JPRS ID: 9759 TRANSLATION PRODUCTION TECHNOLOGY OF MICROELECRONIC DEVICES BY IDEYA ALEKSANDROVNA MALYSHEVA

• Document Type: CREST
• Collection: JPRS
• Document Number (FOIA) /ESDN (CREST): CIA-RDP82-00850R000400010055-6
• Release Decision: RIF
• Original Classification: U
• Document Page Count: 358
• Document Creation Date: November 1, 2016
• Sequence Number: 55
• Content Type: REPORTS

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
_
- FOR OFFICIAL USE ONLY
JPRS L/9759 _
28 May 1981
~
- Tran~lation
PRODUCTION T~ECHNOLOGY OF MICRQELECTRONIC DEVICES
~Y
Ideya Aleksandrovna M~~alysheva -
FB~$ FOREIGN BROADCAST INFORMATION SERVICE
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
i .
NOTE
JPRS publications contain informa.tion primarily from foreign
newspapers, periodicals and books, but also from news agency
transmissions and broadcasts. Materials from foreign-language
sources are translated; those from English-language sources
are transcribed or reprinted, with the original phrasing and
other characteristics retained.
Headlines, editorial reports, and material enclosed in brackets
are supplied by JPRS. Proce~sing indicators such as [Text]
or [Excerpt] in the first line of each item, or following the
- last line of a brief, indicate how the original information was
processed. Where no processing indicator is given, the infor-
' mation was summarized or extracted. -
Unfamiliar names rendered phonetically or transliterated are
enclosed in parentheses. Words or names preceded by a ques- -
tion mark and enclosed in parentheses were not clear in the _
original but have been supplied as appropriate in context.
Other unattributed parenthetical notes within the body of an
item originate with the source. Times within items are as -
given by source.
The contents of this publication in no way represent the poli-
cies, views or at.titudes of the U.S. Government.
~
_ COPYRIGHT LAWS AND REGULATIONS GOVERNING OWi~1ERSHIP OF
MATERIALS REPRODUCED HEREIN REQUIRE THAT DISSEMINATION
OF TfiIS PUBLICATION BE RESTRICTED FOR OFFICIAL USE ONI,Y.
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
JPRS L/9759
_ 28 May 1.981
PRODUCTION TECHNOLOGY OF MICROELECTRONIC DEVICES
Moscow TEKHNOLOGIYA PROIZVODSTVA MIKROELEKTRONNYKH USTRCIYSTV in
Russian 1980 (signed to press 15 Feb 80) pp 1-448
[Book by Ideya Aleksandrovna Malysheva, Izdatel'stvo Energiya, approved
by the USSR Ministry of the Electronic Tndustry as a textbook in the _
_ specialty of microelectronic circuitry production at the middle
specialized schools, 15,000 copies, 448 pages, UDC 621.3.049.77.002(075)] -
CONTENTS
Foreword 1
y
Introduction 3
Chapter 1. General Description of Microcircuit Production (
1-1. Basic Concepts 6
1-2. Cla~sification ~nd General Description of Micr~circuits 8
1-3. Development of Microcircuit Technology, Production Forms
and Records 11
1-4. Basic Processes of IC Production Technology 13
Test Questions and Assignments lg
Chapter 2. General Requirements on Microcircuit Production 21
2-1. General Requirements on the Technological Process 21
2-2. Requirements on Cleanness of Air and C~imatic Parameters 22
2-3. Requirements on the Production Gases and Water 27
2~4. Basic Principles of Electron Vacutm? Hygiene 30
2-5. Basic Production Features of Microcircuits 3],
Test Questions and Assignments
32
- Chapter 3. Manufacture of Semiconductor Plates and Dielectric
~ Substrates for Microcircuits 34 -
3-1. General Information About the Machining of Se~iconductors
and Dielectrics for IC 34
3-2. Abrasive Cutting 36
3-3. Grinding and Polishing Billets for IC Structures 41
3-4. Control of the Plates and Substrates After 2iachining 44
Teat Qvestions and Assignmenta 46
- - a- [ I - US S R- F FOUO ]
FOR OFFZCIAL USE ONLY -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Chapter 4. Chemical Treatm~ent and ~ leaning of the Surface of
Semiconductor Plates and Substrates 49
4~1. General Information 49
_ 4-2. Methods of Liqu~ld Treatment of Plates and Substrates 52
4-3. Intenoification of the Cleaning Processes 55
4-4. Standard Processes of Cleaning Plates and Substrates 57 ~
4-5. Drying Cleaning of Plates and Substrates 59
- 4-6.' Quality Contral of Surface Cleanness of Plates and Substrates 63
_ Test Questions and Assignments G5
Chapter 5. Contact Photolithography 68
5-1. Application andEssence of the Photolithography Process 68
- 5-2. Formation of the Photor.esistive Layer 76
5-3. Formation of a Photoresistive Mask 81
5-4. Obtaining the Configuration of the Elements 84
_ 5�-5. Photomask Production Technology g9,
5-6. Types of Rejects and Qualifiy Control 93
- Test Questions and Assignments 94
Chapter 6. Obtaining the Configuration of IC Film Elements Using
Free Ma.sks 9 7
_ 6-1. Free Mask Method 97
_ 6-2. Free Mask Production Technology for Thin-Film IC 99
6-3. Stenciling Method 103
Test Questions and Assignments 105
Chapter. 7. New Lithography Techniques 106
7-1. Contactless Photolithography 106
7-2. X-Ra.y Lithography 108
7-3. Electron Lithography 112
Test Questions and Assignments 116
Chapter 8. Methods of Obtaining Thin Films 118
8-1. Method of Thermovacuum Deposition 118
8-2. Ion Bombardment Sputtering 122
8-3. Thermal Oxidation 128
8-4. Film Deposition from the Vapor-Gas Phase 130
8-5. Anodic Electrolytic Oxidation 134
8-6. Deposition of Metals from the Electrolytes and Solutions 136
Test Questions and Assignments 137
Chapter 9. Epitaxtal Growth of Semiconductor Layers 140
9-1. Fundamentals of the Epitaxy Processes 140
9-2. Methods of Epitaxial Growth from the Vapor-Gas Phase 144
9-3. Other Methods of Epitaxy 147
9-4. Heteroepitaxy 150
- b -
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440010055-6
FOR OFFICIAL USE ONLY
- 9-S. Local Epitaxy 153
9-6. A1loying of Epitaxial Layers 153
9-7. Defects in the Epitaxial Layers 155 -
9-8. Epitaxial Layer Control 157
Test Questions and Assignments 160
Chapter 10. High-Temperature Diffusion 163
10-1. Fundamentals of the M~ethod of High-Temperature Diffusion 163 -
10-2, Characteristic Features of Diffrsion in Plaxiar Technology 168
10- 3. Methods of Achieving Diffusion 169
- 10-4. Defects and Control of the Diffusion Structures 176
Test Questions and Assignments 177
Chapter 11. Ion Alloying and Other Methods of Obtaining
Semiconductor Elements _ ~79
11-1. Fundamentals of the Method of Ion Alloying 179
11-2. Distribution of the Admixture Concentration in the
Ion-Alloyed Layers 1g2
11-3. Equipment for Implementing the Ion Alloying Process 187
11-4. Advantages and Disadvantages of Ion Alloying 189
11-5. Other Methods ot Obtaining Semi.conductor Elements 191
Test Questions and Assignments 195 _
Chapter 12. Metal Coating of Silicon Structures 197
12-1. General Information 197 -
12-2. Single-Layer Aluminum Coating 198 _
12-3. Multilayer Metallization 203
12-4. Multilevel Metal Coating 205
12-5. P4eta1 Coating of Mounted Active Elements 2pg
12-6. Metallization Defects and Quality Control 211
Test Questions and Assig~ents 214
Chapter 13. Technological Processes of Manufacturing the Structures of
Bipolar Microcircuits 216
13-1. Engineering Design Features of Bipolar Microcircuits 216
13-2. Manufacutirng Technology of the Structures of Bipolar IC With
Insulation by the p-n Junction 219
13-3. Manufacturing Technology of Structures of Bipolar IC With
Dielectric Insulation 223
13-4. Manufacturing Technology of the Structures of Bipolar IC With
Combined Insulation 22g
13-5. Manufacturing Technology of the Structures of Compatible IC 233
Test Questions and Assignments 233
- c -
~ FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
Chapter 14. Technological Processes of Making IrIDS-IC Structures 235
- 14-1. Structural-Technological Features of NIDS-IC 235
14-2. Manufacturing Technology for Structures of Thin-Oxide -
p-Chan~el NIDS-IC 237
1.4-3. Manufacturing Technology of MTOS-IC 239
14-4. Manufacturing Technology of the Structures of NIDS-IC With
Fixed Gates 240
14-5. Manufacturing Technology of CNIDS-IC Structures 243
14-6. Ways of Improving the Quality of NIDS-IC 247
14-7. Protection of Semiconductor Structures 251
14-8. Dielectric Film Control 254
Test Questions and Assignments 256
Chapter l5. Manufacturing Technology of the Structures for Thin-Film
Microcircuits 258
15-1. General Information 258
15-2. Standard Circuits and Basic Steps in the Manufacture of the
Structures of Thin-Fi1m Microcircuits 760
- 15-3. Technology for Manufacturing the Structures of Thin-Film
Microcircuits with the Application of Free Masks 263
15-4. Manufacturing Technology of the Structures of Thin-Film
Microcircuits with the Application of Photolithography 266
_ 15-5. Manufacturing Technology of Structures of Tantalum Thin-F~ilm
Microcircuits 268
15-6. Ma.nufacturing Technology of the Structures of Thin-Film
Microcircuits Using the Beam Processing 271
15-7. Adjustment of the Rated Values and Shielding of the Film Elements 273 -
15-8. Quality Control in the Production of Structures for Thin-Film
Microcircuits 274
Test Questions and Assignments 277
Chapter 16. Manufacturing Technology of the Structure of Thick-Film -
Hybrid IC 279
16-1. General Information 279
- 16-2. Basic Steps in the Manufacturing Technology of the Passive Part
of the Structures of Thick-Film I3ybrid IC 282
16-3. Quality Control When Manufacturing Thick-Film Microcircuits 288
Test Questions and Assignments 289
Chapter 17. Assembly of Microcircuits 291
17-1. Separation of thp Plates and Substrates with the Finished
Structures 291
17-2. Basic Assembly Methods 296
17-3. Installation of Crystals and Plates 298
17-4. Wiring 300
17-5. Wireless Melting 307
17-6. Quality Control in the Welding Process 310
Test Questions and Assignments 311
- d -
- i'OR O~'FICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Chapter 18. Encapsulation of Microcircuits 313
18-1. Microcircuit Cases 313
18-2. Methods of Sealing in a Case [Encapsul.ation] 316
18-3. Various Types of Encapsulation 321
J.8-4. Caseless Encapsulation 324
18-5. Quality Control of Encapsulation 327
18~-6. Finishing Operations of the Manu�acture of Microcircuits 329
- Test Questions and Assignments 330
Chapter 19. Insurance of Production Efficiency and Quality of Microcircuits 332
19-1. Basic Areas and Problems of Microe~ectronics in the Current
' Five-Year Period 332
19-2. Insurance of the Efficier.cy and Quality on the Modern Level of
~evelopment of Microelectronics 333
_ 19-3. Nondestructive Control and Improvement of Technology 336
Conclusion 340
Test Questions and Assignments 343
Bibliography 344 "
Subject Index 346
Annotation 351
- e -
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
~
~ FOR OFFICIAL USE ONLY ~
~
''[Text] FO~EWORD
'~1t.this time microelectronics pla,ys the primary role in the electronics industry.
As it affects other branches, it to a signiffcant degree determines t1~e scientific-
technical and social progress of our country as a whole.
As a result of the constant attention on the part of the party and government a
~owerful scientifi~ research and industrial base for microelectronics has been
created in the Soviet Union. An important condition of the successful production
activity of the enterprises in this branch is constant improvement of production
technology which is possible in the presence of we11-trained personnel. Accord-
Kngly, in a number of the special schools division; have been set up for training
process engineers in the production of microelectronic devices (specialty number
0658). The primary finishing course taught in these divisicros is the subject which
is the title of this textbook. _
7'he subject of "production technology of microelectronic devices" has a number of
specific features which unquest3onably distinguish it both from the subjects in
- the general currl.culum and from many of the sub~ects of the specialized curricula.
~his difference is determined by the enormous difference in the methods and pro-
cesses used and also the exceptionally high rates of development of. production.
A continuous flow of information introduces the necessity for constant supplement-
`ing and revision in the development of engineering theory. Accordingly, when
writing the textbook the author tried ko discover +the essence and the peculiarities
of the basic technological methods and processes of manufacturing various groups
of microcircuits. Attention was given to the theo:retical production problems and
~not the special process formul:as and conditions which can be extremely varied or _
can change as the production facilities and techniques are improved. When dis-
~cussing the material in connection with other sub~ects of the special curriculum,
the author has tried to point out the numerous interrelations of production,
including the choice of materials, ~ptimal process techniques and conditions,
coordination of operations in the process cycle, the equipment used, production
~q uality control and efficiency. _
~ 1
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
_ In Chapters 1 and 2, a general description is presented of the basic steps in the
manufacture of microcircuits of var~ous engineering designs, and the.general
production requirements and characteristic features are presented. The process
of manuf acturing microcircuit structures is discussed in Chapters 3 and 16.
Chapters 17 and 18 familiarize the.reader with the ~se~iy and encapsulation of micro-
circuits. Chapter 19 investigates ways of improving the guality of the microcir-
cuits and their production efficiency on the modern level~.~ The conclusion considers
the prospects for further developraent of microelectronics.
The author expresses her appreciation to docent of the Department of "Radio- i
electronic Production Techno]_ogy" of the MATI Institute [Moscow Avlation ~Technolog-
ical Institute], Candidate of Technical Sciences Yu. G. Obichkin, editor, professor
of the Experimental Department of the MIREA Institute, Doetor~ of -Technica~ Sciences
Y u. A. Kontsevoy, professor, Doctor of Technical Sciences, Ya. A. F~dotov,
CandidatP of Technical Sciences F. P. Pre~s, docent of the Microelectronics
Department of the MIET Institute, Candidat~ of ~!echnical Sciences
0. V. Mitrofanov, Microelectronics Department Instructor of the MIFI Institute -
Candidate of Technical Sciences 0. S. Bulatov, reviewer, i~structor of the STEP
L. A. Levkova, instructors of the MPEP L. I. Konstantinova, M. G. Krutyakova and
A. V. Gaykovich for valuable comments and assistance rendered when working on this
book.
The author will be grateful to the readers for suggestions and comments on the
book which should be sent to the follawing address: 113114, Mosaow, Shlyuzovaya
nab. 10, izd-vo "Energiya." -
Ideya Aleksandrovna Malysheva
[
2 -
FOR aFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
INTRODUCTION =
The goal of the course in "Production Technology of Microelectronic Devices" is
to study the principles of the production technology of the components on ~he
basis of which modern mtcrominiature radioelectronic and computer equipment is
built. -
~he entire period of development of the component base of electronics can be
broken down into four generations: discrete electronics based on electrovacuum
tubes,discrete electronics based on semiconductor devices, integrated micro- -
electronics based on integrated microc.ircuits (IC) , and integrated microelectronics
based on functional microdevices (FMtJ) . -
In the first generation the role of the active components was played by various
electrovacuum tubes. Resistors, capacitors, inductance coils, transformers,
connectors, switches and other discrete radio parts were used as passive elements.
Radioelectronic equipment (REA) was assembled from individual, discrete components
which were mechanically attached to special panels and electrically coanected to
each other by wires using soldering ~or welding techniques. Later, printed circuits
were developed which are more reliable and insure complete reproducibility of the
REA parameters and relative ease of automation of production. The complexity of
the technology of electrovacuum devices, their short service life, significant size
and weight and high energy consumption provided the incentive for building new
active components semiconductor devicea.
