Advances in Electronics and Electron Physics, Volume 28A: Photo-Electronic Image Devices, Proceedings of the Fourth Symposium

ADVANCES IN ELECTRONICS AND ELECTRON PHYSICS VOLUME 28A Advances in Electronics and Electron Physics EDITED BY L. MA...

Author: Jd Mcgee

31 downloads 865 Views 24MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

ADVANCES IN ELECTRONICS AND ELECTRON PHYSICS VOLUME 28A

Advances in

Electronics and Electron Physics EDITED BY L. MARTON National Bureau of Standards, Washington, D.C.

Assistant Editor CLAIREMARTON

EDITORIAL BOARD T. E. Allibone H. B. G. Casimir W. G. Dow A. 0. C. Nier E. R. Piore

M. Ponte A. Rose L. P. Smith I?. K. Willenbrock

VOLUME 28A

1969

ACADEMIC PRESS

New York and London

Photo-Electronic Image Devices PROCEEDING8 O F THE FOURTH SYMPOSIUM HELD AT IMPERIAL COLLEGE, LONDON, SEPTEMBER 16-20, 1968

EDITED BY J. D. McGEE, O.B.E., Sc.D., F.R.S. D. McMULLAN, M.A., Ph.D. E. KAHAN, B.Sc., Ph.D. AND

B. L. MORGAN, B.Sc., Ph.D. Department of Physics, Imperial College, University of London

1969

ACADEMIC PRESS

London and New York

COPYRIGHT

0 1969 B Y ACADEMICP R E S S INC.(LONDON)LTD. ALL RIGHTS BESERVED

NO PART O F T B I S BOOK MAY B E REPRODUCED I N A N Y FORM

B Y PHOTOSTAT, MICROFILM OR A N Y OTHER MEANS WITHOUT WRITTEN PERMISSION FROM THE PUBLISHERS

ACADEMIC PRESS INC. (LONDON) LTD. Berkeley Square House Berkeley Square, London W 1X 6BA

U.S. Edition Published by ACADEMIC PRESS INC. 11 1 Fifth Avenue New York, New York 10003

Library of Congress Catalog Card Niimber 49-7504

SRN 12-014528-6

PRINTED I N GREAT BRITAIN B Y THE WHITEBRIARS PRESS, LONDON A N D TONBRIDGE

LIST OF CONTRIBUTORS H. D. ABLES, Flagstaff Station, U.S. Naval Observatory, Flagstaff, Arizona, U.S.A. (p. 1) R. W. AIREY, Applied Physics Department, Imperial College, University of London, London, England (p. 89) N. ARMAD,Department of Pure and Applied Physics, The Queen’s University of Belfast, Belfast, Northern Ireland (p. 999) H. ANDERTON,Westinghouse Electric Corporation, Electronic Tube Divkion, Elmira, New York, U.S.A. (p. 229) M. ASANO, Department of Electronic Engineering, University of ElectroCommunications, Chofu City, Tokyo, Japan (pp. 309, 381) H. BACIK, Applied Physics Department, Imperial College, University of London, London, England (p. 61) G. S. BAKKEN, Physic8 Department, Rice UniverBity, Houston, Texaa, U.S.A. (P. 907) M. E. BARNETT,Applied Physics Department, Imperial College, University of London, London, England (p. 545) J . R. BASKETT, A.C. Electronics, Defense Research Laboratories, General Motors Corporation, Santa Barbara, California, U.S.A. (p. 1021) c . W. BATES,JR., varian Associates, Palo Alto, California, U.S.A. (pp. 451, 545) W. A. BAUM,Lowell Observatory, Flagstaff, Arizona, U.S.A. (p. 753) W. BAUMOARTNER, Institut f u r Technische Physik, E T H , Zurich, Switzerland (P. 151) J . E. BECKMAN, Physics Department, Queen Mary College, University of London, London, England (p. 801). It, L. BEURLE,Department of Electrical and Electronic Engineering, Nottingham University, Nottingham, England (pp. 635, 1043) R. R. BEYER, Westinghouae Electric Corporation, Electronic Tube Division, Elmira, New York, U.S.A. (pp. 105, 229) 1’. BIED-CHARRETON, Observatoire de Paris, Paria, Prance (p. 27) A. BIJAOUI, Observatoire de Paris, Paris, France (p. 27) M. BLAMOUTIER, Compagnie Franpise Thomaon-Houston, Paris, France (p. 273) A. H. BOERIO, Westinghouse Electric Corporation, Electronic Tube Division, Elmira, New York, U.S.A. (p. 159) A. BOKSENBERG, Mullard Space Science Laboratory, Physics Department, U n i versity College, London, England (p. 297) I. S. BOWEN, M t . Wilson and Palomar Observatories, Carnegie Institution of Washington, California Institute of Technology, Paaadena, Calqornia, U.S.A. (P. 767) P. W. J . L. BRAND,Department of Astronomy, University of Edinburgh, Royal observatory, Edinburgh, Scotland (pp. 737, 783) F. LE CARVENNEC, Compagnie Gknkrals de Tklkgraphie Sans Fil, Paria, France (P. 265) W. N. CHARMAN, Atomic Energy Reaearch Establishment, Harwell, Berkshire, England (p. 705) P. A. CRATTERTON,Department of Electrical Engineering, University of Liverpool, Lancaahire, England (p. 1041)

vi

LIST OF CONTRIBUTORS

M. COHEN, Applied Physics Department, Imperial College, University of London, London, England ( p . 125) P. R. COLLINQS, Westinghouse Electric Corporation, Electronic Tube Division, Elmira, New York, U.S.A. ( p . 105) M . COMBES,Observatoire de Meudon, Meudon, France (p. 39) A. C. CONRAD,JR.,Physic8 Department, Rice University, Howton, Texas, U.S.A. (P. 907) J. M . LE CONTEL,Obaervahre de Paris, Paris, France ( p . 27) R. J. CORPS,Royal Aeronautical Establishment, Farnborough, Hampshire, England ( p . 827) G. I(. L. CRANSTOUN, Inorganic Chemistry Laboratory, University of Oxford, Oxford, England (p. 875) G. W . A. CZEEALOWSKI, Department of Medical Physics, University of Leeds, England ( p . 653) M . V . DANIELS, Department of Electrical Engineering, University of Nottingham, Nottingham, England ( p . 635) R. W . DECKER, Westinghouse Aerospace Division, Baltimore, Maryland, U.S.A. (pp. 19, 357) J. H . M . DELTRAP, Aerojet Delft Corporation, Melville, New York, U.S.A. ( p . 443) E. W. DENNISON, Mt. Wilson and Palomar Obseruatoriea, Paaadena, California, U.S.A. ( p . 767) K . DEUTSCHER, Ernst Leitz G.m.b.H. Optical Works, Wetzlar, West Germany (P. 419) P . DOLIZY, Laboratoires d’$lectronique et de Physique Appliqude, Limeil-Brdvannes, France ( p . 367) B. DRIARD, Compagnie Franpaise Thomson-Houston,Division Tubes .@ectroniques, Paris, Prance ( p . 931) M . DUCHESNE, Observatoire de Paris, Paris, France ( p . 27) M. DvoBAK, Tesla-VUVET, Prague, Czechoslovakia ( p . 347) D. W. EQAN,Jet Propulsion Laboratory, Caliifornia Institute of Technology, California, U.S.A. (p. 801) C. T. ELLIOTT, Royal Radar Establishment, Malvern, Worcestershire, England ( p . 1041) D. L. EMBERSON, Mullard Ltd., Mitcham, Surrey, England ( p . 119) L. ENGLAND, Applied Physics Department, Imperial College, University of London, London, England ( p . 546) G. ESCHARD, Laboratoires d’glectronique et de Physique Appliquke, LimeilBrdvannes, Prance ( p p . 499, 989) J. M. FAWCETT, Westinghouse Defense and Space Celzter, Baltimore, Maryland, U.S.A. ( p . 289) P. FELENBOK, Observatoire de Meudon, Meudon, France ( p . 39) J. R. FOLKES, English Electric Valve Co. Ltd., Chelmsford, England ( p . 375) D. P. FOOTE,Electro-Optical Systems, Inc., Paaadena, Calqornia, U.S.A. (p. 1059) K . G. FREEMAN, Mullard Research Laboratories, Redhill, Surrey, England ( p . 837) B. C. GALE, Department of Pure and Applied Physics, The Queen’s University of Belfast, Belfmt, Northern Ireland ( p . 999) B. R. C . GARFIELD,English Electric Valve Co., Chelmsford, England ( p . 375) R. K. H. GEBEL, Aerospace Research Laboratories, Wright-Patterson A i r Force Bme, Ohio, U.S.A. ( p . 685)


vii

A. GEURTS,N . V . Philips’ Gloeilumpenfabrieken, Eindhoven, The Netherlands (P. 616) R. GIESE,Phyaikaliaches Institut der Univeraitat Bonn, West Germany (p. 919) 0. GILDEMEISTER,Physikalischea Inatitut der Universitat Bonn, Weat Germany (P. 919) G. W . GOETZE, Weatinghouae Electric Corporation, Electronic Tube Division, Elmira, New York, U.S.A. (pp. 106, 169) A. W . GORDON,20th Century Electronics Ltd., Croydon, England (p. 433) J. GRAF, Laboratoirea d’dlectronique et de Physique AppliquBe, Limed - Brdvannes, France (p. 499) M . GREEN, Weatinghouae Electric Corporation, Electronic Tube DivCion, Elmira, New York, U.S.A. (p. 807) G. A. GROSCH, AEG-Telefunken, SoJinger Str. 100, 79 U l m (Donau), Weat Germany (p. 603) P. R. GROVES,Marconi Inatrumenta Ltd., Longacres, St. Albans, Hertfordshire, England (p. 827) J. GUERIN,Observatoire de Meudon, Meudon, France (p. 39) A. GUEST,Mullard Research Laboratories, Redhill, Surrey, England (p. 47 1) A. H. HANNA,Aerojet Delft Corporation, Melville, New York, U.S.A. (p. 443) J. R. HANSEN,Westinghouae Research Laboratories, Pittaburgh, Pennsylvania, U.S.A. (p. 807) W. HARTH, Inatitut f u r Technbche Electronik der Technischen Hochachule Munchen, Weet Germany (p. 636) P. HARTMANN, Laboratoirea de PhotoWectricitB dea Facultha dea Sciencea de Dijon et de Besanpon, France ( p . 409) 8. HASEOAWA, Department of Electronic Engineering, University of ElectroCommunications, Chofu City, Tokyo, Japan (p. 553) G. A. HAY, Department of Medical Phyaica, University of Leeda, England

(P. 663)

W . HEIMANN,Forachungslaboratorium, Wieabaden-Dotzheim, Weat Germany (P. 677) M . HERRMANN, Forachungslaboratorium, Wiesbaden-Dotzheim, West Germany (P. 955) W . HERSTEL,The Radiological Department, Univeraity Hoapital, Leiden, The Netherlands (p. 647) A. V . HEWITT,Flagstaff Station, U.S. Naval Observatory, Flagstaff, Arizona, U.S.A. (p. 1) R. L. HILLS,Department of Electrical and Electronic Engineering, University of Nottingham, Nottingham, England (p. 636) G. W . HINDER,Atomic Energy Research Eatablishment, Harwell, Berkahire, England (p. 966) M . HIRASHIMA, Department of Electronic Engineering, University of ElectroCommunicationa, Chofu City, Tokyo, Japan (pp. 309, 381) T. HIRAYAMA, Electron Tube Division, Nippon Electric Company, Tamagawa Plant, Kawasaki, Japan (p. 189) K . HIRSCHBERU, Ernst Leitz G.m.b.H. Optical Works, Wetzlar, West Germany (P. 419) E . L. HOENE, Forschungalaboratorium, Wieabaden-Dotzheim, Weat Germany (p. 677).

...

Vlll


R. T. HOLMSHAW, Mullard Research Laboratories, Redhill, Surrey, England (P. 471) H . HORI,Toshiba Reaearch and Development Centre, Tokyo Shibaura Electric Co. Ltd., KomuLai, Kawasaki, Japan (p. 253) P. IREDALE, Atomic Energy Research Establishment, Harwell, Berkshire, England ( p p . 689, 965) F . W . JACKSON, Research Laboratories, Electric and Musical Industriea Ltd., Hayes, Middleaex, England (p. 247) V . JAR&, Vacuum Electronics Research Institute, Prague, Czechoslovakia (p. 523) M. JEDLI~KA, Vacuum Electronics Research Institute, Prague, Czechoslovakia (P. 323) G. W. JENKINSON, Department ($ Electrical and Electronic Engineering, Nottingham University, Nottingham, England (p. 1043) A. S. JENSEN,Westinghouse Defense and Space Center, Baltimore, Maryland, U.S.A. (p. 289) J . M . JOHNSON, Research and Development Laboratories, Corning Glass Works, Corning, New York, U.S.A. (pp. 487, 507) J . A. JORDAN, JR., Department of Physics, Rice University, Houston, Texas, U.S.A. (p. 907) E . KAHAN, Applied Physics Department, Imperial College, University of London, London, England ( p . 725) Y. KAJIYAMA, Electron Tube Division, Nippon Electric Company, Tamagawa Plant, Kawasaki, Japan (p. 189) J. S. KALAFUT,Westinghouse Electric Corporation, Electronic Tube Division, Elmira, New York, U.S.A. (p. 105) T . KAWAHARA, Electron Tube Division, Nippon Electric Company, Tamagawa Plant, Kawasaki, Japan ( p . 189) H . KAWAKAMI, Matsushita Research Institute, Tokyo, Kawasaki, Japan (p. 81 ) B. KAZAN, I B M Watson Research Center, Yorktown Heights, New York, U.S.A. ( p . 1069) M . H. KEY, Department of Pure and Applied Physics, The Queen’s Iiniversity of Belfast, Belfast, Northern Ireland (p. 999) M . J . KIDGER,Applied Optics Section, Physics Department, Imperial College, University of London, London, England ( p . 759) Y . KIUCRI, Toshiba Research and Development Centre, Tokyo Shibaura Electric Co. Ltd., Komukai, Kawasaki, Japan (p. 253) T . KOHASHI, Matsuahita Research Institute Tokyo, Inc., Ikuta, Kawasaki, Japan (p. 1073) D. KOSSEL, Erndt Leitz G.m.b.H. Optical works, Wetzlar, West Germany (p. 419) J. K. KRIESER,A E G Telefunken, Soflinger Str. 100, 79 Ulm (Donau) West Germany (p. 603) G. E . KRON, Flagstaff Station, U S . Naval Observatory, Flagstaff, Arizona, U.S.A. (p. 1 ) W. KUHL, Philips Reaearch Laboratories, A’. V . Philips’ Gloeilampenfabrieken, Eindhoven, The Netherlands (p. 616) C. KUNZE,Porachungalaboratorium., Wiesbaden-Dotzheim, West Germany ( p . 955) W. KUNZE,AEG-Telefunken, 2 Hamburg 11, Steinhoft 9, West Germany (p. 629) A. LABEYRIE, Observatoire de Paris, 92-Meudon, France (p. 899) D. L. LAMPORT, Mullard Research Laboratoriea, Redhill, Surrey, England (p. 567) R. LEGOUX,Laboratoires d’&lectronique et de Physique Appliqude, LimeilBrdvannee, France (p. 367)


ix

B. T. LIUDY,Department of Pure and Applied Physics, The Queen's University of Belfaet, Belfaat, Northern Ireland (p. 375) I. D. LIU, AC Electronic.?, Defense Research Laboratories General Motors Corporation, Santa Barbara, California, U.S.A. (p. 1021) B. E. LONG,Mullard Ltd., Mitcham, Surrey, England (p. 119) J. L. LOWRANCE, Princeton University Observatory, Princeton, New Jersey, U.S.A. (P. 861) R. LYNDS,Kilt Peak National Observatory, Tucaon, Arizona, U . S . A. (p. 745) J. D. MCGEE,Applied Physics Department, Imperial College, University of London, London, England (pp. 61, 89) D. MCMULLAN, Royal Greenwich Observatory, Herstmonceux, Suasex, England (PP. 61, 173) H. MAEDA, Matsuahita Reaearch Institute, Tokyo, Kawaaaki, Japan (p. 81) B. W. MANLEY,Mullard Research, Laboratories, Redhill, Surrey, England (p. 471) R. MARTIN,Atomic Energy Research Establishment, Harwell, Berkshire, England (P. 981) H. MESTWERDT,United States A i r Force, Wright-Patterson A i r Force Baae, Ohio, U.S.A. (p. 19) K. MEYERHOFF,A E G Telefunken, 2 Wedel, Holstein, West Germany (p. 629) D. E. MILLER, Physics Department, University College of North Walea, Bangor, Walea (p. 513) K. MIYAJI, Matsuahita Electric Industrial Co., New York, U.S.A. (p. 1073) S . MIYASHIRO,Toshiba Research and Development Centre, Tokyo Shibaura Electric Go. Ltd., Kawaaaki, Japan (p. 191) E. MIYAZAKI,Matsuahita Reaearch Institute Tokyo, Inc., Ikuta, Kawaaaki, Japan (P. 81) B. L. MORGAN,Applied Physics Department, Imperial College, University of London, London, England (p. 1051) S . NAKAMURA, Matsuahita Research Institute Tokyo, Inc., Ikuta, Kawasaki, Japan (p. 1073) T. NAKAMURA, Matsushita Reaearch Inatitute Tokyo, Inc., Ikuta, Kawaaaki, Japan (p. 1073) M. J. NEEDHAM, Department of Physics, Queen Mary College, London, England (P. 129) P. D. NELSON,English Electric Valve Go. Ltd., Chelmsford, Essex, England (P. 209) A. C. NEWTON, Mullard Space Science Laboratory, Department of Physics, University College, London, England (p. 297) T. NINOMIYA, N H K Technical Research Laboratories, Setagaya, Tokyo, Japan (P. 337)

G. NIQUET,' Laboratoires de Photodlectricitd des Facult& dea Sciences de Dijon et de Beaanpon, France (p. 409) M. NOVICE, Westinghouse Electric Corporation, Electronic Tube Diviaion, Elmira, New York, U.S.A. (p. 1087) B. NOVOTN~T, Vacuum Electronics Reaearch Inatitute, Prague, Czechoslovakia (P. 523) Y . NOZAWA, Smithsonian fnatitution, Astrophyeical Observatory, Cambridge, Massachusetts, U.S.A. (p. 891) S. NUDELMAN, University of Rhode Island, Electrical Engineering Department, Kingston, Rhode Island, U.S.A. (p. 677)

