Contributed paper OPTO-ELECTRONICS REVIEW 12(2), 221–245 (2004) (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:7)(cid:8)(cid:9)(cid:10)(cid:3)(cid:10)(cid:5)(cid:3)(cid:11)(cid:12)(cid:13)(cid:8)(cid:14)(cid:11)(cid:12)(cid:8)(cid:14)(cid:11)(cid:5)(cid:6)(cid:7)(cid:8)(cid:2)(cid:7)(cid:6)(cid:15)(cid:10)(cid:8)(cid:6)(cid:12)(cid:12)(cid:6)(cid:16)(cid:13)(cid:1) A. ROGALSKI* Institute of Applied Physics, Military University of Technology 2 Kaliskiego Str., 00-908 Warsaw, Poland Thepaperpresentsprogressinopticaldetectortechnologiesduringthepast25years.Classificationoftwotypesofdetectors (photondetectorsandthermaldetectors)isdoneonthebasisoftheirprincipleofoperation.Theoverviewofopticalmaterial systemsanddetectorsispresented.Alsorecentprogressindifferenttechnologiesisdescribed.Discussionisfocusedmainly oncurrentandthemostrapidlydevelopingfocalplanearraysusing:CdZnTedetectors,AlGaNphotodiodes,visibleCCD andCMOSimagingsystems,HgCdTeheterostructurephotodiodes,quantumwellAlGaAs/GaAsphotoresistors,andthermal detectors.Theoutlookfornear-futuretrendsinIRtechnologiesisalsopresented. Keywords: photondetectors,thermaldetectors,opticaldetectors,focalplanearrays,milti-colourdetectors. (cid:17)(cid:18) (cid:19)(cid:15)(cid:3)(cid:12)(cid:11)(cid:9)(cid:20)(cid:5)(cid:3)(cid:4)(cid:11)(cid:15) Hubble space telescope delivered a deep-space picture, a resultof10day’sintegration,featuringgalaxiesofthe30th Looking back over the past several hundreds of years we magnitude – an unimaginable figure even for astronomers noticed that following the invention and evolution of opti- of our generation. Probably, the next effort will be in the cal systems (telescopes, microscopes, eyeglasses, cameras, big-band age. Thus, photodetectors continue to open to etc.), the optical image was still formed on the human ret- mankindthemostamazingnewhorizons. ina, photographic plate, or films. The birth of photode- This paper is a guide over the arrays of detectors sens- tectors can be dated back to 1873 when Smith discovered ing optical radiation. Optical radiation is considered as a photoconductivity in selenium. Progress was slow until radiation over the range from vacuum ultraviolet to the 1905, when Einstein explained the newly observed photo- far-infrared or submilimeter wavelength (25 nm to electric effect in metals, and Planck solved the blackbody 1000µm): emissionpuzzlebyintroducingthequantahypothesis.Ap- plications and new devices soon flourished, pushed by the 25–200nm Vacuumultraviolet VUV dawning technology of vacuum tube sensors developed in 200–400nm Ultraviolet UV the 1920s and 1930s culminating in the advent of televi- 400–700nm Visible VIS sion. Zworykin and Morton, the celebrated fathers of 700–1000nm Nearinfrared NIR videonics, on the last page of their legendary book Televi- 1–3µm Shortwavelengthinfrared SWIR sion (1939) concluded that: “when rockets will fly to the 3–5µm Mediumwavelengthinfrared MWIR moonandtoothercelestialbodies,thefirstimageswewill 5–14µm Longwavelengthinfrared LWIR seeofthemwillbethosetakenbycameratubes,whichwill 14–30µm Verylongwavelengthinfrared VLWIR open to mankind new horizons”. Their foresight became a 30–100µm Farinfrared FIR reality with the Apollo and Explorer missions. Photolith- 100–1000µm Submillimeter SubMM ographyenabledthefabricationofsiliconmonolithicimag- ing focal planes for the visible spectrum beginning in the (cid:21)(cid:18) (cid:22)(cid:7)(cid:6)(cid:13)(cid:13)(cid:4)(cid:14)(cid:4)(cid:5)(cid:6)(cid:3)(cid:4)(cid:11)(cid:15)(cid:8)(cid:11)(cid:14)(cid:8)(cid:9)(cid:10)(cid:3)(cid:10)(cid:5)(cid:3)(cid:11)(cid:12)(cid:13) early 1960s. Some