The second generation of the component base of electronics appeared with the inven-
tion of the first transistor in 194$ by American scientists Bardin and Brattein.
The invention o'f the first transistor was also preceded by a great deal of work by `
Soviet scientists.
In 1900, Russian scientist and inventor of the radio A. S. Popov used a semi-
conductor device for the first time in. the world to detect radio signals. In 1922
_ 0. V, Losev was the first to amplify and generate electric signals with the help
of a semiconductor. Systematic study of semiconductors on a broad scale in our
coimtry was started in the 1930's by a group of scientists headed by Academician
A. F. Ioffe. The electrical conductivity of semiconductors and the influence of
impurities on it were studied by. .I. -V. Kurchatov, V.P. Zhuz e, M. S. Sominskiy and
a number of other sciehtists. B. I. Davydov and D. I. Blokhintsev developed the
theory of rectifying a concentration n-n+ ~unction. Many other Soviet scientists
and developers also made a significant contribution to the theory and practice of
creating semiconductor devices.
3
FOR OFFICIAL US~ ONLY -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
The first transistor was a point-contact transistor; its p--n �~unctions were
obtained at the point of contact with the~ semiconductor of two electrolytica113~
pointed wires. ~Hawever, the point contacts turned out to be unstab-le and the
devices based on them had low mechanical strength. In I949-1950 the first alloy
transistors were developed in which the p-n ~~ctions were obt~fned on the basis
of interaction of the liguid phase of a conswmmable electrode containing the alloq-
ing element with a solid semiconductor. The a11oy transistors were distinguished
- by large ~ unction areas, low reproducibility of the parameters and impossibility of
obtaining base regions less than 10-microns wide. In 1953 transistors with
diffusion ~unctions were introduced, the~ par~a*neters of~ wfiich are more reproducible,
and the base width can be decreased Co 0.2-0.3 microns.
By comparison with electrovacuum devices, semiconductor devic~s have smaller dimen-
sions and less weight. They consume less energy an.d have greater efficiency,
longer service life and higher reliability. The lawer energy consumgtion and high
efficiency also made it possible to reduce ths size of the passive radio parts
significantly. For example, in the last decade the volume of resistors has been
decreased by 50-75 times, and the volume of capacitors, by 60-70 times. The
necessity for fast processing of a large quantity of information required that the
electrode gaps in the active components be reduced. Therefore the improvement of
semiconductor devices and passive radio parts in parallel with them proceeded along
the path of microminiaturization. This promoted a transition to compact printed
circui~ry and the creation of small assemblies in the form of columnar, two-dimen-
sional and cordwood microminiature modules. The general-purpose cordwood micro-
circuits are 10-15 mm high, they weigh 5-7 grams and hav+e a packing density of
5-20 components/cm3.
- The third generation arose from the development and introduction of planar tech-
nology into semiconductor production in 1957-1958. Planar technology combined with
film technology made it possible to convert to REA-based on theoretically new
components integrated microcircuits. The integrated mierocircuit is a micro-
electronic com~onEnt which performs a defined function of conversion and processing
of a signal and has high packint; density of electrically connected elements (or
elements and components) and (or) crystals which, from the point of view of test
requirements, acceptance, delivery, operation and maintenance is considered as a
unit whole.
Integrated microcircuits contain an enormous number of elements equivalent to the
previously used discrete elements. This decreases the number of connections and
simplifies the process of assembling of REA, it significantly improves the reliabil-
ity of the REA, it decreases size and weight, improves characteristics, expands
the functional possibilities and significantly lowers the cost: ~The~appearance of
IC [integrated microcircuits] played a decisive role in the development of elec-
tronics, establishing the basis for a new phase microelectronics which is a
qualitatively new phase of microminiaturization. However, the complication of the
functions performed by the IC is achieved by increasing the degree of integration,
, which leads to complication of their structural design and has physical liffitations.
The component base of the fourth generation will be functional microdevices and
assemblies in the structure of which it is difficult or impossible to isolate the
elementa or the components equivalent to the traditional discrete elements of the
4
FOR OFFICIAL U3E ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFF[CIAL USE ONLY
first two generations. Functional microelectronics differs theoretically from
all preceding generations. As a result of integration of various bulk and surface
physical phenomena,~the structural complexity barrier has been overcome in fun~
tional microdevices. This makes it poasib le to build more reliable, pawer-intensive -
and economical components for microelectronic equipment. -
The component base of the electronics is developing at fast, continuously grow~~g
rates. Each of the indicated generations, appearing at a defined point in time,
_ has continued to develop in the most ~ustified directions. The development of the
component phase from generat~.on to generation has proceeded in the direction of
functional complication of the e~:ements, improvement of reliabilfty and surface life, -
decreasing size, weight and cost, simpli�ication of technology and improvement of
- the pa~ameters of REA. The most improved base c~raponents at the present time are -
the IC. Many functional microdevices are stilJ. in the development stage. There-
fore the primary goal of the course in "pro~iuccion technology of microelectronic
- devices" is to study the producti'on technology of integrated microcircuits.
5 ~
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
- CHAPTER 1. GENERAL DESCRIPTION OF MTCROCIRCUIT PR~DUCTION
1-1. Basic Concepts
The Term "Tekhnologiya [~echnology, Production Process, Production Engineering]"
This term was derived from two Greek words: "'EEf~or~" meaning art, proficiency,
skill, and "aS~yos" meaning science. In microcircuit production the term is used
to designate the methods, procedures and means of processtng raw materials and
intermediate products and to designate the processing operations themselves and
combinations of them in technological production processes.
The term technology also refers to the descriptions consisting of the production
flow charts, instructions, and so on.
Technology also designates the scientific discipline which deals with the discovery
of the essence of laws of inechan3cal, physical, chemical and other phenomena to
_ improve the existing production processes and to develop and introduce the most
efficient, new technological processes into production.
~~Tekhnologicheskaya operatsiya' [P~roduction Operation
Microcircuit production includes a defined sequence of series and parallel process-
ing operations, on the performance of which the finished products are gradually
obtained from the raw materials. The production operation is the basic component
part of the over.all production or technological process of manufacturing the product.
According to All-Union State Standaxd 3.1109-73, the production operation is the
complete part of the production process performed at one work place. It is
characterized by purposeful alteration of the initial ob~ect (initial billet or
during subsequent operations, initial intermediate product) during the process of
performing the successive operating procedures conversions. The term "tekhnolog-
icheskiy perekhod" jproduction conversion]'designates the completed part of the
= production operation charact~rized by constancy of ~the tools used and the surfaces
formed by the processing or joined during a5semb ly. During the performance of
each pr~duction operation qualitative changes take place in the processed ob~ects:
a semiconductor ingot is converted to plates, the sh,ape of the plates a~d their
surfaces become geometrically more precise; then the plates become more finished,
and so on.
6
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Basic and auxiliary production~operations are distinguished, depending on the
processed object. If ob~ects are processed wtiich, on completion of the production
process, are converted to the required product, then the processing is considered
- Co be part of the basic operations. The production operations performed on
~.uxiliary objecte are called au~d.l~ary operations. These include, for example,
_ the preparation of pickling agents,cleaning of equipment and dryin~ of gases. The
auxiliary operations are an inseparab3.~ part of production and the quality of the
finished TC depends no less on the quality of the performance of the awd liary
operations than on the performance of the basic operations.
The IC production technology includes a large number of different operations with
respect to their physical-chemical mechanism performed in a vacuum, and gases,
liquids and in the air. The number of operations reaches thousands or more in a
- number of cases. If it is considered that larg (LSI) and superlarge (SLSI)
integrated microcircuits contain from 102 to 10~-105 elements, it is clear that it
- is very complicated to manufacture defect-free, reliable microcircuits. An effec-
~ive means of irnproving the quality and the percentage yield of good microcircuits
is to check the absence of defects after performance of the process operations.
Therefore the technological production process includes quality control operations.
The n~ber of quality control operations is deter~mined by the type a~zd the complex-
- ity of the IC.
"Tekhnologicheskiy protsess" [Technological Process]
In accordance with All-Union State Standard 3.1109-73 the technological process is
the "part of the production p rocess containing the operations of changing and sub-
- sequent definition of the state of the ob~ect of production." The technological
process of making microcircuits contains the optimal number of production operations
- arranged in a defined sequence and providing for economLcally substantiated produc-
tion of microcircuits of the given strnctural design with given electrophysical
parameters. Well developed and checked-out standard technological processes which
pravide for the operating reliabilifiy of the integrated circuits are used in mass
and series production. These technological processes are characterized by imity of
content and sequence of the majority of production ~perations and conversions for
_ a group of products with common structural attrib utes.
"Tekhnologicheskiy metod" [Production Method]
It is possib le to solve the same production problem in the manufacture of IC by
using various methods. For example, in order to obtain a g-n~ ~junction it is
possible to perform the production operation by the methods of diffusion, epitaxy,
ionic alloying and so on. The methods of performing the production operations are
characterized by a defined essence of the phenomena occurring during the processing
- and the defined set of operating procedures. The methods of performing the tech-
nological production processes are characterized by a defined set of matched pro-
duction operations, each of which is performed by a defined opt3mal method in the
given production process.
The production methods are classified by different attrib utes. The following
processing methods are distinguished in accordance with the division of production
in.to production sections: machi~ing, chemical processing, heat treatment, photo-
lithographic processing, epit axy, elionics (treatment by electron and ion beams),
assembly, encapsulation, and so on.
7
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE OIYLY .
In accordance with the purpose of the-:technological processes, methods of entry
- control, cutting, grinding, polishing,.degreasing, pickling, washing, drying,
obtaining thir. filma, and so on are distingu3shed. The sepa�r-ation into the methods
of �ilm and semiconductor technology is, generally recognized. This division
reflects the history of -productio~ develapment and emphasize's the specific nature
of each production area. However, ia this classif3cat~nn it ie ~iecessary to con-
sider that a~most all methods of thi~ film fechnology historical~y~manifested earlier,
are also used in semiconductor production. A special role is played by the method
~ of stenciling (in order to obtain only thin-film components) and the free mask
method (to obtain a configuration, given d3mensions and mutual arrangement of thin-
film components). The methods of obtaining rectifying p-n~�and concentration n-n+,
p-p+ ~unctions are specific to semiconductor~production. All the remaining methods
machining to obtain plates and substrates, chemical treat~ent and finishi~g of
the surfaces of the plates and substrates, obtaining the given configuration,
_ dimensions and arrangement of the components, obtaining thin films, assembly, _
encapsulation were used for making all groups of microcircuits. It is true that
they are not all used to an identical degree, and each method has its specific
nature in each specific case of production.
Group and individual production methods are distinguished. In the group methods
not one specimen, but an entire lot are sub~ected to simultaneous processing. The
p rocessing of the lot under identical process conditions permits the dispersion of
_ the parameters from specimen to specimen to be decreased and the efficiency of the
technological process to be increased. Automation of the group processes signif- ~
icantly lowers the cost of the IC.
= 1-2. Classification and General Description of Microcircuits
Classification of Microcircuits
The IC are divided into three groups in accordance with the engineering design
e xecution (Figure 1-1): semiconductor, hybrid, other (film, vacuum, ceramic, and
' so on).
A semiconductor microcircuit is an integrated microcircuit, all the ~lements and
interelement connections of which are made in the body and on the surface of the
~ semicanductor. A semiconductor IC, all the elements of which, b.oth active and -
passi~~e, are in the body of the semiconductor plate is called monolithic. A semi-
conductor IC, the active elements of which are made in the body of the semi-
- conductor and the film passive elements, on the dielectric film obtained on the
- surface of the semiconductor plate, is called compatible.
A hybrid microcircuit is an integrated circuit, which, in addition to the elements,
contains el.ements and (or) crystals. A component is a part of an integrated
- circuit which realizes the funct~ons of any electroradio element, which can be -
isolated as an independent product from the point of view of the test requirements, ,
acceptance, delivery, operation and maintenance. A multicrystal integrated circuit
is a special case of a hybrid integrated circuit. The crystal of the IC is part
of the semiconductor plate, in the body and on the surface of which the elements
of the sem~conductor microcircuit, the interelement connections and contact areas
are formed.
8
,
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONI,Y
A film microcircuit is an iY~tegratetl circuit, all the elememts and interelement -
connectiona of which are executed in the form.of ~ilms. Depending on the thickness
o,f the applied films, the fil~m IC are divided into thin-film~and thick-film IC.
If the film thickness is cc~mparable to the free path .length o� the electrons in them,
the IC is called thin-fil-m. In thiek-film IC the film thickness great'ly exceeds
~he free path length of the electron. In practiee, thin-film IC in the ma3ority of
cases have films no more-than l.micron thick and, correspondingly, thictc-film IC,
more than 1 micron thick. '
From the definitions it is clear that the primary structural attributes tor dividing
the IC into groups are the substrate material and the atructural design of the -
elements. Semiconductor plates in the body of which it is possib le to make active
� I (1~1NI(POCXE616d ~ . � -
l
(2) 3
[lon;~npuauAHmro2~ue ~1~Ha~NJ~HdB (1p04N@
~ m
~ S
~ a u o a x
(5) ` 'v g) (7~ " ~ ~ ~ m
o o o ~8) .
Insuring the Required Purity of Gases -
. Technical gases contain more impurities th an required for IC production. Therefore,
before the gases are supplied to the shops they are centra~ly purified of oxygen,
hydrogen, mechanical particles, moisture, oil, and so on. If the degree of purity
of the gases is insufficient after centralizecl purification, they are also purified _
directly before admission to the operating units (finish purification). Purifica-
tion nf the gases with respect to oxygen is realized by passing the gases~through
a membrane made of a palladium alloy with platinum which absorbs oxygen well or by
28
FC~R OFi~I~C~A~, i1S~ ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400010055-6
FOR OFFIC[AL USE ONLY
- binding the oxygen to hydrogen in the presence.of a palladium catalyst. The
dust is removed from the gases by filtering the gases through various filters
(ceramic, lavs an, electrostatic, and so on).'
Chemical and physical methods are used to dry the gases. In the chemical methods
the water vapor enters into a chemical interac~ion with the formation of hydrates -
or other compounds. The physical methods include adsorptian drying and drying by
freezing. Adsorption drying is realizeii ~ointly with the ~emoval of oxygen or
hydrogen bypassing the gas�throu~h adsorbers filled with material with high
adsorption capacity (silica gel, aluminogel or zeolites). The drying by freezing -
the water vapor, oil and other vapor is done directly before the work place where
the gas is used by passing it through a coil in a cantainer with liquid nitrogen.
_ The candensed liquid precipitates on the coil walls in the form of solid crystals.
In order to prevent transport of~the crystals by the gas flaw, a filter is
required directly after the coil.
After centralized purification before supplying the gases to the shops, they are
subjected to cantinuous automatic monitoring for the oxygen, hydrogen and water
~apor content. At the exit from the finish purification units there is continuous
automatic monitoring of the oxygen and water vapor contents in the gases. The
periodicity of monitoring the impurities at the entrance to the process equipment
and monitoring the dust content of the gases in all phases is estab lished by the
enterprise standards.
The maintenance of the pure gas networks has great significance for insuring a
minimum amount of impurities in the gases. The pure gas networks must be sealed,
noncorrosive, and they must have minimum length. In order to decrease the ~;as
release from the inside surfaces, the lines are made of stainless materials. Before
installing the pipes are degreased, they are blown out with compressed air or
nitrogen, washed with hot water and again blawn out. During operation the inside
walls of the pipes are cleaned regularly.