X


T . W. O’KEEFFE, Weatinghome Research Laboratoriea, Pittaburgh, Pennaylvania, U.S.A. (p. 47) M . OLIVER,Applied Physics Department, Imperial College, Univeraity of London, London, England ( p . 61) J . v. OVERHAGEN,N . V . Philipa’ Qloeilampenfabrieken, Eindhoven, The Netherlands (p. 615) C. H.PETLEY, Mullard Research Laboratoriw, Redhill, Surrey, England (p. 837) J . P. PICAT,Obaervatoire de Meudon, Meudon, France (p. 39) R. POLAERT, Laboratoirea d’glectronique et de Phyaique Appliqude, LimeilBrBvannes, France ( p . 989) L. J . VAN DER POLDER, N . V . Philipa’ Qloeilampenfabrieken, Eindhoven, The Netherlands (p. 237) J . R. POWELL, Kitt Peak Natiortal Obeervatory, Tucaon, Arizona, U.S.A. (p. 745) D. L. PULPREY, Department of Electrical Engineering, Univeraity of Mancheater, England (p. 1041) W . P. RAFFAN, 20th Century Electronica Ltd., Croydon, Surrey, England ( p . 433) R. P. RANDALL, E.M.I. Electronica Ltd., Valve Division, Ruklip, Middlesex, England (p. 713) G. RETZLAFF, AEQ- Telejunken, Hamburg, West Germany (p. 629) G. T. REYNOLDS, Palmer Physical Laboratory, Princeton Univeraity, Princeton, New Jersey, U.S.A. (p. 939) E . A. RICHARDS, Signala Research and Development Eatabliahment, Chriatchurch, Hampshire, England (p. 661) E . W . T . RICHARDS, Atomic Energy Reaearch Eatabliahment, Harwell, Berkahire, England (p. 981) J . H . T . VAN ROOSMALEN, N . V . Philipa’ Gloeilampenfabrieken, Eindhoven, The Netherlanda (p. 281) D. J. RYDEN,Atomic Energy Research Eatabliahment, Harwell, Berkahire, England (p. 589) W .M . SACKINGER, Research and Development Laboratoriea, Corning Glaea Worka, Corning, New York, U.S.A. (pp. 487, 507) F . SCHAFF, C E R N , Geneva, Switzerland (p. 535) P. SCHAOEN, Mullard Reaearch Laboratoriea, Redhill, Surrey, England (p. 393) G. SOHUSTER, Phy8ikaliachea Inatitut der Univeraitiit Bonn, Weat B e m n y ( p . 919) S . SHIROUZU, Toahiba Reaearch and Development Centre, Tokyo Shibaura Electric Co., Ltd., Kawaeaki, Japan (p. 191) M . SCHMIDT, Mt. Wilaon and Palomar Obaervatoriea, Carnegie Inatitution of Waehington, California Inatitute of Technology, Pmadena, California, U.S.A. (P. 767) R. W . SMITH,Applied Optica Section, Physics Department, Imperial College, Univeraity of London, London, England (pp. 1011, 1051) W . A. SMITH,The Rutherford Laboratory, Chilton, Dideot, Berkahire, England (p. 1041) D. W . S . SMOUT,Atomic Energy Reaearch Ealablhhment, Harwell, Berkahire, England (p. 966) M . J . SMYTH,Univeraity of Edinburgh, Department of Astronomy, Royal Obaervatory, Edinburgh, Scotland ( p . 737) A. M . STARK, Mullard Reaearch Laboratoriea, Redhill, Surrey, England (p. 567) C. H. A. SYMS,Services Electronios Reaearch Laboratory, Baldock, Hertfordahire, England (p. 399)


xi

Z. SZEPESI,Westinghouse Electric Corporation, Electronic Tube Division, Elmira, New York, U.S.A. (p. 1087) H . TACHIYA,N H K Technical Research Laboratories, Setagaya, Tokyo, Japan (P. 337) K . TAKETOYHI, N H K Technical Research Laboratories, Setagaya, Tokyo, Japan (P. 337) D. G. TAYLOR, Mullard Research Laboratories, Redhill, Surrey, England (p. 837) M . TEPINIER,Laboratoires de Photodlectricitd des Facultb des Sciences de Dijon et de Besanpon, France (p. 409) R. F . THUMWOOD, Department of Physics, Queen Mary College, London, England (P. 129) G. 0. TOWER, Applied Physsios Department, Imperial College, University of London, London, England ( p . 173) S . TSUJI,Toahiba Research and Development Centre, Tokyo Shibaura Electric Go. Ltd., Komukai, Kawaaaki, Japan (p. 253) A. A. TURNBULL, Mullard Research Laboratories, Redhill, Surrey, England (p. 393) Y . UNO,Matsushita Research Institute Tokyo Inc., Ikuta, Kawaaaki, Japan (p. 81) B. P. VARMA, Applied Physics Department, Imperial College, University of London, London, England (p. 89) J . VINE, Westinghouse Research Laboratories, Pittsburgh, Pennsylvania, U.S.A. (PP. 47, 537) P. VERNIER,Laboratoires de Photodlectricitd des Facultds des Sciences de Dijon et de Besanpon, France (p. 409) S. VERON,Compagnie Gdndrale de Tdldgraphie s a w Pil, Orsay, France (p. 461) K . H . WAGNER,Department of Electrical Engineering, University of Salford, Salford, Lancaahire, England (p. 1033) M . F . WALKER,Lick Observatory, University of California,Santa Cruz, California, U.S.A. (p. 773) J . WARDLEY, Research Laboratories, Electric and Musical Industries Ltd., Hayes, Middlesex, England ( p . 247) G. WENDT,Compagnie Gdndrale de Tgldgraphie Sans Fil, Orsay, France (p. 137) W. L. WILCOCK, Physics Department, University College of North Wales, Bangor, Wales (p. 513) G. A. WILSON,Applied Physics Department, Imperial College, University of London, London, England (p. 1051) H . S . WISE, Atomic Energy Research Establishment, Harwell, Berkahire, England (P. 981) G . WL~RICK, Observatoire de Paris, Meudon, France (p. 787) R. D. WOLSTENCROFT, Royal Observatory, Edinburgh, Scotland (p. 783) A, W. WOODHEAD, Mullard Research Laboratories, Redhill, Surrey, England (P. 667) C . G. WYNNE, Applied Optics Section, Physics Department, Imperial College, University of London, London, England ( p . 759) P. M . ZUCCHINO,Princeton University Observatory, Princeton, New Jersey, U.S.A. (P. 851)

This Page Intentionally Left Blank

FOREWORD Three years have elapsed since we presented in our Volume 22 Parts A & B the proceedings of the Third Symposium on Photo-Electronic Image Devices. It is a great pleasure to publish this record of the fourth one. As Professor McGee points out in his Preface, the important aspect is that this subject is still a growing field. With the development of extra-terrestrial astronomy (orbiting space telescopes, etc.) one would have expected a lessening of interest in aids to terrestrial devices. The contrary happened and, as usual, new developments stimulate new interest in older ones. Professor McGee and his associates succeeded again in organizing an outstanding Symposium and in collecting its material herein. I am certain to express the thanks of all my colleagues, who are users of these volumes, in emphasizing the devotion and amount of work required for this task. It is to be hoped that he and his collaborators will consider to renew their effort in the not too distant future. They have already earned the esteem of the scientific community; their continued efforts make these Symposia into an established feature of our scientific life, which we have come to appreciate at regular intervals. I should like t o use this opportunity to list some of our present expectations of new authors and subjects in our forthcoming volumes.

H. M. ROSENSTOCK

Study of Ionization Phenomena by Mass Spectroscopy J. P. BLEWETT Recent Advances in Circular Accelerators S. AMELINCKX Image Formation at Defects in Transmission Electron Microscopy Quadrupoles as Electron Lenses P. W. HAWKES J. ROWE Nonlinear Electromagnetic Waves in Plasmas G. BENEand E. HENEITX Magnetic Coherence Resonances and Transitions at Zero Frequency N. R. WHETTEN and D. H. DAWSON Mass Spectroscopy Using Radio Frequency Quadruple Fields Ion Beam Bombardment and DopD. B. MEDVED ing of Semiconductors xlii

xiv

FOREWORD

E. R. ANDREWand S. CLOUGH J. A. MERCEREAUand D. N. LANGENBERU S. DATZ

Nuclear and Electronic Spin Resonance Josephson Effect and Devices

Reactive Scattering in Molecular Beams Luminescence of Compound SemiF. E. WILLIAMS conductors Energy Beams as Tools K. H. STEIQERWALI) et al. Electron Precursors RICHARDG. FOWLER The Physics of Long Distance H. A. WHALE Radio Propagation Macroscopic Approach to FerroJ. FOUSEK and V. JANOVIC electricity Sputtering M. W. THOMPSON Plasma Instabilities and TurbuC. KEITHMCLANE lence Electron Polarization STEPHENJ. SMITH F. J. KERRand WM.C. ERICKSON Galactic and Extragalactic Radio Astronomy Light Interaction with Plasma HEINZRAETHER Superconducting Magnets P. F. SMITH Recent Advances in Field Emission L. SWANSON and F. CHARBONNIER Microfabrication Using Electron A. N. BROERS Beams The Measurements of Lifetimes of A. CORNEY Free Atoms, Molecules and Ions Energy Distribution in ThermioniB. W, ZIMMERMANN cally Emitted Electron Beams Information Storage in Microspace S. NEWBERRY Recent Progress on Fluidics H. BURKEHORTON Theory Network L. WEINBERG The Formation of Cluster Ions in W. ROTHand R. NARCISSI Gaseous Discharges and in the Ionosphere

L. MARTON Washington, D .C. June 1969

PREFACE We have pleasure in presenting in these two books, Volume 28, Parts A & B of Advances in Electronics and Electron Physics, the papers read and discussed at the Fourth Symposium on Photo-Electronic Image Devices held at Imperial College from 16th to 20th September 1968. The number of papers presented, and the attendance, at this Symposium are convincing evidence of the interest that is maintained in this field. The four Symposia in the ten years since the first was held have recorded the development of this field of electron physics from a rather limited one to one of world wide interest, in which many of the great laboratories are actively working. Inevitably some of the projects discussed in the earlier Symposia have dropped into oblivion, but many have prospered and these volumes record their progress and, in many cases, ultimate success. We like to think that these meetings, and the lively and objective discussion that they encourage, have done much to advance this subject. It is a considerable advantage that it has been possible to include the proceedings of all four Symposia inJthiswell known series Advances in Electronics and Electron Physics published by the Academic Press. By maintaining a certain uniformity and continuity, the subject matter in this field has been made more readily available to those who are interested. We thank the Editor-in-Chief, Dr. L. Marton, and Academic Press for their co-operation which has made this possible. We have endeavoured to maintain a reasonable uniformity of presentation throughout the volume while retaining the scientific sense as intended by the authors. While we have made every effort to correct accidental errors, the author has the final word as regards subject matter. The Editors wish to thank their colleagues of the Applied Physics Department of Imperial College for their assistance in running the Symposium and Miss Margaret Jones, secretary, for her help and meticulous care in dealing with the papers. We should also like t o put on record our great appreciation of the continuing interest and support from Professor Lord Blackett, O.M., C.H., President of the Royal Society, who opened the Symposium. We also thank all those who participated and contributed papers for the excellent spirit in which the meeting was conducted, making it both very informative and very enjoyable. J . D. MCGEE

D. MCMULLAN E. KAHAN B. L. MORGAN

London June 1969 xv


CONTENTS LIST OF CONTRIBUTORS

.

V

FOREWORD.

xiii

PREFACE.

xv

CONTENTSOF VOLUME B

.

xxi

Electronography A Technical Description of the Construction, Function, and Application of the U.S. Navy Electronic Camera. By G. E. KRON,H. D. ABLESAND A. V. HEWITT

.

Large-image Electronographic Camera. MESTWERDT .

1

By R. W. DECKERAND H. 19

Sur Quelques Progrbs RBcents Apportee B la Camera l%xtronique B Focalisation $lectrostatique et sur son Application en Physique et en Astronomie. By P. BIED-CHARRETON, A. BIJAOUI,M. DUCHESNE 27 AND J. M. LE CONTEL

.

Electronic Cameras for Space Research. By M. COMBES, P. FELENBOK, J. GUERINAND J. P. PICAT. 39 A High-resolution Image Tube for Integrated Circuit Fabrication. T. W. O’KEEFFEAND J. VINE

.

By 47

Further Developments of the Spectracon. By J. D. MCGEE,D. MCMULLAN, H. BACIKAND M. OLIVER 61

.

Cathode-ray Tube with Thin Electron-permoable Window. By Y. UNO, H. KAWAKAMI, H. MAEDAAND E. MIYAZAKI.

81

Image Tubes Cascade Image Intensifier Developments. By J. D. MCGEE, R. W. AIREY AND B. P. VARMA .

89

A Family of Multi-stage Direct-view Image Intensifiers with Fiber-optic Coupling. By P. R. COLLINGS, R. R. BEYER,J. S. KALAFUT AND G. W. GOETZE .

105

Some Aspects of the Design and Manufacture of a Fibre-optic Coupled Cascade Image Intensifier. By D. L. EMBERsoN AND B. E. LONG

119

.

A Proximity-focused Image Tube. By M. J. NEEDHAM AND R. F. THUMWOOD

.

129

INTIC, an Image INTensifying, Integrating and Contrast-enhancing Storage Tube. By C . WENDT

.

.

A Light Amplifier with High Light Output. By W. BAUMGARTNER xvii

137

161

CONTEXTS

xviii

Signal Generating Tubes SEC Camera-tube Performance Characteristics and Applications. By . G. W. GOETZE AND A. H. BOERIO Some Properties of SEC Targets. By D. MCMULLANAND G. 0. TOWLER. Newly Developed Image Orthicon Tube with a MgO Target. By Y. KAJIYAMA, T. KAWAHARA AND T. HIRAYAMA . Electrostatically Scanned Image Orthicon. By S. MIYASHIRO AND S. SHIROUZO .

159

173 189 191

The Development of Image Isocons for Low-light Applications. By 209 P. D. NELSON , Dynamic Imaging with Television Cameras. By H. ANDERTON AND 229 R.R.BEYER . Beam-discharge Lag in a Television Pick-up Tube. By L. J. v. D. POLDER. 237 A 13-mmAll-Electrostatic Vidicon. By J. WARDLEY AND F. W. JACKSON. 247 An Infra-red Sensitive Vidicon With a New Type of Target. By H. HORI, S. TSUJI AND Y. KIUCHI . 253 Recherche d’un Dispositif Nouveau do TBlBvision Thermique. By F. LE CARVENNEC . 265 Un Tube de Prise de Vues Sensible aux Rayons X. By M. BLAMOUTIER. 273 Adjustable Saturation in a Pick-up Tube with Linear Light Transfer Characteristic. By J. H. T. VAN ROOSMALEN . 281 Measurement of TV Camera Noise. By A. S. JENSEN AND J. M. FAWCETT.289 An Electromechanical Picture Signal Generating Device. By A. BOKSENBERU AND A. C. NEWTON . 297 Effects of Caesium Vapour upon the Target Glass of Image Orthicons. By M. HIRASHIMA AND M. ASANO .

Photocathodes and Phosphors

.

Research on Photocathodes in Czechoslovakia. By M. JEDLI~EA Crystal Structure of Multialkali Photocathodes. By T. NINOMIYA, AND H. TACHIYA. K. TAKETOSHI Some Properties of the Trialkali Sb-K-Rb-Cs Photocathode. By M. DVO~AK. Decay of S.20 Photocathode Sensitivity Due to Ambient Gases. By R. W. DECKER A New Technology for Transferring Photocathodes. By P. DOLIZYAND R. LEUOUX Improvements to Photocathodes for Pulse Operation. By B. R. C. GARFIELD, J. R. FOLKESAND B. T. LIDDY Some Getter Materials for Caesiwn Vapour. By M. HIRASHIMA AND M. ASANO New Approaches to Photoemission at Long Wavelengths. By P. SCHAGEN AND A. A. TURNBULL . . , Gallium Arsenide Thin-film Photocathodes. By C. H, A. QYMS

.

.

.

.

309

323 337 347 357 367 375

381 393 399

xix

CONTENTS

fitude de l’fimission Photoblectrique des Structures MBtal-IsolantMBtal. By P. VERNIER,P. HARTMANN, G. NIQUETAND M. TEPINIER. 409 Interference Photocathodes. By D. KOSSEL, K. DEUTSCHER AND K.HIRSCH419

BERG

The Development and Application of Interference Photocathodes for Image Tubes. By W. P. RAFFANAND A. W. GORDON . Image Intensifier System Using Reflective Photocathode. By J. H. M. DELTRAP AND A. H. HANNA . Scintillation Processes in Thin Films of CsI(Na) and CsI(T1) due to Low . Energy X-Rays, Electrons and Protons. By C. W. BATES,JR. Quelques Aspects des Essais de DBp6t de Photocathodes 5.20 et d’ficrans Fluorescents sur Fibres Optiques. By S. VERON .

433 443 451 461

Channel Multipliers and Secondary Emissions Channel Multiplier Plates for Imaging Applications. By B. W. MANLEY, A. GUESTAND R. T. HOLMSHAW . An Analysis of the Low-level Performance of Channel Multiplier Arrays. By W. M. SACKINGER AND J. M. JOHNSON . Quelques Problhmes Concernant les Multiplicateurs Canalis& pour Intensificateur d’Image. By G. ESCHARD AND J. GRAF Effects of Vacuum Space Charge in Channel Multipliers. By W. M. SACKINGER AND J. M. JOHNSON . Statistics of Transmitted Secondary Electron Emission. By W. L. WILCOCK AND D. E. MILLER

.

.

471 487 499

507 513

Electron Optics Two Methods for the Determination of the Imaging Properties of Electronoptical Systems with a Photocathode. By V. JARE& AND B. N O V O T.N ~ 523 Computation of Imaging Properties of Image Tubes from an Analytic Potential Representation. By F. SCHAFF AND W. HARTH . 535 The Design of Electrostatic Zoom Image Intensifiers. By J. VINE . 537 Electron Optics of a Photoconductive Image Converter. By M. E. BARNETT, C. W. BATES,JR., AND L. ENGLAND 545

.


CONTENTS OF VOLUME B Image Tube Assessment Resolving Power of Image Tubes. By S. Hasegawa. Calculation of the Modulation Transfer Function of an Image Tube. By A. M. Stark, D. L. Lamport and A. W. Woodhead. Intensifiers: Detective Quantum Efficiency,Efficiency Contrast Transfer Function and the Signal-to-noise Ratio. By S. Nudelman. On the Quality of Photographic Images Recorded with the Use of Image Intensifiers. By P. Iredale and D. J. Ryden. Leistungsgrenze eines Sichtsystems mit Bildverstiirker. By G. A. Grosch and J. K. Krieser. Information Transfer with High-gain Image Intensifiers. By W. Kiihl, A. Geurts and J. v. Overhagen. The Useful Luminance Gain of Image Intensifier Systems with Respect to Noise Limitations. By W. Kunze, K. Meyerhoff and G. Retzlaff. Image Intensifier Design and Visual Performance at Low Light Levels. By R. L. Beurle, M. V. Daniels and B. L. Hills. The Observation of Moving Structures with X-Ray Image Intensifiers. By W. Herstel. A Quadrature Spatial-frequency Fourier Analyser. By G. W. A. Czekalowski and G. A. Hay. Contrast-enhancement in Imaging Devices by Selection of Input Photosurface Spectral Response. By E. A. Richards. Improvement of Signal-to-noise Ratio of Image Converters with S.1 Photocathodes. By W. Heimann and E. L. Hoene. The Fundamental Infra-red Threshold in Thermal Image Detection a8 Affected by Detector Cooling and Related Problems. By R. K. H. Gebel. Cosmic Rays and Image Intensifier Dark Current. By W. N. Charman. Dark Current Scintillations of Cascade Image Intensifiers. By R. P. Randall. Comparison of the Efficiency of Image Recording with a Spectracon and with Kodak I r a - 0 Emulsion. By E. Kahan and M. Cohen. Linearity of Electronographic Emulsions. By M. J. Smyth and P. W. J. L. Brand. Methods of Increasing the Storage Capacity of High-gain Image Intensifier Systems. By J. R. Powell and R. Lynds.

