measuring errors.For the purposewesendanothersignal ofwavelengthmuchhigherthantheabsorptionband edge through the fiber.The signal position is as shown in fig.Its wavelength being far way form the absorption edge, the semiconductor will not absorb this signal and the transmitted energy will not be affected by the shift in the band edge. However the losses in the fiber and couplers will be the same for both the signals. Whilemeasuring, the ratio of the transmitted intensities of the two signalsis taken to be indicative of the temperature.The system configuration for this sensor is as shown in the figure.Al Ga As LED is the test signal prouder sending pulses of s = 0.88 .In Ga AsP LEDisthereferencesignalprovidersendingpulsesofs= 1.3 .Theseare guided through fibers and couplers 1 and 2 to the sensing element.In the sensing element, the intensity of s is temperature modulated and r is unaffected. On return from the fiber coupler2divertstheoutputsignal to the Ge APD detector, which can be used here since any variation will be felt byboth the signals. Finallythe signal processor normalises the test signal withthe reference signal andsends it toa displayas a temperature value.It is possible to use a number of these configurations together to obtain a multi channel temperature measurement.UsingaGa Assensor,AlGaAsLED, asilicafiberof 100coreandasensor diameter of 2mm, it is possible to achieve an accuracy of better than 2.0C in the range -20 to 150C . Temperature sensor using semiconductor photo luminescence: Wearefamiliar withtheprincipleof luminescencewhereif electronsinthe valencebandareexcitedtohigher levels, whentheyfall backtothegroundlevel, electromagnetic energy is emitted. It the excitation is by electrical means, it is electro luminescence and results in the design of LEDs.We have seen that semiconductors are useful for this purpose due to the proximity of their band gap energy to the energy of electromagneticradiationintheopticalregion.Theexcitationoftheelectronstothe upper levels canalsobeachievedbysupplyingenergyintheformof light. This phenomenon is called photo luminescence. If the light beamincident on the semiconductor surface has a photon energy greater than the band gap energy, electrons in the valence band absorb this energy and jump to the higher levels.When they fall back they will emit light.It should be noted that, whatever be the value of the exciting energy, the emitted photons would always have an energy corresponding to the band gap.Thus thefrequencyofluminescentenergy isalways lessthanthat oftheincidentlight. In terms of the wave length, is always greater than e.We can use this principle for measuring temperature. As we have already seen increasing temperature results in the decrease of band gap energy. Therefore a raise in temperature will result in a raise in the wavelength of light emitted. If we can characterise the luminescent emission, we can find out the temperature.Insteadof tryingtomeasurethewavelengthof light, whichislikelytobea cumbersome process, we try to measure the intensity of different wavelengths.It must be rememberedthatthesensormaterialbeingasolid, theemittedlightconsistsofnota singlewavelengthbut asmall spreadof wavelengthsas showninthefigure. With increasing temperature, this curve will move towards longer wavelengths.If we pass this light through an high pass interference filter whose cut off is not steep but gradual, some of the wave length will be reflected and the rest will be transmitted, as shown in figure.Inposition A, mostofthelight will bereflectedandonlyasmall percentage transmitted. In position B, around half of the light is reflected. In position C only a small amount is reflected and most of the light is transmitted.The ratio of the reflected to transmittedintensityis therefore a measure of the positionof peakwavelengthof emission and consequently of the temperature.Thetemperaturesensor designed on this principle must have a high efficiency, giving out intensity sufficient for the measurement sensitivity to be achieved. For this purpose we use a double heterostructure, where we know that the localisation of charge carriers results in high levels of recombination A GaAs- Ga Al As-Ga As structure is generally used.The sensing element is attached to the end of a 100-core silica fiber.The sourceforexcitationisaLEDwithpeakwavelengthofaround750nm. Thesensor diameter is around 0.6mm. The systemconfiguration is shown in figure. The excitation signal from the LED is brought in through a fiber.At the end of the input fiber a GRIN lens is attached and the light is focussed on to a steep cut off interference filter, IF.IF is transparent to the luminescent wavelengths but not the excitation wavelength. Therefore the excitation signal is reflected back into the GRIN lens which couples it to a second fiber which is connected to the sensor element.IF1isattachedtoasmallglassblock.Theluminescent light fromthesensor comes through the second fiber, meats 1F1, travels through it and refracted at the surface of the glass block, falls on IF2, which transmits some intensity and reflects some. The transmitted intensity is measured by detector 1.The reflected intensity is measured by detector 2.The output of the two detectors are fed to a signal process or which calculates the ratio and possibly also the temperature if sufficient data has been fed into it.Temperature sensors of this configuration can work in the range of 0 to 200 CAnd remote measurement from about 600 to