Academia.eduAcademia.edu

Figure 13 – uploaded by Pierre Castelein

See full PDFdownloadDownload figure
Fig. 15. Histogram plot of NETD measured on one of the demonstrators in the 3-5 um band (MWIR2). The corresponding pixel architecture is presented in Fig. 19. Each pixel consists of two standard n on p photodiodes, located in two Cd,Hg;_,Te layers with different compositions. One is at the bottom of a hole etched in the longest cut-off wavelength absorption layer, while the other is on the surface layer. A common contact on the array side allows the biasing of both lower and upper p-type layers (see the left-hand side of Fig. 19). The barrier The pseudo-planar structure is schematized in Fig. 18. This kind of structure has been grown with

Figure 15 Histogram plot of NETD measured on one of the demonstrators in the 3-5 um band (MWIR2). The corresponding pixel architecture is presented in Fig. 19. Each pixel consists of two standard n on p photodiodes, located in two Cd,Hg;_,Te layers with different compositions. One is at the bottom of a hole etched in the longest cut-off wavelength absorption layer, while the other is on the surface layer. A common contact on the array side allows the biasing of both lower and upper p-type layers (see the left-hand side of Fig. 19). The barrier The pseudo-planar structure is schematized in Fig. 18. This kind of structure has been grown with

Related Figures (28)

Fig. 1. SIMS spectrum obtained on an NPN heterostructure, with two CdHgTe layers absorbing below 3 um and below 5 um. i The longest cut-off wavelength, together with the total thickness of the stack, is extracted from Fourier transformed infrared transmission (FTIR) measurements (Fig. 2). Crystalline quality is probed by high resolution X-ray diffraction
Fig. 1. SIMS spectrum obtained on an NPN heterostructure, with two CdHgTe layers absorbing below 3 um and below 5 um. i The longest cut-off wavelength, together with the total thickness of the stack, is extracted from Fourier transformed infrared transmission (FTIR) measurements (Fig. 2). Crystalline quality is probed by high resolution X-ray diffraction
Fig. 2. FTIR spectrum and fit to the experimental data for a NPN heterostructure (MWIR/MWIR).
Fig. 2. FTIR spectrum and fit to the experimental data for a NPN heterostructure (MWIR/MWIR).
Fig. 3. Mapping of the FWHM of a dual-band epitaxial structure on CZT substrate, with histogram of FWHM values.
Fig. 3. Mapping of the FWHM of a dual-band epitaxial structure on CZT substrate, with histogram of FWHM values.
Fig. 5. SEM pictures of the bottom of structures etched in CdHgTe for dual-band detection manufacturing (/eft detail of a trench 9 um deep and 3 um wide; right detail of a hole 10 «wm in diameter and 5 um deep). We have also managed to control the photoli- thography steps that are necessary to make well defined patterns on nonplanar surfaces. It is known that lithography steps are critical for the successful definition and process of these kinds of patterns (see for example Refs. 13 and 14). An example of post- process SEM analysis is shown in Fig. 6, for a pseudo-planar dual-band pixel. This picture shows the diode contact taken at the bottom of the hole and
Fig. 5. SEM pictures of the bottom of structures etched in CdHgTe for dual-band detection manufacturing (/eft detail of a trench 9 um deep and 3 um wide; right detail of a hole 10 «wm in diameter and 5 um deep). We have also managed to control the photoli- thography steps that are necessary to make well defined patterns on nonplanar surfaces. It is known that lithography steps are critical for the successful definition and process of these kinds of patterns (see for example Refs. 13 and 14). An example of post- process SEM analysis is shown in Fig. 6, for a pseudo-planar dual-band pixel. This picture shows the diode contact taken at the bottom of the hole and
Fig. 4. Scanning electron microscope (SEM) pictures of a deep trench obtained in CdHgTe by dry etching (/eft), and of a bispectral detector array with a 25 um pixel pitch. Trenches are obtained by pure dry etching in an ICP plasma reactor (right).
Fig. 4. Scanning electron microscope (SEM) pictures of a deep trench obtained in CdHgTe by dry etching (/eft), and of a bispectral detector array with a 25 um pixel pitch. Trenches are obtained by pure dry etching in an ICP plasma reactor (right).