Applications in Astronomy A Critical Comparison of Image Intensifiers for Astronomy. By W. A. Baum. The Design of Optical Systems for Use with Image Tubes. By C. G. Wynne and M. J. Kidger xxl

xxii

CONTENTS OF VOLUME B

An Image-tube Spect,rographfor the Hale 200-in. Telescope. By E. W. Dennison,

M. Schmidt and I. S. Boweri. Performance of the Spectracon in Astronomical Spectroscopy. By M. F. Walker. Recent Astronomical Applications of a Spectracon. By P. W. J. L. Brand and R. D. Wolstencroft. Gtudes d’Astres Waibles en Lumiere Totdc avec la

\+

-

IO-~

c

V

TI

. I3 0 V

'; + +'

+'

r

~:=0-68ev 10-8

+'

-

\+\ +\

+\

t

(Temperaturer'

(OK-Ix

lo3)

Fro. 5 . Temperature dependence of the dark conductivity of the Sb-K-Rb-Cs photocathode,

362

M. D V O ~ ~ K

about 1 eV; a further group of energies ranging from 0.4 t o 0.7 eV were found; these were due to thermal electron excitation. Higher energy values, equal to the bandgap energy were not found in the measured temperature range. For still higher temperatures, above 90°C, changes on the surface and inside the photoemissive layer must be anticipated. From a measurement of the sign of the thermoelectric e.m.f., the photocathode was found to be a p-type semiconductor.

ENERGY DISTRIBUTION OF PHOTOELECTRONS The energy distribution of the emitted photoelectrons was determined using the retarding field in a spherical condenser as shown in Figs. 6 and 7. The photocathode was prepared on a small target outside the spherical measuring section which was isolated by a thin glass membrane. Thus the deposition of alkali metals on the spherical inner surface was avoided, this being in our experience a very important

FIG.6. Spherical condenser apparatus.

FIQ. 7. Schematic of spherical condenser. A, Photocathode; B, anode with Sb evaporator; C, alkali generator; D, glass membrane; E, spherical condenser; F, quartz window.

SOME PROPERTIES O F Sb-K-Rb-Cs

PHOTOCATHODES

353

condition for obtaining reliable measurements. Following the processing of the photocathode and the sealing-off of the tube from the pump system, the photocathode was moved into the spherical section after breaking the glass membrane. The photocathode was illuminated through a quartz window in the spherical section. Using this apparatus, current/voltage characteristics from zero values of the photocurrent up to saturation were measured (see Fig. 8), the

Collector potential ( V )

FIQ.8. Photoelectric current vs. collector potential characteristics measured using the spherical condenser with monochromat,ic illumination (photon energies: 3.4, 4, 4.3 and 5.15 eV).

photocathode being illuminated by monochromatic radiation. The energy distribution of the photoelectrons is given by the curves in Fig. 9 which were obtained by differentiation of the current/voltage characteristics. For a photon energy of 3.4eV most of the emitted photoelectrons have energies around 0.5 eV. With increasing photon energy a rise in the number of fast photoelectrons having energies from 0.8 to 1.6eV could be observed. The photoelectric work function derived from this measurement is 1-8eV, which is in good agreement with the value calculated from the long-wave threshold (Ao = 7000 A).

354

M. D V O ~ ~ K

Photoelectron energy ( e V )

FIG.9. The energy distribution of the photoelectrons for various photon energies.

CONCLUSION

A new alkali-antimonide photocathode composed of Sb-K-Rb-Cs was investigated and its properties determined. The discrepancies of the luminous and spectral sensitivities between illumination through the glass or from the vacuum side can be attributed to the fact that the photocathodes were thicker than is common with semi-transparent photoemissive layers. When illuminating the photocathode through the glass substrate, blue light is absorbed and electron excitation occurs near the substrate so that photoelectrons are unable to reach the surface. This absorption is thus photoelectrically ineffective as the layer behaves for these wavelengths as a filter and the spectral sensitivity is low in this region. Light of longer wavelength is absorbed t o a lesser extent and penetrates nearer to the surface of the layer and the photoelectrons can escape more easily. This results in increased sensitivity for the red region of the spectrum. For illumination from the vacuum side no filter-effect for the blue light occurs and the photocathode has maximum sensitivity in this region. We believe that a decrease in the thickness of the layers would result in thin semitransparent photocathodes having properties similar to those we have found for thicker layers illuminated from the vacuum side. Such photocathodes would be suitable for use in the blue and in the near ultra-violet region with an input window of uviol glass or quartz.

SOME PROPERTIES O F Sb-K-Rb-Cs

PHOTOCATHODES

355

REFERENCES 1. Sommer, A. H., Rev. Sci. Inetrum. 26, 726 (1955). 2. Dvo?ak, M., Slaboproudd Obzor 24, 377 (1963).

DISCUSSION Can you explain the fact that sensitivity in terms of pA/lm differs by a factor of 3 whilst in terms of spectral efficiency it differs by a factor of more than 20, considering the area of integration when illuminating from different sides? M. DVOBAK: The difference is due to the different position of the maximas of both curves. Because of the relative spectral energy distribution of the tungsten source (2850"K),the value of the luminous sensitivity will not be much influenced by the high quantum efficiency a t 3300A. The increase of the luminous sensitivity for the illumination from the vacuum side is caused by the higher quantum efficiency values of the photocathode in the region of 4000-5400 A. w. E. TURK: When this new photocathode is used in image orthicons, is the sticking effect eliminated? M. DVOBAK:This photocathode has not yet been used in image orthicons. R. w. AIREY: What were the optical transmissions of the layers of antimony evaporated during cathode activation? M. DVOBAK: Three layers of antimony were evaporated. The optical transmission of the fist layer was 75%. The second and third antimony layer were evaporated before sensitizing with rubidium and caesium respectively, their thickness being such as to cause the sensitivity of the photocathode to drop to zero in both cases. E. H. WAGNER:


Decay of S-20 Photocathode Sensitivity Due to Ambient Gases R.W. DECKER Westinghouse Aerospace Division, Baltimore, Maryland, U.S.A.

INTRODUCTION As described in another papert in this volume this laboratory is developing$ a large-image electronographic camera for recording highresolution images directly on photographic film with photoelectrons. The criterion for the life of the camera and indeed whether the camera will even operate initially is: can the partial pressures of contaminating gases be maintained a t such a low level that the photoresponse of the photocathode will remain high? I n a study program conducted over the last two years, the contaminating effects of pure gases on the response of the trialkali 5.20 photocathode were measured. These effects were then related to the outgassing of materials used in the camera. All photosensitive surfaces that are responsive t o low-energy photons in the visible spectrum are very reactive. The sensitivity of a photosurface is altered by two general forms of contamination. First, deposits of foreign material on the substrate on which the photosurface is formed will cause a poor photocathode. I n this study it was assumed that if a good photocathode is formed and remains stable, the effect of substrate contamination is negligible. Second, the sensitivity of a photosurface can be changed by gaseous materials depositing on the surface, these coming from the outgassing of materials in the device, the vapor pressure of materials, and leaks. We have measured the outgassing of materials, and in the design of the electronographic camera have reduced this by a suitable choice of materials and processing technique. The materials tested were processed in the manner appropriate to their operation in the camera, and the outgassing was measured with

t See p.

19.

2 Under contrmt from the U.S. Air Force. 367

358

R. W. DECKER

a residual gas analyzer. Stainless steel, Viton-A, Teflon, Mylar, and Kodak SO- 159 film were studied and the predominant gases found were hydrogen, nitrogen, methane, oxygen, carbon dioxide, carbon monoxide, and water vapor; those having the highest partial pressures being hydrogen and water vapor.

EXPERIMENTAL PROCEDURE

A special vacuum system, shown in Fig. 1, was constructed for measuring the residual gases. This system consists of three separate pumping stations, a gas bottle manifold,l n test chamber, and a residual gas analyzer, all interconnected by bakeable valves. A diagram of the arrangement is shown in Fig. 2. Each atation can be isolated from the

FIQ.1. Vacuum system for measuring residual gases.

DEUAY OP S.20 PHOTOCATHODE SENSITIVITY

359

others, the bakeable leak-valve between the gas manifold and the test torr against chamber being able to maintain a vacuum of atmospheric pressure. First the system was evacuated and baked t o obtain a low background pressure, typical total pressures obtained being in the range of

0 Valve Standard leaks

FIQ.2. Diagram of vacuum system.

to tom. At such pressures the most persistent gas was hydrogen. Second, a photosurface was formed in the photodiode attached t o the test chamber; a photodiode assembly is shown in Fig. 3. After the 5-20photocathode was formed, it was cooled and its stable response was measured using a calibrated light source. The sensi-

FIQ.3. Photodiode assembly.

360

R . W. DECKER

tivity was typically > 120pA/lm. I n theinitial studies only the luminous sensitivity was measured, but later the spectral response was also measured with a grating monochromator. The spectral response of the S.20 was found to be a more sensitive measurement of the contamination than the total response because the response in the red decreases more rapidly than that of the mid-range of the visible spectrum. The response of the photocathode was then monitored continuously as the partial pressure of the pure gas admitted was increased in steps. The gases used wcre hydrogen, nitrogen, methane, oxygen, carbon dioxide, carbon monoxide, chlorine, and water vapor. The purity of the gas that was leaked into the system from the gas manifold was measured by the residual gas analyzer. I n the first experiment the gases admitted as contaminants were hydrogen, nitrogen, and methane, the photoresponse of the photodiode was initially 190pA/lm, and the background pressure was below torr total pressure in the measuring system. The first gas admitted through the leak valve was hydrogen. The residual gas analyzer was used to measure the partial pressure as well as the purity of the gas: at a pressure of 2 x torr of hydrogen the system pressure was essentially due t o pure hydrogen. The pressure was then increased in steps by opening the leak valve, and a t each step 5 min was allowed for the gas to react with the photosurface. The response of the photocathode was monitored by recording the cathode current while it was being illuminated by a 0.1-lm calibrated light source. From two to four measuroments were made with each gas t o make sure that the results were reproducible.

RESULTS The reaction a t a surface is proportional to the number of gas molecules that strike it. The number of monolayers of gas striking a surface in a 5-min period as a function of pressure of the gas is represented in Fig. 4, and this indicates that a t pressures greater than 8 x lo-* torr, in 5 min a t least one monolayer of gas has struck the photosurface t o react with it.2 Since semi-transparent 5.20 photocathodes are very thin, any change in response due to contamination of the surface should be noticeable within 5 min. A composite graph of the reactions of hydrogen, nitrogen, and methane with the 190-pA/lm S.20 photocathode is shown in Fig. 5 . The hydrogen and nitrogen pressures were increased to 2 x torr without a permanent change in the luminous response, although the response was lower while the pressures were high. With methane the response decayed when the pressure reached 1 x torr (a contamination rate of 2000 monolayers/min).

DECAY O F 9.20 PHOTOCATHODE SENSITIVITY

36 1

A similar graph showing the reaction when oxygen, carbon monoxide, and water vapor were introduced to a 160-pA/lm S-20 phutotorr of cathode is shown in Fig. 6. At a partial pressure of 5 x oxygen the response decreased to 110 pA/lm in 10 min, but recovered

Pressure ( torr 1

4. Vacuum contamination rate as a function of pressure.

Fra. 5 . Photocathode contamination by hydrogen, nitrogen and methane. Continuous line partial pressure; broken line, photocathode sensitivity. Initial response 190 eA/Im.

to 128 pA/lm after the oxygen was removed. Contamination by carbon monoxide commenced at a pressure of 5 x 10 - 5 torr, the initial response of 182 ,uA/lm falling to 145 pA/lm. When water vapor was introduced, the contamination began a t a much lower partial pressure, and in order to obtain a more sensitive measure of the effect on the photo-

362

R. W. DECKER

response, each partial pressure was maintained for 1 h. Figure 6(c) shows that the decay in the response began at 2 x torr. More complete data on each gas contaminant is given in Table I, which lists the gas contaminants, the initial photoresponses, and the partial pressure at which a change in response occurred for each experiment. As the Table shows, the partial pressure a t which the response first begins to decrease is quite reproducible for each contaminant. In the study of vacuum materials it became obvious that all the gases in the electronographic camera except water vapor can easily be held

E

2

1

I

I

I

I

\

I

I

1

-- .

I

I

1-200;

t

a

"-*O

'

~

O

Time (min) Oxygen

'

~

' 0~

'O

~'

~

Time (min) Carbon monoxide

b ' ; 0

'

'

a'

6

Time ( h ) Water

FIG.6. Photocathode contamination by oxygen, carbon monoxide and water vapor. Continuous line, partial pressure of contaminant; broken line photocathode sensitivity.

to a much lower partial pressure than is required to prevent the photocathode response decaying. Of particular interest for the electronographic camera, the gases evolved from Mylar and Kodak SO159 film are only water vapor and hydrogen, and since the latter is not detrimental to the photocathode, the water vapor is the problem. The detrimental effect of this may be due to the OH radical because neither hydrogen nor oxygen affect the photocathode to the same extent. I n the electronographic camera water vapor is difficult to remove, but we have found that by vacuum-preprocessing SO-159 film, the partial pressure of water can be maintained at 5 x torr with small holding pumps. To determine more about the effects of gaseous contaminants on 5-20 photocathodes, we used a grating monochromator to measure the

363

DECAY O F 5.20 PHOTOCATHODE SENSITIVITY

spectral response of the photocathode a t each partial-pressure increment. The results obtained when a sensitive photocathode was exposed t o water vapor are shown in Fig. 7. Note that it is the longer wavelength response that is first affected by the contaminant. The response TABLEI Summary of contaminant tests

Contaminant

Initial photoresponse (pAllm)

Decay first noted (torr)

x

10-4

x

10-4

Remarks

100 190 63

2

Nitrogen

100 190 182

5 x 10-5 2 x 10-4 2 x 10-4

Response recovered

Methane

108 180 182

I x 10-4 1 x 10-4 1 x 10-5

Slight recovery

92 155

1 x 10-6 6x

Partial recovery

Carbon dioxide

96 165 130

2 x 10-8 2 x 10-8 5 x 10-8

No recovery

Carbon monoxide

165

i 30

5 x 10-6 5 x 10-5

164 140 125 87

8 7 5 5

165

2 5

Hydrogen

Oxygen

Chlorine

Water vapor

158

2 x 10-4 2

Response recovered

Slight recovery

x lo-‘ x 10-7 x 10-7 x 10-7

x

x

10-7 10-8

Slight recovery

No recovery

before contamination (labelled “pre-contamination” in the figure) was stabilized for 10 days a t a background pressure of less than 2 x torr. The life of an S.20 photocathode can be estimated from these data. At a partial pressure of water vapor of 1 x lo-’ torr, the fall in the peak response will not exceed 10% over a period of a t least 10h.

R. U’. DECKER

364

However, at the same partial pressure of water vapor the spectral response curve A (Fig. 7 ) shows that the red response a t 8500 A has changed from 3.8 mA/W to 2.4 mA/W in 45 min, a change of 33%. At this wavelength the decay rate is 1.8 mA W-’ h - l , and from Fig. 4 the contamination is 132 monolayers in 1 h a t lo-’ tom, so that the decay can be expressed as 1-4 x mA/W per monolayer. With this figure one can then compute an estimated life for the photocathode if one knows the partial pressure of water vapor. For example, let a response of 1 mA/W a t 8500 be the minimum response that will -

-

Partial Time(min1 pressure (torrl

Pre- contamination

-

A 45

IxIO-~ . _c

-

[r

L

l 4000 l

l 5000l

l 6000 l

l7000 l

8000 l l

9000 l

l

IC

I00

Wavelength

(8. )

FIG.7. Spectral response of photocathode after exposure to wattor vapor at various partial pressures and for various times. Luminous eensitivity: initial, 350 pA/lm; after 45 min at torr, 330 pA/lm.

give the sensitivity required and let the initial sensitivity of the photocathode be 4 mA/W a t 8500 A. If the partial pressure of the water vapor were 5 x l o d 8 torr, the contamination rate (from Fig. 4 ) would be 72 monolayera per hour. Then using the decay conversion figure previously derived (1.4 x mA/W per monolayer) the decay rate is 72 x 1-4 x M 1 mA/W per hour or a 3-h life for the photocathode. It is important t o note that the life of the photocathode depends on the spectral region of the incoming radiation. I n the above example the red response has been taken as the criterion; however as can be seen from Fig. 7 the decay in the remainder of the visible spectrum is much less and in fact the luminous sensitivity would still be 90% of the initial value.

DECAY OF 5-20PHOTOCATHODE SENSITIVITY

365

The most significant result of this study is that the sensitivity of a photocathode is selectively degraded by contaminants and is most degraded in the longer wavelengths. Hence contaminants may be only a small problem for images in a particular region of the visible spectrum but a much larger problem in other regions. ACKNOWLEDGMENTS The author wishes to thank Mr. J. S. Knoll and Mr. W. H. Beck I11 for their work on the fabrication of test photosurfaces and measurements of residual gases.

REFERENCES 1. Hastings-Raydists Inc., “Calibrated Gas Leaks”. Specification Sheet 904, June 1966. 2. Dushman, S. and Lafferty, J. M., “Scientific Foundations of Vacuum Technique”. Wiley, New York (1962).

DISCUSSION w. o. TRODDEN: The decay curves shown appear to represent a dynamic rather than a static equilibrium state, in that poisoning gases were pumped out while the photoemission was still decaying rapidly. It might be more realistic to show the steady otate corresponding to a gas partial pressure. Have you any such data? R. w. DECKER: The system is a dynamic system at equilibrium. That is, the gas flow in is equal to the gas being pumped out. The partial pressure is measured and maintained a t the constant values reported for the period of time of 5 min or 1 h as shown in Figs. 5 and 6. M. HIRASHIMA: Could you tell me about what kind of gas is most predominant among the residual gases in electronic image devices, apart from cesium vapor? If I remember correctly, it is believed that CO is most predominant in ordinary receiving tubes. R. w. DECKER: The type of gas present in a particular tube depends on the construction. In all stable tubes the partial pressure of the gas present must be far below the level that caused contamination as reported here. In metal tubes, hydrogen and helium are present with hydrogen predominating. In tubes with an electron gun, CO predominates. E. ZIEMER: In your work on photocathode decay as a result of the introduction of various gases, did you study the effects of materials such as Viton A and Teflon. I f so, what was the effect? R. w. DECKER: The degassing of Viton A and Teflon was studied. The details of the degassing are extensive and very dependent on the preparation of the materials prior to exposing to a photocathode. If properly prepared, Teflon will not affect an 5-20photocathode. Viton-A has always caused the photocathode to decay. H. BACIK: Do you know of any similar work being done on S.1 photocathodes? R. w. DECKER: Y e s , it is now under way but there are no results to report.


A New Technology for Transferring Photocathodes P. DOLIZY and R. LEGOUX Laboratoirea d’&lectronique et de Phyeique Appliqude, L~med-Brduannea,Prance

INTRODUCTION The conventional method of preparing the photoemissive layer in a photoelectric tube is by the thermal evaporation of antimony and alkali metals from sources mounted inside the tube. Some of these sources must be located either directly in view of the photocathode, so as to achieve uniform evaporated layers, or in such a manner that the molecular diffusion rate is relatively high. It is easily understood that such sources prohibit the use of quite a number of electrode configurations, particularly those in which the electrodes are placed close to the photocathode. Moreover, with the conventional method, the alkali vapour permeates throughout the tube during processing, and very often interacts chemically with the surface of the electrodes. This can enhance undesirable effect,s such as leakage currents, field emission, and parasitic photoemission. It can be said that, as a general rule, the performance of the photocathode is dependent to a large extent on the internal structure of the tube. These troubles, often encountered with conventional tube processing methods, can be avoided by the so-called “transfer technique”. The main features of this technique are as follows. (1) The substrate, on which the photoemissive layer is to be formed, is isolated from all other component parts of the photoelectric tube, so that it alone is exposed to the materials evaporated during the processing of the photocathode, (2) The several evaporation sources are grouped together in an auxiliary enclosure, which is located in front of the substrate and is maintained in that position throughout the processing of the photocathode, after which it is removed. (3) The photoemissive layer is only transported to its final position in the tube, if, after stabilization and cooling, it has the required photoelsctrio properties. 367

368

P. DOLIZY AND R . LEGOUX

DESIGNO F THE “TRANSFER” TUBES The design of tubes to be made by the transfer process is strongly influenced by the way they are to be pumped and processed in the special chamber. The envelope of a tube to be made by this process is divided into two parts, each of them containing elements of the tube itself. One of these two parts is often the input window of the tube, and in many msea, this is used as the photocathode substrate. The two parts are eventually joined together by metal rings (Pig. 1). However, the tube is first pumped and the photocathode processed with the two parts separated. The two parts are then joined by an indium seal which is made by pressing the two rings together, one of these having a groove Glass window and Dholocathode substrate

Metal rings

FIG.1. Indium compression seal a t front of phototube. The electrode structure in the lower half of the tube is not shown.

to hold the indium and the other a tongue. Indium is particularly suitable for the purpose because of its great malleability, its low vapour pressure at high temperature, and its great resistance to oxidation. The seal-off tip, which is usual on convendional tubes, is of course no longer required.