of these early developments were in- tended for a picturephone, other efforts were for television Progress in optical detector technology is connected mainly cameras, satellite surveillance, and digital imaging. Infra- with semiconductor IR detectors which are included in the red imaging has been vigorously pursed in parallel with classofphotondetectors.Inthisclassofdetectors,theradia- visible imaging because of its utility in military applica- tion is absorbed within the material by interaction with elec- tions. More recently (1997), the CCD camera aboard the trons. The observed electrical output signal results from the changed electronic energy distribution. The photon detectors show a selective wavelength dependence of the response per *e-mail:[email protected] unit incident radiation power. They exhibit both perfect sig- (cid:1)This paper was presented during the XVII School of nal-to-noise performance and a very fast response. However, Optoelectronics: Photovoltaies-Solar Cells and Detectors, KazimierzDolny,October13–17,2003. toachievethisininfrared(IR)spectralregion,thephotonde- Opto-Electron.Rev.,12,no.2,2004 A.Rogalski 221 Opticaldetectorsforfocalplanearrays tectors require cryogenic cooling. Cooling requirements are resolution below 0.1 K is reached because effective noise the main obstacle to the more widespread use of IR systems bandwidthslessthan100Hzcanbeachieved. based on semiconductor photodetectors making them bulky, heavy, expensive and inconvenient to use. Depending on the (cid:23)(cid:18) (cid:24)(cid:10)(cid:12)(cid:14)(cid:11)(cid:12)(cid:25)(cid:6)(cid:15)(cid:5)(cid:10)(cid:8)(cid:13)(cid:2)(cid:10)(cid:5)(cid:4)(cid:14)(cid:4)(cid:5)(cid:6)(cid:3)(cid:4)(cid:11)(cid:15)(cid:13) nature of interaction, the class of photon detectors is further sub-divided into different types (see Table 1). The most im- Toprovideeasycomparisonbetweendetectors,certainfig- portant are: intrinsic detectors, extrinsic detectors, uresofmerit,computedfromthemeasureddata,havebeen photoemissive(metalsilicideSchottkybarriers)detectors,and defined. quantumwelldetectors. Thevoltage(oranalogouscurrent)responsivityisgivenby Thesecondclassofdetectorsiscomposedofthermalde- Q tectors. In a thermal detector, the incident radiation is ab- R (cid:2) u, (1) P sorbedtochangetemperatureofthematerial,andtheresul- tant change in some physical properties is used to generate where Q is the output quantity supplied by the detector u an electrical output. The detector element is suspended on (e.g., the current I , the voltage V , or any other physical u u lags which are connected to the heat sink. Thermal effects quantity)andPistheincidentradiantpower. are generally wavelength independent; the signal depends At equal responsivity, the detector with the smallest upon the radiant power (or its rate of change) but not upon output noise Q on the useful signal is the most sensitive. u itsspectralcontent.Inpyroelectricdetectorsachangeinthe Therefore,thefirstfigureofmeritforadetectoristheNEP internal spontaneous polarization is measured, whereas in – noise equivalent power defined as the ratio of output thecaseofbolometersachangeintheelectricalresistanceis noisetoresponsivity measured. The thermal detectors typically operate at room g temperature. They are usually characterized by modest sen- NEP (cid:2) n. (2) R sitivity and slow response but they are cheap and easy to use.Thegreatestutilityininfraredtechnologyhasfoundbo- So, the NEP represents the input power that gives a unity lometers,pyroelectricdetectorsandthermopiles. signal to noise ratio, S/N = 1 at the output; that is, a mar- Up till