`
`by
`
`Iliana L. Fujimori
`
`M.Eng. in Electrical Engineering and Computer Science,
`Massachusetts Institute of Technology,
`June 1997
`
`B.S. in Electrical Engineering and Computer Science,
`Massachusetts Institute of Technology,
`June 1997
`
`Submitted to the Department of Electrical Engineering and Computer Science
`in partial fulfillment of the requirements for the degree of
`
`Doctor of Philosophy in Electrical Engineering and Computer Science
`
`at the
`
`MASSACHUSETTS INSTITUTE OF TECHNOLOGY
`
`February 2002
`
`@Iliana L. Fujimori, 2002. All rights reserved.
`
`BARKER
`
`MASSACHUSEtTS INSTITUTE
`MASSACHUSETTS INSTITUTE
`OF TECHNOLOGY
`
`APR 1 6 2002
`
`LIBRARIES
`
`Author ...........
`
`Certified by.
`
`Department of Electrical Engifeering and Computer Science
`February 1, 2002
`
`Charles G. Sodini
`Professor of Electrical Engineering
`Thesis Supervisor
`
`A ccepted by .................
`
`Arthur C. Smith
`Chairman, Department Committee on Graduate Students
`
`...................
`
`We 0hWr hwe9sbv s1 M
`if
`pWTn s*kn to reproduce nd t
`dftbute WuW papr and
`09ect rn coes of
`thi
`k
`dbcum,.iwn v b or in pat.
`
`Magna 2045
`TRW v. Magna
`IPR2015-00436
`
`
`
`2
`
`
`
`CMOS Passive Pixel Imager Design Techniques
`by
`
`Iliana L. Fujimori
`
`Submitted to the Department of Electrical Engineering and Computer Science
`on February 1, 2002, in partial fulfillment of the
`requirements for the degree of
`Doctor of Philosophy in Electrical Engineering and Computer Science
`
`Abstract
`CMOS technology provides an attractive alternative to the currently dominant CCD tech-
`nology for implementing low-power, low-cost imagers with high levels of integration. Two
`pixel configurations are possible in CMOS technology: active and passive. The active pixel
`requires a minimum of three transistors to convert light to voltage. The passive pixel, on the
`other hand, consists of a single transistor, and its output is in the form of charge. Column-
`parallel opamps are used to amplify the charge to a voltage output. The main advantage of
`the passive pixel is a higher fill factor in a given pixel geometry. This advantage becomes
`increasingly important as we scale to smaller pixel sizes. The higher fill factor comes at a
`high cost as the charge output on the high impedance node of the column line is susceptible
`to disturbances, namely a parasitic current and temporal noise. The goal of this thesis is
`to determine the source and effects of the disturbances on the image sensor characteristics
`and the repercussions for scaling to high-density arrays.
`A signal-dependent parasitic current composed of optically-generated carrier diffusion,
`blooming and subthreshold currents contaminates the pixel output. This parasitic current
`is detrimental to the imager because a few bright pixels can affect the rest of the pixels
`on the column line, resulting in bright vertical stripes on the image. A correlated-double
`sampling circuit in a differential architecture is used to remove the effects of the parasitic
`current. Column fixed-pattern noise is maintained below 1.5% for the linear illumination
`range of the imager.
`A noise analysis reveals the opamp read noise is the dominant source of temporal noise.
`The effects of the sample-and-hold readout circuit on the output-referred opamp read noise
`are modeled and closely match the measured noise. The output read noise power is directly
`proportional to the vertical resolution of the imager and inversely proportional to the pixel
`area, resulting in a strong dependence between noise and pixel density.
`This co-dependence is further analyzed in a scaling model where the fill factor, noise
`and dynamic range are observed for varying pixel size and vertical resolution over three
`fabrication technologies. The fill factor decreases with pixel size, and is highest for the
`technology with the smallest feature size, 0.18pm. The noise increases with decreasing
`pixel size and increasing vertical resolution and has the best performance in the 0.18pm
`technology. The dynamic range decreases with pixel density, but has a strong dependence
`on the power supply voltage of the technology.