THETRANSFER EQUIPMENT A photograph of transfer equipment capable of handling photocathodes up to 120 mm in diameter is shown in Fig. 2. A cross-section of the cylindrical transfer enclosure is shown in Fig. 3. This enclosure is made of stainless steel and the top, which is in the form of a bell-jar of glass or metal, is removable. The two halves of the tube to be processed can be seen in the figure. One half of the tube, pre€erably that including the photocathode substrate, is fastened to the upper part of

NEW TECHNOLOGY BOR TRANSFERRING PHOTOCATHODES

369

the enclosure with a clamping ring held by two columns. The other is set up in the lower part of the enclosure on a movable table guided by the columns and connected to a hydraulic press. At a later stage this lifts up the lower half of the tube for the seal to be made to the upper half. All the evaporation sources necessary for the preparation of the photocathode are grouped together in an auxiliary enclosure which is open at the top and can be moved sideways by a second mechanism.

FIG.2. The transfer apparatus.

The system is pumped first with a cryogenic pump (zeolite and liquid nitrogen) to about 6 x l o T 4torr and then either with an oil diffusion pump having two refrigerated baffles in series or with a Penning ion pump. The ultimate vacuum as measured by the ionization gauge shown in Fig. 3 is generally of the order of torr. The entire equipment is outgassed for 16 h during each pumpiiig cycle by heating coils at a temperature of 260°C, while the transfer bell-jar is brought to between 400°C and 460°C. When the bake-out is terminated, the photoemissive layer is processed in the conventional way but with modifications to allow for the large volume of the processing chamber. The pressure is then about torr.

370

P. DOLIZY AND R. LEQOUX

FIQ.3. Croae-section of the transfer encloaure.

NEW TECHNOLOGY FOR TRANSFERRING PHOTOCATHODES

37 1

During the processing, the sensitivity of the photocathode is monitored using the light from a tungsten filament lamp placed above the bell-j ar, As soon as the photocathode has cooled to a temperature of between 40°C and BO'C, and has stabilized, then, providing that the required photoelectric properties have been achieved, the tube is ready to be closed. The pressure at this stage is torr. In the closing process the sensitizing enclosure is first moved to a lateral position in the belljar leaving room for the tube to be lifted up (Fig. 4). The two parts of

Processing enclosure

1 I

FIG.4. Showing how the processing enclosure is moved to one side and the tube is closed.

the tube are then joined by compressing the indium seal. The photocathode is now in its operating position. Air is let into the bell-jar and the tube can now be removed from its mechanical supports and is ready for use. OF THE METHOD ADVANTAGES The advantages of the method are numerous. Apart from those that have already been noted, it avoids the poisoning of component parts of the tube by the physical and chemical action of alkali vapours during the photocathode processing. This applies in particular to the following; the fluorescent screens of image tubes, surfaces which must be highly insulating, and electrodes which must be free from field emission or photoemission. Photocathodes made using the method are comparable to those

372

P. DOLIZY AND R. LEQOUX

obtained by conventional methods. Trialkali photocathodes with sensitivities higher than 200 pA/lm have been achieved on glass and metal substrates. The uniformity of the photocathodes is also improved because the evaporation sources can be placed at greater distances from the substrate than are possible in conventional tubes.

FIQ.5 . High current photodiode. Anode t o cathode spacing 2 mm.

After a photocathode has been processed it can be tested before its introduction into a tube, and if inadequate it can be rejected, thus avoiding throwing away the entire tube. A most important aspect of the process is that it makes possible the construction of tubes in which the dimensions and shape are such that photocathode processing would not be practicable by conventional methods. As an example of such a tube, Fig. 5 shows a high current

FIR.6. High-speed shutter tubes.

NEW TECHNOLOGY FOR TRANSFERRING PHOTOCATHODES

373

photodiode in which the distance between the cathode and anode is 2 mm. Figure 6 shows a family of fast shutter tubes for which the photocathode diameters range from 40 mm to 120 mm, and the cathode to phosphor distances from 2 mm to 10mm. These are more fully described in another paper in this volume.?

DISCUSSION s. MAJUMDAR: 1. What current can you draw from the high current photodiodes? 2. In your image tubes, does the cathode performance deteriorate after many hours of operation? J . GRAF: 1. These diodes deliver a linear rosponse up to 10 A for an applied voltage of 3 kV and a pulse of 1 psec. The saturation current is 20 A. 2. We have not seen any change in cathode performance on a tube tested for 5 h continuous running. I t must be noticed that these types of tubes aro designed only for pulse operation. M. ROME: What improvements are found in dark current by the use of the transfer technique? Would you please compare the dark current of tubes with the same photocathode, (e.g. type 5-20),of similar sensitivity, which differ only that some are conventionally prepared and others by the transfer method? J. GRAF: For the same 5.20 sensitivity in two photomultiplier tubes, one conventionally prepared and the other by the transfer method, the transfer tube has a dark current nearly hundred times lower than the conventional tube. R. DECKER: 1. Is the tube isolated or just shielded from the photocathode processing chamber? 2. Is it possible to process more than one cathode a t one time? 3. How far do the bellows have to deflect to make a seal? J . GRAR: 1. The body of the tube is only separated from tho sensitizing enclosure containing the alkali dispensers. There is not a tight separation between the dispensers and the body of the tube. 2. Yes, it is possible to process simultaneously several photoemissive layers in the transfer equipment. 3. The deflexion of the bellows depends upon the height of the sensitizing enclosure which is itself a function of the diameter of the cathode to be processed. The maximum dellexion is 40 cm. R. AIREY: Have you attempted to effect an indium seal by bringing the parts together in the presence of the molten metal, thus eliminating the need for a high pressure hydraulic ram? J . GRAF: The sealing of tubes by means of a molten metal joint can be done provided that the gases desorbed by the joint inside the tube do not spoil the characteristics of the photoemissive layer and of the tube structure.

t See p. 989.


Improvements to Photocathodes for Pulse Operation €3. R. C. GARFIELD and J. R. FOLKES

Engliah Electric Valve Co. Ltd., Chelmeford, England and

B. T. LIDDYt Department of Pure and Applied Phyeics, Queens University, Belfast, Northern Ireland

INTRODUCTION For pulse-operated image tubes it is necessary to deposit the photocathode on a transparent, low resistance substrate to prevent image distortion due to saturation effects.$ Because of its high overall sensitivity the trialkali photocathode is a desirable one to use. Unfortunately the usual substrate material, tin oxide (Nesa),is unsuitable for use with the trialkali photocathode as it is rendered highly resistive by reaction with sodium. I n addition, the resulting photosensitivities are usually lower than normal. As an alternative to Nesa as a conducting substrate a fine mesh embedded in the glass face-plate has been investigated. The resistance, sensitivity, spectral response and pulse performance have been measured for cathodes deposited on such substrates. TRIALKALI PHOTOCATHODE NESASUBSTRATES An investigation was made to determine at what stage during the processing the Nesa was attacked. The results, which are given in Table I, show that the substrate resistance was relatively unaffected by potassium, but that after the introduction of sodium it had increased by several orders of magnitude. The photo-cathode sensitivities were low and the deep orange colour of transmitted light suggested that the antimony component in the Nesa had been attacked.

t Temporarily at the English Electric Valve Co. Ltd. $ See p. 999.

376

B . R. C . OARFIELD, J. R . FOLKES AND B. T. LIDDY

376

TABLEI Variation of resistivity during trialkali processing on an antimony-doped Nesa substrate Initial Resistance

Resistance after K

Resistance after Na

a/0

Q/ 0

a/0

6

68

72

3.2 x 106

10

45

50

1-6

Tube No.

x lo6

Final resistance Q/

0

4 x 105 1.8

x 106

Photocathodes were also processed on Nesa substrates prepared from a formulation containing no antimony. The results, which are listed in Table 11, show that a major resistance change again occurred during the sodium stage, indicating that the tin oxide itself was attacked. TABLEI1 Variation of resistivity during trialkali processing on an undoped Nesa substrate Initial resistance

Resistance after K

Resistance after Na

Final resistance

a/0

Tube No.

a/0

14

140

150

7.6 x 104

2 x 106

15

150

164

2.1 x 104

4 x 104

16

140

140

2 x 105

5.4 x 105

0

Q/O

MESH SUBSTRATE For this substrate a fine metallic mesh is embedded in a Pyrex glass disc -3 em in diameter and the photocathode is formed on this. Copper meshes of 750 mesh/in. and 55% transmission, having square apertures of length -25 pm and bar thickness of -9 pm are used. The glass disc is prepared by sandwiching it with a stretched mesh between two flat carbon blocks and applying a pressure of approximately 20g/cma. The assembly is heated in a reducing atmosphere (90% N,, 10% H,) t o a temperature of between 750 and 800°C and is then cooled slowly ( 3 to 4 h) to room temperature. Scanning electron microscope studies indicate that during the forming process the glass flows up through the interstices thus firmly embedding the mesh. The surface resistivity of this substrate is less than 0.1 Q/o.

377

PHOTOCATHODES FOR PULSE OPERATION

Trialkali photocathodes were prepared on these substrates by standard processing techniques. Sensitivities were in the range 60 to 120 pA/lm, which corresponds to 100 to 200 pA/lm for the cathodes in the clear apertures, after allowance is made for the light obscured by the mesh. The spectral response was normal.

PULSE PERFORMANCE OF A TRIALKALI PHOTOCATHODE ON A MESH To investigate the pulse performance of these mesh photocathodes, a number of small test diodes of a co-axial design were prepared. I n these, the substrate disc was mounted close to the glass window and after photocathode processing a flat metal anode disc was moved into position a few millimetres from the photocathode surface and secured. For pulse evaluation the photocathode was illuminated by the light from a ruby laser (having a pulse half-width of about 50 nsec).

/+

I I I

i

I

I 005

I 0.10

I 0.15

I 0.2c

Relative light intensity

FIQ.1. Pulse performance of trialkali photocathode on glass substrate.

Uniform illumination was ensured by placing a diffusing screen a t an appropriate distance from the photocathode while the light intensity was varied by inserting suitable neutral-density filters. I n order t o correct for variations in laser output between pulses, a fraction of the output was arranged t o be incident on a second reference diode (E.M.I. Type 9648B with an S.10 photocathode on a metal substrate),this being operated a t a current density of less than 1 mA/cm2 t o ensure linear operation. The output signals from both diodes were fed simultaneously into a Tektronix 556 double-beam oscilloscope via 5042 terminations. The peak current obtained from the test diode was correlated with the P.E.1.D.-A

14

378

B. R. C. QARFIELD, J . R. FOLKES AND B . T. LIDDY

corresponding incident light intensity. Figures 1 and 2 show the performance of two such diodes. It can be seen that in the case of the trialkali on glass, saturation effects are evident a t current densities of less than 1 mA/cm2,whereas for the trialkali on a mesh these appear a t a current density of about 300 mA/cm2. Calculations using the ChildLangmuir equation indicate that the saturation evident a t the highest current densities was probably due to space-charge limitations.

c

m

t

u

Relative light intensity

FIQ.2. Pulse performance of trialkali photocathode on mesh substrate.

ANALYSISOF MEsIr PHOTOCATHODE Assuming a circular cathode configuration, a uniform emission current density and a uniform surface resistivity, analysis shows that the potential drop V a t the centre of the photocathode is given by the formula:

v=- p l4R2’ where p is the surface resistivity of the photocathode, I is the emission current density and R is the radius of the photocathode. Thus for p = lo7 Q/o, R = 10 pm and I = 1A/cm2, the potential drop a t the centre of each photocathode element is N 2.5 V.

PHOTOCATHODES FOR PULSE OPERATION

379

MESH PHOTOCATHODE IN A PRACTICAL DEVICE A mesh trialkali photocathode of 100 pA/lm sensitivity has been prepared in an English Electric Valve Company shutter tube, type P 8 5 6 . I This has been operated under pulse conditions in a Hadland ‘Imacon” camera. No detailed tests have as yet been made but preliminary observations are favourable.

CONCLUSIONS There are now two substrates available for processing semi-transparent photocathodes with saturation thresholds at high current densities: the conventional low resistance Nesa substrate, and the mesh as has been described. For use with trialkali photocathodes, the mesh substrate is necessary as the resistivity of the Nesa substrate is seriously affected by reaction with the alkali metals used in the photocathode processing. ACKNOWLEDGMENTS The authors would like to thank Mr. R. A. Chippendale of English Electric Valve Company and Prof. D. J. Bradley of the Queen’s University of Belfast for encouragement during the course of this work, and tho Managing Director of the English Electric Valve Company for permission to publish this paper. One of us, B. T. Liddy, is supported by a Postgraduate Studentship from the Northern Ireland Ministry of Education.

REFERENCE 1. Huston, A. E. and Walters, F., In “Advances in Electronics and Electron Physics”, ed. by J. D. McGee, W. L. Wilcock and L. Mandel, Vol. 16, p. 249. Academic Press, New York (1962).

DISCUSSION J. D . MCCIEE: 1. Has the

use of the mesh to improve conductivity of the trialkali cathode improved the stability of the cathode? 2. Have you observed electrolytic effects in photocathodes? B. R. C . GARFIELD: 1. It is known that the trialkaliphotocathode,in common with most other types of photoemitter, fatigues during continuous operation at high current densities ( > N 1 pA/cma). The probable mechanism of this is electrolysis (due to the voltage drop across the layer produced by the photocurrent) of the constituent alkali metals through the layer, leading to departure from stoichiometry in the bulk, and loss of caesium from the surface (Miyazawa, H. and Fukuhara, S., J . Phys. SOC. Japan 7,645 (1962);Garfield, B. R. C. and Thumwood, R. F., Brit. J . Appl. Phye. 17, 1005 (1966)). Thus fatigue is substantially reduced when photocathodes are formed on conducting substrates (Nesa or solid metal) (Linden, B. R., I n “Advances in Electronics and Electron Physics”, Vol. 16, p. 311. (1962)). The mesh substrate is expected to be equally effective in this respect. 2. In the case of pulse opcration a t current densities of the order of 1 A/cm2, it seems likely that i t is the total charge drawn in any given time

380

B. R . 0. OARFIELD, J. R. FOLKES AND B . T. LIDDY

(hours or days) which will determine the extent of fatigue. In the present work the cathodes had only minimal pulse usage, and therefore electrolytic effects and fatigue were not expected to be appreciable. This appears to be the case as sensitivities have remained stable to within about 10%. M. K. KEY: Is it possible that the non-linear behaviour observed with the trialkali photocathode that was not deposited on a low resistance substrate is due not only to the voltage drop associated with the surface resistivity, but also to a fundamental solid-state property of the photocathode? This suggestion arises from observations with image converter cameras in our 1aboratory.t A. s . JENSEN: As a partial answer to the question by M. H. Key, in our laboratory about four years ago, we were working with a photostorage tube having a grating storage target (Jensen, A. S. et al., I n “Advances in Electronics and Electron Physics”, Vol. 22A, p. 155. (1966)) which is essentially a transmissive storage grid. When we illuminated the photocathode with a small spot of light, the current passing through the storage grid was such a function of the light intensities as could be explained by assuming that the photocathode surface potential pattern was a positive peak. Thus while this does not say that there is no solid-state effect, at least the positive swing of the photocathode voltage as a result of the voltage drop across its surface is a part if not all the explanation of the behaviour. B. R . C. GARFIELD: Our ideas concerning the non-linear behaviour observed with the trialkali photocathode on a high resistance substrate, are in agreement with those of A. S. Jensen, i.e. the current is space-charge limited due to a reduction of the effective applied potential between anode and cathode, resulting from the transverse voltage drop across the surface. More recent measurements of saturation current as a function of applied voltage, tend to confirm this. This does not, however, rule out the possibility that internal saturation effects may occur in photocathodes a t extreme levels of illumination and very fast pulse conditions.

t See p. 999.

Some Getter Materials for Caesium Vapour M. HIRASHIMA and M. ASANO Department of Electronic Engineering, University of Electro-Communicatione, Chofu City, Tokyo,Japan

INTRODUCTION Following the discovery that gold can be used as a getter material for caesium vapour,l efforts have been made t o find other suitable materials. I n general, materials to be used as getters for caesium vapour should be either metals that can react readily with the vapour or compounds containing elements that can so react. I n this connexion, Kienast and Verma2 have reported on the results of their exhaustive experiments with compounds of alkali metals and either copper, silver, or gold. As for caesium compounds, however, only AuCs is mentioned in their paper. Among the many compounds, oxides seem to be the most promising for gettering caesium. Not all oxides can, however, be used asgetters for caesium vapour, because the first requirement is that they must be reduced readily by caesium a t a temperature which is relatively low, but is not lower than the baking temperature. Furthermore, the final compound of caesium should preferably not be a good photoemitter. The value of the free energy of formation is a measure of the usability of oxides as getter materials for caesium vapour. To be precise, the absolute value of the free energy of formation of any oxide must be smaller than that of caesium oxide, and in fact, the smaller, the better. For this reason the useful oxides seem likely to be limited t o those tabulated in Table I.3 Of these, the oxides of tin and lead were probably used by Zworykin and his co-workers for gettering excess caesium when processing the 1conoscope.* It is also known among tube engineers that an Aquadag coating can adsorb a large quantity of caesium, but this has some drawbacks as a getter. I n the present paper the results of some experiments carried out with the oxides of nickel and iron will be described briefly. Copper oxide 381

382

M. IIIRASHIMA AND M. ASANO

TABLE I Free energy of formation of oxidest Element

Oxide

CAESIUM Tin Iron Nickel Cobalt Lead Copper Carbon Palladium Silver Gold

cs20 SnO FeO NiO coo PbO cu,o

co

PdO Ag2O Au203

25OC

Free energy 500°C

koal -79 - 61 - 59 - 52 - 51 - 45 - 35 - 33 - 20 -2 19

koal - 74 - 49 -51 -43 -43 - 34 - 28 -43 -18 +5 $31

+

1ooo"c kcal - 69

-36 - 45 - 32 - 34 - 22 -21 -54 -15 12 44

+ +

t After E ~ s t e i n . ~ was also studied without obtaining a satisfactory result. The possibility of using carbon monoxide as a caesium getter will also be briefly discussed, for this is one of the most common residual gases found in electronic tubes using oxide-coated cathode^.^

EXPERIMENTAL TECHNIQUE Oxides of Nickel and Iron As was the case with gold,l throughout the present experiments a silica spring balance was used t o measure the change in weight of each specimen during oxidation in an atmosphere of oxygen and also during the reduction of the oxide by caesium vapour. The sensitivity of the silica spring balance used was 0.2 mg per mm elongation. The silica spring was hung in the tubular quartz reaction chamber which was 30 mm in diameter and about 600 mm in length, as shown in Fig. 1. To the bottom end of the spring was hooked a strip of thin plate, in the case of the nickel, and a piece of thin wire, in the case of the iron. The quartz tube was connected to a conventional hard glass tube by a graded seal. It was heated from outside by means of a long electric furnace consisting of four sections, one of which was provided with a longitudinal slit through which the elongation of the spring could be measured using a telescope. The only differences from the experiments with gold were that a vessel containing potassium permanganate was connected to the reaction chamber via a tap so that oxygen could be introduced into the

GETTER MATERIALS FOR CAESIUX VAPOUR

383

reaction chamber in the oxidation stage, and pure hydrogen could also be fed into the reaction chamber through a small palladium tube provided with an electric heater. The hydrogen was necessary for removing the oxides that were almost always formed on the surface of the iron specimen during annealing in vacuum, even though the vacuum torr. pressure was only N

Electric furnaces

c

./ Sealed off after Cs admission

-

To H c linder 2

Y

To rotary pump

-

FIG.1. Schematic diagram of equipment for measuring the oxidation of nickel and iron, and the reduction of their oxides.