Fig. 7. SEM picture of 15 um pitch indium bumps. For pseudo-planar architecture, where two in- dium bumps per pixel are necessary, a 25 um pixel means that the indium bump pitch is near 17 ym. Even if the 15 um pitch hybridization is well con- trolled, as the pixel topology is more complex than for a single color, planar device, this hybridization step can be considered to be as challenging as the 15 wm planar pixel case. For our 256 x 256 pixel FPA demonstrators, we had no problem with the hybridization steps, and the hybridization yield was as good as the one for a similar size planar FPA. We are working on further improvements of our
Fig. 7. SEM picture of 15 um pitch indium bumps. For pseudo-planar architecture, where two in- dium bumps per pixel are necessary, a 25 um pixel means that the indium bump pitch is near 17 ym. Even if the 15 um pitch hybridization is well con- trolled, as the pixel topology is more complex than for a single color, planar device, this hybridization step can be considered to be as challenging as the 15 wm planar pixel case. For our 256 x 256 pixel FPA demonstrators, we had no problem with the hybridization steps, and the hybridization yield was as good as the one for a similar size planar FPA. We are working on further improvements of our
Fig. 6. SEM picture, taken after focused ion beam (FIB) preparation, of a diode contact at the bottom of a hole, in the pseudo-planar architecture (upper metal layer is added for FIB sample preparation).
Fig. 6. SEM picture, taken after focused ion beam (FIB) preparation, of a diode contact at the bottom of a hole, in the pseudo-planar architecture (upper metal layer is added for FIB sample preparation).
Fig. 8. NPN CdHgTe epitaxial structure. MWIR 1 refers to the short wave layer with cut-off wavelength around 3.0 um. MWIR 2 refers to the long wave layer with cut-off wavelength around 5.0 um. N+ doping is due to indium incorporation. The 25 wm pitch NPN pixels have breakdown voltage and maximum impedance close or equal to those usually measured on conventional planar
Fig. 8. NPN CdHgTe epitaxial structure. MWIR 1 refers to the short wave layer with cut-off wavelength around 3.0 um. MWIR 2 refers to the long wave layer with cut-off wavelength around 5.0 um. N+ doping is due to indium incorporation. The 25 wm pitch NPN pixels have breakdown voltage and maximum impedance close or equal to those usually measured on conventional planar
Fig. 11. Normalized 77 K spectral response. Cut-off wavelength are 3.1 um and 5.0 «~m (MWIR1 and MWIR2). The cut-on at 2 4m is due to the Ge dewar window. CEA-LETI has designed a 256 x 256 pixel, two- color, readout integrated circuit (ROIC) with a 25 um pitch suited to the stacked dual-band detec- tors. It makes use of a high performance direct injection (DI) input circuit. A specific storage capacitor is provided for each band. This mode of operation presents the major advantages to allow easy separation of the two photocurrents and to enable independent optimizations of the two charge wells and, thus, of electro-optical performances. The integration is snapshot for each band, but, because the detector has only one bump per pixel, simulta- neous integration of the two bands is impossible. Nevertheless, the delay between the integrations is minimized, as shown on the chronogram in Fig. 12. The charge handling capacity of each band has been optimized for a focal plane temperature of 77 K and a dynamic range of 290-350 K in the 3-5 wm range and above 450 K in the 1-3 ym range. The readout
Fig. 11. Normalized 77 K spectral response. Cut-off wavelength are 3.1 um and 5.0 «~m (MWIR1 and MWIR2). The cut-on at 2 4m is due to the Ge dewar window. CEA-LETI has designed a 256 x 256 pixel, two- color, readout integrated circuit (ROIC) with a 25 um pitch suited to the stacked dual-band detec- tors. It makes use of a high performance direct injection (DI) input circuit. A specific storage capacitor is provided for each band. This mode of operation presents the major advantages to allow easy separation of the two photocurrents and to enable independent optimizations of the two charge wells and, thus, of electro-optical performances. The integration is snapshot for each band, but, because the detector has only one bump per pixel, simulta- neous integration of the two bands is impossible. Nevertheless, the delay between the integrations is minimized, as shown on the chronogram in Fig. 12. The charge handling capacity of each band has been optimized for a focal plane temperature of 77 K and a dynamic range of 290-350 K in the 3-5 wm range and above 450 K in the 1-3 ym range. The readout
Fig. 10. I(V) (top) and R(V) (bottom) characteristics. Scales are volts, amps, and ohms. Negative bias: MWIR2 diode, positive bias: MWIR1 diode. Fig. 9. Schematic cross-section of a single bump npn dual-band pixel.