Water Requirements and Insur~nce of Them
In microcircuit production water is used in large quantities to make various solu-
tions, for washing the substrates, the finish structures, the case parts and also
as a reagent and protective medium. Natural water contains a large number of
- mechanical parts (hydrosols), dissolved mineral salts, admixtures of copper, si'lv~r,
gold, bacteria, and so on. Therefore water which has been purified of a11 pollution F
is used in IC production. The index of the water purity is its specific resistance.
The resistance of water depends strongly on temperature; therefore the body of the
specific resistance is given at 20�C. The natura3::.specific resistance of ideally
pure water is 25 mohms-cm. In micror_ircuit and semiconductor device production, -
first, second and third degrees of purity of water are distinguished, which
correspond to specific resistances of 18, 10 and 1 moh~cm (A, B and V type water,
respectively). In addition to the speeific reaisrance, the silicic acid, micro-
_ particle, microor~anism, copper, iron and so an content in the water is determined. ~
For microcircuit production the tap water is sub~ected to preliminary and final
purification. The preliminary purification of water to remove suspended and
colloidally-dissolved~particles is carried out by the methods of distillation and
sorption using special filters, reagent coagulation, electrocoagulation, and so on.
The previously purified water has soluble salts and other admixtures. For final
29
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
purification of water to remove soluble materials at the present time purification
by ion exchange resins (deioniza'tion)~and~the.method of inverse osmosis are used.
In order to obtain especially pure water~with a specific resistaace of 18-20 mohms-Cm,
the following purification systems are used: distillation-deionization; electro-
~ coagulation-deionization; electrocoagulation-ultrafiltration-inverse osmosis;
ultrafiltration-deionization. These systems also include preliminary filtration,
for e~sample, through activated charcoal to remove chlorine, orgariic pollutants,
turbidity, large and fine solid particlea. -
The pure water is checked for specific resiatance by an instrwnent with a bridge I
compensation electric circuit. The presence of silicic acid in the~water is
determined comparing the color of a special solution prepared on the basis of the _
checked water with the standard scale colors (the colorimetric method). Organic
impurities are deterndned by their capacity to be oxidizecl on introduction of an
oxidizing solution of potassitan bichromate, ccm centrated sulfuric acid and
crystalline silver sulfate into the water which plays the-~role of a catalyst. The
suspended particles are analyzed by the filtration~-method, microphotography of the
water sample, the optical method and other methods. -
2-4. Basic Principles of Electr~n Vacuum Hygiene
The purity of the air in the production facilities t~at:manufacture~IC is on such _
a high level that the service personnel become sources of pollution. Therefore when
possible the presence of service personnel is limited, and all of the personnel
participating in the manufacture of integrated microcircuits are sub3~et to the
- electron-vacuum hygiene (EVH) rules.
EVH is made up of the general hygiene of the production fac~~.ity, individual
hygiene of the process equipment and the personal hygiene of the service personnel.
Each enterprise has a special EVH service which realizes organization and control
of the effective instructions for observing the EVH rules. The EVH service accounts
- for the state of the EVH in the enterprise subdivisions, it monitors the co~ndition
of the microclimate in the production sections, it develops the operating conditions
and the conditions of maintenance of the facilities, the requirements on the work� ,
places, instruments, production forms and reports and also the requirements imposed
~ on the people working in the production facilities.
In the clean rooms for maintaining the laminar state and, consequently, purity of
the air f1ow, the equipment is arranged so that the spacing between work places
- will be no less than 1 to 1.2 meters, and the distance from the walls to the equip- -
ment, no less than 0.5 to 0.8 meters. In order to decrease the effect of the heat
released by the equipment on the convective transport of dust particles the equip-
ment is built into wall panels so that only the charging devices protrude into the
clean room. This arrange.ment makes it possib le to carry out preventive cleaning of
the equipment outside the clean room. _
In clean rooms, in addition to monitoring the dust content o~f the air atmosphere
and microclimate, the conditions of labor are also monitored.
30
FO~2 OFF[CYAL USE ONILY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OH'FICIAL USE ONLY
~he air of the clean rooma contains particles ~rom hundredths to several tenths of
a micron in size which are not trapped by the finish filters of the purifiers and
are not monitored when analyzing the dust content. These particles can settle and
accumulate int~e facility. For.removal of them.periodic wet cleaning is needed,
sometimes with the addition ot 5% glycerine in order that the dust not separate f rom
the surfaces. The water for wet cleaning of the floors, wa].ls and eeilings must be
pure, better deionized with a specific resistance of no less than 0.5 mohm-cm.
The sources of pollution from the personnel are primarily the skin of man and
clothing. Water vapor, salt, tat and other materials are released through the skin
of man. As a result of constant renewal of the outer laye~ of skin, the dead
particles flake off. Depending on the type of activity the n~ber~~of dust particles
removed from human clothing can vary within broad limits. The~efmre work done in
- the clean room is done in special work clothing made from matei*tal with minimum
dust generation. Before going into the clean area, the workers go through inter-
mediate areas, dressfng rooms and b low-off locks. Observation~of ~he personal
EVH rules is a necessary condition of produdng high-quality microcircuits.
2-5. Basi c Production Features of Microcircuits
The production of IC differs to a great extent from other industrial production and
has a number of specific features. Let us consider the basic ones of them:
1. Almost all the piienomena and proeesses known to science and practice are used in
IC production: mechanical, physical, chemical, various types of treatments, includ-
ing electronic, ionic, laser beams, uarious methods of ineasi~rement and control with
the appli cation of radioelectronic equipment, x-ray television, electron, laser
microscopes and so on.
2. IC production has developed and is being improved at exceptionally fast rates,
influencing, at the same time, the development of in practice all branches of the
national economy.
- 3. IC production uses an enormous amount of materials of various properties, much
of which is in a special class of materials "for semiconductor production" and must
correspond to the requirements of exceptionally high purity. This defines the high
requirements on the microclimate and purity of the production fac~lities, the
production environments, fixtures, and so on.
4. When manufact uring IC b road use is made of group production methods. Deviations
in the pro cess conditions of even one operation can lead to re~ection of an entire
lot.
5. The processed units are distinguished by micron and submicron dimensions, which
requires high precision and stability of the production treatments and conditions,
for otherwise dispersion of the p arameters is~possible not only from substrate to
substrate, but also within a single substrate. In addition, the compleXity of
handling micron size ob~ects requi.res manual labor to be replaced by automated
labor (machines).
31
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
6. Many of the production processes of ~mariufacturing IC are distinguished by
comparative complexity of control, for the~parameters of the elea~ents obtained are
simultaneously influenced by a large .number of factors, the number of which are
difficult to consider. The multifactor nature of the dependence.of t~ie parameters
of the processed ob~ects on the~ conditions of performance of the~ production opera-
tions imposes defined difficulties on the~mathematical-simulation of the technolog-
ical~processes and makes it complicated to write simple programs for computer con-
trol.
7. All of the operations of making IC are regulated and must be performed with '
observation of EVH and in exact correspondence to the production reports and forms
of the microelectronic industry. The observation of the production discipline
maintenance of all of the operating canditions,~app lication of materials only of
the needed types, the All-Union State Standards, fechnical speeifications..and so on
is mandatory for everyone engaged ~n the development and manuf acture of IC. Any
alterations of the process can be made only after approval of the new documents.
8. Among the variety of materials~used, the various types of energy earriers, gas
media and reagents there are toxic, explosive and flammable materials. This
requires observation of strict engineering safety rules specifi c to each work place.
9. Integration of knowledge of many materials, a b road class of phenomena occurring
during the technological processing, the operating principles of complex (and some-
times unique) equipment, knowledge of the organization and economics of production,
the physics of the operation of microcircuit elements and systems engineering is
charaateristic of the technologists. The development and manuf acture of IC is not
within the grasp of a single specialist this is the.work of a collective. There-
fore the technologist specializing in the production of microelectronic devices
must not only have a defined set of skills, but must also be a production organizer.
Test Questions and Assignments
1. Enumerate the basic requirements imposed on the technological production process
- of manufacturing IC. ~
2. Briefly formulate the requirements on the production cycle.
3. What is a production environment? .
4. What groups are the production environments divided into?
5. What determines the requirements on the air environment of the production
facilities?
6. Why and by what attributes is production broken dawn into p ro duction sections?
7. What parameters characterize the quality of the air environment and why?
8. How many types o~ facilities are there with.respect to.temperature-humidity .
parameters and dust content?
32
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
FOR OFFIC[AL USE ONLY
9. What 3s a clean room?
10. What types of aerodynamic systems.for clean rooms do you know?
11. Com~are clean rooms of different aerodynamic spstems. -
12. What is the role of 1oca1 work spaces?
13. Compare the local work space knbwn to us.
14. Explain the means of monitoring temperature, humidity and dusf content in the
air.
15. Briefly formulate the requirements imposed on process~�gases.
r6. What is the role of the gases in IC production?
17. What measures are taken to insure the required purity of gases? .
18. What is the role of water in IC production?
Y9. What distinguishes the final purity of the water from the preliminary purity?
2'0. Cor~are distilled water with deionized water. ~
21. What is the specific resistance of water used for the production of semi-
conductor devi ces and microcircuits?
22. Ha�,a is the water purity quality controlled? -
- 23. Enumerate the basic EVH principles.
24. What are the basic IC production features?
33
FOR Ol~ F~CIAL USE ONI~Y
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
CHAPTER 3. MANUFACTURE OF SEMICONDUCTOR PLATES AND DIELECTRIC SUBSTRATES FOR
MICROCIRCUITS
3-1. General Information About the Machining of Semiconductors and Dielectrics for
IC
~ Abrasive Machining
Semiconductor materials and monocrystalline dielectrics (sapphire and spinel) go
to the machining production section in-the form~of ingots; ~inorphous and other
dielectric materials (glass, pyroceram, polykor, ceramics) in the form of
sheet billets. Semiconductors~and dielectrics are distinguished by high hardness
and brittlenes~s; therefore the methodo used for mach3ning metals cannot be used to
obtain bill~ets (semiconductor plates and dielectric substrates) for the manufacture
of IC structures. In order to obtain semiconductor plates and dielectric substrates,
abrasives of various types are used. The essence of abrasive machining consists in
tne mechanical effect of a harder and less brittle material the abrasive on a
less hard and more b rittle material. The mechanical pressure from the tool is
transmitted to the ab rasive grains and f rom them to the machined materiala The _
ab rasive grains produce local microdestructions of the surface of the machined
material, the released microparticles of which are removed from the machining zone.
Abrasive Materials
B asically synthetic abrasives ~.re used in microcircuit~production: diamands, sili-
con carbide and electrocorundum.
Syn.thetic diamonds are not inferior to natural diamonds with respect to their
II:echanical properties. A diamond is the hardest o~ a11 known materials. The Mohs
hardness of diamonds is 10. Among the abrasive materials diamonds are in a special
class.
Sili~~on carbide (SiC carborundum) has different color depending on the amount of
impurities from light green to black. In IC production most freq~ently green _
- silicon carbide (KZ) is used. The Mohs hardness o� silicon is 9.5-9.7.
WhiCe synthetic corimdum is crystalline aluminian oxide (A].20 3) with different
impurities (0.5- 1.5~6). Synthetic cordundum is inferior to silicori carbide with
respect to h ardness,~but the strength of the synthetic corundum is hi~her. The
Mohs hardness is 9-9.2.
34 -
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
- Table 3-~].. ~ Char�acter�istics��o�� Eaurde~ced~.Abrasives�~and~~Diamonds r
~ Grairi'size'""""~~"'
Group According to Accordi~lg to ~rain size
AlI-Union All-Union of basic
State~�Standa~d - -State~�Standard~ fr�action,
. . 3647=7I . 9206=70 . . . .inicroris
12 - 160-125
10 - 125-100
Abrasive grinding 8 - 100-80
pawders 6 - 80-63
5 - 63-50
4 - 50-40 -
3 - 40-28
~ M63 - 63-50
M~~ - 50-40
M40 - 40- 2 8
Abrasive M28 - 2~20
_ mi cropowders M20 - 20-14 -
. M~-4 - 14-10
Ab ras ive M10 - 10- 7
fine micropowders M7 - 7_,5
~ - 5- 3
- 60/40 60-40
- 40 /2 8 40-2 8
Diamond micropowders - 28/20 28-20
- 20/14 20-14
- 14/10 ~14-10
- 10/7 10-7
- 7/5 7-5
- 5/3 5-3
- - 3/2 3-2
- 2/1 2--1
- 1/0 2-1
- 1 or less
Abrasive powders are divided into four groups depending on the grain size: grind- -
ing grain, grinding powder, micropowder and fine powder. The three last groups
of powders are used in IC producffion. Powdered diamonds are divided into grinding
grain and micropowders with respect to grain size, and with the micropowders are
used in IC production. The designations for the grain sizes of abrasive and
diamond powders with different grain.size in the basic fraction are presented in
- Table 3-1.
35
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
_ The type of material and the gra~n size' are part of the . des.ignation for abr.asive
or diamond powders. For example,~EBM14 is white synthetic corundum with basic
fraction grain size from 14 to 10 microns; ASM 10/7 is synthetic diamond with
basic fraction grain size from ~3.0 to 7 microns. ~
- Abrasive Tools -
Ab rasive materials are used in the free (suspension, paste) and bound (disc) state.
_ Abrasive and diamond suspensions are mechanical s~spensions of the corresponding
powder in water, oil or another liquid. The pastes are bas~ca~ly prepared from
diamond micropowders and surface-active material~ wliich proieofie improved quality of
machining the surface. The discs are of two types: simple metal and metal with
diamond-containing cutting edge. Metal discs with diamond- cutting layer bound to
their end or peripheral parts are used for grinding. T~e strength of the discs
depends on their dimensions, the type of diamend pawder, its concentration and the
- machining conditions. The binder which binds-~the diamond grains to the cutting part
must provide for self-sharpening of the tools, th at is, it must hold the diamand _
grains and at the same time not interfere with removal of the dulled grains from
the diamond-containing layer.
Machining Operations ~
Machining includes the follawing operations: cutting, grinding, polishing.
Crystallographic orientation which serves to determine the angle of devl.ation of the _
plane of the end of an ingot from the given crystallographic plane, is used before
- cutting monocrystalline ma.terials. Cutting is used to separate the ingots and
sheet billets into plates and substra.tes of the required size and also to separate
semiconductor plates or substrates with finished structures of the microci~cuits '
into individual crystals or plates. Before cutting into p lates the ingots of non-
uniform cross section are sized, and the large size ingots are laid out in bars.
Th e separation of the sheet materials into substrate billets is done with an
allowance for subsequent grinding of the ends.
The grinding and polishing are finishing operations, and they are used to insure
accuracy of the dimensions and quality of the plate and substrate surfaces.
It is necessary to attach the samples in special holders to perform the machining
operations. The fastening is most frequently done by gluing using a mastic based
on ED-5 or EI~-6 epoxy resin, BF adhesive, shellac, glyptal resins, and so on.
= During bonding it is necessary to insure cantinuity and uniformity of the thickness
of the bonding layer, where the thickness of the bonding layer must be as thin as
passible. The quality of attaching the ingots in sheet billets infi~ences the
planarity and parallelne~s of the sides of the plates�and substrates, the thickness
of the surface layer with disturbed crystal structure, and so oa.
?-2. Abrasive Cutting
~o~id Abrasive Cutting. Discs with inside and outside diamond cutting edge are used
- for bnund abrasive cutting.
36
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FUR OFFICIAL USE ONLY
_ Cutting by a disc with an inside~ cutting edge is basically used for cutting ingots
into plates or bars (Figure 3-1~. The cutting is distinguished by high output
c~pacity (60-80 mm/min for silicon), and it insures good quality of machining
_ (eighth to ninth class for silicon). The waste is sma11: with a thickness of the
disc base of 0.1 mm the width of the silicon cut does not exceed 0.~25 The
recovery of the material from the waste is comparat~~ly easy.