The preparation and treatment of the specimens were carried out in the following manner. As mentioned above, the nickel specimen was a strip. This wag 0.053 mm thick, 2 mm wide, and 15 to 20 mm long; it weighed about 17 mg, and the purity of the nickel was 99.99%. The specimen of iron was a wire 0.199 mm in diameter and -50 mm in length, weighing about 12 mg, and the purity was as high as 99.9987%. N

384

M. HIRASHIMA A N D M. ASANO

The ends of the iron wire were welded together to form a ring, and it was thus possible to heat it by induction in an atmosphere of hydrogen in order t o remove surface contamination prior t o oxidation. After degreasing, all the specimens were annealed for 30 min in a vacuum of torr a t the respective recrystallization temperatures, 760°C for nickel and 720°C for iron. The specimen of nickel was then oxidized a t 680 to 700°C in an atmosphere of oxygen a t a pressure of 63 torre%When the desired thickness of nickel oxide was attained, the oxygen was pumped out, and the pressure reduced t o about lo-? tom; meanwhile the electric furnace was kept at the same temperature in order t o bake the tube and was then lowered t o room temperature. Caesium was admitted into the reaction chamber from an appendage in which there was a nickel capsule containing a measured quantity of caesium chromate mixed with silicon as a reducing agent. The reaction chamber was then sealed off from the vacuum system. The temperature of the electric furnace was again raised t o a temperature of around 200"C, and was kept a t a constant value during the reduction period. I n the case of iron, the specimen was heated by induction from outside the quartz tube t o a temperature of about 780°C for 16 min in hydrogen a t a pressure of 3 torr. This removed surface contamination. The hydrogen pressure was then reduced to torr; oxygen was admitted into the reaction chamber a t a pressure of 63 torr, and the specimen was oxidized a t 540°C.' After the oxidation the oxygen was pumped out, and the temperature of the furnace was lowered t o room temperature; caesium was admitted into the reaction chamber, which was then sealed off from the vacuum system, as in the case of nickel. The temperature of the furnace was then raised t o the desired temperature for the reduction period.

Carbon Monoxide An ionization gauge? was used as the reaction chamber to see whether carbon monoxide could be used as a getter for caesium. As shown in Pig. 2 the ionization gauge was connected to a sealed ampoule containing caesium, and to an appendage which could be cooled by running water; an exhaust tube led to a carbon monoxide reservoir and t o a vacuum pump via suitable taps. The ionization gauge was evacuated to a good vacuum, and the envelope and electrodes thoroughly degassed. Then a small quantity of carbon monoxide was introduced into the gauge while the appendage was cooled with running water. When the pressure of carbon monoxide (measured by the gauge) was of the order of torr, the ionization gauge, the caesium ampoule and the appendage were sealed off, as indicated in Fig. 2. After the

t Toshibe, W-1.

GETTER MATERIALS FOR CAESIUM VAPOUR

385

caesium had been admitted into the gauge from the ampoule (by breaking the small glass tip with the iron ball), the whole apparatus was heated in the electric furnace to about 50°C, and was kept at this temperature for 10min. The gauge was kept at this temperature and the appendage was then cooled with running water, thus letting the caesium vapour in the reaction chamber condense into the appendage. After a further 10min the pressure of the carbon monoxide was measured, and it was found that the pressure had decreased as a result of the chemical reaction between the carbon

Iron

A

To

CO reservoir

FIG.2. Apparatus for measuring the reaction of caesium with carbon monoxide.

monoxide and caesium. The same procedure was repeated four times, the total reaction time being about 40min. I n conducting this experiment, however, it was difficult to keep the tube temperature constant a t a value as low as 50°C, because the tube temperature rose owing to the heat radiated from the filament of the ionization gauge while the pressure of carbon monoxide was being measured.

EXPERIMENTAL RESULTS Nickel Oxide ( N i O ) A typical example of the oxidation of a specimen of nickel and its reduction in an atmosphere of caesium is shown in Fig. 3; the weight

386

M. HIRASHIMA AND M. ASANO

gain of the specimen was 0.593 mg/cm2 after oxidation for 250 min. The number n, of oxygen atoms that have combined with the nickel per unit area of the specimen t o form nickel oxide can be calculated from the gain in weight: n, M 2.24 x 1019. On the other hand, the weight gain of the same specimen during reduction by caesium for 205 min was 5.8 mg/cm2, and the calculated number of caesium atoms reacting with the nickel oxide to form

0

1

2

3

4

0

1

2

3

Time ( h )

FIG.3. Typical curves of oxidation and reduction runs on a specimen of nickel.

caesium oxide and nickel is n2 M 2.63 x 1019 per unit area of the specimen : NiO 2 Cs + Cs20 Ni. (1)

+

+

Since two atoms of caesium combine with one atom of oxygen to form a molecule of caesium oxide, Cs,O, the number of the oxygen atoms that have been effectively used is equal to n2/2, assuming that all the caesium atoms have been converted into caesium oxide. Thus we can see that the availability of the oxygen for gettering caesium in this case is given by n 2 / 2 n , w 59%. I n Fig. 4, there are shown the curves for five specimens of nickel, the reduction by caesium being carried out a t different temperatures. I n the case of specimen No. 5 the oxide layer became detached after the reduction run had started. The experimental data obtained with the five specimens of nickel out of 15 specimens prepared a t the start are tabulated in Table 11. The fraction of the oxygen atoms reacting with caesium atoms to form caesium oxide (the valuea in the last

387

Nickel

----------N0.12

(

193OC)

Time ( h )

FIG.4. Reduction by caesium of specimens of niclrel oxide at various temperatures,

column of Table 11)is plotted as a function of the reaction temperature in Fig. 5 . It is seen from this figure that, as expected, the higher the reaction temperature the larger this fracbion becomes, but if the temperature is too high the oxide layer may become detached as was the case with the specimen a t 252°C (see Fig. 4). TABLEI1 Experimental data for five specimens of nickel FracWeight tion Thickgain Oxida- Reducof Surface nesst after Speci- Weight tion tion oxygen after of oxide dearea temp. temp. effect(mg) men (cm? OxIdation layer oxida("C) ively ("C) (pm) tion used (mg/cm2) (mg/cm2) Weight gain

(YO)

NO.3 NO.5 NO.6 No. 8 NO. 12

16.54 17.65 17-45 17.25 16.79

0.7294 0,7746 0.7642 0.7428 0.7227

0.593 0.384 0.186 0.279 0.317

t Calculated using the value:

3729 2410 1169 1750 1990

5.800 0.694 1.198 2.566

690 690 690 677 677

200 252 170 180 190

58.9 22.5 25.9 48.3

1 pg/cma = 62.9 A (Gulbransen and Andrews).

388


p

zc

6ot

;I -5

Nickel

60-

4040

0

C

.t

2020

t

0

,x

Iron

-Xd-

I

1 I70

I0

I

I

190

210

Reaction temperature

("C)

FIG.5. Fraction of oxygen atoms reacting with caesium to form Cs,O, as function of reaction temperature.

Oxides of Iron (FeO and Fe,O,) A typical example of the oxidation and reduction of a specimen of iron is shown in Fig. 6. A comparison of the reduction curve of Fig. 6 with those in Figs. 3 and 4, shows that there is a marked difference in the shape of the curves. This difference may be partly attributed to the different form of the specimens of the two materials, viz. a strip of thin plate in the case of nickel, and a thin wire in the case of iron. However, the reason why the reduction curve for iron has two different slopes may be explained by the fact that two kinds of iron oxide have been formed8s0during the oxidation run, namely, FeO and Peso,; the first part of the reduction curve corresponds to the reduction of the TABLE111 Experimental data for four specimens of iron

Specimen

No. 2 No. 3 No. 8 NO. 10

Surface area

Weight Oxidation gain after temperaoxidation ture

(em2)

(mg/cma)

("C)

0-313 0.311 0.330 0.335

0.903 0.820 0.491 0.658

540 540 540 539

Fraction Weight Reduction gain7 of oxygen tempera- effectively after ture reduction used (mglcm") ("C) (%I 1.391 0.831 0.510 0.580

t After reduction for about 130 min.

174 155 145(?) 148

9.3 6.1 6.3 5.3

389

GETTER MATERIALS F OR CAESIUM VAPOUR I

I

Iron

I

I

I

Oxidation run at 700°C in 63torr 0

c

I

I

2

If

-

Reduction run at 17OoC in 2 x lo-, torr Cs

0

I

2

3

4

Time ( h )

FIa. 6. Typical curves of oxidation and reduction runs on a specimen of iron.

Fe,O, layer and the second part of the curve to the reduction of the FeO layer, which lies beneath the former. When the reaction temperature was less than 170°C, this effect could not be observed so clearly (see Fig. 7 ) .

Time ( h )

Fro. 7. Reduction by caesium of specimens of oxides of iron at various temperatures.

390


The experimental data are summarized in Table I11 for four successful specimens out of the nine initially prepared. The fraction of the oxygen atoms comprising the oxides of iron which is effectively used t o form caesium oxide is plotted as a function of the reaction temperature in Fig. 5 . The chemical reactions are:

+ 8 Cs

Fe,O, and FeO

--f

+ 2 Cs

+ 3 Fe, + Fe.

4 Cs,O

+ Cs,O

(2)

(3)

Carbon Monoxide? I n Fig. 8, there is shown an example of the results of the reaction of carbon monoxide with caesium vapour. The pressure of the carbon monoxide just after the seal-off of the reaction chamber was 3 x lo-, torr, and on cooling the appendage with running water the I

after seAl-off

‘

I

I

I

I

the appendage was cooled

.. 00

30 50

$0

2 ?? ? .

f

W

t

?O 0

10

20

30

40

50

60

70

80

p: 9

10 I

Time(min1

FIG.8. Change in the pressure of carbon monoxide following reaction with caasium vapour a t about 50°C.

t Just after the present paper was read on Sept. 18, 1968, Dr. S. J. Hellier of S.A.E.S. Getters, Milan, Italy, kindly brought to our notice a paper entitled: “Alkali Metal Generation and Gas Evolution from Alkali Metal Dispensers” by P. della Porta, C. Emili and S. J. Hellier, which was presented at the Conference on Tube Techniques held in New York on Sept. 17-18. 1968. 111 this paper similar results to ours are described.

GETTER MATERIALS FOR CAESIUM VAPOUR

391

pressure decreased to 1 x tom. After the carbon monoxide had reacted with the caesium vapour,

co + 2 cs -+ cs,o + c,

(4)

the pressure of the carbon monoxide decreased t o torr. And after the four successive reactions mentioned above, the total reaction time being about 40 min, the pressure of the carbon monoxide was found to decrease finally to 9 x torr a t about 50°C. I n the above experiment, it should be understood that the caesium vapour was being used as a getter material for carbon monoxide, but from the other viewpoint the carbon monoxide could also be used as the getter material for the caesium vapour. The latter usage seems to be more advantageous than the former, since two atoms of caesium react with one molecule of carbon monoxide to form a molecule of caesium oxide, leaving an atom of free carbon which is thought t o be active as a getter for gases other than caesium vapour.

CONCLUSION I n view of the fact that different forms of specimen were used in the present experiments for the two cases of nickel and iron, it seems t o be a little dangerous to draw a conclusion concerning the relative merits of nickel oxide, the oxides of iron, and also gold,l as caesium getters. To be precise, the comparison should be made with specimens of similar form and similar size. If specimens in the form of wire are to be used, for instance, they should be wires of the same diameter. I n fact', although the nickel oxide appears from Fig. 5 t o be better than the oxides of iron as a getter material for caesium vapour, a comparison of the curves shown in Fig. 4 with those in Fig. 7 shows that the absolute value of the weight gain due to the absorption of caesium is larger in the case of the oxides of iron for the same reaction temperature (see curve No. 6 in Fig. 4 and curve No. 2 in Fig. 7). I n order to decide whether these materials might be used with advantage as caesium getters in the production of photocathodes, i t seems to be necessary first to obtain information concerning the effects of various gases upon the sensitivity of photocathodes. ACKNOWLEDGMENTS The writers wish to take this opportunity to express their hearty thanks to Professor S. Sakui and Dr. K. Satoh of the Department of Mechanical Engineering for their generosity in supplying tho specimens of pure nickel and iron used in the present experiments. The writers are also very grateful for the ever-willing assistance in various phases of the present research given by T. Yoshino, K. Utagawa, Y. Takegawa, N. Takasaki and others.

392


REFERENCES 1. Hirashima, M. and Asano, M., In. “Advances in Electronics and Electron Physics”, ed. by J. D. McGee, D. McMullan and E. Kahan, Vol. 22A, p. 643. Academic Press, London (1966). 2. Kienast, G. and Verma, J., 2. Anorg. Allg. Chem. 310, 143 (1961). 3. Epstein, L. F., Ceramic Age, p. 37 (April, 1954). 4. Zworykin, V. K. and Morton, G. A., “Television”, p. 272. Wiley, New York (1948). 5. Thompson, B. J. and North, D. O., RCA Rev., p. 373 (Jan. 1941). 6. Gulbransen, E. A. and Andrew, K. F., J . Electrochem. SOC.101, 128 (1954). 7. Gulbransen, E. A., Tram. Electrochem.SOC.81, 327 (1942). 8. Paidassi, J., J . Metals 4, 536 (1952). 9. Kubaschewski, 0. and Hopkins, B. E., “Oxidation of Metals and Alloys”, p. 4. Butterworths, London (1963).

DISCUSSION B. R.

c. GARFIELD: Did you observe any changes in the weight measurements

during oxidation due to thermomolecular flow effects around the silica spring microbalance? M. HIRABHIMA: No, we did not observe such thermomolecular flow effects. We admitted oxygen from the vessel containing KMnO, into the reaction chamber gradually; it took about 1.5 min to attain a pressure of 63 tom.

New Approaches to Photoemission at Long Wavelengths P. SCHAGEN and A. A. TURNBULL Mullard Reeearoh Laboratories, Redhill, Surrey, England

INTRODUCTION Gradual improvements in processing techniques during recent years have made it possible to produce photocathodes of the alkali-antimonide type with high quantum yields to visible radiation. A requirement still exists, however, for highly efficient photocathodes with thresholds at longer wavelengths. An important example of the need for such cathodes is in image intensifier tubes used for visual observation at night, where, in the absence of moonlight, the only illumination is provided by the night sky. Extensive measurements of the intensity and spectral distribution of this radiation, illustrated in Fig. 1, have indicated strong emission components in the near infra-red, originating in the airglow.

$i 10"

I

I

I

I

1

I

1

5

1.8

Wavelength (pin)

FIG.1. Typical example of distribution of radiation in the night-sky. 393

i 0

394

P. SCHAQEN AND A . A. TURNBULL

An image intensifier tube, using a photocathode with a high quantum yield in the infra-red, would thus in principle achieve a considerably improved performance. A limiting factor in this respect is, however, the increase in background current which accompanies such a shift of photocathode threshold into the infra-red. This is caused not only by thermionic emission but also, and even more important, by theincreasing fraction of the black-body radiation from the scene and equipment to which the photocathode will thus respond. Such currents reduce the apparent contrast in the image, and calculations have shown that most of the possible improvement in performance would already be obtained if a photocathode could be employed with high quantum yield out to a threshold of about 1.25 pm. Very little would be gained by increasing the threshold to 1.6 pm, whereas from then on a rapid deterioration in performance would take place under typical nocturnal conditions. For this important application it therefore seems that a photocathode which is highly efficient out to a threshold of 1.25 pm would suffice.

APPROACHES TO THE DESIGN OF PHOTOCATHODES Figure 2 shows schematically the simplified energy band diagram of a p-type semiconductor ph0tocathode.l As a result of absorbing a quantum of the incident radiation, electrons from the valence band are

FIG. 2. Simplified energy band diagram of p-type semiconductor photocathode.

lifted into the conduction band, from where they must be able to escape into the vacuum. It is essential for any efficient photocathode with a certain desired threshold that there should be strong absorption of light out to that wavelength. This is ensured by selecting a semiconductor with a bandgap that is smaller than that corresponding to this threshold, the actual threshold in this case being determined by the surface barrier. The conditions for a high quantum yield are: firstly, the escape depth of the excited electrons must be large compared with the absorption length; secondly, a large fraction of the absorbed quanta must excite

PHOTOEMISSION

Ar LONG WAVELENGTHS

395

electrons to energy levels above that of the vacuum; thirdly, the probability of escape of electrons from the surface must be high. These three requirements for the efficient extraction of excited electrons into vacuum lead to two distinctly different approaches. 1. Reduced Surface Barrier

The first approach is to lower the surface barrier by providing the semiconductor with a suitable surface coating of low work function. Some examples of this approach are illustrated in Figs. 3 and 4.

FIQ.3. Surface barrier reduction by Cs monolayer coverage: (a) the general case, (b) the special case GaAs-Cs.

Ec

t

I-4eV

t-fc

LEF E"

E" (a)

(b)

FIG.4. Surface barrier reduction by application of Cs-0 layer: (a) GaAs-Cs-0: C s and 0 coverage by few monolayers, (b) GaAs-Cs-0; coverage of GaAs by Cs-oxide.

One of the most effective methods of reducing surface barriers is to apply a coating of approximately one monolayer of caesium to the atomically clean surface. This reduces the effective surface barrier, indicated as a drop in the vacuum level in Fig. 3 (a),to a value approximately 1.4 eV above the Fermi level. Figure 3 (b) shows the special case where a p-type semiconductor has been chosen with an energy gap equal to the reduced work function. An example of this is the wellknown system, gallium arsenide and caesium, first reported by Scheer and van Laar2 in 1965, and now a subject of study in many research laboratories. Because the vacuum level coincides with the bottom of

396

P. SCHAOEN AND A. A. TURNBULL

the conduction band, excited electrons still have sufficient energy for escape, even when they have thermalized. For this reason the escape depth may be as long as the diffusion recombination length. An even lower work function is possible by the application of both oxygen and caesium. This principle, applied to tungsten as a substrate, was first described by Kingdon3 more than 40 years ago. Applied to GaAs, this results in a negative electron affinity, illustrated in Fig. 4(a), due to a work function of about 1-1 eV. As expected, the principle can be applied to materials of smaller bandgap than GaAs to obtain high yields at even longer wavelengths.* The lack of stability with time of such thin surface coatings creates a serious problem in a practical device. As one of the authors has r e p ~ r t e d this , ~ can be improved by using caesium oxide in the form of a layer, a few tens of Angstroms thick, which appears t o leave the escape probability unaffected. Such a layer also reduces the effect of remaining contaminations on a semiconductor surface which is not atomically clean, because use is now made of the inherently low work function of caesium oxide itself (about 1.3 eV), as illustrated in Fig. 40)). It appears quite likely that other coatings will be found in the near future which reduce the effective work function to a value as low as about 1.0 eV. If this could be applied in practice to a suitable semiconductor with a bandgap energy of also 1.0 eV, a photocathode could result with its spectral response extending to 1.24 pm. Such a cathode, as has already been indicated, would be very attractive for image intensifiers used for nocturnal observation. 2. Applied Internal Field

The second approach to the problem of enabling the escape of the excited electrons, is the application of a controlled internal field. The electrons, which are drawn across this region, should gain enough energy from the field to exceed the work function. If they do not lose too much of this energy again on their further way to the surface, they will be emitted. Depending on the construction of the field layer, one can distinguish between photoemission based on p-n junctions, and on the tunnel effect. Some examples of field-assisted emission are shown in Fig. 5. I n diagram (a) a p-n junction is biased in the reverse direction. Excited electrons are thus accelerated by the field across the space-charge region towards the surface. Figure 5 (b) illustrates how, at least in principle, a hetero-junction could be employed, such as for example zinc selenide on a germanium substrate. The large bandgap of ZnSe would tend to minimize the internal reverse current. A further example,

397

PHOTOEMISSION AT LONG WAVELENQTHS

shown in Fig. 5 (c), is of a low bandgap semiconductor separated from a thin metal coating by a thin insulating layer. I n this case, as in the previous examples, the photoexcited electrons would be accelerated by the controlled internal field in the intermediate region, but tunnelling may here be one of the processes involved. I n all three cases the main problem likely to be met will be the losses in electron energy due to collisions. These will reduce the effective escape probability and lead to a low quantum yield, thus making this approach less promising than the first one.