`
`Thesis Supervisor: Charles G. Sodini
`Title: Professor of Electrical Engineering
`
`3
`
`
`
`4
`
`
`
`Acknowledgments
`
`After ten and a half years at MIT, I feel indebted to a multitude of extraordinary individuals.
`
`This page is by no means complete, as it allows me to thank only a small fraction of those
`
`who helped shape my decade at the 'tute.
`
`I would first like to thank my advisor and mentor, Prof. Charlie Sodini, for his guidance
`
`and support during the length of this project. His advice and encouragement proved more
`
`valuable than any CAD program in the world! I also wish to acknowledge my thesis readers,
`
`Prof. Harry Lee and Marc Loinaz, for taking the time to read my thesis. Their many
`
`suggestions and challenging questions helped shape this thesis.
`
`Many thanks go to my contemporaries of the Sodini group, Ching-Chun Wang, Pablo
`
`Acosta-Serafini, and Don Hitko, for listening, humoring, commiserating, and going to Tosci's
`
`at a moment's notice.
`
`I am also thankful to the rest of the Sodini/Lee groups for their
`
`support and friendship: Ayman Shabra, Kush Gulati, Dan McMahill, Susan Dacy, Mark
`
`Spaeth, Mark Peng, and Andy Wang.
`
`I am thankful to Pat Varley for showing me the ropes around the MIT procurement
`
`system. I thank Marilyn.Pierce and Monica Bell for their friendly reminders about depart-
`
`mental requirements and deadlines. Many thanks go to Myron (Fletch) Freeman and Mike
`
`Hobbs for all their help with computer emergencies.
`
`I thank my grad school buddies for making this academic experience a little less aca-
`
`demic: Samara Firebaugh, Alice Wang, Joel Voldman, Jeanie Cherng and Debb Hodges-
`
`Pabon. I also wish to thank all my 6.012 students (Spring 97 and Fall 99) for inspiring me
`
`to seek a career in teaching someday.
`
`I am extremely grateful to my fiance Zony Chen for his love and support, especially
`
`during the final stages of this thesis. He played a crucial role in the conclusion of this thesis
`
`by cooking for me, driving me to school and staying up with me to ensure I met deadlines.
`
`Most of all I thank my parents, Pedro and Elizabeth Fujimori, for all the sacrifices they
`
`made to get me to this point in life. Our journey to America was worth it all!
`
`The author was funded by an NSF fellowship, and a Lucent Technologies and GRPW
`
`fellowship. This project was funded by DARPA contract DAAL-01-95-K-3526 and the MIT
`
`Center for Integrated Circuits and Systems.
`
`5
`
`
`
`6
`
`
`
`Dedication
`
`I dedicate this thesis to Uchan and Twooey.
`
`7
`
`
`
`8
`
`
`
`Contents
`
`1 Introduction
`
`1.1 M otivation
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`19
`
`19
`
`1.2 Thesis Organization
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`2 Passive Pixel
`
`2.1 Background
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`21
`
`23
`
`24
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`2.2 A rchitecture .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`2.3 Pixel Operation - From photons to electrons .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`2.3.1 Quantum efficiency .
`
`.
`
`. . . .
`
`2.3.2 Photon Count .
`
`. .
`
`. .
`
`. . ..
`
`2.3.3 Photon-to-electron pixel gain
`
`. . . .
`
`24
`
`25
`
`26
`
`27
`
`27
`
`28
`
`2.3.4 Total pixel charge
`
`. .
`
`2.4 Pixel Readout - From electrons to V(
`
`Its
`
`.
`
`.
`
`.
`
`2.4.1 Output circuit reset
`
`. .
`
`. . .
`
`2.4.2 Pixel charge transfer . .
`
`. . .
`
`2.4.3 Opamp settling .
`
`. .
`
`. .
`
`. . .
`
`2.4.4 Pixel reset .
`
`. . .
`
`. .
`
`. .
`
`. . .
`
`2.4.5 Offset voltage cancelling . . .
`
`2.5
`
`Imager Performance
`
`. .
`
`. .
`
`. .