the nineties last century, thermal detectors have ginalconditionofdetection. been considerably less exploited in commercial and mili- Thebetterthedetectorperformanceis,sincethesmaller taryinfraredsystemsincomparisonwithphotondetectors. theNEPis.Thereforeitismoreconvenienttodefineitsin- The reason for this disparity is that thermal detectors were verse as a merit figure. In addition, it should be taken into popularly believed to be rather slow and insensitive in consideration that whatever the noise source is, it can be comparison with photon detectors. As a result, the world- expectedthatthenoisequadratictotalvalueisproportional wide effort to develop thermal detectors was extremely toobservationbandwidth(cid:1)fandthedetectorareaA.Thus, smallrelativetothatofphotondetector.Inthelastdecade, is even better to take, as the intrinsic noise parameter of a however, it has been shown that extremely good imagery detector,theratioNEP/(A(cid:1)f)1/2normalizedtounitareaand can be obtained from large thermal detector arrays operat- bandwidth.Inordertosimplifythecomparisonofdifferent ing uncooled at TV frame rates. The speed of thermal de- detectorsandtohaveaparameterthatincreasesastheper- tectors is quite adequate for non-scanned imagers with formanceimproves,thedetectivityD*(calledD-star)isde- two-dimensional (2D) detectors. The moderate sensitivity finedas ofthermaldetectorscanbecompensatedbyalargenumber ofelementsin2Delectronicallyscannedarrays.Withlarge (A(cid:1)f)12 D*(cid:2) . (3) arrays of thermal detectors the best values of temperature NEP Table1.Photondetectors. Type Transition Electricaloutput Example Intrinsic Interband Photoconductive AlGaN,Si,GaAs,PbSe,InSb,HgCdTe Photovoltaic AlGaN,Si,InGaAs,InSb,HgCdTe Capacitance Si,GaAs,InSb,HgCdTe PEM InSb,HgCdTe Extrinsic Impuritytoband Photoconductive Si:In,Si:Ga,Ge:Cu,Ge:Hg Freecarriers Intraband Photoemissive PtSi,Pt Si,IrSiSchottkybarriers 2 GaAs/CsO Photoconductive InSbelectronbolometer Photon-drag Ge Quantumwells Toand/orfromspatially Photoconductive GaAs/GaAlAs,InSbnipi quantisedlevels Photovoltaic InAs/InGaSbSLS 222 Opto-Electron.Rev.,12,no.2,2004 ©2004COSiWSEP,Warsaw Contributed paper This is the fundamental figure of merit used for detectors. trated.ThereadercanconverttotheD*valuesappropriate Itcanbetransformedtothefollowingequation to the photoresistors and photovoltaic detectors by multi- plyingthedetectivityvalueillustratedbythesquarerootof (A(cid:1)f)12 S the detector area. The signal fluctuation limit shown in the D*(cid:2) . (4) figureisindependentofarea[seeEq.(5)]. P N Table2.Thermaldetectors. The ultimate performance of detectors is reached when the detector and amplifier noise is low compared to the Detector Methodofoperation photonnoise.Thephotonnoiseisfundamentalinthesense Bolometer Changeinelectrical thatitarisesnotfromimperfectioninthedetectororitsas- Metal conductivity sociated electronics but rather from the detection process Semiconductor itself,asaresultofthediscretenatureoftheradiationfield. Superconductor The radiation falling on the detector is a composite of that Ferroelectric fromthetargetandthatfromthebackground. Hotelectron When photodetectors are operated in conditions where Thermocouple/Thermopile Voltagegeneration,causedby the background flux is less than the optical (signal) flux, changeintemperatureofthe the ultimate performance of detectors is determined by the junctionoftwodissimilar materials signal fluctuation limit (SFL). It is achieved in practice with photomultipliers operating in the visible and ultravio- Pyroelectric Changesinspontaneous letregion,butitisrarelyachievedwithsolid-statedevices, electricalpolarization which are normally detector-noise or electronic noise lim- Golaycell/Gasmicrophone Thermalexpansionofagas ited.TheNEPofdetectorsoperatinginthislimithavebeen derivedbyanumberofauthors(seee.g.Kruseetal.