- The disc is stretched in the radial direction and it is fastened by the peripheral
p~rt on the head of the spindle of a cutting tool. The fastening around the
_ periphery insures low vibration when the disc ~otates at 3000 to 5000 rpm. Before
b~ginning cutting the ingot glued to the mandrel is rotated using the machine tool
rotary device so that the plan of the cut will be parallel to the giveri crystallo-
graphic plane.
.
.2. �
3 "
4
5 ~
Figure 3-1. Diagram of a cut made by a dis c with inside diamond-
containing cutting edge.
_ 1-- nozzle supplying the cooling and lubricating fluid; 2-- ingot; "
3-- disc base; 4-- mandrel for fastening the ingot; 5-- cutting
edge of the disc
The diamond grains bound to the cutting edge press with large force against the
machined surface when the dis c is rotated forming scratches from which cracks
penetrate deep iz~to the ingot. The intersections of the set of cracks make punc-
tures, and the particles of the material removed f rom the ingot. In addition to
_ brittle fracture of the material, the protruberances from its surface are also cut
away by the diamond grains .
A jet of liquid is fed to the cutting zone which serves to remove the heat
generated during cutting, to remove the particles of cut- away material and the
deteriorated diamond grains, to decrease friction, and to lessen the impact-
- vibration forces. On feeding the fluid to the microcracks of the machined material,
as a result of capillaxy wedging it promotes rupture of the material. A 3 to 5%
aqueous solution of calcined soda or NIIALMAZ fluid (0.6% tr.isodium phosph ate,
0.3% borax, 0.25% calcined soda, 0.1% sodium nitrate and 98.75% water) is used as
the cooling and lubricating fluid. The spent cooling and lubricating fluid with
waste material is discharged by centrifugal force outside through the lateral
openings in the drum.
37
' FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE OIVLY
The cutting conditivns the disc rpia, the ingot feed rate, the cooling and
lubricating fluid constmmption depend on the' properties of the machined material,
its dimensions, thickness of the~ cut plates, the reqvire~nents on finish and pre-
cision of the machining. ~ncreasit~g the~ clisc rpm inereases the cutting rate, but
vibrations at temperature rise and, consequently, the precision-~and finish of the
machining diminish. With an increase in feed rate of the ingot, the precision and
quality of the machining decrease. In addiCion, at high feed rates�the disc bends
which influences -the shape of the cut-off platea or leads to breaking of them. The -
cooling and lubricating fluid consumption is selected so that it will exit through
the side openings of the drum and the cut plates will not break under the pressure
of the pawerful jet when hitting the collector.
Thus, in order to insure high q~ality of the plates with maximum output capacity,
the cutting conditions are made optimal.
, 1
, 2
! ~ 1 \ .
- - - - 3
~ 4
S
~ s
. '
Figure 3-2. Diagram of disc cutting with outside diamond-containing
cutting edge.
1-- cooling and lubricating fluid feed nozzle; 2-- cutting edge of -
the disc; 3-- disc base; 4-- cut plate; 5-- bonding material;
6-- mandrel for attachment of the plate
The recommended cutting condittons are as follaws : ~Linear velocity of the disc
17-22 m/sec corresponding to a tool spindle rpm of 4000 to 5000; ingot feed no -
higher than 30-40 mm/min for silicon and 40-50 mm/min for germanium and gallium
arsenide; cooling and lubricating fluid consumption 2.5-4 liters/aninute.
The cutting by discs with inside diamond cutting edge also has deficiencies: com- -
plexity of pulling up the disc, depende~ce of the quality and precision of
- machin~ng on the fastening of the ingot, strengthened quality of the tool.
Cutting by disc with outside diamond eutting edge is realiz~d in accordance with _
the diagram presented in Figure 3-2. The cutting mechaniam is analogous to the
disc cutting mechanism with inside cutting edge. The disc with the outside cutting
edge is fastened to the machine tool spindle by its central part. This fastening
does not insure high rigidity o~ the disc. ~Under the effect of cutttng forces
the disc ~an bend sharply in the transverse diiee~ion and vibrate in the axial
direction, which promoCes the formation of defects- on the surface of the cut sample -
and also an increase in width of the cutting. In order to increase rigidity, the
discswith outside cutting edge are made on a thicker base (0. 3-0. 7~) . The
38
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
m3nimwn width of cut of silicon is about 0.45 mm, the machin~ng of the surface
corresponds to roughness class 6=7. ~It is inexpedient to use such discs to cut
l,arge-diameter ingots.
Discs with outside cutting edge are used to cut t~e substrates of the requi~ed -
dimensions from sheet billets of glass, pyroceram, ceramic, polycor and also to
separate semiconductor plates and dielectric substrates with finished structures ~
into crystals and plates. In this case the depth of the cut is small, and discs
with a thinner b ase can be used. In order to increase the output capacity f re-
quently several cutting discs separated by inserts are used.
Free Abrasive Cuttings
When cutting by means of a free abrasive the mec~anical forEes are transmitted to -
_ the suspension grains using a moving tool: d~sc, steel web, wire.
Cutting by a disc using abrasive suspension is less efficient, and the recovery
of the materials from the waste is inore complicated than when cutting by diamond
discs. The basic advantage of the method is the better quality of the machining
of the surface, for the abrasive micropowders are not as hard as diamands.
The abrasive suspension is continuously fed to the working zone. With a ratio of
liquid and solid phase of the suspension of 3:1 or 4:1, all of the abrasive grains
are in the suspended state, and they are uniformly distributed in the cutting zone.
Special cooling of the disc is not required in this case, for the heat is removed by
the liquid part of the suspension. The disc delivers the abras3ve grains to the
cutting zone. Turning at high speed, it e~ects the grain, and the grain hits the
surface of the material with great force. Brittle rupture takes place, b ut in
pr�ctice there is no cutting off of the protruberances of the machined surface. -
The cutting by steel b lades (Figure 3-3) fastened by means of inserts to a special
- holder and undergoing reciprocal motion, is comparatively rarely used, for it does
not provide high output capacity or high quality of machining. The cutting speed
depends on the speed of movement and length of stroke of the blades, the size,
shape and hardness of the abrasive grains, the abrasive concentration in the suspen-
sion, the hardness of the macl~iined material, the pressure of the blades of the ingot,
the number and thickness of the steel blades. For cutting silicon ingots usually a
suspension b ased on silicon carbide with M20 or M28 grain size is used; for cutting
silicon plates, M10-M20. .
_ 1 2 j,~~ 3 4
,
5 , -
6
'
8 . 7~ p~
Figure 3- 3. Diagram of steel blade cutting.
1-- holder; 2-- steel blade; 3-- suspension feed nozzle; 4-- cut
plate; 5-- insert; 6-- table; 7-- weight; 8-- lever
39
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
The rigidity of the blades is 1aw. During the.cutting process the blades wear,
and their tension.decreases. The paired`inserts determining the thickness of the
cut plate or crystal width are difficult to make~in identically exact sizes.
Wire cutting is used to cut plates and substrates. The wires wear quickly. There-
fore in addition to reciprocal.movement the wire also is rewound. The method is
distinguished by low efficieney and breaks in the wire, The advantages of the
method include the following: better quality of cutting than when cutting with
blades, the possibility of cutting without damagYng the films applied to the blades
and substrates; small width of the cut (0.08 to 0.2 mm); the possibility of rapid
replacement of the worn wire.
Ultrasonic cutting is carried out using abrasive s~spension and a tool that.tmder-
goes reciprocal'vibrations at ultrasonic fregueney. These v~brations are received _
from a magnetostrictive emitter (Figure 3-4), which is a core with a winding
assemb led from sheets of ferromagnetic material (nickel, permallay, and permendur).
AC voltage of ultrasonic frequency is fed to the-core winding. The alternating
current flowing through the core windings creates a variable magnetic field causing
the magnetostriction phenomenon, that is, the convers;on of the elECtromagnetic
field oscillations to mechanical vibrations of the core. In order to increase the
amplitude of the core vibrations and the energy concentration on the tool, a
concentrator is attached to the core. The vibrations are transmitted from the
concentrator to the tool. The suspension is continuously fed to the machining zone.
The cutting speed depends on the frequency of the ultrasonic vib rations (the tool
vib ration:amplitude) and also the parameters of the ab rasive suspension and the
machined material.
During ultrasonic cutting, the shape and size of the tool are copied on the
machined sample, and as a result it is possib le to cut out crystals of complex
configuration, for example, circular ones, and also to perform embossing and
obtaining lines, grooves, craters and holes which is impossible in other methods of
_ mechanical cutting. During the cutting process the tool must be raised and lawered
periodically, for as it penetrates into the machined material the exchange of
abrasive suspension and removal of waste become comp~icated.
_ . _ . '
.v3r
Z ~ -
~
. ~ 3 �
5
_ ~"O 6
Figure 3-4. Ultrasonic cutting system: ~
1-- magnetostrictive emitter; 2-- ultrasanic voltage generator; -
3-- concentrator; 4-- tool; 5-- machined~plate; 6-- table;
7-- abrasive suspension feed nozzle
Key :
a. ultraeonic generator -
40
FOR OFFICdA~. US~ ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
Ultrasonic cutting is used to separate plates and substrates and to obtain shallaw
reliefs. Ultrasonic cutting is highly efficient, but it does not provide high
quality machining.
3-3. Grinding and Polishing Billets for IC Structures
Grinding. After cutting, the plate and substrate billets have dispersion with
respect to thickness, errors in shape and significant surface layer with disturbed _
structure with respect to thickness. In order to improve the precision and quality
of machining the billets, the following finishing operatiorts are performed grind-
ing and polishing. Grinding is the finishing of the billets by machining on solid
~ disc grinders made of cast iron, steel, glass and other materials by suspensions
with abrasive grain size from 28 to 3 mic mns or using diamond grinding discs with
grain size of 5 microns. The billets are ground i~? several steps with successive
application of finer and finer abrasive grains and, correspondingly, with gradual
. improvement of the quality of surface machining. During the grinding of sili con,
class 7 to 12 surface roughness is obtained.
' I P
~ ,2
/ ~ ~ / 3
i y
~ a~ -
' ~ .
' 4 -
/ 5
' _ 6
_ ~ ~
b)
Figure 3-5. Diagram of one-way grinding of substrates (a) and
the head location of the grinder (b).
1--- suspension feed nozzle; Z-- subetrate; 3-- grinder;
4-- roller; S-- weight; 6-- grinding head
Grinding with a free abrasive d'epending on the type of machine tool can be
= accomplished either one-sided or simultaneously from two sides.
,In one-sided grinding (Figure 3-5) the billets are glued to special heads th at move
freely and are held against the sur~ace of the grinder only by the rollers. In
- order to obtain a tighter fit of the heads against the grinder surf ace, wei~hts are
used. An abrasive suspension is ~ed to the mach~.ning zone from a batching unit.
Under the effect of frictional forces, the motion of the grinder is transmitted to
the grinding heads. An interlayer of abrasive suspension is formed between the
- surfaces of the machine billets and the grinder. ~During the rolling and turning,
the suspension grains act on the machined surface, form-Lng cracks. The chip size
41
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
from the material and, consequently, the grinding apeed depend on the aize and -
hardness of the abrasive grains and the pressure on them from the grinder. During
- ~rinding the surface of the grinder and the~grinding head became worn; therefore
they are periodically checked, and if riecessary the surfaces are polished. For
one-sided grinding without boriding the billets are sf acked in recesses in special
= separators. It is possible to grind billets of different thickness simultaneously
by this meti~od. Grinding without bor~ding provides better shape precisionand high
quality of surface machining, but as ~a result of splattering of the suspension it
cannot be used at high grinder rpm. '
p
- - 1 _
.
2
3
_ 3
~ 2 1
~
o- ,
Figure 3-6. Diagram of diamond disc grinding.
1-- diamond disc; 2-- substrates; 3-- grinding head
nao-sided grinding by a free abrasive is accomplished using two grinders. The
lower grinder is usually stationai*y, and the upper grinder freely self-adjusts to
_ the machined billets placing in separators. The separators are rotated by pinions
around their own axes and the grinder axis. The billets entrained by the
separators undergo complex movement with respect to the grinder surfaces. Two-
sided grinding by a free abrasive is more efficient, it insures high surf ace
machining precision and does��not require bonding. The bending of the billets is
decreased during two-sided ~rinding, for the residual mechanical stresses are more
imiformly distributed.
Diamond disc grinding is performed in accordance with the diagram presented in
~ Figure 3-6. The table to which the~ ~billets are ~astened and the~ diamond discs are _
driven by different electric motors. A coolant is.fed to the table surface. Bond
ab rasive grinding is the most efficient method insuring high roughness class of the -
surface, but as a result of the impact ef~ect,o~ the grains against the machined
material it leaves a quite deep surf~ce layer with distu~rbed structure. ~n
addition, the rigid fastening of the tab le and grinder shafts leads to wedge
shaping of the plates and substrates if the uaachine tool wears.
42
FOR OFFIC~AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Polishing. The plate and substrate billets are polished~using soft finishing
polishinR discs. Fabr3.cs~cambric, .velour, chainois, felt, and synthetic fabrics) are
stretched on an ordinary grinding disc and fastened by a clamp~for this purpose.
Polishing is carried out in several steps, ~ust as is grinding.
Preliminary polishing of silicon plates is realized by diamond suspension with grain
size of no more than 3 microns using polishers b ased on fab rics with bulges. The
- fabric must not wrinkle when it is stretched on the disc. Polishing is accompanied
by plastic deformations of the surface layers of the plates and ~igh heat generation.
In order to prevent burning of the liquid part of the suspension, softening of the
adhesive and rupture of the plate, the polishing is done at low rpm of the po lisher.
The surface of the machined plates has a fine network of lines ("dia~nond b ackground")
arising under the effect of the sharp cutting edges of the diamond grain. In order
to remove the "di ~nond background" and decrease the surface roughness, �inal fine
polishing is carried out by mechanical, chemical-mechanical or chemical-dynamic
methods.
Fine mechanical polishing is accomplished by soft polishing compounds b ased on
aluminum, silicon, chromium, zirconium and other oxides with grain size of less than
= 1 micron using polishers with sleazy materials in which submicron powder grains are
"submerged." This decreases the working surface of the grains and improves the
quality of the surface machinin g of the plate.
Chemical-mechanical polishing is distinguished by the fact that in addition to the
ordinary abrasive mechanical effect the machined surface is subjected to the additional
effect of chemical that intensified the process of removing the material.
Polishing of silicon with the application of aerosyl (silicon dioxide) or zirconium
dioxide as the abrasive is carried out using water suspension with the addition of
alkali (pH=10 to 11). At pH~9, the abras~ve effect of the suspension predominates,
and the quality of the pol3shing becom~s worse. At pH>,11, th~ silicon surface is
_ ~ggravated. In order to insure more uniform distribution of the micropowder, small
amounts of surface-active-materials (ethylene glycol, liquid glass) are added to the
suspension.
Chemical-dynamic p~lishing is a method in which the abrasive effect on the machined
surface is entirely excluded. As an example it is possible to consider the po lishing
of silicon substrates by bivalent copper ions. A solution containing copper nitrate
and a~onium fluoride is continuously fed to the chamois polisher. A layer of pure
copper is precipitated from this solution on the machined surface
Cu~'+2e->Cu~b ; ( 3-1)
and the silicon dissolves simultaneously with the formation of water-solub le
silicates:
Si-~S i~'4-F4e; }
}
~ Si+4+6~"-~SiF6-23 } (3-2)
SiF6-2+2(NH4)+~(NH4)2SiF6.}
43
FOR OFFICIAL U5E ONII~Y
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
The copper layer is removed mechanically by the moving polisher-, and the copper
is again precipitated on the cl~an machined surface, and so on.~ The polishing of
- the silicon by copper ions is highly efficient (100 microns/hour), but it is dis-
tinguished by difficulty of removal of the~copper residue fram the surface of the
plates.