`
`. . .
`
`2.6 Pixel Structure
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`. . .
`
`2.6.1 Photodiode Implementation .
`
`2.6.2 Pixel Capacitance
`
`.
`
`. .
`
`. . .
`
`2.6.3
`
`2.6.4
`
`Parasitic Column Line Capac
`
`itance .
`
`Feedback Capacitance
`
`. .
`
`.
`
`.
`
`9
`
`28
`
`29
`
`30
`
`31
`
`32
`
`33
`
`35
`
`36
`
`36
`
`37
`
`39
`
`40
`
`
`
`2.7 Amplifier Specifications
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. . . . . . . . . . . . . . . . . . . . .
`
`2.7.1 DC Gain ............
`
`2.7.2 Bandwidth
`
`. .
`
`2.7.3 Output Swing .
`
`.
`
`.
`
`.
`
`.
`
`2.8 Passive vs. Active Pixels .
`
`. . . . . . . . . . . . . . . . . . . . .
`
`. . .
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. . . . . . . . . . . . . . . . . . . . .
`
`. . . . . . . . . . . . . . . . . . . . .
`
`. . . . . . . . . . . . . . . . . . . . .
`
`2.8.1 Active Pixel Sensor .....
`
`. . . . . . . . . . . . . . . . . . . . .
`
`2.8.2
`
`Passive Pixel Advantages
`
`. . . . . . . . . . . . . . . . . . . . .
`
`3 Parasitic Current
`
`3.1 Sources of parasitic current
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`.
`
`.
`
`. .
`
`.
`
`.
`
`4 1
`
`4 1
`
`4 2
`
`4 2
`
`4 3
`
`4 3
`
`4 4
`
`47
`
`3.1.1 Optically-generated carrier diffusion
`
`3.1.2 Blooming
`
`. .
`
`.
`
`. .
`
`.
`
`. . . .
`
`. .
`
`. .
`
`3.1.3 Subthreshold current .
`
`3.2 Effects of parasitic current .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`. .
`
`.
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`. .
`
`.
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`4 7
`
`4 8
`
`4 8
`
`4 8
`
`3.2.1
`
`Implementation of Test Chip
`
`3.2.2 Test Chip Results
`
`.
`
`.
`
`3.3 Removal of parasitic current .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`. .
`
`.
`
`.
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`4 9
`
`5 0
`
`5 3
`
`3.3.1 Architecture level .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`3.3.2 Circuit level .
`
`. .
`
`. .
`
`. . .
`
`. .
`
`. .
`
`. .
`
`3.4 Results.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`. . .
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`5 5
`
`5 5
`
`5 5
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`.
`
`. .
`
`.
`
`. .
`
`.
`
`. .
`
`. .
`
`5 9
`
`4 Noise in Passive Pixels
`
`4.1 Temporal noise sources .
`
`4.2 Opamp read noise
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. . .
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`. .
`
`. .
`
`63
`
`6 3
`
`4.2.1 Thermal Noise
`
`.
`
`. .
`
`. . .
`
`. .
`
`. .
`
`4.2.2 Flicker Noise
`
`. .
`
`. .
`
`. . .
`
`. .
`
`. .
`
`4.2.3 Sample-and-hold noise model
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. . .
`
`.
`
`. .
`
`. .
`
`. . .
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`4.2.4 Output-referred opamp read noise
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`6 4
`
`6 4
`
`6 4
`
`6 5
`
`4.3 Reset noise
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`4.3.1 Pixel reset noise
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. . .
`
`. . .
`
`.
`
`.
`
`. .
`
`. .
`
`.
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`. .
`
`. .
`
`7 3
`
`7 4
`
`. .
`
`. .
`
`4.3.2 Feedback capacitor reset noise
`
`.
`
`4.3.3 Other sources of reset noise .
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`. . .
`
`. . .
`
`4.4 Dark current shot noise
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`.
`
`7 4
`
`7 6
`
`7 6
`
`7 7
`
`10
`
`
`
`4.4.1 Pixel dark current shot noise
`
`.