[1,2]). Table3.AreasofdetectorsillustratedinFig.1. The NEP in the SFL is given (when Poisson statistics areapplicable)by Detector Area(cm2) CdSphotoconductor(PC) 1.00 2hc(cid:1)f NEP (cid:2) , (5) CdSephotoconductor(PC) 1.00 (cid:2)(cid:3) SiSchottkybarrierphotodiode 0.03 wherehisthePlanck’sconstant,cisthelightvelocity,(cid:2)is Sip-njunctionphotodiode 0.25 thequantumefficiency,and (cid:3)isthewavelength. Siphotoconductor 0.25 Thepracticaloperatinglimitformostinfrareddetectors is not the SFL but the background fluctuation limit, also Siavalanchephotodiode 0.07 known as the background limited infrared photodetector Gephotoconductor(PC) 0.20 (BLIP) limit. In this approximation the NEP is given by Geacbiasphotoconductor(PC) 2.4×10–5 [1,2] Photomultipliers(PM) 1.00 12 (cid:3)2A(cid:4) (cid:1)f(cid:6) NEP (cid:2)hv(cid:5) b (cid:8) , (6) Forinfraredfocalplanearrays(FPAs),therelevantfig- (cid:4) (cid:2) (cid:7) ure of merit is the noise equivalent temperature difference (NEDT). Noise equivalent difference temperature of a de- where (cid:4) is the total background photon flux density b tectorrepresentsthetemperaturechange,forincidentradia- reaching the detector and (cid:1)f is the electrical bandwidth of tion, that gives an output signal equal to the rms noise the receiver. The background photon flux density received level.NEDTisdefined by the detector depends on its angular view of the back- V ((cid:5)T (cid:5)Q) (cid:1)T ground and on its ability to respond to the wavelengths NEDT (cid:2) n (cid:2)V , (7) containedinthissource. ((cid:5)V (cid:5)Q) n (cid:1)V s s Typical D* values for available optical detectors are shown in Figs. 1 and 2. Figure 1 shows the spectral where V is the rms noise and (cid:1)V is the signal measured n s detectivity of optical detectors responding in 0.1–1.2 µm for the temperature difference (cid:1)T. It can be approximated region.NotethatdetectivityisnotD*,butratherreciprocal that of NEP for a 1-Hz bandwidth. This figure of merit is em- ployed to include photomultipliers whose noise does not NEDT (cid:2)(C(cid:2)BLIP Nw)(cid:9)1, (8) depend in all cases upon the square root of the photo- cathode area. Table 3 lists the areas which Seib and where C is the thermal contrast, N is the number of photo- w Aukerman[3]statearepropertothevariousdetectorsillus- generatedcarriersintegratedforoneintegrationtime,t int Opto-Electron.Rev.,12,no.2,2004 A.Rogalski 223 Opticaldetectorsforfocalplanearrays N (cid:2)(cid:2)At Q . (9) w int B whereQ isthephotonfluxdensityincidentonthedetector B areaA. PercentageofBLIP,(cid:2) ,issimplytheratioofphoton BLIP noisetocompositeFPAnoise 12 (cid:3) N2 (cid:6) (cid:5) photon (cid:8) (cid:2) (cid:2) . (10) BLIP (cid:4)(cid:5) N2photon (cid:10)NF2PA(cid:7)(cid:8) It results from the above formulas that the charge han- dlingcapacityofthereadout,theintegrationtimelinkedto the frame time, and dark current of the sensitive material becomes the major issues of IR FPAs. The NEDT is in- versely proportional to the square root of the integrated charge and therefore the greater the charge, the higher the performance. (cid:26)(cid:18) (cid:27)(cid:11)(cid:5)(cid:6)(cid:7)(cid:8)(cid:2)(cid:7)(cid:6)(cid:15)(cid:10)(cid:8)(cid:6)(cid:12)(cid:12)(cid:6)(cid:16)(cid:8)(cid:6)(cid:12)(cid:5)(cid:28)(cid:4)(cid:3)(cid:10)(cid:5)(cid:3)(cid:20)(cid:12)(cid:10)(cid:13) Theterm“focalplanearray”(FPA)referstoanassemblage