Fig. 10. I(V) (top) and R(V) (bottom) characteristics. Scales are volts, amps, and ohms. Negative bias: MWIR2 diode, positive bias: MWIR1 diode. Fig. 9. Schematic cross-section of a single bump npn dual-band pixel.
Table I. Main Characteristics of the Dual-Color Readout Circuit
Table I. Main Characteristics of the Dual-Color Readout Circuit
Table II. Operating Conditions of the MWIR1-MWIR2 Focal Planes that integration times are short in both bands (around or less than 0.5 ms), with F/2 apertures.
Table II. Operating Conditions of the MWIR1-MWIR2 Focal Planes that integration times are short in both bands (around or less than 0.5 ms), with F/2 apertures.
Fig. 13. Simplified architecture of the NPN demonstrator (/eft), and view of one of the dual-color demonstrators mounted on its laboratory connecting networks. More complete and _ statistical electro-optical characterization is available through the testing of 256 by 256 pixel hybridized FPAs. The general operating conditions that have been used for the characterizations presented below are summed up in Table II. The integration time was adjusted for each component to reach half of the well capacity for a scene temperature of 50°C in the 3-5 wm range and at 220°C in the 2-3 um band. It should be noted The following figures present some representative results obtained with this kind of FPA. More detailed results have been presented in Refs. 17 and 18. We present here only the most striking data, concerning noise equivalent temperature difference (NETD) measurements. Figures 14 and 15 present the NETD distribution measured in each band, with a mean value around 20 mK in the upper band and 35 mK in the lower band. The dead pixel operability is very high, with only 0.1% of defects for the MWIR1 band and 0.3% for the MWIR2 band. The population of pixels with two-times the mean NETD is 99.8% for the MWIR1 band and 98.7% for the MWIR2 band, and the noise is almost limited by the background theoretical noise. It is important to underline that these operabilities are very high, with a very harsh requirement on the NETD. If we take into account a commonly used criterion of five-times the mean NETD, operability is even better, going up to 99.9% for MWIR1 band and 99.7% for MWIR2 band.
Fig. 13. Simplified architecture of the NPN demonstrator (/eft), and view of one of the dual-color demonstrators mounted on its laboratory connecting networks. More complete and _ statistical electro-optical characterization is available through the testing of 256 by 256 pixel hybridized FPAs. The general operating conditions that have been used for the characterizations presented below are summed up in Table II. The integration time was adjusted for each component to reach half of the well capacity for a scene temperature of 50°C in the 3-5 wm range and at 220°C in the 2-3 um band. It should be noted The following figures present some representative results obtained with this kind of FPA. More detailed results have been presented in Refs. 17 and 18. We present here only the most striking data, concerning noise equivalent temperature difference (NETD) measurements. Figures 14 and 15 present the NETD distribution measured in each band, with a mean value around 20 mK in the upper band and 35 mK in the lower band. The dead pixel operability is very high, with only 0.1% of defects for the MWIR1 band and 0.3% for the MWIR2 band. The population of pixels with two-times the mean NETD is 99.8% for the MWIR1 band and 98.7% for the MWIR2 band, and the noise is almost limited by the background theoretical noise. It is important to underline that these operabilities are very high, with a very harsh requirement on the NETD. If we take into account a commonly used criterion of five-times the mean NETD, operability is even better, going up to 99.9% for MWIR1 band and 99.7% for MWIR2 band.
Fig. 16. Spatial maps of NETD measured on one of the demonstrators in the 1-3 um band (/eft) and in the 3-5 um band (right).
Fig. 16. Spatial maps of NETD measured on one of the demonstrators in the 1-3 um band (/eft) and in the 3-5 um band (right).
Fig. 18. Multilayer CdHgTe structure for pseudo-planar dual-band FPAs. The MWIR and LWIR layers are separated by an SWIR barrier. between upper and lower layers prevents electrical shortage between the diodes. On each pixel, two independent contacts for the two diodes allow one to address the two detectors simultaneously (see the right-hand side of Fig. 19). This architecture thus provides both temporal and spatial coherence, which can be needed for some applications. Any composition combination may be realized, depend- ing on the application: MWIR/MWIR, MWIR/LWIR, or LWIR/LWIR. We have primarily focused our
Fig. 18. Multilayer CdHgTe structure for pseudo-planar dual-band FPAs. The MWIR and LWIR layers are separated by an SWIR barrier. between upper and lower layers prevents electrical shortage between the diodes. On each pixel, two independent contacts for the two diodes allow one to address the two detectors simultaneously (see the right-hand side of Fig. 19). This architecture thus provides both temporal and spatial coherence, which can be needed for some applications. Any composition combination may be realized, depend- ing on the application: MWIR/MWIR, MWIR/LWIR, or LWIR/LWIR. We have primarily focused our
Fig. 17. MWIR1/MWIR2 images taken with the dual-color NPN demonstrator.