(a I

(bl

(cl

FIG.6. Examples of the concept of field-assisted photoemission: (a) biased Si p-n junction, (b) biased Ge-ZnSe hetero-junction, and (c) biased Si-insulator-metal sandwich structure.

Other possibilities for obtaining photoemission at longer wavelengths that could also be considered, are conventional photoemitters in series with biased photoconductorsa or in combination with wavelength c o n v e r ~ i o n ,and ~ * ~with these the threshold could in principle be moved further still into the infra-red. For the important application considered here, however, the approaches already discussed appear to be more promising.

EXPERIMENTAL RESULTS I n surveying possible future developments in the field of long wavelength photoemission, several approaches have been discussed. In our laboratories, work has been proceeding for some yeam on one of these approaches, that of reducing the surface barrier of a semiconductor, namely GaAs, by the use of caesium, and caesium and oxygen. Considerable progress has been made in this work which is aimed, initially at least, at the development of a practical photocathode based on GaAs. Stable high-yield photoemission has been observed not only

398

P. SCHAOEN AND A . A. TURNBULL

from vacuum-cleaved samples of GaAs, but also from air-cleaved and vapour-deposited epitaxial samples. Figure 6 shows the spectral response of photoemission from GaAs doped with zinc carriers/cm3) and coated with caesium oxide. The GaAs was in two forms: a vacuum-cleaved crystal (curve A), and a vapour-deposited epitaxial layer (curve B).

p-

'"4!0

115

i.0

i.5 3:o Photon energy ( e V )

3!5

3

FIG.6. Examples of spectral dependence yield of GaAs-Cs-0. A, Vacuum-cleaved crystal of GaAs, Zn doped with 2 x 10lB carriers/cni3. B, 24-pm-thick layer of GaAs, Zn doped, with carriers/cm3, grown epitarially on GaAs substrate by vapour transport.

Further work is in progress on the study of photoemission from thin polycrystalline layers of GaAs on transparent substrates. Photoemissive yields are still a factor of some five times lower than those from good examples of single-crystal GaAs. In conclusion it can be said that, although such layers show promise for application in devices, it is not yet possible at this stage to estimate their ultimate potential.

REFERENCES Spicer, W. E., J . Appl. Phya. 31, 2077 (1960). Scheer, J. J. and van Laar, J., Solid State Commun. 3, 189 (1965). Kingdon, K. H., Phy.9. Rev. 24, 510 (1924). Bell, R. L. and Uebbing, J. J., Appl. Phye. Letters 12, 76 (1968). Turnbull, A. A. and Evans, G. B., Brit. J . AppZ. Phye. 1, 155 (1968). Auphan, M., Boutry, G. A., Brissot, J. J., Dormont, H., Perilhou, J. and Pietri, G., Injra-red Phys. 3, 117 (1963). 7. Kruse, P. W., Pribble, F. C. and Schulze, R. G.,J. Appl. Phya. 38, 1718 (1967). 8. Phelan, R . J., Proc. Inat. Elect. Electronics Engrs. 55, 1501 (19B7).

1. 2. 3. 4. 5. 6.

Gallium Arsenide Thin-film Photocathodes C. H. A. SYMS Services Electronics Research Laboratory, Baldock, Hertfordshire, England

INTRODUCTION Many experimental GaAs-Cs photocathodes have been prepared in recent years by cleaving, under vacuum, apiece of acceptor-doped singlecrystal gallium arsenide. A little caesium is then allowed t o condense on the freshly exposed face, which, when illuminated, yields a very high photocurrent. Values between 500 pA/lm and 1000 pA/lm have been reported.1-4 Such a photocathodeis thusmany timesmore efficient than the multialkali type with, moreover, a sensit,ivity extending into the longer wavelength region of the spectrum with high efficiency. The long wavelength threshold for GaAs is approximately 0.9 pm (1-4eV) but sensitivity can be further extended towards 1 pm by the use of semiconducting compounds with slightly smaller energy bandgap, for example In,Gal-,As. The high conversion efficiency and the infra-red sensitivity are of great importance in device development. Photocathodes formed from cleaved single crystals are not, however, very suitable for incorporation into photomultipliers and are virtually excluded from use in image tubes by the difficulties that would be encountered in the design of a practical folded optical and electron optical system. Therefore, a programme has been undertaken a t SERL to determine whether satisfactory photocathodes can be formed from thin films of GaAs deposited on t o transparent substrates. I n the course of the experimental work it has been shown that GaAs layers can be deposited on polished sapphire substrates. Some of these layers have then been caesiated t o provide photocathode emission efficiencies comparable with present commercial devices.

EXPERIMENTAL ARRANGEMENT Figure 1 shows the experimental arrangement and the disposition of the zinc metal used t o provide the p-type doping, the gallium metaI source, and the substrate position within the heated resction tube. 3QQ

400

C . H. A . SYMS

The gallium is held a t 950°C and the substrate a t approximately 675°C. Variation of the position o f the zinc changes the resistivity of the condensed layer, in a controlled way, in the range o f 0.01 t o 1000 Rcm. Deposition on the substrate commences when the hydrogen is bubbled through the AsCI, t o react with the hot gallium. The volatile components then condense in the cooler region of the reaction tube near the substrate to provide a layer growth rate of 1 pmlmin.

To fume cupboard exhaust

FIQ.1. Vapour-phase reaction tube and gas flow system.

After chemical etching, the layers are located on a mounting bebween two angled stainless steel mirrors in a vacuum chamber, as indicated in Fig. 2. Reflexion and transmission spectral response measurements may thus be made by reflecting the incident light on to the specimen with either of these mirrors, one mirror acting as an electron collector. The whole system is vacuum-baked overnight, and the caesiation process is then initiated by crushing a capsule of caesium metal in the copper-tube side-arm. The arm is warmed to increase the caesium vapour flow while monitoring the photoemission current. When the current has been substantially enhanced, the specimen chamber is separated from the main system t o allow spectral response studies. A small ion pump a t the rear of the system makes tfhe specimen chamber self-contained and portable. Further quantities o f caesium, and then oxygen, are applied t o the specimen during observation of the spectral sensitivity, until

GALLIUM ARSENIDE THIN-FILM PHOTOCATHODES

401

maximum emission is achieved. Normally, only a small (1 mm diameter) area of the photocathode is examined at a time in this way. Quantum efficiency curves are then evaluated for the specimen by reference t o the response of a calibrated silicon photodetector over the same spectral range. The luminous efficiencies are then computed. Some of these results are shown in Fig. 3. The upper curve is characteristic of that to be expected from a cleaved single-crystal specimen

FIG.2. Specimen vacuum chamber.

and is included for reference. The lower group has been obtained from several polycrystalline layers examined as reflexion photocathodes. The long wavelength threshold of emission in each case is near the optical absorption edge of GaAs a t 1.4 eV, indicating thc effectiveness of surface caesium in lowering the surface work function. The numbers here represent the pA/lm values of the respective specimens. These values are approximately the same as bhose obtained a t SERL from chemically cleaned and subsequently caesiated single-crystal GaAs specimens. This indicates that the photoemission is limited by surface

402

C. H. A. SYMS

/ Cleaved single crystal

/

1.3 1.4 1.5 1.6

Incident 1.8 2radiation 0 2.2(eV)

24

2.6

3

FIQ.3. Spectral resporise (reflexion) of five caeskted polycrystalline GaAs layers on sapphire substrates and of a caesiated cleaved single-crystal of GaAs.

effects and that the internal properties of the layers are reasonably satisfactory.

TEANSMISSION PHOTOCATHODES The problem with transmission photocathodes is the preparation of very thin, uniform samples. Chemical etching techniques have been used to prepare layers with thicknesses of only a few microns, but it is difficult t o achieve a uniform thickness over an area of 1 cm2, the peripheral area usually being considerably thinner than the central area. Figure 4 indicates some preliminary measurements of the spectral response of thinner layers operated as transmission photocathodes. The diagram shows the results for two such samples. Both the transmission and reflexion spectral response for the same small area of each sample is indicated. The long wavelength limit is again a t the optical absorption cdge of bulk GaAs. Tho overall yield is, however, much

GALLIUM ARSENIDE THIN-FfLM PHOTOCATHODES

403

lower. Note that a t the longer wavelengths the emission is higher for the transmission curve, which is possibly the result of the better optical impedance match from vacuum into the GaAs through the sapphire substrate. I

I

l

l

I

I

I

I

I

w, Transmission

lo-' B

Incident radiation ( e V )

Fra. 4. Spect'ralreaponse (reflexion and transmission) of two caesiated polycrystelline GaAs layers on sapphire substrates (A and B) and of a caesiated cleaved single crystal of GaAs (reflexion only).

THEORETICAL MODEL As the semiconducting properties of GaAs are relatively well understood it is possible to consider the photoelectric yield of a photocathode of this material from a fundamental theoretical standpoint. A simple model for the processes of photon absorption and subsequent electron emission has been examined. Incident light generates electrons throughout the GaAs. Some will diffuse to the caesiated surface and escape, others will recombine within the layer. The photoexcited electron density in the GaAs can bo

404

C. H. A. SYMS

calculated as a function of depth into the material from the GaAs surface on which the light is incident. Consider a layer of GaAs of thickness dx a t a distance x below the illuminated surface, under conditions of steady illumination. The electron density n(x) then has the equilibrium value given by the continuity condition,

where j ( x ) is the net electron current density in the thin layer, g(x) is the electron generation rate in the layer and e is the electronic charge. The three terms in Eq. (1) then represent, respectively, the electron diffusion (as there is no appreciable electric field in the layer the current is diffusion limited), electron generation and finally electron recombination (T is the electron lifetime). Now

and g(x) = (1 - R)Au exp(-ux) ,

(3)

where D is the electron diffusion coefficient, R is the optical reflexion coefficient, A is the incident light flux in photons, and u is the absorption constant of the GaAs. Substitution of Eqs. (2) and (3) in Eq. ( 1 ) yields a second order differential equation which may be solved for the electron density n(x). For a complete solution appropriate to a GaAs layer of finite thickness the boundary conditions a t the surfaces must be determined. I n the simple model this has been done by allowing the surface electron densities to recombine a t acceptor surface states a t a certain rate which is represented by appropriate values of the surface recombinationvelocity coefficient. Thus, a t the illuminated surface, j(0)= en(O)S,, and a t the electron emitting surface, j ( t ) = en(t)S,, where t is the total thickness of the layer and S , and S , are surface recombination-velocities. The solution to the equation is then ?%(X) =

(1 -B)AaT ( U Z L 2 -1 )

(8,+ 4 ( S , + D / L ) e - ' " ( S , +D/L)(S,-D/L)etiL

-k

(8,+uD) (8,--D/L)etiL -(S,-D/L) (8,+aD)e-at - (8,+ D / L ) (8, -D/L)etiL

[@,--D/L) (8,+D/L)e-i'L

UALLIUM ARSENIDE THIN-FILM PHOTOCATHODES

405

This expression has been evaluated for n(t)wherc t = 2-5 pm and using for the parameters the representative values for GaAs in Table I. TABLEI

R, reflexion coefficient A , photon flux from a black body source a t 2850°K between co and 1.350%' giving 5.95 x lo3 lm/cma a, optical absorption coefficient T, minority carrier lifetime D , diffusion coefficient L = (DT)'/', diffusion length S1= Sp,surface recombination-velocities5

=

0.16

= 4.9 x 1020photons cm-2sec-1 = 2 x 104cm-1 = 10-9sec = 150 cma/sec =

3.87 x cm x lo6 cm/sec

=2

The value of the photocurrent can then be calculated by using an effective surface recombination-velocity due t o emission, Si, with an assumed value of 2 x 104cm/sec. This value has been estimated from measurements of the photoemissive quantum yield near 2.0 eV for reflexion photocathodes (Fig. 3) and with the above value of S,, assuming that the photogenerated electrons either recombine through Average crystal li le

/

FIG.5 . Idealized polyorystalline layer.

the surface states or are emitted. Then j ( t ) = en(t)S, = en(t)(8; +&) where Si represents the actual recombination rate through the surface states. As &/S; 4 1 the emission current, j e ( t ) ,M en(t)Sh which, after evaluation, gives a photocathode efficiency of 1184 pA/lm. Figure 5 shows how the effects of polycrystallinity can be included by formulating an expression derived from the geometrical aspects of an average crystallite. In this very simple model the electrons are considered to be generated a t the centre of the base of the cylindrical crystallite. Solid angles subtended from this point then give the probability of the electron reaching the emitting surface. Recom-

406

C. H. A . SYMS

bination, and therefore, loss, of the electron a t the orystallite boundary is assumed. The function, t , f(b)= 1 (ba t 2 ) l i 2 is the required probability and the product j ( e ) f ( b )gives an indication of the yield expected from a polycrystalline thin film. Table I1 gives the yields for practical values, near unity, of the ratio blt.

+

TABLEI1

0.5 1 3 6 10

130 343 810 990 1070

The conclusion, therefore, to be drawn from the experimental work is that a thin film GaAs-Cs photocathode is possible, the present results giving efficiencies of about 1 pA/lm in transmission and 100 pA/lm in reflexion. Theoretical calculations indicate that there is no reason why a transmission photocathode of this material should not have a greater yield and a sensitivity to longer wavelengths than the S.20 multialkali type. ACKNOWLEDGMENTS This paper is published by permission of the Ministry of Defence.

REFERENCES 1. Scheer, J. J. and van Lam, J., Solid State Cornrnun. 8, 189 (1965). 2. Turnbull, A. A. and Evans, G. B., Brit. J . Ap p l. Phys. 1, 155 (1968). 3. Eden, R. C., Stanford Electronics Laboratory Technical Report No. 5221-1 (May 1967). 4. Uebbing, J. J., Bell, R. L. and Spicer, W. E., private communications. 5. Vilms, J., Stanford Electronics Laboratory Technical Report No. 5107-1 (Nov. 1964).

DISCUSSION w. E. TURK: 1. What is the reason for choosing sapphire as a substrate? 2. Can the material be prepared on glass? 3. What is the purpose of the chemical etch before caesiation? C. H. A . S Y M S : 1. Sapphire was chosen as the substrate material as it has good optical transmissionoverthe wavelength rangeof interest, it is inert a t thesubstrate temperatures used for deposition of the layer and may be readily obtained prepared to optical qualities. There is the possibility of epitaxial deposition on to the material. 2. The material can be prepared on glass. 3. Chemical etching is used to remove the surface layer of &As which is likely to be heavily oxidized

GALLIUM ARSENIDE THIN-FILM PHOTOCATHODES

407

in the period between layer preparation and caesiation. Samples aro also chemically thinned for transmission photocathode studies. c . w. BATES, JR: What chomical etcher did you use on your GaAs? c . H. A. SYMS: Generally a solution of 0.2% Rr in methanol. E. EBERHARDT: Were your films exposed to air before caesiation and, if so, for how long? c . H. A . SYMS: All the films are exposed t o air between preparation and caesiation; some for several weeks. All the samples that are to be caesiated are etched immediately beforehand, but there is still an effective exposure t o air for at least several minutes before the pressure in the vacuum system is reduced. B . R. GARFIELD: The semiconductor band model for OaAs on which the high photoemissive yield is usually explained has zero effective electron affinity a t the surface. This would suggest that the thermionic emission will be high compared with conventional alkali-antimonide photoemitters which have a finite electron affiity. Have you made measurements of the dark current? c . H. A. SYMS: Our routine current measurements are limited by the current meter sensitivity which is about 10-13A. We have not been able to measure any dark current a t this sensitivity. It is not yet clear from where the major contribution to the dark current will originate and the zero effective electron affiity model is not appropriate for the evaluation of the influence of localized surface states and other defects of the physical structure. I n principle, it is possible to adjust the energy bandgap of a compound like InGaAs so that a finite electron affinity remains for the structure as a whole. Without doubt, it is important that reliable measurements of the dark current of a GaAs-Cs photoemitter be made.


6tude de I%mission Photoklectrique des Structures M6tal-Isolant-M6tal P. VERNIER, P. HARTMANN, G. NIQUET et M. TEPINIER

Laboratoires de Photodlectricik! des Facultds des Sciences de Dijon et de Besanpon, Prance

INTRODUCTION Nous avons eu l’occasion, lors du troisiltme Symposium,’ de montrer que la camBra Blectronique est un instrument de choix pour l’etude des faibles courants Blectroniques. Lifshitz et Musatova ont montrB que le courant Bmis par des structures m&al-isolant-m6taI, convenablement polarisBes, pouvait &re renforcB par un Bclairement de la structure. Nous avons entrepris d’htudier systBmatiquement ce phBnom&neB l’aide de la camera Blectronique. La sensibilite des plaques photographiques et la sBcuritB apportde par la connaissance des points d’oh sont Bmis les Blectrons procurent des QlBmentsnouveaux. PREPARATION DES STRUCTURES

EMISSIVES

torr, une Nous Bvaporons, en premier lieu, sous un vide de bande d’aluminium de 2 mm x 60 mm et dont 1’6paisseurest de l’ordre de 1000 d. Nous formons ensuite une couche d’alumine par oxydation, d’abord par mise zt l’air et transport dans une Btuve portke zt 100” C. Lea rBsultats ainsi obtenus &ant irrbguliers, par la suite, nous avons operB cette oxydation en faisant une rentrBe d’oxygltne sec et en chauffant zt 150°C la couche d’aluminium dans la cloche oh Btait eEectuBe 1’6vaporation de l’aluminium. L’Bpaisseur de la couche d’alumine obtenue varie de 60 d zt 160 d suivant la tempdrature et le temps d’oxydation. Nous Bvaporons ensuite une contre-electrode d’or sur l’aluminium oxydB. Nous formons ainsi un condensateur. Pour limiter les risques de courts-circuits et pour Btudier I’influence de 1’6paisseurde la contre-Blectrode, nous avons rBalisB sur une meme bande d’aluminium oxydh une aerie de couches d’or d’un millimlttre P.E.1.D.-A

409

15

410

P.

VERNIER, P. HARTMANN, a. NIQUET

ET M. TEPINIER

de largeur, isolees les unes des autres (Fig. l(a)). Les rksultats de ces experiences feront l’objet d’une autre publication. Pour les Btudes photo6lectriques’ nous nous sommes au contraire efforcds de realiser des structures aussi grandes que possible (Fig, l ( b ) ) .

-

/ “ 0 ,-

2

A

Fro. 1. Structures M-I-M: (a)avec six contre-6lectrodeset (b) avec une grande contreBlectro de.