ofindividualdetectorpictureelements(“pixels”)locatedat the focal plane of an imaging system. Although the defini- tion could include one-dimensional (“linear”) arrays as well as two-dimensional (2D) arrays, it is frequently ap- plied to the latter. Usually, the optics part of an optoelec- Fig. 1. Detectivity vs wavelength values of 0.1–1.2 µm photo- tronic images device is limited only to focusing of the im- detectors.PRindicatesaphotoresistorsandPMindicatesaphoto- age onto the detectors array. These so-called “staring ar- multiplies.DetectorareasaregiveninTable2(afterRef.3). rays”arescannedelectronicallyusuallyusingcircuitsinte- Fig.2.ComparisonoftheD*ofvariouscommerciallyavailableinfrareddetectorswhenoperatedattheindicatedtemperature.Chopping frequencyis1000Hzforalldetectorsexceptthethermopile(10Hz),thermocouple(10Hz),thermistorbolometer(10Hz),Golaycell(10 Hz)andpyroelectricdetector(10Hz).Eachdetectorisassumedtoviewahemisphericalsurroundatatemperatureof300K.Theoretical curvesforthebackground-limitedD*foridealphotovoltaic(PV)detectors,photoresistors(PR)andthermaldetectorsarealsoshown. 224 Opto-Electron.Rev.,12,no.2,2004 ©2004COSiWSEP,Warsaw Contributed paper Fig.3.Increaseinarrayformatsizeoverthepastthirtyyears(afterRef.4). (cid:26)(cid:18)(cid:17)(cid:18) (cid:29)(cid:11)(cid:15)(cid:11)(cid:7)(cid:4)(cid:3)(cid:28)(cid:4)(cid:5)(cid:8)(cid:6)(cid:12)(cid:12)(cid:6)(cid:16)(cid:13) gratedwiththearrays.Thearchitectureofdetector-readout assemblies has assumed a number of forms which are dis- cussed below. The types of readout integrated circuits In general, the architectures of FPAs may be classified as (ROICs) include the function of pixel deselecting, anti- monolithicandhybrid.Whenthedetectormaterialiseither blooming on each pixel, subframe imaging, output pream- siliconorasiliconderivative(suchase.g.platinumsilicide plifiers, and may include yet other functions. Infrared im- PtSi),thedetectorandROICcanbebuiltonasinglewafer aging systems, which use 2D arrays, belong to so-called (seeFig.4).EffortstodevelopmonolithicFPAsusingnar- “secondgeneration”systems. row-gapsemiconductorshavefailed.Thereareafewobvi- Development of detector FPA technology has revolu- ous advantages to this structure, principally in the simplic- tionized many kinds of imaging in the past twenty five ity and lower cost associated with a directly integrated structure. Common examples of these FPAs in the visible years[4].From(cid:11)raystotheinfraredandevenradiowaves, andnearinfrared(0.7–1.0µm)arefoundincamcordersand the rate at which images can be acquired has increased by more than a factor of a million in many cases. Figure 3 il- digital cameras. Two generic types of silicon technology provide the bulk of devices in these markets: charge cou- lustrates the trend in array size over the past thirty years. pled devices (CCDs) and complementary metal-oxide- Imaging FPAs have developed in proportion to the ability -semiconductor (CMOS) imagers. CCD technology has of silicon integrated circuit (ICs) technology to read and processthearraysignals,andwithabilitytodisplaythere- achieved the highest pixel counts or largest formats with the numbers approaching 108 (see Fig. 5). CMOS imagers sulting image. FPAs have nominally the same growth rate are also rapidly moving to large formats and are expected as dynamic random access memory (DRAM) ICs (which to compete with CCDs for the large format applications have had a doubling-rate period of approximately 18 within a few years. Because the CCD imager market is months;itisaconsequenceofMoore’slaw,whichpredicts much smaller than that for CMOS devices in general, it