3-4. Control of the Plates and Substrates After Machining
- Defects and Criteria for Evaluating the Quality of Plates and Substrates
The surface condition of plates and substrates, varfations in its shape~and size
have a significant influence on.the subsequent technological operations, and to a
significant degree they determine the quality and percentage yield of usab le micro-
circuits.
The nonparallelness of the size of plates and substrates~is estimated by difference
in thAir thickness hl-h2 in a given-.length k(Figure 3-7, a). Nonplanarity eh is
the greatest distance from the points of the real surf ace to the ideally p lane su~-
face (Figure 3-7, b). The bending f is the greatest distance from the points of the
actual profile in radial croas section to the corresponding flat surface of the
, ad~ acent profile (Figure 3-7, c) .
The quality of surf ace machining is characterized by the depth of inechanically
disturbed layer and roughness.
- t ~
~
h �
- dh
% /,,i
b)
f
' .
- Figure 3-7. Deviationa o� the substrates from the precise form.
a-- nonparallelness o~ aides wedge ahape; b-- nonplanarity;
c bending
A mechanically disturbed layer consists o~ thr-ee parts (Figure 3-8): the disturbed
relief layer has randomly arranged protrusions, cracks and pimctures; a cracked
layer has individual noncrumb ling puncturea and microcracks running deep into the
layer; the deformed layer has dislocation pileups, continuations of microcracks
and the mech anical stress zones arranged around tliem.
44
_ ~'OR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
~ :Roughness (Figure 3~9) can be estimated by the arithmetic mean deviation of the
~ ;profile Ra, that is, the arithmetic .mean absolute values of the profile . deviations
::within the limits of the base line !Ct
~ ,
_ R.= ~ ~~y~~ (3-~3)
~_i
or the height o,f the unevennesses in the profile RZ with respect to ten points :
_ _
5. _ _ . ._6 -
1
RZ= 5 ~ H1 max-}' H1 mIn � ~~4~
t=~ i=?
Depending on the sizes of the parameters Ra and RZ, the quality of the surface
machining is estimated by the corresponding roughness class.
Control of Plates and Substrates. The controllabZe parameters are the thickness of
the substrates, planarity, nonparallelness of the sides (wedge shape), bending,
, thickness of the mechanically disturbed layer, roughness, the presence of
. scratches , chips and lines.
_ \
_ _ ,
I
, II
- ' ` ' ~
~ M
Figure 3-8. Mechanically disturbed layer of the plate surface.
I-- relief layer; II cracked layer; III deformed layer;
IV undisturbed structure of the plate
Z
(1).
. Jlurfua Be~cm noB �
yt
~ ~ e
N
Z Z
~ C
~ ~ E y~
~ Jlunun Bnaauy z~
F
z ~ ~2~
Figure 3-9. Qro~ile of the rough sur~ace o~ a suhstrate
Key:
1. protrusion 13ne
2. depression line
45
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
7'he thickness is measured by a clock type indicator which is fastened to a stand.
The substrate is placed on the table. The indicator is ad~usted.so that its
zero will correspond to the position of the measurit~g pr~be bn the surface of the
table. The thickness is measured at several points, and then the mean value is
determined. The precision of the irid~cator measurement is 1 micron. Higher pre-
cision (0.5 micron) is provided by the IZV-2 optical length gauge. The clock
indicator and length gauge can also be used to determine the wedge shape of the sub-
strates.
The nonplanarity is determined by the plane-para11e1 standard glasses wbich are
- applied to the polished surface of the attached plate or substrate. In the sections
where there is an air space, an interference pattern arises as a result of super-
position of light beams reflected from the controlled surface of the adjacent
surface of the standard glass, by which the nonplanarity is ~udged.
It is possib le to measure the bending of plates and substrates by a toolmaker's
straightedge during observation by a microscope. The size of the clearance between
_ the straightedge and the bent plate can be measured with precision to 1.5 microns.
The thickness of the mechanically disturbed layer can be measured by various methods,
- the simplest of which is based on using the dependence of the speed of chemical
pickling on the degree of the disturbance of the crystal. -As the mechanically dis-
turbed layer is pickled away, the pickling speed decreases. The time at which the
pickling rate of the single crystal becomes constant signals removal of the entire
disturbed layer.
The surface roughness within the limits of cl.ass 12-14 is determined ~sing the
MII-4 microinterferometer. At the locations of the microunevennesses, the inter-
ference bands are distorted, and it is possib le to determine the magnitude of the
irregularities by the degree of distortion. -
The presence of scratches, lines, chips and traces of contaminattbn of the surface
can be detected by observation using the MBS-l.and MBS-2 stereoscopic m~�roscopes
or the MI1~7 metallographic udcroscope. The dimensions of the indicated defects
can be meas ured using the scale o~ the eyepieces. The microscopes have,a compara-
tively sma11 field of view and do not permit the entire surface to be monitored,
especially the defects such as the smooth trregularities with small height grsdients,
sags along the edges of the~p~ates and the b ends. These defects can be monitored by
_ irradiation of the polished surfaces by parallel or diverging laser beams and sub-
sequent analysis of the interference pattern. _
Test Questions and Assignments
- 1. Repeat the questions-from the ad~acent topics: "Materials for IC substxates,"
"Materials for machining," "Requirements on IC substrates."
2. What is abrasive machining? What is the role of the tools during abrasive
machining?
3. What abrasive materials are used for machining?
4. What explains the fact that the grain size of."the powders in Table 3-1 is
designated in accordance with dffferent All-Union State Standards?
- 46
FOR OFFICIAIL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR O; ~ICIAL USE ONLY
5. Decipher the abrasive designations FBM20, KZM28, ASMZ/2. -
6. Explain what the free and bound s~tate of an abrasive is.
_ 7. Which production operations :belong to machining?
8. For what materials is preliminary crystallographic orientation requi:ted? Why?
9. What is the role of grinding and polishing during machining of billets?
10. What is the role of the quality ~f attaching the ingots and plates? _
11. What are the advant~ages of cutting by a disc with inside diamond cutting edge?
What are they caused by?
12. What is the cutting mechanism by a disc with inside diamond cutting edge?
- 13. What is the practical implementation of cuttirig by a disc with inside diamond
- eutting edge?
14. What is the role of the cooling and lubricating fluid in the cutting process?
15. Which parameters determine the disc cutting conditians?
16. From which arguments are the optimal values of the disc rpm, ingot reed rate
and cooling and lub ricating fluid flow rate selected?
- 17. What are the deficiencies of cutting by a disc with external cutting edge?
18. Compare the methods of cutting by a disc with external anc~ internal cutting
edges.
19. What are the differences in the cutting by free abrasive from cutting ~,y -
bound abrasive?
20. Which tools are used for free abrasive cutting?
21. Why is the cutting by blades and wire r.arely used in IC production for cutting _
ingots?
22. What is ultrasonic cutting?
23. What is the practical implem~ntation of ultrason3.c cuttin$?
24. What is the uniqueness of ultrasonic cutting by comparison with other methods?
25. Compare all the cutting techniques from the point of view of application,
efficiency and quality.
26. What is mechanical grinding and polishing? Is there a theoretical difference
between them?
- 47
F7R OFFICIAL USE ONLY -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY -
27. Why are mechanical grinding and polishing.performed in several steps?
28. What is the practical implementation of free abrasive grinding?
29. Give a description of the method of diamond disc grinding.
' 30. What distinguishes a polisher from a grinder?
- 31. WhaC is preliminaYy polishing and how is it done? -
32. Which methods are used for final polishing?
~
33. Compare chemical-mechanical polishing with precision mechanical polishing.
34. Haw is chemical-m~chanical polishing of silicon realized?
35. Compare the chemical-mechanical polishing of silicon by aerosyl with silicon _
polishing by copper ions.
36. Wb.at are the basic shape and surface defects of platies and substrates?
37. Enumerate the monitored substrate parameters after machining and explain how
they are determined.
38. Enumerate the factors influenca.ng the efficiency and qualfty of machining.
39. What are the deficiencies of the machining method?
F~
, 8
F~R OE'F~CIAI. 1USE 03~IlLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFF(CIAL USE UNLY ~
QHAPTER 4. CHEMICAL TREATMENT AND CLEANING OF THE SURFACE OF SEMICONDITCTOR _
PLATES AND SUBSTRATES
4-1. General Informatinn
~oncept of the "Clean Surface." The basic volume of cleaning operations ln micro-
circuitry technology pertain to preparation of the surface of the plates (sub-
strates) for the structure manufacturing process. Atoms of the plate (substrate)
~aterial located on the surface have many more unsaturated bonds than atoms in its
body. This gives rise to high chemical activity on the surface. It is in practice
impossible to obtain an ~deally clean surface, that is, a surface without foreign
impurities.
Under microcircuit production conditions the plates and substrates are in contact
v~ith various media, and it is impossib le to completely shield them against the
aclsorption of impurities; therefore the concept of "clean surface" used has a rela-
tive nature. A surface which has an impurity concentration that does not prevent
reproduction of the given values and stability of the microcircuit parameters is
considered to be technologi ca11y clean. Even in the case of nonrigid requirements
the imp urity concentration on such a surf ace must not exceed 10" 8 to 10-~ g/cm2.
- The processes of cleaning plates (substrates) are intended for the removal of con- _
tamination to a level corresponding to the technologically clean surface. The
goals of cleaning monocrystalline plates (substrates) also include the removal of
the surface layer with the structure disturbed during the machining process. The
presence of a mechanically disturbed layer does not allow high-q uality and reproduci-
b1e sem~conductor and film structures to be obtained on semiconductor plates,
sapphire and spinel substrates.
Sources of Contamination of Plates and Substrates. The b asic sources of contamina-
tion of the surf aces of plates and substrates are the following: abrasive and
~ adhesive used during machining; dust in the air of the production ~acility;
objects that have come in contact with the plates and substrates (equipment, tools,
fittings, packaging, transportation and storage); pxoduction environments; organic
and inorganic reagents, water, and so on.
~Contamination of the plates and substrates is possible in practice in all operations
of the manufacture of microcircuits. Therefore throughout the entire technological
production process the cleaning of the surfaces of the plates amd~substrates is
realizecl again and again. The most important processes are the processes of
49
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY -
cleaning the surface after machining, before thermal processes in which the
diffusion of the impurity atoms and the probability of the formation of undesirable _
chemical compounds and alloys before the applicatidn of various types of coatings,
films, layers, and so on inc-rease.
~ Types of Contamination. The technology of chemical treatment and cleaning of
plates and substrates is determined in each step b3~ the nature of the possible con-
tamination and requirements~ imposed on the surface. It is expedient to classify �
a the possible contamination on the surface of plates and substrates by their physi--
cal-chemical properties, for the latter basically determine the choice of inethods
of removing the contamination:
Organic contaminants are primarily greasy nonpolar contamination by adhesives, the
oil from the machine tools, the operator.s' hands, and so on.
Water-soluble polar contaminants include salts, acids, pickling agent residues,
fluxes, and so on.
Physical ~ontaminants include dust particles, hair, abrasive and other foreign
- particles not chemically bound to the surface of the plates and substrates.
' Contaminants chemically bound to the surface of the plates and substrates include
oxide films and other compounds.
Gases adsorbed by the surface.
Man_y types of contaminants can be present on the surface of the plates and sub-
strates simultaneously, The most difficult to remove are the organic and some con-
taminants chemically bound to the machined surface. Among the physical contaa~i.nants _
it is most difficult to remove small abrasive grains introduced into the surface -
layer. Among the water solub le polar contaminants it is dif ficu~t to remove the
mobile metal ions which change the electrical conductivity of the surface causing
current drift and the appearance of inversion semi.conductor layers and at the same
time having a harmful influence on the stability of the IC parameters.
Classification of the Cleaning ~Iethods. From the point of view of the mechanism of
the processes, al~ of the cleaning methods can be provisionally divided into
= physical and chemical methods (Figure ~?-1). I~ the physical methods the contam-
inants are r~moved by solution, annealing and also treatment of the surface by ions
of in~rt gases accelerated to high energy. In cases where it is impossible to
- remove the contaminants physically, chemical methods are used in which the con-
taminants located on the surface or in the surface layer are converted to new chem--
ical compounds and then ~asily removed. The cleaning given by the removal of the
- surface layer of the plate and substrate is called pickling.
- In accordance with the applied means, the cleaning can bE divided into liquid and
dry. It is vezy complicated to select a liquid ~neditun which siinultaneously removes _
_ all possible surface con.taminants; therefore the liquid cleaning includes a number
of successive operations. Water-insoluble greasy organic conta~h3.nants make the
surface hydrophobic, that is, poorly wet by water and th~ majority of solvents. -
For uniform cleaning the surface of substxates must be converted to hydrophilic,
that is, a state that is well wet by water. The opera~tibn~ of removing the greasy
50
~~R '''~~~'I~Cd~.~. 115~ ONd.`~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
- contaminants accompanied by canversion of.the.surface from the hydrophobic state
to the hydrophilic state is called degreasing.
;Dry cleaning, as a rule, is the finish cleaning, for it is done di~ectly before
- ~erforniance of the subsequent technologiEal processes, for exampie, before spraying
.pn the films, oxidat3on, photolithography, and so on. Dry cleaning includes
annealing, gas, .ion and p.lasma-ctiemi.cal pickling. These methods exclude the
necessity for the application of expensive liquid reagents dangerous to work with
and also the problems of the interogerations storage of the plates and substrates
~ and cleaning of the waste water which are of little importan~ce when using liquid
_ ;cleaning means. In addition, the dry cleaning processes are more controllab le and
-are easily automated.
o~aC~iA t~neCa~u ;i lroA~o~c -
, . . YHA1t00TH88 . ~ CY%~HH .
mH3N49CNBH XNYN49CK8ft � mM3riY6CR8R , XNYN4BCIS8iI
x M '
~C a
U '~i O � ys ~
5 SJ U ~ . ~ 7XQ S G
y~ 'U tp O ~ . . ' ~ ~ m~ i7 {~p
~ F~i 4' O ~R ~4 fl ~ R1 ~ N F
~ R7~1`. RI 71 ~ 4f W O~
I ~ N Iq ~ IC H . Lpa, F ~ F O
w~L 10 ~ W N M N O U ' `
S PI 5 ~ ~ ' O ~7 O Y
~ ~ ~ ~ ~ ~1 ,
G ~ ~ 9 ' ~
U ~
{ q O ~ ' / ~ .
[ ~ F
(6)' (7) , (8) (9) (10) (11) (~12) (13)
F_ ~ure 4-1. Classification of inethods of cleaning plates and
substnates.
Key:
"l. Cleaning of plates and substrates.- 8� Degreasing in soap and aimnonium
2. Liqu3d peroxide solutions -
3. Dry 9. Acid pick~.ing
4. Physical 10. Aiznealing
5. Chemical 11. Ion pickling
6. Degreasing in organic solvents 12. Gas pickling
7. Washing in water 13. Plasma-chemical pickling
51
FOR OFFICIAL USE ONLY _
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
- FOR OFFICIAIL USE ONLY
4-2. Methods of Liquid Treatment o~ Plates and.Substrates
Degreasing. Almost all greasy contaminants are.effectively dissolved in organic
solvents (carbon tetrachloride, benzene, toluene, isopropyl aloohol and.others).