`
`4.4.2 Column dark current shot noise
`
`4.5 Total temporal noise
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`4.6 Temporal noise measurements.....
`
`4.6.1 Output noise spectrum.....
`
`4.6.2 Varying sample frequency
`
`4.6.3 Varying Cline . .
`
`4.6.4 Varying Cout
`
`4.7 Fixed-pattern noise .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`..
`
`.
`
`.
`
`.
`
`.
`
`4.7.1 Pixel fixed pattern noise .
`
`4.7.2 Column fixed pattern noise
`
`4.8 Performance comparison .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`77
`
`78
`
`80
`
`81
`
`82
`
`82
`
`84
`
`85
`
`87
`
`87
`
`89
`
`90
`
`93
`
`5 Scaling of passive pixels
`
`5.1 Scaling parameters
`
`.
`
`.
`
`.
`
`.
`
`5.1.1 Spatial resolution .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`5.1.2 Pixel size
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`. . .
`
`5.1.3 Technology
`
`5.2 Effects of scaling
`
`.
`
`Fill factor .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`93
`
`93
`
`95
`
`96
`
`96
`
`5.2.1
`
`5.2.2 Output noise
`
`.
`
`5.2.3 Dynamic range
`
`.
`
`.
`
`.
`
`.
`
`5.2.4 Scaling summary .
`
`5.3 Fundamental limits .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`6 Conclusions
`
`6.1 Thesis Contributions .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`.
`
`97
`
`97
`
`101
`
`103
`
`104
`
`107
`
`107
`
`6.2 Future Work
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`. .
`
`.
`
`.
`
`108
`
`A Appendix A - Implementation of Diff.
`
`Architecture with
`
`CDS circuit
`
`109
`
`. .
`
`. .
`
`.
`
`. .
`
`.
`
`. .
`
`A.0.1 Pixel Design
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`A.0.2 CDS Design .
`
`. .
`
`. .
`
`. . .
`
`. .
`
`A.0.3 Programmable Analog Buffers
`
`.
`
`.
`
`.
`
`.
`
`.
`
`11
`
`.
`
`.
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`. .
`
`. .
`
`.
`
`.
`
`110
`
`113
`
`113
`
`
`
`A .0.4 Analog M ux . .
`
`. .
`
`.
`
`. . .
`
`. .
`
`. .
`
`. .
`
`. .
`
`. .
`
`.
`
`. .
`
`.
`
`.
`
`. .
`
`.
`
`. .
`
`.
`
`. .
`
`.
`
`115
`
`A.0.5 Row Decoder
`
`....
`
`...............
`
`..
`
`...
`
`..
`
`..
`
`..
`
`. 116
`
`12
`
`
`
`List of Figures
`
`2-1
`
`2-2
`
`2-3
`
`2-4
`
`Simple Passive Pixel Architecture.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`Cross-section of passive pixel with N-well to P-substrate photodiode.
`
`Single-Ended Passive Pixel Readout Circuit and Clock Phases.....
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`25
`
`26
`
`29
`
`Output Circuit Reset Phase. .
`
`2-5
`
`Pixel charge transfer. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`30
`
`30
`
`.
`
`.
`
`.
`
`.
`
`2-6
`
`2-7
`
`2-8
`
`Opam p settling.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`Pixel output for different ratios of
`
`Active reset to prevent image lag.
`
`+
`C
`+Cfb+Ci
`Cline+Cfb+Cpt'
`.
`.
`.
`.
`.
`.
`
`.
`
`.
`.....
`
`..
`
`.......
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`31
`
`32
`
`33
`
`2-9
`
`Reset phase with opamp offset voltage ..
`
`.
`
`2-10
`
`Sample phase with opamp offset voltage. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`34
`
`34
`
`2-11
`
`Pixel signal and noise sources over varying light intensity.
`
`.
`
`.
`
`2-12
`
`2-13
`
`Passive Pixel Cross Section with N-diffusion to P-substrate photodiode.
`
`Passive Pixel Cross Section with N-well to P-substrate photodiode.