theabilitytodoubletransistorintegrationoneachICabout maybedifficultforCCDtoremaincompetitiveinthelong every 18 months) but lag behind in size by about 5–10 term. years. ROICs are somewhat analogous to DRAM-only CCD technology is very mature in respect to both the readouts, but require a minimum of three transistors per fabricationyieldandtheattainmentofnear-theoreticalsen- pixel,comparedtooneforeachmemorycell.Readoutsare sitivity. Figure 6 shows the schematic circuit for a typical alsoanalogousintermsofanemphasisonlownoiseinputs CCD imager. The monolithic array is based on a and generally maximum charge storage capacity. Charge metal-insulator-semiconductor(MIS)structure.Incidentra- coupled devices (CCDs) with close to 100 M pixels offer the largest formats. PtSi, InSb and HgCdTe have been fol- diationgenerateselectron-holepairsinthedepletionregion lowing the pace of DRAM. In the infrared, 4 M pixel ar- of the MIS structure. The photogenerated carriers are first integrated in an electronic well at the pixel and subse- raysarenowavailableforastronomyapplications. Opto-Electron.Rev.,12,no.2,2004 A.Rogalski 225 Opticaldetectorsforfocalplanearrays Fig.4.MonolithicIRFPAs:(a)all-silicon;(b)heteroepitaxy-on-silicon;(c)non-silicon(e.g.,HgCdTeCCD);and(d)microbolometer. quently transferred to slow and fast CCD shift registers. 0.18 µm are in production, with pre-production runs of This is done step by step as the gate biases are clocked to 0.13µmdesignrulesalreadyunderway.Asaresultofsuch movethechargewithminimalloss.Thefigureofmeritfor fine design rules, more functionality has been put into the the effectiveness of this process is called the charge trans- unit cells of multiplexers and smaller unit cells, leading to ferefficiency(CTE).Channelstopsbetweencolumnshelp large array sizes. Figure 5 shows the timelines for mini- topreventchargesfromstrayinglaterally.Attheendofthe mum circuit features and the resulting CCD, IR FPA and CCD register, a charge carrying information on the re- CMOSvisibleimagersizeswithrespecttoimagingpixels. ceived signal can be readout and converted into a useful Along the horizontal axis is also a scale depicting the gen- signal. eralavailabilityofvariousMOSandCMOSprocesses.The The configuration of CCD devices requires specialized ongoingmigrationtoevenfinerlithographieswillthusen- processing, unlike CMOS imagers which can be built on able the rapid development of CMOS-based imagers hav- fabricationlinesdesignedforcommercialmicroprocessors. ingevenhigherresolution,betterimagequality,higherlev- CMOShavetheadvantagethatexistingfoundries,intended els of integration and lower overall imaging system cost for application specific integrated circuits (ASICs), can be than CCD-based solutions. At present, CMOS having with readilyusedbyadaptingtheirdesignrules.Designrulesof minimum features of (cid:12) 0.5 µm is also enabling monolithic 226 Opto-Electron.Rev.,12,no.2,2004 ©2004COSiWSEP,Warsaw Contributed paper Fig.5.Imagingarrayformatscomparedwiththecomplexityofmicroprocessortechnologyasindicatedbytransistorcount.Thetimeline designruleofMOS/CMOSfeaturesisshownatthebottom(afterRef.4). visible CMOS imagers, because the denser photolithogra- A typical CMOS multiplexer architecture (see Fig. 7) phy allows low-noise signal extraction and high perfor- consists of fast (column) and slow (row) shift registers at mance detection with the optical fill factor within each the edges of the active area, and pixels are addressed one pixel. The silicon wafer production infrastructure which by one through the selection of a slow register, while the has put personal computers into