Fig. 17. MWIR1/MWIR2 images taken with the dual-color NPN demonstrator.
Fig. 20. SEM pictures of pseudo-planar bispectral detector arrays, with a 30 «um pixel pitch and two indium bumps per pixel. Fig. 19. Schematic cross-section of a pseudo-planar dual-band pixel.
Fig. 20. SEM pictures of pseudo-planar bispectral detector arrays, with a 30 «um pixel pitch and two indium bumps per pixel. Fig. 19. Schematic cross-section of a pseudo-planar dual-band pixel.
Fig. 21. I(V) (top) and R(V) (bottom) characteristics. The scales are volts, amps and ohms. Left figure is for the MWIR diode, while the right one is for the LWIR diode. Quantum efficiency measured in both bands is above 50% (with antireflection coating). This result corresponds to an optical fill factor of 100% in the lower band and 70% in the upper band. At the moment, quantum efficiency is limited by the absorption layer thickness for the LWIR diode. For the MWIR diode, as explained before, it is limited by controlled by the diffusion length dispersion itself. As the contact and implantation size are limited by the small hole bottom diameter (about 6-8 ym), control of the minority carrier diffusion length is crucial to maximize the optical filling factor. We are confident that this dispersion will be reduced in our future devices, by our adjusting doping levels and minority carrier diffusion length. On the right-hand side, one can see that the LWIR diodes present very
Fig. 21. I(V) (top) and R(V) (bottom) characteristics. The scales are volts, amps and ohms. Left figure is for the MWIR diode, while the right one is for the LWIR diode. Quantum efficiency measured in both bands is above 50% (with antireflection coating). This result corresponds to an optical fill factor of 100% in the lower band and 70% in the upper band. At the moment, quantum efficiency is limited by the absorption layer thickness for the LWIR diode. For the MWIR diode, as explained before, it is limited by controlled by the diffusion length dispersion itself. As the contact and implantation size are limited by the small hole bottom diameter (about 6-8 ym), control of the minority carrier diffusion length is crucial to maximize the optical filling factor. We are confident that this dispersion will be reduced in our future devices, by our adjusting doping levels and minority carrier diffusion length. On the right-hand side, one can see that the LWIR diodes present very
Table ITI. Main Characteristics of Dual-Band Readout Circuit
Table ITI. Main Characteristics of Dual-Band Readout Circuit
Fig. 22. Normalized 77K spectral response. The cut-off wave- lengths are 4.8 um and 9.7 wm for the MWIR and LWIR diodes, respectively.
Fig. 22. Normalized 77K spectral response. The cut-off wave- lengths are 4.8 um and 9.7 wm for the MWIR and LWIR diodes, respectively.
Fig. 23. Operating mode of the dual-band demonstrator.
Fig. 23. Operating mode of the dual-band demonstrator.
Fig. 24. NETD histogram in the MWIR band for a P-P dual-band demonstrator measured at 77 K.
Fig. 24. NETD histogram in the MWIR band for a P-P dual-band demonstrator measured at 77 K.
Fig. 25. NETD histogram in the LWIR band for a P-P dual-band demonstrator measured at 77 K.
Fig. 25. NETD histogram in the LWIR band for a P-P dual-band demonstrator measured at 77 K.
Table IV. Comparative List of System Needs More Adapted to NPN or Pseudo-Planar Dual-Dand Architectures
Table IV. Comparative List of System Needs More Adapted to NPN or Pseudo-Planar Dual-Dand Architectures
Fig. 26. MWIR/LWIR images taken with the dual-band P-P demonstrator.
Fig. 26. MWIR/LWIR images taken with the dual-band P-P demonstrator.
Related topics:
Electronic MaterialsMolecular beam epitaxyTemporal coherenceSpatial CoherenceElectrical and Electronic EngineeringFocal Plane Array
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved