Polycrystalline Lead Selenide: the Resurgence of an Old Infrared Detector
Total Page：16
File Type：pdf, Size：1020Kb
OPTO-ELECTRONICS REVIEW 15(2), 110–117 DOI: 10.2478/s11772-007-0007-7 Polycrystalline lead selenide: the resurgence of an old infrared detector G. VERGARA*, M.T. MONTOJO, M.C. TORQUEMADA, M.T. RODRIGO, F.J. SÁNCHEZ, L.J. GÓMEZ, R.M. ALMAZÁN, M. VERDÚ, P. RODRÍGUEZ, V. VILLAMAYOR, M. ÁLVAREZ, J. DIEZHANDINO, J. PLAZA, and I. CATALÁN Centro de Investigación y Desarrollo de la Armada (CIDA), Arturo Soria 289, 28033 Madrid, Spain The existing technology for uncooled MWIR photon detectors based on polycrystalline lead salts is stigmatized for being a 50-year-old technology. It has been traditionally relegated to single-element detectors and relatively small linear arrays due to the limitations imposed by its standard manufacture process based on a chemical bath deposition technique (CBD) devel- oped more than 40 years ago. Recently, an innovative method for processing detectors, based on a vapour phase deposition (VPD) technique, has allowed manufacturing the first 2D array of polycrystalline PbSe with good electro optical character- istics. The new method of processing PbSe is an all silicon technology and it is compatible with standard CMOS circuitry. In addition to its affordability, VPD PbSe constitutes a perfect candidate to fill the existing gap in the photonic and uncooled IR imaging detectors sensitive to the MWIR photons. The perspectives opened are numerous and very important, converting the old PbSe detector in a serious alternative to others uncooled technologies in the low cost IR detection market. The num- ber of potential applications is huge, some of them with high commercial impact such as personal IR imagers, enhanced vi- sion systems for automotive applications and other not less important in the security/defence domain such as sensors for ac- tive protection systems (APS) or low cost seekers. Despite the fact, unanimously accepted, that uncooled will dominate the majority of the future IR detection applications, today, thermal detectors are the unique plausible alternative. There is plenty of room for photonic uncooled and complemen- tary alternatives are needed. This work allocates polycrystalline PbSe in the current panorama of the uncooled IR detectors, underlining its potentiality in two areas of interest, i.e., very low cost imaging IR detectors and MWIR fast uncooled detec- tors for security and defence applications. The new method of processing again converts PbSe into an emerging technology. Keywords: uncooled, low cost, infrared sensors, lead selenide. 1. Introduction mal or lately, quantum devices (QWIPs)) and technological limitations associated with the CBD method deviated the Polycrystalline PbSe was one of the first IR detectors used efforts of scientists and technicians towards other direc- successfully during the WW II. After that, important efforts tions, relegating lead salts to a few elementary, and very were made along the 50’s and 60’s trying to develop better specific applications. Along the 70’s, the research activity and more complex devices and to understand their physics on lead salts decreased considerably and only a very lim- [1–10]. At this time, exhaustive works were done for opti- ited number of applications have kept alive the old PbSe mizing preparation methods of sensitive PbSe thin layers. technology during the last 30 years, mainly at the US. To- The result is the current standard PbSe technology based on day, its physics is still bad understood and the standard pro- a CBD method [11,12]. cesses used for manufacturing detectors have evolved so PbSe polycrystalline films exhibited a number of draw- little that, at present, the biggest format commercially backs which limited their applications in the IR imaging available is a linear array of 256 elements. systems: Recently, an innovative technology for processing • significant 1/f noise, polycrystalline PbSe has been developed at CIDA. The • poor long term stability, most remarkable advantages associated with it [13–15]: • bad photoresponse uniformity, • good reproducibility, • high dielectric constant. • good uniformities, Poor understanding of the physics involved in the photoconduction process, the appearance of new and more • long term stability, promising materials sensitive to IR (InSb, SiPt, CMT, ther- • fully compatible with plain (no textured) Si substrates, • viability studios show that it is compatible with existing Si CMOS technology, *e-mail: [email protected] • simple and affordable technology, 110 Opto-Electron. Rev. 15, no. 2, 2007 • compatible with complex multilayer structures such as reaching an adequate performance/affordability ratio. The interference filters (spectral discrimination feature advances along the last five years have been impressive, monolithically integrated). but still will be necessary to overcome important obstacles. Big efforts are being made at CIDA for processing the The road towards affordability started more than twenty first imaging device of PbSe monolithically integrated with years ago. Along eighties and nineties, the important its read out electronics