When the specimens are submer-ged in the solroent, the greaee molecules are gradually
separated from their surface, which as a result of diffusion are uniformly dis-
- tributed over the entire volume of the solvent. The separatfon of the grease
molecules from the treated surface arises from their natural v~bration movements
- and attraction on the part of the solvent molecules. Simultaneously with solution, !
the opposite process oc~urs adsorption of the grease-molecules by the cleaned
surface. In order to decrease the return contamination, the�~largest possible lot of
' solvent is used for the given.~iumber of plates or substrates. The process is
carried out successively in several lots of fresh solvent. If the solvent is
distilled, the degreasing is carried out in the last step in the so~vent vapor. The
solubility of the greases increases with an increase in t~mperature, and therefore
the degreasing is carried out in hot or boiling organic solvents. For a given
number of samples and a given i~t of a specific solvent the controlled parameters
of the degreasing process are the solvent temperature and the treatment time.
The plates and substrates are very effectively cleaned in organic solvents. Never-
theless, the application of organic solvents in production is undesirable. The
- grease molecules go into solution without chemical destruction and can again get
on to the cleaned surfaces. Therefore multiple cleaning is required accompanied by
large lots of solvent. As a result of the high requirements on the purity of the
solvents themselves, they are subjected to multiple redistil}.ation which also
increases thz cost of the process. Many of the solvents have law boiling points
and high vapor pressure; therefore the degreasing is accompanied by large losses
of solvent; in addition, organic solvents are frequently toxic and fire-hazardous.
Freons which have naw been widely introduced into production are advantageously
distinguished from the above-presented organic solvents. Freon-113 a liquid with
_ a boiling point of 47.6�C and a density of 1.57 g/cm3 is the most frequently
used. Freon is incombustible and nontoxic; it provides high cleaning efficiency
and makes it possible to do away not only with many of the organi c solvents, b ut
also the application of succes~ive operations of chemical pickling and washing in
deionized water.
Chemical degreasing is done in eompoimds that destroy the grease molecules and do
not affect the treated material. The absence of grease molecules in the treated
~ solv'.~on and, consequently, the absence of the probability of return contaminaticm
theoretically distinguish chemical degreasing.
In microcircuit production sometimes the surfaces are treated in soap solutions,
which is used to convert the saponified greases to soaps which are water-soluble
salts. The latter are re~moved from the surf aces of the plates and substratea by
washing. The saponified greases 3nclude all of the vegetable and animal oils which
- are complex esters of glycerine and high-molecular organic acids (stearic, nleic,
palmitic, and so on).
52
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400010055-6
FOR OFF[CIAL USE ONLY
- ~For example, the process of saponification of stearin takes place by the
equation -
~C17H35C00)3C3H5+3Na0H=3C17H35COONa+C3H~(OH)3, (4-1) ~
where C17H35~OON a is sodium stearate (soap); C3H5(OH)3 is glycerine. Both
materials are easily soluble in water.
At the present time hot (75-80�C) "universal" amm~onia peroxide solution is used for
che~.cal degreasing o~ silicon plates. It~consists of an aqueous solution of a
mixture of H2O2 Perhydrol and NH40H alkali. It removes saponified and unsaponified
greases.
I~uring degreasing Perhydrol decomposes with the release of atomic oxygen:
H2O2->Ofi+H2O. . (4-2)
The release of atomic oxygen increases with an increase in temperature. Atomic -
oxygen oxidizes both organi~ and inorganic contaminants.-The NH40H alkali
_ accelerates the reaction of the composition of Perhydrol, and it also combines the
co~ounds of certain met~ls of first and secand group of the periodic table into
- well-solt~hle complexes.
By comparison with physical degreasing in organic solvents the mechanical degreasing
is distinguished by lower toxicity, low cost of the reagents and less labor consump- -
_ tion of the processes.
Pickling. The pickling of the plates and substxates, as has already been noted, is
_ accompanied by the removal of the surface layer. -together with which the contanr-
inants available on the surface are also re~oved. The pickling is usually carried
out after degreasin~, for only in this case does the pickling agent ~et the entire
surface well and the upper layer is uniformly removed. The pickling is a mandatory
production operation when preparing monocrystalline plates and substrates, for which
it is done not only to clean the surface, b ut also to remove~ a layer with mechan-
ically disturbed crystalline structure. -
Acid pickling of the semiconductors proceeds in several steps in accordance with
chemical theory: the diffusian of the region to the surface, adsorption of the
reagent by the surface, surface chemical reactions, desorption of the reaction ~
products and diffusion of them from the surface.
The pickling agents in which the diffusion stages are the slowest stages -
determining the overall pickling process are called polishing pickling agents.
- These pickling agents are insensitive~~to p~ysical and chemical nonuniformities of _
the sur�ace; they smooth out the roughness, leveling the microrelief. The pickling
rate in polishing pickling agents essentially depends on the viscosity and mixing
of the pickling agent and depends little on temperature.
The pickling agents in which the slowest reactions are the aurface chemical
reactions are called selective. ~The pickl3.ng rate in selective pickl3ng agents
53
FOR OFFICIAL USE ONLY
~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
depends on the temperature, the.structure and the crystallographic orientation of
the surface, and it does not depend on the'vlscosity or mixing of the pickling
agent. The selected pickling agents with great difference in�-pickling rates and
different crystallographic directions are called~anisotrcpic.
Polishing pickling agents with high and stable pickling rate, low~impoverishment
and stability during storage are used to prepare the.pistes. Pickling agents
based on a mixture of~nitric acid and hydrofluoric acid have such properties for
silicon and germanium.
- ;
In accordance with chemical theory, the surface chemical reactions take place in
- two stages during polishing picklirig: oxidation of the surface layer and conver-
sian of the oxide to an easily soluble salt.
cr The role of the oxidizing agent is played by nitric acid: -
Si+4HN03-~Si02+4N02+2H2O. (4-3)
Hydrofluoric acid is a complex former which converts the silicon dioxide to silicon
- tetrafluoride:
Si02+4HF-~SiF4+2H2O. ( 4-4)
- The compositions of the pickling agents in the ma3ority of cases are selected _
empirically. The composition of the pickling agent, in addition to the basic
components, nitric acid and hydrofluoric acid, include various additives. For -
example, the most frequently introduced acetic acid retards the chemical reactions
(4- 3) and (4-4) and, consequently decreases the overall pickling rate. In the
absence of acetic acid in a pickling agent the reactions proceed too rapidly, and
the pickling process is difficult to control.
Let us note that the pickling is also used for dimensional treatment of semiconductor
plates, for example, to bring the thickness to a given value, to obtain local
depressions or through holes. Selected pickling of semiconductors is used to
investigate surface defects, to discover the p-n ~unctions, for controlled formation _
of the depression of a defined shape. Selecting the defined surface orientation
and also the orientation of the hole in the contact mask, it is possible to obtain
completely dQfined shapes of depressions correaponding to them.
- Washing the P~ates and Substrates. During the manufacture of the microcircuits, the
surfaces of the plates and substrates are washed several times. Especially pure
deionized or distilled water is used for the washing. Careful washing of the
silicon or dielectric substrates is nECessary after degreasing them in soap solu-
tions or ammonium pero~dde solvent. Certain orgariic solvents remove only nonpolar
greasy contaminants from the surface of the substrates; others-~can remove polar
(ionic) cantaminants also. Trichlo~oethylene, trichloroethane, perchloroethylene,
trichlorotrifluo~thane, for example,.remove only nonpolar molecules; therefore ~
after degreasing subsequent washing o~ ti~e polar contaminants in water is required.
54
- FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL (1SE ONLY "
The forced method is used to wash silicon after pickling. During forced washing
the pickling agent is not completely drained of~ so that the plates do not contact
the air; deionized water is added to the bath, diluting the remains of the pickling
a$ent and gradually forcing it out. Otherwise, th at is, after pouring off the
pickling agent completely or removing the plate, chemical reacti~ns wi11 continue
- at the points where the drops of pickling agent remain. As a result, the surface -
of the plates will not be unifo~nly smooth. During forced flushing it is necessary
to consider the possibility of splattering of the acetic pickling agent. -
4-3. Intensification of the Cleanin~ Processes
Classification of Zntensification Techniques. In order to insure efficiency and
-quality of the treatment of the surface plates and substrates, various intensifica-
tion procedures are used. For intensificatioi; the slawer stages of the process
(for example, tha_ supply of fresh reagent to the treatment zone, the removal of the
products of the chemical reactions from the ~treated surface) are aceelerated, the
desorption of the atoms or ions is insured, solution is accelerated, fast penetra-
tion and subsequent removal of the treating solutions from the microer3cks, and so
on are insured. The intensifying means can be divided into phqsical, chemical and
combination. The first-mentioned include heating, boiling, treatment with a jet, _
hydrocirculation, flushing, hydromechanical cleaning, centrifuging, ultrasonic
treatment, industrial-frequency vib rations, and plasma. Chemical me ans include
surf ace-active mater~als, complex formers and catalysts.
Physical Methods of Intensification. Treatmen~ during heating or boiling in
organic solvents, acids, deionized water are carried out in exhaust hoods or
enclosures using a quartz or fluoroplastic dish or bath. For example, the UTU-1 -
pickling unit has three baths: one for pickling with water heating, the other two
for preliminary and final washing in running deionized water. A deficiency of the
- cleanin~ by heating and I~oiling is treatment in a constant volume of reagent which
- leads to repeated contamination of the substrates; to ~reat losses of the reagents
as a result of evaporation, and, in addition, the intensification of the cleaning
processes is inadequate.
Jet cleaning can be carried out using pneumatic or centrifugal jets operating at a
pressure of 4 to S atmospheres. The advantage of jet cleaning is continuous replace-
ment of the reagent and acceleration of the removal of the reaction products as a �
result of the ;~ydraulic effect of the jet. A deficiency of the method is the high
rate of consumption of reagents. The reagent consu~ption rate can be lawered by
circulati.ng, filtering and regeneration of them, that is, utilizing a closed tech- -
nological pr.ocess. The effectiveness of ttie treatment can be improved by increasing
the j et pressure.
H~ydromeclzanicaJ_ cleaning consists in mixing a solution or water by mixers, brushes,
and the mechanical effect of them on the treated sur~ace. Al1 of th is promotes
r.enewal of the solution directly on the surface, it increases the wettability of
the surzace and mechanicall}r knor_ks out the greast molecules and other impurities,
acting as mechanical "scrapers" on the cleaned s-urface. The deficiencies of hydro-
mechanical cleaning include tfie p~ss~bili.t5~ of return transfer of the contaminants
- from the brushes to the substrates, the probability of contamination as a result of
wear of the brushes, the necessity for periodic careful cleaning of the brushes
and mixers themselves.
55
FOR ~OFi~'ICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAI. USE ONLY -
~ Centrifuging promotes mixing and.renewal of the liquid near the txeated surfaces.
The substrates or plates are fastened in a holder which; in turn, is fastened to
the spindle of the centrifuged turning at abou~ 200 rgm. Increasing the
- centrifuging rpm improves the cleaning quality. Dzying is possib le at highe r.
speeds (2400-2500 rpm). The centxifuged diying removes the drop moisture well,
which is necessary for removal of the infinitesimal pr.oportion of the salid sedi-
ment in deionized water with the water drops.
Ultrasonic cleaning is~the most effective of the indicated physical methods. The
bath of the ultrasonic unit is attached to the coneentrator of the magnetostrictive
- e~nitter. The vibrations of the concentrator are transmitted to the bath walls, and
_ from them, to the liquid medium. Elastic waves arise in the liquid medium (punch-
ing and rarefaction as a resu"lt of shifting of the liquid particles). At the
' rarefaction points in the liquid sma11 bubbles appear which are filled with liquid
vapor called cav3.tation bubbles. Under the effect of forces which try to return
the shifted particles to the initial position, these bubbles collapse after brief
_ existence. In the case of 3:ntense vib xations and collapse of the cavitation
. bubbles, shock waves arise coimnunicating high accelerations to the liquid molecules.
The liquid molecules co}.lide with a force against the surface of the treated sub-
strates and knock the ccmt�aminant particles off them. As a result- of cavitation,
the liquid is capable of penetrating into the deep pores, channels, depressions and -
cracks which remain uncleaned when using ordinary methods.
The effectiveness of ultrasonic cleaning depends on the frequency of the intensity
of the ultrasonic vibrations, the location of the cleaned surface with respect to
the concentrator, the temperature and pressure of the cleaned liquid vanor, and
- the cleaning time. Ultraso~iic cleaning is appreciably more effective if the -
surfaces of. the plates or substrates are arranged perpendicular to the direction
of the ultrasonic wave propagation front.
However, as a result of strong mechanical eTfects, the ultrasonic cleaning must be
carefully carried out for thin and brittle plates and substrates and also for
finished structures in the precavitation mode, for a short period of time, locating
= the treated specimens far from the bottom of the bath. The vibrations of industrial
frequency used, for example, in the RVKhO-GS60-1 group pickling unit are less _
~Iangerous for surface rupture.
Chemical Intensification. The introduction of catalysts, surface-active materials
and complex formers into the cleaning process is one of the prospective ways of
improving the quality of preparing plates and substrates and the effectiveness of
cleaning processes, and in a number of cases it permits elimination of toxic and
_ fire-hazardous organic solvents and also concentrated acids. -
The complex formers form stable, complex campounds with the surface impurities or
- with the harmful products of chemical reactions, which go into solution and remain .
in solution. The complexes mUSt have the smallest possib le dissociation constants; -
othextaise the impurity found in t~ie co~mplex can be again- adsorbed f~om solution by
`L~>~-
the treated surface. For example, copper or silver ions, the presence of which on
~he surface of the semiconductor substrates increases the reeambin~tion of the
minority carriers and the b ack curxents of the p-n ~unctions, and they are easily
bound into complexes by anunonia:
56
F'fl1R ~FfI~I~SJL i1S~ ~Nd~Y
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400010055-6
FOR OFFICIAI. USE ONLY
Cu+~ ~kN!-i, [(Cu (NH,).~+'
Ag+ 2NH,-? (Ag (NH,),~*� } ~ (4-5)
During pickling of the semiconductors potasaium bichromate binds t~e reaction
products into easily'and .quickly soluble complexes. Acetonitrile, ethylenedi-
aminotetraacetic acid (trilon B), and so on have good complex-forming properties.
Combined Intensification. The comb~ned methodsof intens~fication are quit~e fre- -
quently used in practice: the treatment with a hot jet, ultrasonic treatment in
s,olvents, treatment by a hot jet with centrifuging, pick.ling in a hot solvent with
the application of complex-forming additives, and so on.
4-4. Standard Processes of Cl~aning Plates and Substrates
Prelj.minary Cleaning of Plates and Substrates. It was noted earlier th at it is
in practice impossible to select a universal~ composition for wet cleaning of the
surface of plates and substrates. Tn addition, any technological operations,
including the cleaning operatiot~s themselves, can be sources of pollutants. When
_ manufacturing the microcircuits, the billets for their structure are trested more
~han once, and each cleaning is complex, that is, it includes a number of operations
_ for the removal of all possible cont~inants. The nature of the treatment in each
operation, the sequence of operations, the means and methods used are determined
by the phase in which the cleaning is carried out, the material of the billet and
rhe elements entering into the structure and also ttie requirements on the quality
of cleanin g. _
After machining, the semiconductor plates and substrates are contamined with
polishing suspension, paste, adhesives and oil from the machine tools. The pre-
~ liminary cleaning of the plates and substrates is carried out directly on a polish-
ing disc using an aqueaus solution of "Nega" or "i~otos" pawder, and they ~re
flushed with pure water. The water is removed by centrifuging. Then the discs are
h,eated to melt the adhesive, and the plates or substra~tes are removed from the
disc.
Fin~l cleaning of th~ plates and substrates is realized by different paths, the
basic operations of which are degreasing, polishing pickling, washing and drying.
_ Processes of Cleaning Silicon Plates. During the production cy cle of manufaeturing
_ semiconductor IC structures, the substrates are cleaned mc~rP than once: after
machining, be.fore masking the surface and local treatment, and after photo-
lithography, Let us consider one o~ the standard process flow charts for cleaning
silicon plates.