`
`2-14
`
`Passive Pixel Cross Section with P-diffusion to N-well photodiode.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`36
`
`37
`
`37
`
`38
`
`41
`
`2-15
`
`Opamp transfer function ..
`
`.
`
`2-16
`
`Active pixel sensor schematic.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`2-17
`
`2-18
`
`Layout of passive pixel with N-well to P-substrate photodiode.....
`
`Layout of active pixel with N-well to P-substrate photodiode.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`43
`
`45
`
`45
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`3-1
`
`3-2
`
`Optically-generated electrons diffuse to column line. .
`
`Blooming mechanism.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`48
`
`49
`
`3-3
`
`3-4
`
`Subthreshold conduction currents.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`Column-parallel differential passive pixel architecture for test chip.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`49
`
`51
`
`3-5 Parasitic current effects on column 256 much higher than those of column 1 with
`
`single opam p readout.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`51
`
`13
`
`
`
`3-6 Layout mismatch of differential architecture to accentuate effects of parasitic current.
`
`3-7 Effects of diffused carriers on pixel output with no pixels selected, pixel on column
`
`2 selected and pixel on column 1 selected.
`
`3-8 Differential Passive Pixel Architecture. .
`
`3-9 Correlated Double Sampling Circuit. .
`
`.
`
`.
`
`.
`
`3-10 Column to Column Fixed Pattern Noise. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`52
`
`54
`
`56
`
`57
`
`3-11 Sample image taken with passive pixel imager with CDS.
`
`3-12 Imager Die Photo.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`4-1
`
`4-2
`
`Temporal noise sources in passive pixel.
`
`Opamp thermal noise.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`61
`
`62
`
`62
`
`64
`
`.
`
`.
`
`.
`
`Opamp flicker noise.
`
`Auto-zero readout.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`4-3
`
`4-4
`
`4-5
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`65
`
`65
`
`66
`
`4-6
`
`4-7
`
`4-8
`
`Sample-and-hold noise model in time doma
`
`sample-and-hold noise model in frequency (
`
`Auto-zero and loop gain transfer function.
`
`Output-referred opamp read noise.....
`
`Sampled noise.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`67
`
`68
`
`69
`
`70
`
`4-9
`
`4-10
`
`Held noise.
`
`.
`
`.
`
`.
`
`4-11
`
`Pixel reset noise
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`71
`
`72
`
`.
`
`.
`
`.
`
`4-12
`
`4-13
`
`Feedback capacitor reset noise
`
`Pixel dark current shot noise .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`7 5
`
`7 7
`
`4-14
`
`4-15
`
`4-16
`
`4-17
`
`4-18
`
`Column dark current shot noise
`
`Output noise power.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. . . . . . . . . . .
`
`Output noise power for different values of sampling frequency. .
`
`Output noise power for different values of Cline. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`Output noise power for different values of Cline and fs......
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`7 8
`
`7 9
`
`82
`
`83
`
`83
`
`84
`
`4-19
`
`Output noise power for different values of Cout.
`
`4-20
`
`Rms output noise as a function of Cout.
`
`4-21
`
`Amplifier power as a function of Ct.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`4-22
`
`Rms output noise and power as a function of C0 st.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`5-1 Parasitic column line capacitance components. .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`85
`
`86
`
`88
`
`88
`
`94
`
`14
`
`
`
`5-2
`
`5-3
`
`5-4
`
`5-5
`
`Pixel fill factor as a function of pixel pitch.....
`
`Noise components as function of pixel pitch. .
`
`Output noise as function of pixel pitch.
`
`Output noise as function of pixel pitch.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`Noise components as function of spatial resolution.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`97
`
`99
`
`99
`
`100
`
`5-6
`
`5-7
`
`5-8
`
`5-9
`
`Output noise as function of spatial resolution.
`
`Output noise as function of spatial resolution.
`
`Dynamic range as function of pixel pitch.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`Dynamic range as function of spatial resolution.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`100
`
`101
`
`102
`
`103
`
`5-10
`
`A-i Block diagram of differential passive pixel imager with CDS circuit.
`
`A-2 Signal path of analog passive pixel imager with CDS circuit .