GTUDEDE LA

CONDUCTION DES STRUCTURES

Une mesure de capacitB nous fournit d’abord une Bvaluation grossibre de 1’6paisseur de l’alumine, en admettant pour celle-ci une constante didlectrique relative E~ = 8. La mesure de la caracteristique courant-tension entre 1’6lectrode d’aluminium et la contre-Blectrode d’or a BtB effectuke A l’aide d’un oscillographe cathodique en utilisant le montage schematise sur la Fig. 2. ris !rice

Adoptoteur d‘imp&dance‘-

I

I

Oscillogrophe

.

* *

I

H 0

II

A, space charge acts to lower gain and to saturate the output current. The normalized transit time change due to space charge is dependent upon the emission energy and angle, and upon the potential difference due to the space charge between channel center and edge. The actual channel diameter should have no effect. A series of calculations was to lO-'cm, and the made for channel diameters in the range results are plotted in Fig. 4. The normalized transit time change due to a current of I = l o v 4A per channel is plotted vertically. No variations are seen.

511

SPACE CHARGE IN CHANNEL MULTIPLIERS

1-

?Ole L

E f

10

.3 e

-

B

I

Pulse amplitude

FIG.3. Spectrum from same dynode as in Fig. 2, but with primary electron energy 2.7 keV. 1000

-

I

Al2O3(5001)+ A L ( 2 0 0 % ) + K C l (500;) 10.0 keV:

-

n

= 2.19

Pulse amplitude FIG.4. Spectrum from Name dynode a8 in Fig. 2, but with primary electron energy 10.0 keV.

519

STATISTICS OF T.S.E. EMISSION

Pulse amplitude

FIG.6. Spectrum from dynode of bulk-densityKCI 2600 d thick. Energy of primary electrons 6.1 keV: mean secondary yield 9.9.

large pulses are present, corresponding to the emission of electron groups of high multiplicity. We are able to resolve these puIses as peaks up to n = 24, but the distribution continues beyond this in exponential form as far as we have been able to observe (n M 60). The occurrence of events involving such large values of secondary multiplication is not, 100

5.1 keV:

-

n

11.5

P ( 0 ) =0.22

0 x

0

10 > .+ 0 a4

LT

I

Pulse amplitude

FIG.6. Spectrum from dynode of bulk-density CsI, 600 d thick. Energy of primary electrons 6.1 keV: mean secondary yield 11.5.

520

W . L. WILCOCK AND D. E. MILLER

as has been suggested, in some way associated with space charge, because a freshly prepared dynode gives the equilibrium proportion of large pulses as soon as the primary bombardment begins. Figure 5 illustrates the effect, or rather the lack of effect, on the distribution which results from an increase in the thickness of the potassium chloride layer. Similarly, Figs. 6 and 7 are sample spectra from dynodes I

AL20,(5008)t

A L ( Z O O 8 ) t C s I (2000%)

6.0keV:

._ n

= 13.0

P ( 0 ) ~0.27

n= I

fB

6

g0

a

Pulse amplitude

FIG.7. Spectrum from dynode of bulk-density CsI, 2000 A thick. Energy of primary electrons 6.0 keV; mean secondary yield 13.0.

in which the emitting material is caesium iodide. These examples, all of which relate to primary energies near that for maximum secondary yield, have mean values of n much higher than are obtained from potassium chloride dynodes of the usual structure (cf. Fig. 2); but the form of the distribution is essentially the same in all cases, and remains so over the whole range of primary energies we can explore. To sum up, as far as our observations go we find the extraordinary

52 1

STATISTICS OF T.S.E. EMISSION

result that the most likely outcome of a primary encounter, if there is any emission a t all, is always the emission of a single secondary; and the chances of the emission of larger numbers of secondaries then follow in descending, and approximately geometrical, progression. This result does not fit easily into the framework of the customary model of the secondary emission process. This model, which can be made to explain well enough the observed dependence of yield on primary energy, assumes that the dissipation of this energy leads to the production of secondary electrons in the interior of the dynode, each of which diffuses and may escape from the surface. Fluctuations of the number of secondaries emitted then arise through the number v of internal secondaries produced, and the probabilities pl,.. .pv that these secondaries will escape. Our observations appear to require an approximately geometric probability distribution for either v or the average of the p’s. There are undoubtedly fluctuations of v, but such data as are available on total energy losses in the passage of electrons through thin films does not suggest that the distribution of v has the required insensitivity to primary energy and dynode thickness. Similarly, an explanation in terms of fluctuations of the mean p seems to call for severe, and implausible, inhomogeneity of the dynode. I n short, we are unable to offer a convincing explanation of these observations, and if, as seems likely, they are evidence of some basic underlying physical process, we are not yet in a position to identify it.

REFERENCES 1. Wilcock, W. L., I n “Advances in Electronics and Electron Physics”, ed. by J. D. McGee, D. McMullan and E. Kahan, Vol. 22, p. 629. Academic Press, London (1966). 2. Delaney, C. F. G. and Walton, P. W., IEEE Trans. Nticl. Sci. NS-13, No. 1, 742 (1966).

DISCUSSION G. W. GIOETZE: Have you made any measurement on “low-density” TSE films? We know that these films give much higher average yields (50 to 100) and one might suspect that in those cases where the average yield is much closer to the maximum number of secondaries generated by the primary electron the distribution is much narrower, or should at least approach more and more closely the distribution of the “generating” process. W. L. WILCOCH: We have not made measurements with low-density films, and i t would certainly be interesting to do so. However, judging from the results reported by Dietz, Hanrahan and Hance (Rev. Sci. Inatrum. 38, 176 (1967)), I would be surprised if we did not find a nearly geometric distribution from this ilm also. I wonder, too, if it is reasonable to expect the distribution in type of f this case t o approach that of the generating process. It is true that the average yield of low-density films is closer to the maximum number of secondaries generated, but not, I believe, significantly so. My estimate is that these numbers P.E.1.D.-A

19

522

W. L. WILCOCK AND D . E. MILLER

differ by a factor of order 10 for low-density films, compared with a factor of order 100 for bulk-density films. G . T. REYNOLDS: This is very interesting work, for its own sake, and also very important for all types of multiplying structures, including channel devices. Professor Wilcock is to be congratulated for a very beautiful piece of work. J. D. M ~ Q E E : Would the author comment on the origin of the large values of TSE gain of order 10 and on the relation of these results to the noise characteristics of the TSE image tube? w. L. WILCOCK: I think it is now quite widely known that, by proper selection of thickness and material, it is possible to prepare transmission-type dynodes of average yield appreciably higher than the value of about 5 obtained with 500 A thickness of bulk-density KC1, which was the recipe used in TSE image tubes. Unfortunately our results show that no improvement in noise characteristics is to be expected from the use of such dynodes: the single-electron response of the tube would still be quasi-exponential.

Two Methods for the Determination of the Imaging Properties of Electron-optical Systems with a Photocathode V. JARES and B. NOVOTNY Vacuum Electronic8 Reaearch Imtitute, Prague, Czechoslovakia

INTRODUCTION The determination of the imaging properties of electron-optical systems for transferring the electron image from the photocathode t o the target of a TV camera tube or to the luminescent screen of an image intensifier is a rather complicated problem. Conventional methods, based on the distribution of the electrostatic field in the system of electrodes to be investigated, are very tedious and they do not yield results of the desired accuracy. Measurements on complete experimental samples of tubes give the required values. This method, however, is tiresome and expensive. The present paper briefly describes two methods for determining the imaging properties of the electron-optical system of an X-ray image intensifier with variable magnification. These are the experimental method, based on the use of a demountable model of the tube t o be investigated, and the computational method, utilizing programmes prepared in advance for the determination of imaging properties of the given electron-optical system with the aid of a National-Elliott 503 computer.

MEASUREMENTOF

IMAGING PROPERTIES OF AN X-RAY IMAGE INTENSIFIER

THE

The Demountable Model Method To determine the imaging properties of the electron-optical system of the X-ray image intensifier, a n experimental demountable model of the intensifier was constructed, the input section of which (the fluorescent screen and the photocathode) was replaced by a concave alnminium disc with several apertures (see Fig. 1). Tungsten cathodes were mounted behind the apertures which were covered with metal grids having a variable pitch. The cathodes were spot-welded t o 623

524

v. JARES AND

B. NOVOTNI?

metallic holders embedded in ceramic supports. To eliminate the scattering of the emitted electrons through adjacent apertures and to prevent the imaging of the tungsten filaments, the ceramic supports with the thermionic cathodes were inserted in metallic cylinders which were provided with additional metal grids (see Fig. 1). The metallic cylinders and grids were fixed under the aluminium disc. The electrical

FIG.1. Diagrammatic section of aluminium disc D showing one of the apertures with variable pitch metal grid MI, metal grid Mz,and tungsten filament K.

connexions to the thermionic cathodes were by metal pins sealed on the flank of the glass envelope. A general view of the aluminium disc and the grids is shown in Fig. 2. Figure 3 shows how the disc was accommodated in the envelope. The fine metal grids with variable pitch were produced electrolytically from a glass matrix. Their geometrical configuration is illustrated in

Fro. 2. Photograph of metal disc with grids.

IMAGING PROPERTIES O F ELECTRON-OPTICAL SYSTEMS

525

FIU.3. Metal disc assembled in demountable model.

Fig. 4. The grids are secured in the plane of the apertures above the tungsten cathodes by mean8 of an outer disc made of aluminium foil. The focusing electrode of the intensifier consists of an aluminium foil on the wall of the glass envelope. The anodes, in common with the output screen, are secured in the upper, narrowed section of the glass

526

v. JAREB

AND B. NOVOTN+

bulb; their positions may be adjusted. A diagrammatic cross-section of the demountable model of the X-ray image intensifier with variable magnification is shown in Fig. 5(a). The electrode configuration is illustrated in Fig. 5(b). This configuration, obtained after a series of measurements on the demountable model, was chosen for its very good imaging properties. The resolution of the experimental model of the intensifier was investigated for various electrode arrangements by viewing the demagnified electron images of the metal grids through a micro-

FIG.6. (a) Cross-section of the demountable model of X-ray image intensifier with variable magnification showing microscope M, luminescent screen S, anode A, correctoranode A,, focusing electrode F, and aluminium disc K with thermionic cathodes. (b) Cross-section of the electrode configuration showing luminescent screen S, anode A, corrector-anode A,, and focusing electrode F.

scope. The imaging characteristics are shown in Fig. 6. The resolving power of the electron-optical system of the intensifier can be determined both in the centre and on the periphery of the cathode from the number of distinguishable grid meshes. The image distortion is apparent from the geometry of the electron images of the grids. The curvature of the image plane can be estimated from the ratio of the voltages required to focue the images in the centre and at the periphery of the cathode. I n a similar way the influence upon the sharpness and geometry of the image of both the anode curvature and the size of the anode aperture can be analysed.

IMAQING PROPERTIES OF ELECTRON-OPTIUAL SYSTEMS

I 200 -

-9

527

Un=20 kV

>

100

-

0

I

I

I

5

10

15

Up(kV)

FIG.6. The variation of focusing potential U,, (upper curve) and dernagnification 1/M (lower curve) corrector-anode potential U p .

The Computational Method

To determine the imaging properties of the electron-optical systems by means of a computer, a programme was prepared for the solution of the problem of both the distribution of the electrostatic field and the form of the electron trajectories in a rotationally symmetrical electrostatic field. The distribution of the electrostatic potential in an electron-optical system with rotational symmetry is given by the solution of Laplace’s equation. The numerical solution of this equation is usually derived from methods based on the approximation of derivatives by finite differences. The difference form of the Laplace’s equation suitable for numerical computation is given by the expression1 2Vl 2v2 hz) hl(h1 hz) + h,(h,

+

+

___ 2vo _ [h, - h2

+

+

+

(2r

(2r h4)V3 + (2r - hdV4 h3(h3 h4) h4(h3 h4)

+

+

+

] =o,

- h3)V0 h3h,r

h4

(1)

where V ois the successively approximated potential at the point which is at a distance r from the axis of symmetry, and V , t o V , are the values of potentials at the adjacent points of the network which are a t distances h, t o h4 from the original point. For some special cases, the general equation Eq. (1)can be simplified. If, for instance, the adjacent points of the network are a t equal distances, h, = ha = h3 = h, = h,

528

v. J A R E ~AND

B.

NOVOTNJ

Also for points on the axis the following simplified relations can be used:

and

Vl

+ Vz + 4V3 - 6Vo

==

0.

(4)

Equations (1) to (4)are the basic relations which were used for the solution of the axially symmetrical field. The program was written in the symbolical language ALGOL for computation on the NationalElliott 503 computer. The procedure, used for computing the field, can be described as follows. A fine network of squares is drawn in the r-z plane of the system of electrodes to be investigated. The lower edge of the network is identical with the axis of symmetry, while the remaining edges overlap the outer contours of the electrodes and form rectangles. By using this structure the tables of fixed potentials can be prepared, serving, in common with the control parameters, as the input values for the calculation on the automatic computer. A programme, which is recorded in the memory of the computer, controls, in common with the input values, the successive approximation of the potential in individual regions of the network. The Young-Frankel super-relaxation method was used to accelerate the convergence of the iteration procedure. The equations defining the motion of electrons in the axially symmetrical electrostatic field were solved by the use of the predictor-corrector method.2 A flow-diagram showing these procedures for the solution of fields and trajectories is given in Fig. 7. Rectangular framing of the instructions means the instruction: “put into the computer”; round framing expresses the condition for a conditional transfer instruction and, finally, oblique framing denotes the text displayed by the highspeed printer or the content of the printed data. The solution of the problem begins with the computation of the electrostatic field. By repeating the whole cycle the difference between successive values of the potential at an arbitrary point is diminished to the pre-selected value. As soon as this value has been attained at all points in the field, the calculation is terminated. The machine then reads the input values of the trajectory from the data tape, prints the heading and the input values of the trajectory, and carries out the computation of the first four points by means of the Runge-Kutta formulae. Then follows the printing of the values of all four points and the computation of further co-ordinates by the predictor-corrector method. As soon as the calculation reaches the plane of the phosphor screen the printer prints the word “END”. The value of the focusing voltage is determined for the two trajec-

IMAQINQ PROPERTIES OF ELECTRON-OPTICAL SYSTEMS

529

w

Ilntroducing tape of initial values .]

*

lReoding of initial values for eleC.1 (Field calculation I

I

(Counting d 20 cycles of network rec.1

0

~~

FIG.7. Flow-diagram for computation of the imaging properties of the device.

tories with particular input values. The computer follows the effect of 1 % changes of the given focusing voltage on the positions a t which the two trajectories strike the phosphor screen. When the distance between these points of intersection is less than 0.1 mm the focusing voltage is determined and the calculation of the next trajectory is begun. The calculation is finished by printing out the value of the focusing voltage, the values of the potentials a t all points, and by

FIG.8. The computed equipotential lines and the trajectories of electrons.

531

IMAGING PROPERTIES OF ELECTRON-OPTICAL SYSTEMS

drawing the equipotential lines. The magnitudes of the potential along the equipotential lines are given by the values of the potential at points on the axis of the system. The pre-set programme enables an automatic repetition of the complete calculation for other values of the variable parameters. TABLEI Input Values of Trajectories Trajectory

1 2 3 4 5 6 7

ro(mm) 65 131 65 131 65 131 0

zdmm)

dro(mm/sec)

dzo(mm/sec)

7, 11 30 7, 11 30 7, 11 30 0

-1.286 X 10' -2.59 X lo8 0 0 -2.59 X 10' -5.336 x 108 1.285 x lo8

6.79 X 10' 6.336 x 10' 5.53 x 10' 5.93 x lo8 5.336 x 108 2.59 x lo8 5.79 x 108

The programme described above has been used to compute the distribution of the electrostatic field and the shape of electron trajectories in the electron-optical system of the image intensifier shown in Fig. 5(a). The computed equipotential lines and the trajectories of electrons with initial velocity of 5.93 x lo5 mlsec are plotted in Fig. 8. The potential of the corrector anode is Up = 19 kV,and the

20

15

12 0 3 3 -

B

1"

7

6

FIG.9. The parts of the trajectories near the output screen (output radius 65 mm from the axis).

532

v. J A R E ~AND B. NOVOTNP

image is assumed to be demagnified by a factor of 7. The input values of individual trajectories are given in Table I. Some equipotential lines for a corrector-anode potential of U p = 7 kV (for which a 12 x image demagnification is assumed) are plotted for the zone in front of the output screen. Figure 9 shows those parts of the trajectories near the output screen (radius 65 mm) for the two values of the voltage U p . A similar graph for an output radius of 131 mm is illustrated in Fig. 10. UD = 7kV

-2

up/

\U0=20kV

\ \ \

FIG.10. The parks of the trajectories near the output screen (output radius 131 mm from the axis).

The computed results correspond closely with the measurements on the experimental model. The method using the model yields data which describe how electrons strike the output screen, while the computational method enables the trajectories of electrons during their passage through the electron-optical system to be followed. Both methods are an effective aid for the design of new concepts for imaging systems using wide electron beams.

REFERENCES 1. Weber, C., Philips Tech. Rev. 24, 130 (1962). 2. Carr6, B. A. and Wreathall, W. M., Radio Electronic Engr 27, 446 (1964).

DISCUSSION P. FELENBOK:

are you using?

How much computer time do you need and how many memories

IMAGING PROPERTIES OF ELECTRON-OPTICAL SYSTEMS

533

v. J A R E ~ :Calculating the potentida at one thousand points with the desired accuracy of better than 0.1 V, takes on average 6 min for about 100 approximations; the calculation of one trajectory takes about 16 sec. For the calculation of the fields and trajectories only the inner memory of the computer is used. D . R. CHARLES: 1. Are the meshes to simulate the initial velocities of the electrons? 2. Is your programme able to take account of the initial velocities of the electrons and of space charge? V. J A R E ~ :1. The metal meshes with variable pitch do not simulate the velocities of the electrons. The ah adow images of these meshes modulate the electrons from the individual tungsten cathodes so that, after imaging by the system under investigation on to the output screen of the intensifier, the imaging properties of the system may be measured. 2. The programme takes account of the initial velocities of the electrons; the influence of space charge is not considered.


ABSTRACT

Computation of Imaging Properties of Image Tubes from an Analytic Potential Representation? F. SCHAFFt and W. HARTH Institut fur Technische Elektronik der Technischen Hochschule Munchen, West Germany

Numerical methods of investigating the imaging properties of image tubes by computation of electron trajectories have so far used field tables t o represent the imaging potential distribution. These tables have t o be computed from a given electrode geometry and to be stored in the computer memory. If the curvature of the photocathode is large enough, however, as it is in some commonly used image diodes, it has been shown that it is also possible to represent the potential distribution analytically, i.e. by a linear combination of potential functions. For that purpose the so-called flat-ring co-ordinate system is used both t o establish the potential eigenfunctions and to replace the electrode device by more simple geometrical and electrical boundary conditions. For one image diode, imaging distance, magnification, distortion, tangential and sagittal image surfaces were evaluated and compared with similar results obtained for the same tube by another author who had used the field table method. An effort was then made to calculate back from the potential representation to the electrode geometry which had been used before to determine the boundary conditions. This gave an idea of the influence which various parts of the electrode system exert on the imaging field. Variations in the potential representation were then introduced by changing the boundary conditions, and the effects on the imaging properties studied. As an example, it was noticed that both image distance and magnification are closely connected t o the field strength on the cathode surface and depend hardly a t all on the field distribution elsewhere. Finally, from a potential representation that yielded a higher resolution of the image, the corresponding shapes of the focusing electrodes were computed. This work has shown how the image tube system considered initially could be modified to produce some improvement in the imaging properties, 1 For full paper see 2. Angew. Phy.9. 23, 64 (1967). $ Present address: CERN, Geneva, Switzerland. 535


The Design of Electrostatic Zoom Image Intensifiers J. VINE Westinghouse Research Laboratories, Pittsburgh, Pennsylvania, U.S.A.