The cleaning of silicon plates before the ~irst thermal oxidation includes the
fol"lowing operation:~:
1. Degreasing fn a hot (75-$0�C) ammonia p~roxide solution.
2. Washing in flawing deionized water to remove the products of the chemical
r'eactions of the preceding treatment.
57
FOR OFFTCIAL USE ONI.,Y -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
- FOR OFFICIAL USE ONLY
3. Treatment in hot (90-100�'C) concentrated nftric.acid for oxidation of silican
to the required depth and partial removal of the inetal io~s.
4. Flushing in a flaw of deionized water to remove the acid.residues.
5.; ;~Hyd'rodynamic treatment of the plates with 13me brushes in a~et of deionized
water.
6. Drying of the plates in a centrifuge in a jet of purified dry air.
7. Treatment of the plates in a solution of hydrofluoric acid-with acetone to
remove the oxide film obtained dur-ing treatment in nitric acid, and together with
it, also the surface contandnants.
Acetone is introduced into the solution to decrease the contamination of the
silicon plates by the solid residue of hydrofluoric acid whic~ is formed for ar~
excess of chlorine ions in the solution as a result of chemical reactions:
SiO, 9HF : SiI=, -~-21-I,O; '
S i F, 2f= S i I= 2; ( 4-6 )
� SiF~ 2 -~-;2[-I"~ 1-I~S iI=a. -
Acetone has a lo~w dielectrict constant; therefore on addition of it, the dissocia-
tion of the hydrofluoric acid decreases. In additian, acetone forms complexes
with certain cations.
- 8. Washing in a f law of deionized water until the specific resistance at the
exit of the device becomes equal to the resistance of water and the entrance to
the device.
9. Ultrasonic treatment in several lots of aimnonia peroxide solution for more
complete removal of organic and inorganic contaminants left in the ~icrocracks and -
surface pores.
10. Washing in running deionized water.
11. Drying in a centrifuge.
Standard Process of Cleaning Substrates Made of Glass, Sapphire and Ceramic.
Good results are obtained when using the clean3ng process including the following
operations:
1. Ultrasonic washing in a s olvent at room temperatuxe.
2. Ultrasonic washing in a solvent heated to a temperature o~ 70�C and in solvent
vapor.
3. Washing in running water.
58
~OR O~F~~~AI. &JSE ONL`~ -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAI, USE ONLY
4. Boiling in a Perhydrol solution.
5, Washing in hot distilled ~aater.
6. Boiling in distilled water.
7. Drying in a flow of pure nitrogen.heated to a temperature o~ 110�C.
Standard Process of Cle~ning Pyroceram Substrates. The pyroceram composition
includes amorphous and crystalline phases. Different piakling rates of the differ-
ent phases lead to the ~ormation of relief on the substrate surface. In order to
decrease the roughness, neutral and acid solutions a~e~~used which form slightly
solub le compounds an the surface of the pyroceram preventing surface deterioration.
The standard prc~cess of cleaning pyroceram includes the following operations:
l. Begreasing by b~iling in ammonia peroxide solution for 15-20 minutes.
2. Washing in running deionized or distil7.ed water.
3. Washing by t~oiling in distilled water for 5-1Q minutes.
4. Drying in isopropyl alcohol vapor (~0 mi.~utes) or in a flaw of argon or
nitrogen heated to a temperature of 320+30�C.
Freon Cleaning. Cleaning using freon is universal. It can be used for plates and
substrates made of any material. Several standard technological processes have been
developed for ireon cleaning. For exanaple:
1. T reatment with freon emulsion (water+freon+a surface-active material), in which
- inor.ganic salts are dissolved in the disperse water drops, and organic contaminants,
in freon.
2. T reatment in a mixture of freon with other solvents (isopropyl alcohol, methanol,
and methylene chloride) removes the remains of surface-active materials from the
substrate surface.
- 3, Treatment with pure freon.
4. Treatment in freon vapor.
- The treatment of silicon using ~reon-113 in the Soviet "Freon" unit accelerates the
- cleaning p rocess by about 15 times and insures the~same quality as during careful
treatment by degreasing, acid pickling and washing.
4-5. Drying Cleaning of Plates and Substrates
Heat T reatment (Annealing). Heat treatment is used to remove the impurities
adsorbed by the surface, for decomposition of surface contaminants and evaporation
of volatiles. As a rule, annealing is carried out in vacuum and thermal units
directly before aarrying out the thermal processes. For example, when growing
~ masking films on silicon, the gases and rnoisture are removed from ttie surface by
59
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
FOR OFFICIAL USE ONLY
heating the plates to the ox~:dizing temperature. When annealing semiconductor
plates in a vacuum, moisture, carbon dioxtde and li~~ht hydrocarbons are easily
desorbed from ~heir axi dized surface at a temperature o� 400�C. The effectiveness
of the cleaning increases with an increase in temperature, but tbe treatment
temperature is limited by the melting point o� the cleaned~~materials or diffusion
of the alloying admixtures. Lower-temperature cleaning processes are used in these
cases .
- Gas Pickling. The essence of the gas pickling process cansists ix~ chemical inter-
action of the plate material with gaseous substances and the formation of easily
removed volatile compounds when this happens. During gas pickling the contaminants
_ are removed together with the surface layer of the plates.
Halogens, hydrogen halides, sulfur compoundG and water vapor are used as the reagent
- gases for pickling silicon plates. Small amoimts of these gas~s are added t.o the
gas carrier (hydrogen or helium) and are transported to~ ~the chamber of the w1it,
in the temperature zone of which the plates are located.
- The pickling of the silicon by hydrogen fluoride is widely used before grawing the -
silicon layers on the plates:
Si(solid)+4HC1(gas)-~SiC14~h (gas)+2H2 (gas) . (4-7)
Hydrogen fluoride vapor is delivered by hSTdrergen to the reaction chamber of the
epitaxial growing unit where the silicon plates heated to a temperature of 1150-
- 1250�C are located.
In silicon tetrachloridP vapor the pickling of the silicon plates is accompanied t+y
the reaction
SiC'14(gas)+Si(solid) f2SiC12(gas). (4-8)
In the case of chlorine pickling helium is used as the gas carrier. The pickling
takes place at temperatures of about 1000�C and with a chlorine content in the '
helium of no more than 0.2%. On deviatio~ from the optimal conditions, the pickling _
~ agent loses the palishing characteristics, and irregularities appear on t1.~ surface
of the silicon. Pickling takes place in accordance with the reaction
siCsolia>+c12 (gas>-~sicl2(gas> . (4-9>
During pickling in water vapor the ~o1lew~ng xeactions take place:
Si(solid)-i-H2O(gas)-~SiO(gas)-I-H2 (gas); _
- Si (solid)+2H2O(gas)-~Si02 (solid)~2H2 (gas) ; (4-10)
Si (solid)-I-Si02 (solid)-}2Si0 (gas) .
With a water vapor content in the hydrogen o~ more than 5�.10-2%, a silicon dioxide ,
film is formed on the surface of the silicon, and pickling stops.
60
E~R OFFI+CaAL US1E ONLX -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400010055-6
FOR OFFI~CIAL USE ONLY
It is also possible to carry out the pickling of the silicon in hydrogen sulfide -
or sulfur hexafluoride vapor. During pickling in hydrogen sulfide? large rates -
of removal of the surface layers are obtained (to.15 microns/minute). However, -
~the hydrogen sulfid~ is toxi.c. ~u1f~r hexa~luoride, on the contrary, is not
to~ic and insures good quality of the silicon surface and also sapphire.
- fihe pickling of the silicon is accamp anied by the reaction
= 4Si(solid)+2SF6(gas)-}SiS2(solid ar liquid)+3SiF4~gas), (4--11)
'Gas pickling makes it possible to obtain cleaner surfaces than liquid picklin~. -
- Hawever, in any case gas pickling has limited application as a result of- high
treatment tempexatures and the necessity for using especially pure gases.
Ionic Pickling. Ionic pickling is the process of removing the contaminants together
'with the surface layer of treated material sprayed in a vacuinn.
For spraying, the surface of ~he ~lates or substrates is bombarded with accele rated
'p ositive ions of inert gases. Most f requently argon is used for spraying, for
it is cheap, plentiful and allows effective spraying. The ac~elergted ions pene- -
t rate the surface layer and, colliding with the atoms of the treated plate or s ub-
s trate, transfer their energy to them, If the transmitted energy is sufficient,
th~ atoms are shifted from the nodes and can transmit energy to other atoms. Thus,
~zones of shifted atoms radiation disturbances in the substrate structure are
formed alon; the trajectoiy of motion of the ion. Spraying takes place if part of
~the shifted atoms reach the surface an d if the energy of these atoms is of the
work function of the substrate material.
From investigation of the mechanism of the effect of the accelerated ions on the -
s ubstrate it is clear that the pickling begins with defined values of the energies
s ufficient for spr3ying. The silicon is pickled with ion current densities of ~ore
than 10 a~s/m2 and an ion energy of 1=10 kev. For very high energies the ions
penetrate deeply into the treated substrate and spraying is not observed. -
The eff.ectiveness of the spraying and, consequently, pickling, is characterized by
the spraying coefficient which is numerically equal to the number of atoms of cleaned
s~bstrates sprayed by one bombar:ding ion. The spraying coefficient S depends on
the mass of treated material ml, the mass mz, energy E and angle of incidence 6
of the ion and also the physical state of the surface:
nt, ~a, ( 4-12 )
S k ~ (E) , (m, mo) E,
_ where k is the coefficient characterizing the sur�ace state, a(E) is the mean free
>p ath length of the ion in the treated material, whirh depends on 8. -
~~epending on the structure of the devices and the method of gene.rating the ions,
p lasma ion pickling and ion beam pickling are distinguished.
P lasma ion pickling is ca~ried out in vacuum spray~chambers. The plate or sub-
s trate holder is located in a gas dis charge plasma. (hl feeding a negative
61
~OR OFFICIAL USE ON]LY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
k'OR OFFICIAL USE ONLY
potential to the holder, positive ions are drawn out of the plasma which are
accelerated by an electric field and bombard the surface of the substrate, clean-
ing it. The pickling of poor-conducting or nonconducting substrates with constant
potential on the holder is compli.cated and~stops as a result o~ the accumulation
of a positive ion charge on the subs~.rates. Therefore high-frequency AC voltage
is used to pic~Cle them. With a~e?ative:~pctential the substrates are ~ickled ~
with - a positive potential, ~~ectrons are drawn out of the plasma, and the
accumulated positive ion charge is neutralized. -
The rate ef plasma ion picklii~g is regulated by varying the vottage on the elec- ~
trodes, the gas~ disch arge current, the inert gas pressure and the time of~perfornr
ing the process. The pickling ~ate of silicon reaches 0.2 microns/~i.r.ute, the ~
substrate temperature is comparatively law (100-120�C) during pickling. A signif-
icar~t advantage of plasma iDn pickling is inertialessness, ~or the pickling process _
stops immediately after removal of the potential from the substrates. A unique
property of plasma ion pickling is the fact that its rate is directed alon g the -
~ normal to the cleaned surface. Tlzis permits treatment of strictly defined 1oca1 _
- sections.
Ion beam pickling 3s carried out in ion injection units ion beam accelerators
(ILU) where the ion beam is formed by a special gas discharge source, a system
of extracting, accelerating and focusi~.g lenses. The beam is direete3 in the
direction of a rotating incli,ned table, ox~. the surface of which the plates or
substrates are located. To campensate for the positive charge accumulated on the
_ treated surfaces, a neutralizer a heated cathode emitting electrons is used. ~
During ion beam pickling the plates or subs~t~ates are located outside the p~asma
_ gas discharge gap. This permits regulation of the process parameters, the ion
energy, the current density of the ion beam and angle of incidence o~f the ions on
the surface of the plates and substrates independently of each other.
Ion pickling is universal. It is possible to clean the surface of any materials
to remove impurities of any type. It is possible to use ion pickling to treat
multilayer fil~us with properties of fihe layers which are incompatible from the _
point of view of liquid chemical cleaning. T_on cleaning insures high quality with-
out deep alterations of the treated surface layer, high precision of removal of the
layers (+0.03 microns) and excludes the interoperation expenditures of time, for
it is possib le to perform the subsequent operations (oxidation, film deposition)
directly in the same vacuum chamber. Thus, -i.4n pickling is used as finish cleaning
of L-he plates and substra~es before the process is perform~ed ii~ a vacuum and also
for treatment of materials which are difficult to clean by other methods, for
example, for substrates made of sapphire. The absence of lateral components of the
picklir~g rate permits the application of ion pickling for precision dimensional _
~ local treatment.
- Plasma Chemical ~ickling. In contrast to plasma pickling, plasma chemical pickling
- is realized by ions of actzve g a~es instead of inert gases. During bombardment
these active gases enter into chemical react3ons with the treated ~aterial, fox~m-
ing volat:iles. As a result of the-electrical activation of the gases with the
formation of reactive radicals, plasma ch~m.ical pickling can be r_arried out at _
significantly lower temperatures than ordinaYy gas pickling. -
- 62 -
Ff~I~ O~~
e~-
f,~
� g~
h
~
f)
Fi~ure 5-8. Diagram of the method of direct photolithography with
a sublayer.
a-- substrate with titanium film; b-- application of the sub layer
(gold); c-- formation of the photoresistive layer; d-- exposure;
e-- development; f-- pickling for transfer of the pattern to the
sublayer; g-- removal of the photoresistive mask; h-- transfer of
the oattern from the Au-mask to the titanium film; i-- removal of
' the contact mask
Inverse photolithography with a sublayer is free of the danger of he ating. The
contact mask with the p~ttern that is reversible with respect to the given pattern
for the film element is obtained in a film of easily pickled material applied to
th:: substrate, for example, copper, aluminum and bismuth oxide. The element film
- is app~ied over the contact mask. During pickling the film lying an the mask is
removed together with the cantact maslc (see Figure 5-12).
In order to ohtain the inverse contact mask, just as tha direct one, it is possib le
to use organic materials which are removed by compounds less aggressive to the
materials entering into the microciicuit structures.
86
~ ~'O~ O~'FA('T~u~, U5~' ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY _
SiOz
,~--i~~r..~ . .
i .
(111 S i, _
a)
� ,
(111 S i, p
b)
ii
� ~~7~,i
(111) Si p
~ c)
: ~ :
(111 Si
- a>
= (~�)sti ~
_ Figure 5-9. Diagram of the formation of inesa-structures.
a-- application of a double layer of Si02-photoresist; b-- forma-
tion of a photoresistfve mask; c-- transfer of the p attern to the
Si02; d-- plate pickling; e-- removal of the double contact mask
Removal of the Phofiores3stive Mask. In the finishing operations, as a rule, it is
necessary to insure not only rupture and removal of the photoresistive mask playing
- its role, but also good cleaning of the surface to remove cont amination introduced
by the entire process of photolithography, for the surf ace state influ~~nces the
quality of the subsequent operations. At the present time three methods of remov-
ing the photoresist are used: chemi cal destruction (decomposition) in sulfuric
acid or solutions based on 3.t, treatment in organic solvents, destr.uction by oxida-
tion in oxygen or in oxygen-containing gas mixtures.
Chemical destruction is accompanied by the reaction of decomp osition of the photo-
resist with the formation of less complex short molecules with small molecular
mass which then are easily washed away with water. For acceleration of destruction,
the concentrated sulfuric acid is heated to 160�C. The removal of the photoresist
3;s accompanied by fast exhaustion of the acid; therefore it is consumed in large
quantities. .
87
FOR OFFICIAL USE ONLY ,
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFF[CIAL USE ONLY
~
- (~oo) sL
_ a) a) '
~
. .