`
`A-3 Passive pixel layout with N-well photodiode
`
`A-4 Section of passive pixel array .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. .. ..
`
`..
`
`.
`
`.
`
`.
`
`.
`
`.
`
`103
`
`110
`
`111
`
`112
`
`112
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`A-5 CDS Schematic Diagram.
`
`.
`
`.
`
`.
`
`A-6 Programmable Analog Buffers. .
`
`A-7 Analog multiplexer.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`A-8 Row select decoder and driver circuit.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`114
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`. . . . . . . . 115
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`116
`
`118
`
`15
`
`
`
`1616
`
`
`
`List of Tables
`
`3.1 Passive Pixel Imager Characteristics.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`61
`
`4.1 Passive pixel noise power for single-ended output with single readout and differential
`
`output with double readout.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`4.2 Sources and effects of column FPN for passive pixel.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`4.3 Performance comparison of passive pixel, APS and CCD. * Quoted value for green
`
`light was multiplied by factor of three to approximate monochrome responsivity.
`
`5.1 Design features for 0.6pm, 0.35pm, and 0.6Lm .
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`.
`
`80
`
`89
`
`91
`
`96
`
`17
`
`
`
`1818
`
`
`
`Chapter 1
`
`Introduction
`
`1.1 Motivation
`
`The increased demand for affordable consumer digital cameras has led camera manufactur-
`
`ers to look beyond the mature, but costly, Charge-Coupled Devices (CCD) technology to
`
`alternative methods in solid-state imaging, such as CMOS image sensors [1],
`
`[2]. CMOS
`
`imagers can be integrated with analog and digital functional blocks in a standard CMOS
`
`process leading to significant cost reductions [3]. Additional benefits include random access
`
`readout and low power consumption [4].
`
`In addition to low cost and low power, the consumer digital camera market demands
`
`high spatial resolution [5] in order to match the quality of high-definition images produced
`
`with film cameras. Since the cost of an imager is directly proportional to silicon area, the
`
`combination of low cost and high resolution specifications can be achieved with a high-
`
`density imager containing more pixels per unit area. The reduction in pixel area, however,
`
`results in lower fill factor, or ratio of photodiode to pixel area for a given technology. Most
`
`CCD and CMOS imagers currently use microlenses to improve the collection efficiency of
`
`shrinking pixels. While they provide a temporary solution, microlenses are costly and their
`
`efficiency is beginning to reach limits for small pixels [6].
`
`Recent efforts to achieve high fill factors in small pixel area have focused on reducing
`
`the number of transistors and contacts per pixel [7],
`
`[8]. Active pixel sensors (APS) using
`
`capacitively coupled bipolar transistors are also a viable option for high density arrays [9],
`but require a costly BiCMOS process and suffer from fixed-pattern noise (FPN) values as
`
`high as 10% [10].
`
`19
`
`
`
`The CMOS passive pixel presents a promising alternative to the active pixel for achieving
`
`a high fill factor in small pixel geometries. Similar to the history of the single-transistor
`
`DRAM cell [11], the passive pixel has the potential to make high-density imaging arrays
`
`available at a lower cost. For a given technology and fill factor, the passive pixel can achieve
`
`a higher pixel density. Conversely, for a given technology and pixel density, the passive pixel
`
`yields a larger fill factor. Consisting of a single transistor for readout and row select, the
`
`passive pixel has the added advantage of a lower pixel FPN and a linear transfer function.
`
`Fill factors as high as 80 % have been reported for a CIF format passive pixel imager [12].
`
`The advantages of using a passive pixel for its higher fill factor come at a high cost.
`
`Since charge-to-voltage conversion does not occur within the pixel, the charge signal becomes
`
`extremely sensitive to disturbances on the column line. Additional implants and fabrication
`
`steps can be used to reduce these disturbances [13],
`
`[14], but the increase in cost and
`
`complexity does not justify the use of a passive pixel imager. The goal of this thesis is to
`
`identify the source/s of the disturbances, quantify its effects and determine the limitations
`
`it poses on scaling to high-density arrays.
`
`One of the disturbances manifests itself in the form of a leakage current, or parasitic
`
`current, at each column line. Three mechanisms are identified as the causes of this current:
`
`optically-generated carrier diffusion, blooming and subthreshold currents. The effects of
`
`this current are catastrophic to the output signal as a few bright pixels can contaminate
`
`the output of all pixels on the column line, leading to bright vertical streaks in the image.
`
`Column FPN measurements demonstrate a correlated double-sampling (CDS) circuit in a
`
`differential architecture is effective in removing the effects of the parasitic current.
`
`An additional problem that limits the signal-to-noise ratio (SNR) of passive pixels for
`
`large imaging arrays is the temporal noise. In this thesis, the temporal noise is decomposed
`
`into the basic components: read noise, reset noise and dark current shot noise. Measure-
`
`ments indicate that the opamp read noise is the main contributor to the noise. The read
`
`noise of the amplifier, which is not present in active pixels, is amplified by the loop gain of
`
`the output circuit,
`
`Cf b
`is the feedback capacitance. Since Cline is proportional to the number of rows per column
`
`, where Cline is the parasitic capacitance of the column line and Cfb
`
`and Cfb is designed to match the pixel capacitance, the closed-loop gain which amplifies
`
`the column amplifier noise is unusually high for passive pixel output circuits, i.e. > 50, and
`
`therefore limits the SNR for low light levels and the dynamic range of the imager.
`
`20
`
`
`
`The severity of the temporal noise becomes obvious when scaling to higher density
`
`arrays. An increase in the number of rows leads to an increase in the loop gain through
`
`the line capacitance. A smaller pixel size further exacerbates the problem by decreasing
`
`the feedback capacitance and thus increasing the loop gain. The scaling of both parameters
`
`therefore result in higher noise. Designing in a smaller fabrication technology (i.e. 0.18pm)
`
`can alleviate the noise problems by increasing the pixel capacitance through a higher fill
`
`factor, and decreasing the line capacitance with smaller row select transistors. The lower
`
`power supply voltage, however, significantly reduces the dynamic range.
`
`1.2 Thesis Organization
`
`Chapter 2 provides a brief background on the passive pixel and a derivation of the expression
`
`relating photons and output voltage. The passive pixel structure and amplifier specifications
`
`are also discussed. A comparison between the active vs. passive pixels is given.
`
`Chapter 3 introduces the sources of parasitic current and illustrates its effects on the
`
`pixel output. A CDS circuit in a differential architecture is then presented. FPN measure-
`
`ments quantify the efficiency of the CDS circuit and an image demonstrates the effects of
`
`the parasitic current have been completely removed.
`
`The temporal and FPN sources are listed in Chapter 4. The effect of each noise source
`
`on the output voltage is explained and noise measurements are presented for a number of
`
`circuit parameters.
`
`The limits of the passive pixel when scaling for high-density arrays are illustrated in
`
`Chapter 5. The effects of decreasing pixel size and increasing vertical resolution on the pixel
`
`fill factor, temporal noise and dynamic range are observed for three fabrication technologies.
`
`Chapter 6 concludes this thesis with a list of contributions and ideas for future work.
`
`21
`
`
`
`2222
`
`
`
`Chapter 2
`
`Passive Pixel
`
`This chapter begins with a background of previous work done on the passive pixel. A
`
`typical CMOS passive pixel architecture is then presented. The pixel operation is divided
`
`into two parts: light-to-charge conversion, and charge-to-voltage amplification. The final
`
`result is an expression between the photon flux and the output voltage. Second order
`
`effects, such as image lag, finite opamp gain and opamp offset voltage, are also considered.
`
`The implementation of a CMOS passive pixel is described in Sec. 2.6. Advantages and
`
`disadvantages of three different photodiode implementations are discussed. An explanation
`
`of the effects of pixel and parasitic line capacitance on opamp specifications is also presented.