INTRODUCTION Electrostatic image intensifier design is an excellent subject for the application of large computers, as has been illustrated by several published papers.la2 The main reasons for this are firstly, that the mathematical model of the tube is good, so that the electron-optical problem can be accurately presented t o the computer and the computed results readily interpreted, and secondly, that experimental design study is impeded by many problems that do not arise from the electron optics. The most important single development in electrostatic image tube design was the introduction of the spherical cathode surface by Morton and Ramberg,3 an innovation which reduced geometrical aberrations to a considerably lower level than had previously been attainable with a flat cathode. The three important geometrical aberrations are image curvature, astigmatism, and distortion, and the main problem of electron-optical design is to minimize these while keeping the tube dimensions within reasonable limits. The remaining important aberration is chromatic aberration, due to the spread in the emission energies of the electrons, which sets the ultimate resolution limit of the tube. It is well known that this resolution limit is directly proportional to the electric field strength a t the cathode surface. DESIGNTRENDSIN DIODES Figure 1 shows three important design parameters in a schematic representation of a basic image diode. They are the radius of curvature of the photocathode Rc, the cathode-anode separation x, and the diameter of the anode aperture d. Varying these parameters one a t a time produces the following effects on the focal length f and magnification M : (i) increased R, increases f end M ; (ii) increased d increases f 537

538

J. VINE

and M ; (iii) increased x decreases f and M . It is also important to consider the variation of two parameters simultaneously to maintain constant magnification, in which case a reduction of either d or R, results in a reduction o f f . Both these trends produce higher field strength a t the cathode, providing higher ultimate resolution capability.

FIG.1. Basic design parameters.

However, it is usually found that the shorter the focal length for a given magnification, the worse are the geometric aberrations, particularly distortion. Thus, a shorter tube will generally have higher center resolution but poorer image uniformity. The introduction of a control electrode between cathode and anode

produces a triode capable of being focused. Figure 2 shows a configuration in which this electrode has been made to coincide approximately with an equipotential surface of the basic diode. Some computed principal rays are shown, and the image surfaces for three values of focus voltage V,. The typical properties illustrated are that f and M

ELECTROSTATIC ZOOM IMAGE INTENSIBIERS

539

increase with V,; the image surfaces tend to scale directly with M , with little change in shape and the principal rays do not change significantly.

ZOOMTUBEDESIGN

A zoom tube can be produced by introducing into the triode a fourth electrode either in the anode space or in the cathode space. The distinction is not necessarily clear cut, but the terminology is convenient and its implications will be made clearer in what follows. Anode Space Zooming This is most simply achieved by separating the screen from the anode, as shown schematically in Fig. 3. The essential feature of such a tube is that it comprises two independent parts, the cathode lens and the zoom lens, separated by a field-free space within the anode. This

Screen

I

FIG.3. Variable magnification tube employing a screen lens.

separation simplifies the design problem considerably, since the basic properties of the two parts can be computed separately, and the results of combining them in any manner are then calculable by the simple formulae of Gaussian optics. The required properties of the image triode are described by two curves, namely the magnification M and focal length f as functions of focus voltage V,. The simple zoom lens shown in Fig. 3 can be termed a “screen” lens,? since the output screen forms an essential element of the lens. Figure 4 shows the action of the lens schematically by means of ray diagrams utilizing the cardinal points. The lens is either (a) reducing, or (b) magnifying, according as the screen voltage V , is greater or less than the anode voltage V,. The cardinal points of the lens have been computed for a range of values of the ratio VJV,. These data are not presented here. Since

t This terminology should not be confused with screen lenses formed by parallel wire screens.

J. VINE

540

Screen

Screen

Imoge

--_

--F 2 F :

5

2.0

u--

v--

(b)

FIQ.4. Imaging action of the screen lens. (a) V , < V,,, demagnifying; (b) Vs> V,, magnifying.

the image is required in practice to lie at the screen, only one conjugate pair is of interest for each value of VJV,. Therefore the cardinal point data are reduced to the data of practical interest represented by the two curves shown in Fig. 5. These show the object position Zoblwith respect to the screen and the magnification M , as functions of the voltage ratio. It can be seen that the voltage variations are large, and that there is a range of M over which the object position varies very

Anode

E

.c c .-

'c

ul c

r"

1.2 -

1.0 -

080.6

-

04 -

0.02

0.05

0.1

0.2

0.5

1'0

2.0

5.0

10

20

V,/G FIQ.5. Properties of the zoom lens (computed).

50

ELECTROSTATIC ZOOM IMAUE INTENSIFIERS

d

541

J. VINE

542

little, although it should be remarked that Zobfis proportional to the lens diameter D. Figure 6 shows a design example developed from the triode of Fig. 2. Principal rays and image surfaces are shown for: (a), M = 0.91 and (b),M = 0.45. It is noted that there is a 10 : 1 voltage variation for a 2 : 1 change in M . The effect of the screen lens on the image surfaces is slight, but would be greater if the lens were of smaller diameter. Since Zob,is proportional to D, a smaller diameter screen lens would necessitate less adjustment of the focus. If the required voltage-ratio variation is achieved by changing V a ) then the resolution capability at the photocathode varies proportionately, On the other hand, if the screen voltage V , is varied then the tube gain varies in a way that accentuates the output brightness change that occurs due to the magnification change itself. Limited variations of both voltages might be employed in practice to optimize the resolution and gain changes. A practical tube similar to this type has been described by Woodhead, Taylor and S ~ h a g e n . ~

Cathode Space Zooming This type of zoom image intensifier is illustrated by an actual design example in Fig. 7. The design problem is more complicated because the system may not be considered as two independent parts. Study of the four-electrode system places greater demands on the computational techniques employed.

i

Cathod

M.051

M= 1.0

tangential focus. FIQ.7. Cathode space zoom tube showing image surfaces. 0, 0, sagittal focus.

The mode of use of a zoom tube is such that the output diameter remains fixed, while the input diameter varies inversely with M . This is illustrated by the two principal rays shown in the figure. Thus, the

ELECTROSTATIC ZOOM IMAGE INTENSIFIERS

543

most exacting requirements on the lens occur a t low M , since it is in this condition that the beam diameter within the lens field is largest. Therefore, the initial diode design should be satisfactory a t low magnification; then as M increases, the beam diameter reduces, helping to minimize the effects of field distortions. If in addition the tube is made long, the geometrical aberrations a t high magnification will be further reduced, as was mentioned above. Thus the device shown in Fig. 7 is basically a low magnification diode of large focal length, the latter being achieved by the use of a large anode aperture. The limitation on focal length is set by resolution requirements, because the greater length is accompanied by reduced field a t the cathode, with corresponding increase in chromatic aberration. To form the zoom tube, two additional electrodes are introduced between cathode and anode. Their positions correspond approximately with equipotential surfaces, so that when operating a t low M the tube essentially reproduces the performance of the basic diode. The performance a t high M is dependent upon the choice of equipotentials. I n this condition the potential of G2 equals the anode potential so that GZ becomes essentially the anode of the tube, the final electrode being situated in field-free space. Thus, the diode formed by the cathode and GZ must achieve high magnification in a relatively short focal length, necessitating the use of a small aperture for G2. The field distortion necessary to achieve a focus by adjusting the potential of G1 is then minimized. These principles provide guide lines that help to optimize the tube performance overall, but much still depends on the location and detailed shaping of the electrodes. A trial and error study of these effects can be economically conducted by computation. The principal rays and image surfaces shown in Fig. 7 are typical results of such a process. I n this configuration the field a t the cathode is largely determined by V,,, so that a variation in limiting resolution approaching 4 : l might be expected over the magnification range shown. Referred to the output this variation would be only about 2 :1. On the other hand, curvature of the image varies quite markedly with M , with the result that edge resolution tends to fall as the center resolution rises. REFERENCES 1. Vine, J., IEEE Trans. Nucl. Sci. ED-13, 544 (1966). 2. Wreathall, W. M., I n “Advances in Electronics and Electron Physics”, ed. by J. D. McGee, D. McMullan and E. Kahan, Vol. 22A, p. 583. Academic Press, London (1966). 3. Morton, G. A. and Ramberg, E. G., Physics 7, 461 (1936). 4. Woodhead, A. W., Taylor, D. G. and Schagen, P., Philipe Tech. Rev. 25, 88 (1963).


Electron Optics of a Photoconductive Image Converter M. E. BARNETT, C. W. BATES, Jr.t and L. ENGLAND Department of Applied Physics, Imperial College, University of London, England

INTRODUCTION The feasibility of direct view image tubes, using electron mirror read-out from infra-red sensitive photoconductive layers, is well known.la2 Figure 1 shows the simplest version of such a tube, reported some twenty years ago.2 Light falling on a photoconductive layer creates a potential relief on the surface facing the beam. The potential of the layer is held near gun-cathode potential in such a way that the layer acts as an electron mirror, the majority of the beam being reflected without striking the layer. The reflected beam is modulated by the potential relief, and an image is formed on the output phosphor screen. The conversion of the light signal into a potential relief has been quite well discussed in the original references, but the electron optics of the device has never been treated quantitatively and hence rational design and device assessment has not been possible.

GEOMETRICAL OPTICS Figure 2 shows equivalent electron-optical representations of an electron mirror tube. Approximate expressions for the important design parameters can be derived as follows. The mirror anode can be treated as a thin aperture lens, and the retarding field approximated by a uniform field in which the trajectories are parabolic. The focal length f of the aperture lens, treated as thin, is given from Hoeft’s modification of the small aperture formula3 by

t Academic Visitor at Imperial College, London. Associates, Pa10 Alto, California, U.S.A. 645

Permanent address: Varian

546

M. E. BARNETT, C. W. BATES, JR. AND L. ENGLAND

&

Glass objective

I

Fluorescent screen

Magnifying glass lens Light -optical mirror

FIG.1. Sohaffernicht’seleotron mirror tube.

where V A is the anode potential and V o the potential of the point at the centre of the anode hole, R is the radius of the anode hole and d the distance between the mirror and the mirror anode. V o is given by4

this equation being a good approximation for the range 0

< -Rd < 2-3‘

The anode aperture can thus be replaced (Pig. 2(b)) by a diverging lens of focal length 1--. (3) ( 3 Owing to the parabolic trajectories in the retarding field, the effective

f=4d

plane of reversal (virtual mirror plane) in the equivalent optical

ELECTRON OPTICS O F A PHOTOCONDUCTIVE IMAGE CONVERTER

547

representation of Fig. 2(b) is a t a distance 2d from the anode plane. The combination of diverging lens and plane mirror can be replaced (Fig. 2(c)) by a convex mirror of focal length fm, whose principal plane lies behind the photoconductive mirror plane. It is quite easy to show geometrically that, within the range of Rld quoted above, the focal length of the mirror is well represented by

">

- 0.42-d ,

f,,,= :(l and that

(4)

h = f-.m 4

1;'

fl

1---

h

y-*----I

\

I I

I

I I I

-I-

I

I

I

(C)

Fm. 2. Equivalent representations of an electrostatic electron mirror.

Equation (4) becomes exact when Rld -+ 0, but for larger values of Rid is only accurate to about 10%. Using the equivalent representation of Fig. 2(b), the magnification y/x (Fig. 2(a)) can be shown to be 4d 4d 4d 4d I+-+m=Y=[$)[ 2 + - + z + 4z d f ) + ' ] _____ f L 2d X 2d l+-+l, I+-+-

i;"

f

L

f being given by Eq. (3). I n the tube of Pig. 1 the approximate values of the ratios were R -_ 5 _

_d - - 3

D

1 D - 12' L - 26' L a 2' implying an electron-optical magnification of about x 4.

_-

548

M. E . BARNETT, C. W. BATES, JR. AND L. ENGLAND

A further matter of interest is the question of what proportion of the output field of view is taken up by the aperture in the output screen which allows passage to the incident beam. This aperture, acting as the field stop for the system, needs t o have a radius ro just large enough to permit the whole output screen (radius yo)to be filled. The important relationship in this case is

which, for the geometry of the tube shown in Fig. 1, comes to about 13, implying that less than 1% of the field of view is lost. The limit on the useful field of view at the mirror is set by the magnitude of the transverse velocity components at the edge of the field. Transverse velocity components increase the distance of closest approach of the reflected electrons with the result that the image in the peripheral region is degraded when compared with that in the axial region. The ratio ofthe transverse energy V,eV to the beam energy VeV is given by

If an input field of view of 1-mm diameter is required in the geometry of Fig. 1, a transverse energy of about 2 eV is found for an electron reflected a t the edge of the field of view (beam energy 5 keV). This is clearly a great obstacle to obtaining a device of reasonable sensitivity over a useful field of view. It seems from Eq. ( 7 ) that a long device working at a low beam voltage is desirable. The situation can be improved somewhat by incorporating a weakly converging lens between the anode and screen, as in the test system described later.

IMAGE FORMATION Unlike conventional photoemissive tubes, the photosurface is not electron-optically conjugated with the output screen S. It is clear from Fig. 2(b) that such conjugation is not possible owing t o the presence of a diverging lens, since for such a lens, object and image space always coincide, whereas the photosurface and screen are on opposite sides of the lens. In fact, for the example under discussion, the screen is conjugated with a plane S' which lies between S and the anode aperture. Thus the electron density distribution a t the output is identical with the virtual electron density distribiition in S', apart from a scale factor due to magnification.

ELECTRON OPTICS O F A PHOTOCONDUCTIVE IMAGE CONVERTER

549

The modulation of the reflected beam is caused by the transverse electric fields associated with the signal-induced potential relief on the photosurface. These cause angular deflexions in the electron trajectories which, projected over the distance zo (Fig. 2(b)), are seen as lateral displacements in the virtual object plane S‘. If the displacements of electrons in s’ due to the application of the light signal are small compared with the wavelength of the fundamental spatial harmonic of the light pattern, then a “differentiated” image is obtained, i.e. the electron density distribution in the output is determined by the first derivative of the transverse field at the mirror surface. A true differentiated image can only be obtained however if the image contrast is low, i.e. for small modulation. For large modulation when the displacement in S‘ is large compared with the spatial wavelength, electrons from different parts of the input field become mixed up, with the result that the image bears little apparent relation to the object and the device is effectively useless. Between the limits of large and small modulation the image suffers from a characteristic type of distortion6 in which the unilluminated areas appear larger in the output than they really are, and illuminated areas correspondingly smaller. Further unusual features of the mechanism of image formation are due to the fact that the modulation arises from transverse field components. At a given amplitude of the voltage variation, the magnitude of the transverse field increases with increasing spatial frequency. Thus it might be thought that the sensitivity of the device should actually increase at higher spatial frequencies. However, with increased spatial frequency the range of the field associated with the potential relief decreases so that an increased retarding field E is required to ensure that the slowest electrons approach near enough to the layer to experience the increased transverse field. The angular deflexion due to the potential relief is inversely proportional to E , and furthermore there is a practical upper limit on E . Thus the modulation transfer function (which only has meaning at low image contrast, where the device operates linearly) is moderately complicated. Analysis using a simplified theoretical model5 shows that at a given E , there is a spatial frequency at which the response is a maximum and conversely for a given spatial frequency the response can be maximized by choice of E . Fortunately the most favourable spatial frequency range for operating the device appears to be loa to lo3 lp/mm, which means that restrictions on the field of view are less damaging. It seems that in principle it may be possible at such frequencies to detect potential variations of millivolt amplitudes.6 The most fundamental drawback in the mechanism of image formation, however, lies in the nature of the restriction on the dynamic

550

M. E . BARNETT, 0 .

W.

BATES, JR. AND L. ENGLAND

range for linear operation. Using the simplified model referred to above it can be shown that the condition that a pure spatial harmonic in the potential relief should give rise to a pure spatial harmonic in the output current density distribution is

V o being the amplitude and X the spatial wavelength of the potential relief and - V being the mean d.c. potential of the mirror surface. The problem here is that this condition contains the input parameters V , and A, which for arbitary inputs are unknown and distributed according to the Fourier spectrum of the input signal. Equation (8) implies that correct adjustment of the tube requires prior knowledge of the input! This is clearly a most undesirable feature in practice. A PRACTICAL TESTSYSTEM Figure 3 shows a demountable system which has been built in order to investigate the electron optics of the photoconductive image converter. A weak magnetic lens is included which provides a means of magnification control. An important practical consideration is the choice of a suitable photoconductive layer. The layer surface needs to be very smooth, otherwise the effect of surface topography appears in the output image. It should have a thickness of the order of one micron and its resistivity should be not less than l O W cm. These requirements appear to be fulfilled by a mixture of selenium and bismuth in the proportion 95% Se, 5% Bi by atomic eight.^ This combination yields a layer having a peak response at a wavelength of about 1 pm and a long wavelength limit at about 1.6 pm.7 Co-evaporatorr gives a vitreous tion of these elements at a pressure of lod6to layer. Selenium-bismuth layers of this type have been prepared on conducting substrates and have proved to be suitable for use in the test system. The photoconductivity of these layers has not been measured directly but absorption measurements indicate that the bismuth content shifts the long wavelength cut-off for the response of selenium (normally N 7000 8 )into the infra-red as expected. Using the demountable system and evaporated photoconductive layers of the type described it has been possible to obtain image conversion in the infra-red, the filter employed at the input having a narrow pass band centred on 9600 8. Typically, the beam energy is 5 keV and the layer substrate is held at about lOV positive with respect to the cathode, the beam current being adjusted so that the layer surface exposed to the besm stabilizes

ELECTRON OPTICS OF A PHOTOCONDUCTIVE IMAGE CONVERTER

551

+

a t a potential of to 1 V negative with respect t o the cathode. The image obtained using a test grid pattern shows the predicted form of distortion. The build-up of insulating contamination due to the poor vacuum interferes with the proper functioning of the layer, and it seems I Infra-red filter

I

I

Quartz window

A, -

pqq-

Se-Bi photoconductor

Magnetic lens

I

I

I

Viewing window

Phosphor Screen

I

I I

I

FIQ.3. Schematic diagram of demountable infra-red converter.

likely that prolonged tests of the image converter require a sealed-off system. The preceding analysis of the electron optics will enable us t o make a rational choice of geometry for such a system,

REFERENCES 1. Orthuber, R., 2. Angew. Phys. 1, 79 (1948). 2. Schaffernicht, W., “Fiat Review of German Science, Electronics”, Vol. 1, p. 100 (1948). 3. Hoeft, J., 2. Angew. Phys. 11, 380 (1969). 4. Fry, T. C., Amer. Math. iVonthZy 39, 199 (1932). 6. Barnett, M. E. and England, L., Optik 27, 341 (1968). 6. Barnett, M. E., AppZ. Phys. Letter8 12, 229 (1968). 7. Schottiniller, J. C., Bowman, D. L. and Wood, C., J . AppZ. P h p . 39, 1663 (1968).

552

M. E. BARNETT, C. W. BATES, JR. AND L. ENGLAND

DIscussIoN s. JEFFERS: By how much does the addition of bismuth extend the long wavelength cut-off of the layer? L. ENOLAND: Addition of more than 6% of bismuth (by atomic weight) extends the long-wavelength cut-off to approximately 1.6 pm. The wavelength of maximum response does not change significantly from 1 pm for concentrations of bismuth up to 30% by atomic weight.

Advances in Electronics and Electron Physics, Volume 28A: Photo-Electronic Image Devices, Proceedings of the Fourth Symposium

Advances in Electronics and Electron Physics, Volume 64B: Photo-Electronic Image Devices, Proceedings of the Eighth Symposium

Advances in Electronics and Electron Physics, Volume 71 (Advances in Imaging and Electron Physics)

Advances in Electronics and Electron Physics, Volume 80 (Advances in Imaging and Electron Physics)

Advances in Electronics and Electron Physics, Volume 85 (Advances in Imaging and Electron Physics)