circuitry. The imaging detector un- hi-tech companies around the world began a rush for devel- der processing will incorporate the last advances in elec- oping and commercializing less expensive and better unco- tronics, including active pixel sensor (APS) concept with oled IR sensors. All of them focused their efforts on ther- analogue/digital signal mixing. Without any doubt, such at- mal imaging sensors. As a consequence, today, uncooled is tainment will turn PbSe into a major actor among the synonymous of thermal. At present, numerous companies uncooled IR detectors. Meanwhile, this new technology still spend big amount of resources on developing enable has made possible to process devices such as the first 2D technologies to increase sensor performances and to reduce FPAs of polycrystalline PbSe on plain (no textured) silicon manufacture costs. [16,17]. Thermal FPAs are monolithic or hybrid devices and This technology overcomes some of the most important their evolution have been tightly linked to the microelec- inconveniences presented by the standard CBD method. On tronics golden rule, Moore’s law [22,23]. the other hand, the effects associated to the material nature During this time, the performance increment and the (high 1/f noise frequency knee or high dielectric constant) cost reduction strategies have been based on the following can, today, be minimized using specific circuitry and the points: high processing capabilities supplied by modern electron- • Processing on larger areas. At present, the most ad- a ics. All these facts open new and important perspectives, vanced uncooled technologies (VOx and -Si micro- ³ reviving PbSe sensors and converting it again, more than bolometers) use large area Si wafers ( 8”) as standard 50 years later, into a very promising IR sensitive material. substrates. It is an important driven factor for reducing manufacture costs. Section 2 reviews the panorama of uncooled and low Reduction of pixel and array sizes. In addition to per- cost imaging IR detectors and their issues in terms of cost • formance improvements there are further benefits reduction and increased performances. Section 3 is dedi- linked to the pitch size shrink, smaller detector sizes, cated to give a brief description of the CIDA’s PbSe tech- meaning more detectors per wafer, higher yields and nology (sec. 3.1), to present its potential advantages in smaller and more affordable optics. terms of low cost and affordability (sec. 3.2) and finally, to • Increase process yields. Improvements in wafer han- propose PbSe as a candidate for filling the existing gap in dling, process reproducibility, etc. the field of uncooled photonic MWIR detectors. • Testing reduction. The goal was to reduce testing for 2. Uncooled thermal detectors. A journey minimizing touch labour improving reliability of each process and diminishing scrap. towards affordability After the efforts done along all these years, the results are excellent in terms of both performance and cost reduc- Along the last 30 years, the technology of IR detectors has tion. So, today the largest format commercially available is grown very fast, achieving an outstanding degree of devel- a microbolometer of 640×512 elements, but all important opment, inconceivable few years ago. Without any doubt, actors commercialize systems or devices with a standard the most outstanding event has been the revolution led by format of 320×240 elements. For volume applications, all thermal uncooled IR detectors. Thanks to the last techno- manufacturers have unanimously chosen a standard format logical advances in this field, the dream of creating devices of 160×120 elements, in some cases with a TEC-less fea- able to work at room temperature, close to the BLIP condi- ture, integrated in the read out integrated circuit. Regarding tion, is now a reality [18,19]. cost, during the last years it was notably reduced but it is The global IR detector industry has been shaken by still out of the affordability threshold for volume consump- thermal uncooled devices. This market is growing annually tion and the market continues dominated by defence and by ~25% [20] and huge efforts are being made for improv- security applications. ing sensor performances and reducing manufacture costs. The manufacture cost reduction must go on, but the ac- The objective is to bring low/medium performance IR tivity has reached a point where the cost reduction rate is de- imagers to a volume market in a short period of time. The creasing. Up to now, the main driven forces (wafer size in- applications are multiple: personal imagers, medical ther- crement, pixel size reduction, testing reduction and yield in- mography, automotive enhanced vision systems, atmo- crease) are close to their limits. Shrinkage of pixel size also spheric pollutant control, homeland security, smart envi- reduces the number of photons collected by each pixel, it in- ronmental comfort systems for domestic use, etc. The pre- creases the statistical noise level and reduces the available dictions foresee an enormous widespreading of very low signal dynamic range. At present, in the microbolometer cost imaging IR sensors in a near future [21].