`
`A comparison between the active and passive pixel is given. Three advantages, mainly
`
`a larger fill factor, lower pixel fixed-pattern noise (FPN) and a linear transfer function,
`
`are cited as the motivation for this thesis. A larger fill factor is usually achieved with the
`
`passive pixel since it requires a single transistor, leaving plenty of pixel area for photon
`
`sensing. The lower pixel FPN is attributed to the simplicity of the passive pixel.
`
`In
`
`contrast with the active pixel, the passive pixel does not have any amplification or level
`
`shifting within the pixel, making it independent of device mismatches from pixel to pixel.
`
`Finally, the passive pixel uses a voltage-independent poly-to-poly capacitor, rather than the
`
`photodiode depletion capacitance, for the charge-to-voltage conversion, leading to a linear
`
`transfer function.
`
`23
`
`
`
`2.1 Background
`
`The passive pixel has a long history, longer than CCD's and the active pixel, dating back
`
`to 1967. G. P. Weckler was the first to implement a photodiode with an MOS process and
`
`operate it in integrating mode [15]. P. J. W. Noble followed him a year later with a 10-by-10
`
`array with an integrating charge amplifier [16]. Both Weckler's and Noble's designs were
`
`limited by the dark current and high parasitics, not to mention FPN and blooming, of the
`
`technology at the time.
`
`Almost thirteen years later, Hitachi reported the implementation of a 320 x 244 passive
`
`pixel sensor with high sensitivity and a FPN suppressing circuit. Using an additional P-
`
`implant, S. Ohba et. al. successfully reduced the FPN by 20 dB [13]. Hitachi later combined
`
`the passive pixel with a three-phase horizontal bulk charge-transfer device, similar to the
`
`ones used by CCD's at the time. Using a hybrid circuit made up of enhancement-mode
`
`and depletion-mode transistors, H. Ando et. al. were able to isolate the high column line
`
`capacitance in order to limit the read noise of the output amplifier [14].
`
`Profitting from the advances in VLSI technology in 1989, R. H. Wyles integrated a
`
`full column-parallel architecture in which one capacitive feedback transimpedance amplifier
`
`is used to read each column [17]. Prior to this point, column-parallel architectures had
`
`been difficult to implement since the readout circuits were too bulky to pitch-match the
`
`pixel. Shortly thereafter, D. Renshaw and P. B. Denyer reported on the integration of
`
`computational and control functions with 786 x 576 passive pixel arrays in a CMOS 0.8pm
`
`technology [18].
`
`More recently, W. Hoekstra et. al. published a CIF format (352(H) x 288(V)) passive
`
`pixel array in a CMOS 0.5pm technology, achieving a fill factor of 80% [12]. Despite the high
`
`fill factor it represents, the passive pixel has been dismissed as "noisy" and "unscalable"
`
`for visible range imagers [19],
`
`[20].
`
`2.2 Architecture
`
`Fig. 2-1 shows the architecture of a simple passive pixel imager, consisting of a pixel array,
`
`output circuit and row/column decoders [18]. The pixel array is read out in a typical raster-
`
`scan fashion. An entire row is selected for readout simultaneously and the corresponding
`
`charge from each pixel appears on the respective column line. The multiplexer then selects
`
`24
`
`
`
`one column at a time for charge-to-voltage conversion by the capacitive-feedback amplifier.
`
`_0
`0
`
`JL
`
`_JL
`
`rowsel
`
`col
`
`Multiplexer
`
`Figure 2-1: Simple Passive Pixel Architecture.
`
`2.3 Pixel Operation - From photons to electrons
`
`Pixel photocurrent arises when photons exhibiting an energy higher than the bandgap
`
`energy of the material (1.1 eV for Silicon) generate electron-hole pairs, as shown in figure 2-
`
`2. The electric field at the edge of the depletion region of the photodiode then separates
`
`the electron-hole pairs and the corresponding electrons and holes are collected in the N-well
`
`and P-substrate respectively.
`
`The amount of charge collected in the N-well depends on the location of the generated
`
`carriers. If the carriers are generated right at the edge of the depletion region, they will be
`
`immediately separated and collected. If, on the other hand, the photon is absorbed in the
`
`N-well or P-substrate, the carriers will have to diffuse to the depletion region bef