many homes is now en- fast register scans through a column, and so on. Each ablingCMOS-basedimaginginconsumerproductssuchas photodiode is connected in parallel to a storage capacitor digitalstillandvideocameras. locatedintheunitcell.Acolumnofdiodesandstorageca- Fig.6.ArchitectureoftypicalCCDimager(afterRef.5). Opto-Electron.Rev.,12,no.2,2004 A.Rogalski 227 Opticaldetectorsforfocalplanearrays Fig.7.CMOSmultiplexingreadoutwithCTIAdetectorinterface(afterRef.6). pacitors is selected one at a time by a digital horizontal scan register and a row bus is selected by the vertical scan register. Therefore each pixel can be individually ad- dressed. CMOS-basedimagersuseactiveorpassivepixels[6–8] asshown,insimplifiedform,inFig.8.Incomparisonwith passive pixel sensors (PPSs), active pixel sensors (APSs) apart from read functions exploit some form of amplifica- tionateachpixel.PPSshavesimplepixelsconsistingofas few as two components (a photodiode and a MOSFET switch).Asaresult,circuitoverheadislowandtheoptical collection efficiency [fill factor (FF)] is high even for monolithicdevices.AlargeopticalFFofupto80%maxi- mises signal selection and minimises fabrication cost by obviating the need for microlenses. Microlenses, typically used in CCD and CMOS APS imagers for visible applica- tion,concentratetheincominglightintothephotosensitive region when they are accurately deposited over each pixel (see Fig. 9). When the FF is low and microlenses are not used, the light falling elsewhere is either lost or, in some cases,createsartifactsintheimagerybygeneratingelectri- calcurrentsintheactivecircuitry. APSsincorporatetransistorsineachpixeltoconvertthe photogeneratedchargetoavoltage,amplifythesignalvolt- age,andreducethenoise.Addingthesecomponents,how- ever, reduces the FF of monolithic imagers to about 30–50% in 0.5-µm processes at a 5–6-µm pixel pitch or in 0.25-µmprocessesata3.3–4.0-µmpixelpitch[6]. Fig.8.Passive(a)andactive(b)pixelsensor(afterRef.6). 228 Opto-Electron.Rev.,12,no.2,2004 ©2004COSiWSEP,Warsaw Contributed paper (cid:26)(cid:18)(cid:21)(cid:18) (cid:30)(cid:16)(cid:31)(cid:12)(cid:4)(cid:9)(cid:8)(cid:6)(cid:12)(cid:12)(cid:6)(cid:16)(cid:13) Ultraviolet and infrared imagers are most commonly built with a hybrid structure. Visible hybrids have also been built for specific applications. Hybrid FPAs detectors and multiplexers are fabricated on different substrates and matedwitheachotherbyflip-chipbondingorloopholein- terconnection (see Fig. 10). In this case, we can optimise the detector material and multiplexer independently. Other advantages of the hybrid FPAs are near 100% fill factor and increased signal-processing area on the multiplexer chip. Indium bump bonding of readout electronics, first demonstrated in the mid-1970s, provides for multiplexing the signals from thousands of pixels onto a few output lines, greatly simplifying the interface between the sensor andthesystemelectronics. The detector array can be illuminated from either the frontside(withthephotonspassingthroughthetransparent silicon multiplexer) or backside (with photons passing through the transparent detector array substrate). In gen- eral,thelatterapproachismostadvantageousasthemulti- plexerwilltypicallyhaveareasofmetallizationsandother opaqueregions,whichcanreducetheeffectiveopticalarea of the structure. When using opaque materials, substrates must be thinned to 10–20 µm in order to obtain sufficient quantumefficienciesandreducethecrosstalk. Hybrids readouts are usually built with silicon, al- thoughafewdemonstrationsofreadoutsusingothermate- rialshavebeenexperimentallystudied.Shiftregistersclock Fig. 9. Micrograph and cross-sectional drawing of microlensed the signals from each row in turn, while all columns are hybridFPAs(afterRef.6). typically read in parallel. The readout may have as few as Fig.10.HybridFPAwithindependentlyoptimizedsignaldetectionandreadout:(a)indiumbumptechniques,(b)loopholetechnique. Opto-Electron.Rev.,12,no.2,2004 A.Rogalski 229 Opticaldetectorsforfocalplanearrays Fig.11.QuantumefficiencyofUV,visible,andinfrareddetectorarrays(afterRef.4). one, and as many as 64 outputs, depending on the format HgCdTe detectors – is the high-density vertically-integra- size and frame rate. A typical output can provide tedphotodiode,orloopholephotodiode[13]. 5–20MHzdatarates[4]. Key to the development of ROICs has been the evolu- Awidevarietyofdetectormaterialshavebeenadapted tion in input preamplifier technology. This evolution has to the monolithic and hybrid format [9–12]. Figure 11 been driven by increased performance requirements and shows the quantum efficiency of some of the detector ma- silicon processing technology improvements. A brief dis- terials used to fabricate arrays of ultraviolet (UV), visible cussionofthevariouscircuitsisgiveninRef.14. and infrared detectors. AlGaN detectors are being devel- opedintheUVregion.Siliconp-i-ndiodesareshownwith (cid:18) (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:7)(cid:8)(cid:14)(cid:11)(cid:5)(cid:6)(cid:7)(cid:8)(cid:2)(cid:7)(cid:6)(cid:15)(cid:10)(cid:8)(cid:6)(cid:12)(cid:12)(cid:6)(cid:16)(cid:13) and without antireflection coating. Lead salts (PbS and PbSe) have intermediate quantum efficiencies, while PtSi Detectorarraysareavailableinwidespectralrangeofelec- Schottky barrier types and quantum well infrared photo- tromagnetic spectrum. A variety of detector array formats detectors(QWIPs)havelowvalues.InSbcanrespondfrom areelaboratedinthevisibleandinfraredregions.Feverop- thenearUVoutto5.5µmat80K.Asuitabledetectorma- tions are available in the shorter or longer wavelength re- terial for near-IR (1.0–1.7-µm) spectral range is InGaAs gions. Below we will survey a sample of the types that latticematchedtotheInP.VariousHgCdTealloys,inboth have been built to address the various portions of the opti- photovoltaic and photoconductive configurations, cover calspectrum. from 0.7 µm to over 20 µm. Impurity-doped (Sb, As, and Ga) silicon impurity-blocked conduction (IBC) detectors (cid:18)(cid:17)(cid:18) !"(cid:12)(cid:6)(cid:16)(cid:8)(cid:9)(cid:10)(cid:3)(cid:10)(cid:5)(cid:3)(cid:11)(cid:12)(cid:8)(cid:6)(cid:12)(cid:12)(cid:6)(cid:16)(cid:13) operating at 10 K have a spectral response cut-off in the rangeof16to30µm.Impurity-dopedGedetectorscanex- tendtheresponseoutto100–200µm. InthecourseofthepasthundredyearstheX-raydetection has migrated from film to digital cameras for dental and UV, visible, and infrared arrays most commonly em- ploy a photodiode structure. Photodiodes are preferred to medical applications. Several classes of X-ray sensor ar- photoconductors because of their relatively high imped- rayshavebeendevelopedincluding[4]: ance,whichmatchesdirectlyintothehighinputimpedance (cid:127) phosphors, stage of an FET readout circuit and also allows lower (cid:127) scintilators, power dissipation. Mesa photodiodes are used in AlGaN, (cid:127) microchannelplates, InSb, and HgCdTe detectors, whereas planar photodiodes (cid:127) silicon detector arrays (CCDs, hybrid p-i-n structures, areusedinSi,PtSi,Ge,HgCdTe,InGaAs,andInSbdetec- thinfilmsiliconpanels),and tors. A third photodiode structure – used exclusively with (cid:127) CdZnTehybriddetectorarrays. 230 Opto-Electron.Rev.,12,no.2,2004 ©2004COSiWSEP,Warsaw