~
b )
c)
c)
Figure 5-10. Diagram of obtaining Figure 5-11. Diagram of the method of
V-grooves in (100) silicon. inverse photolithography (the "explo-
a-- formation of a double contact sion" method) .
mask; b-- local pickling; c-- removal a-- substrate with photoresistive mask;
of the mask b-- application of film; c-- removal
of photoresistive mask
Q~ -
/ r
b~
s,' / rs , i
�,i
F~
S~'~
e)
f )
�g)
h)
Figure 5-12. Diagram af the method of inverse photolithography with
a sublayer.
a-- substrate with a layer of contact mask m~terial; b-- appZication
and drying of the photoresist; c-- exposure; d-- development;
e-- local pickling of the contact mask film; f-- removal of the
- photoresistive mask; application of the element film; h-- removal
of the contact mask (sublayer)
8$
FOR O~'FICIAI, USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFiCIAL USE ONLY'
The better qualitq oF removal even at lawer temperatures (70-160�C) insures a
mixtLre of concentrated sulfuric acid with 30% hydrogen peroxide in a volumetric
ratio of 3:1. The method is not applicable for the removal of photoresist from
_ the ~etallized substrate.
The treatment in organic solvents is successfully used to remove the photoresist
f rom the me~allized substrates. The substrates are held in solvents (acetone,
methyletiiylketone, cellosolve, dimethylformamide). The quality of the process is
improved on adding organic alkalis ethanolamir.es to the organic solvents. Then '
the sc~elling relief is removed by a mechanical tampon, and the treatment is
_ repeated in fresh lots of the solvent.
The deficiencies of the liquid methods of removing photoresist are the multistage
nature, labor consumption, uncontrolled contamination of the surface by impurities
from the solutions, aggressiveness of the reagents, complexity of inech anization and
automation.
Depending on t~he method of activatibn of the oxygen molecules, oxidation destruc-
tion is divided into heat treatment in an oxygen atmosphere and plasmochemical
destruction. _
Heat treatment in an. oxygen atmosphere is xeali.zed at a temperature of 800�C at -
which the photoresist and the contamination are simultaneously destroyed and -
removed as a resu'lt of annealing and oxidation. Unfortunately, high temperatures
lead to irreversib le structural changes connected with oxidation, sub Iimation and
_ burning in of the residual contamination. The oxygen of the air can ~~e activated _
by u_ltraviolet radiation, which makes it possible to reduce the substrate treatm~nt
temperatures to 2.50�C. 'The process of removing the photoresistive mask can also
be accelerated by introducing about 2% ozone into th e air.
Plasmochem~.ca1 destruction is treatmer_~ in a high-temperature high-frequency
oxygen plasma at a pressure of 1.2� (102 to 103) Pa.
In a high-frec~uency oxygen plasma the excited oxygen molecules, atomic oxygen and
ozone are citemically active. Decomposition or photoresist in the o~gen plasma
is of a cha3n nature; t~ie r.e5ultant produc~s with lau,* molecular weight volatilize,
and on being subjected to further oxidation, they decampose to the end products of -
carbon dioxi.de C02, nitrogen oxide NO2 and water H2O. It is possible to use
hycirogen, nitrogen and moisture additives as the catalyst accelerating the process
of remaving the photoresistive mask. Inorg~ic contaminanrs do not form volatile
compounds duri.ng oxidation; for removal of them, carbon halides, for example,
freon, are added to thP plasma.
5-5. Photomask Productior~ Techno7_ogy
In order to obtain an exact pattern of the Photomask, first an enlarge image of
~
one module i.s drawn. A module is a single image of the IC elements or the set of
tt~em exeruted on the co-rresponding scale with respect to the ~limensi.ons of t~ie
elements of the topologic drawing of tt?e given IC pr.ocess layer. The image of the
module is successively diminished and multiplied, th at is, iC is repeated a '
=nultiple number of times ~a~~h respect to the working zone of the standard photomask.
89
FOR OF~'T~'~~,~, iJSE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIA~. USE ONLY
For series production of microcircuits using eontaEt photolithography, as they are
used the photomasks wear (defects accumulate). Therefore working copies are
made from the standard photomasks which, in case of damage or wear, can be repro- -
duced,
. Let us consider the basic steps in the production technology of working photomasks . -
- The primar~ original is made on a special device coordinatograph. The primary
original is a layot~.t of the microcircuit moclulp made on an enlarge (1000:1, 500:1,
- 200:1) scale intended for the manufacture of photomasks by the method of successive
reducing and multiplication.
~ In order to make primary originals, substrates 600x600 to 1200x1200 mm2 made of _
plate glass or polyester film 6-10 or 0.05-0.2 mm thick, respectively, are used.
_ A thin continuous vaseline layer is applied to the surface of t~ie substrate, and
then by multiple spraying, a layer of lacquer or nitro enamel 30-50 microns thick
_ which dries in several hours at a teinperature of 30-40�C. _
After formation of the film it is cut along the outline of the pattern on the
coordinatograph using a cutting tool with a diamond or tungsten carbide tip
fastened in a rotazy head. The displacement of th e cutting tool with respect to
a given trajectory and with the requ~re3 precision is controlled by special
mechanisms driven manually, by .-otation of lead screws ~r automatically using a
program unit. Before the oper~tion of cutting thQ .film, adj ustment to optimal
cutting depth is needed so that the substrate will not be cut through and the
cutting tool wi11 not be dulled if the original is made on a glass subs~rate.
After cutting through the outline of the layout, the excess sectionsin the lacquer
- film are removed. The coordin atographs insure accuracy of cutting out the pattern
within th e limits of +SG microns.
An intermediate photamaster is ma.de by the method of photography. The intermediate
photomaster is an image of the or1_ginal with element sizes, intermediate between
the element sizes on the master and the corresponding dimensions given by the
topologic drawing.
'ihe intermediate photomaster can be obtained by copying the p~imary master using a
r.educing camera (Figure 5-13), which is a photographic device for precise photo-
_ copying of flat objects with 10-50fold reduction.
If the intermediate master is made with image reproduction, precise shifting of the
holder with the photographic plate in two mutually perpeiLdicular directions is
provided for in the reducing camera. After exposure of the image of the primary
master, the photographic plates are processed: they are developed�, washed, fixed,
� go through secondary washing and dryin�;. The tedious and multistage process of
making a prim~ry intermediate photomas~~er using a coordinatograph and reducing
camera can b e r.enlaced by one process or successive ph.otoprinting of the elements
of the layout on the photographic pla1_e which can be done by two metYiods:
Scanning of a focused light or eler_tron beam computer-controlled by a given program; -
90
~~~i '~Ja' n' l~ltl~,~. U.~~ -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFl~ICIAL USE ONI.Y
Photocomposition in which th~� entire pattern is divided into elementary rectangles
with different ratio of sid~as and defined orientation of them with respect to
angle; then these rectanglus are pro~ected successively on the require3 locations
~ of the photoplate using r,he coordina~e tahle and diaphragms.
In Soviet practice the method of photocomposition is more frequentiy used.
~ Microphotocompositors EN4~�508, E1~519, EAf-549 are distinguished by high precision
and output capacity.
1 ~Z 3 i 4
5 6 7 8
~ _ - - - ~
.
~ '
~ .
Figure 5-13. Diagram of a reducing camera.
_ 1-- reflecting shield; 2-- light source; 3-- dispersing glass;
_ 4-- primary master holder; 5-- light filter; 6-- objective;
7-- holder with photographic plates; 8-- microscope
1
The manufacture of_ the standard photomask is one of the responsible steps in the
~ technological process. The standard photomask is the fir.st photomask in the
technological manufacturing process with element dimet?sions corresponding to the
dimensions of the topologic drawing of the given technological layer. The
sta.ndard photomask is designed for subsequent manufacture of the working photo-
masks.
The standar.d photomask is made by reducing the image of the intermediate master by
the dimensions of the drawing of the working photomask and multiple rep2tition of
- this im~.ge for the defined working zone of the light sensitive plate. The basic
parameters and compatibility of the set of photomasks are realized in this step.
At the present time basically the me thod of successfv~ multiplication of a single
image using precision photocopiers (step-repeate�r cameras) is used to make
standard photomasks. The phot~~opier (Figure 5-14) consis ts of a projection
opticat exposure system, a precision coordination table and control panel. Before
reproduction, a final check is made, foreiga particles are removed and the
interr~~ediate photomas~er is retouched. Ther~ the intermediate photomaster is
installed with the image down on the base frame, and it is fixed by the fast-
drying adhesive.
The reproduction process consists of successive alternation of. exposure and
- shifting of the photographic plate on the coordinaticn ta~le to the next coordinate
position. The required coordinate positians are determined by the precision
- coordinate system, and they are given by a special reproduction program. In
addition to the single-position photocopiers there are also multiposition photo-
copiers in which it is possib 1~ to make a number of photomasks on several photo-
graphic plates simultaneously (and wifih a high degree of matching) .
91
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
' 1
~ z
" " 3
~
4
il 5
~i~ ,y 6 ~
10 9 ~
8
,
Figure 5-14. Diagram of successive reproduction using a single-
position photocopier
1-- light source; 2-- condenser; 3-- light filter; 4-- inter-
mediate master; 5-- base frame; 6-- objective; 7-- image on the
photographic plate; 8-- photograpfiic plate; 9-- coordination table; _
10 measuring system; I1 exposure module control system
Then the photographic plates with the latent image of the photomask pattern go _
th rough further operations: development, washing, fixing and 3Yying.
The manufacture of wo rking photomasks is an ordinary photolithograph ic process
(the standard photomask performs the role of an ordinary photomask), and it is used
for reproducing the standard photomasks. A working photomask is a photomask
designed directly for matching and exposure in photolithographic processes when _
making basic products.
The process of making working photomasks begins with preparation of the base _
the glass plate (or f lexib le polymer film), The plate is carefully cleaned and
~ activated to improve adhesion of the film of the pattem material applied to it.
Then a layer of chrom:i.um, iron-oxide, silicon monoxide, chalcogenide glass or others
is applied to the surface of the glass base. Then a phoCoresistive layer is formed,
~ exposure takes place, for example, in the EM-523 Soviet unit, and further opera-
tions ~re performed to transfer the pattern to the corresponding films.
For the manuf acture of photomasks it is necessary to insure that a minimum amount
of dust particles get on the billets, the surface of th e optical objectives, the
mer_hanical assemb lies of the coordination tables. The basic equipment for making
- photomasks is placed in a class 2 cleaning facility. The operations of preparation,
" application and drying of the photoresi~t, the multiplication, photochemical treat-
_ ment of the substrates are perf~rmed in the class 1 clean rooms with no more than ,
four dust partfcles per liter of air.
A three-step s~stem for making photomasks (the master in~~rmediate master
standard photomask) p resented by us has been quite we11 worked out and satisfies
the demands of microcircuit developers when a medium degree of integrat~on and
element sizes of no less than 2 microns aY�e required. At the presedt time a
. two-step (master and standard photomask) and ane-step (master) process flora charts
� 92
FOR OF~6CI~,L i1~E ~O1NI.~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFr ICIAL USE ONLY
for making photomasks have been developed and are being introduced into produc-
tion. In the two-step system the tedious operation of large-ecale drawi_zg and
cutting out of the complex edges is excluded. For technical implementation of the
~ingle-step flaw chart the most prospective are electron beam devices with multi~
- beam scanning or lasers with inertialess deflecting systems. The double and
single-step systems make it possible to construct computer-controlled, completely
automated production lines for the manufacture of photomasks.
5-6. Types of Rej ects and Quality Con.trol of Photolithography
The performance of the photolithographic process is accompanied by a number of
control operations. The basic control steps are as follows:
Quality control of the photomasks;
Substrate surface control;
Photoresistive layer control; ~
Photoresistive mask control;
Control ot the obtained pattern.
The rejection criteris and the admissib le def ects are regulated by the correspond-
ing production control charts.
Al1 dist urb ances of the quality of the image transfer during contact photo-
lithography can be provisionally divlded into 1oca1 defects, inaccurate transfer
- of the dimensions of the pattern elements given by the photomask-and inexact
matching.
Local defects punctures {defects in the form of through holes or in the form
of excess islets), craeks, scratches, foreign inclusions, ruptures of the conduct- _
ing tracks, projections and depressions with resp ect to the boundaries of the ,
_ pattern elements are individual. They do not pertain to all structures
~modules) of the substrates, but only to individual ones, but they are very danger-
ous, f.or the appearance of even one local defect within the limits of the critical
region will lead to rejection of the entire integrated cireuit. The area of the
- module on which the excessive number of defects is inadmissible is called the
critical region.
The causes for the appearance of local defects are as follows:
N atural contamination of the photoresist, the developer, the pickling agent,
deionized water; external contamination from the atmosphere, from contact with
packaging, fixtures, equipment and operators; defects in the ph otomasks punc-
_ tures, the remains of opaque film on the transparent elements, chips inthe glass,
frosted spots, dust, dirt, photoresist residue; surface defects of the substrate,
the presence of bumps on the film; defects of the photoresistive layer punctures,
~pech anical rupture as a result of the solid particles getting between the phota-
~ask and the photoresistive layer.
93
FOR OFFICIAL USE ONLY ~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFIC[AL USE ~NLY
Inex.~ct transfer of the dimension~ by the photomask, including imeve~ess of the
edge, can occur basically from the follocaing causes: inexact sel.ection of the
optimal methods and conditions of performing the production operations and also
deviations fram the given conditions; incorrect selectior, of the photoresist at~d
(or). picklix~g agent for the substrate of the given type; unawidabi~ity of the
clearance between the working surface of the photomask and the surface of the
photoresistive layer and as a result af flefects in the shape of the substrate or
photomask, nonuniformity of thickness of the photolayer, nonuniformity of the films
applied to the substrate, foreign part3cles between the photolayer and the photo-
~nask; imperfection of the equipment, and so on.
Inexact matching of the elements of rhe pattern can be obtained as a result of
' ine xact matching of the photomasks of the set used, as a result of imperfection of
the matching symbols or equipment, and in the case of visual matcrYng, as a result
_ of individual peculiarities and operator fatigue. The inexactness of the matching -
_ is estimated by using the NU-2E microscope (300-400X) by measuring the clearsnce
between the boundartes of the matching patterns and calculation of the mismatch of
their centers.
It is possible to improve the quality of photolithography only with a complex -
app roach to th is problem. A high yield of usab 1e products of increased complexity
can be obtained with simultaneous satisf action of a number of eonditions: the
application of a united purification and filtration system (YeSOF) of the photo-
resists, water, all gases, air, and so on; periociic monitoring, proper organiza-
tion of the cleaning of the photomasks, and as they wear out, replacement of them;
the development of optimal technological processes and control of all photo-
. lithography steps; automation of the charging and transport operations.
The conversion to contactlesG photolithography is a radical solution to many of the
prob'lems when ob*_aining the configuration of microcircuit elements.
Test Questions and Assignments
1. What is the role of photolithography in IC production? List the examples of
the apu lication of photolith~graphy known to us.
_ 2. What con~ponents enter into the composition of a photoresist? What is their
purpose?
3. What are ne$ative photoresists? What are positive photoresists?
4. Which photochemical processes ta~a.e place in the case of actinic irradiation in
photoresist based on PVC, elastomers and NKhD?
5. How do negative and positive photoresists transfer the image of the photamask?
5. List the basic characteristics of p:~otoresists. Explain them.
~ 7. Compare nega~tive and positive photoresists fram the point of view of resistancP
to acids and bases; from the point of view of resolution.
= 94
FOR ~F~ffC[AL [JSE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFIC[AI. USE ONLY
8. What basic steps can the process of contact photolithography be divided into?
Explain the production operations of each step for photolithography with respect
. ~o ~t silicon dioxide ~ilm in planar technology.
9. Formulate th~ basic requirement on the surface state of a substrate made for
application of a pho~oresist layer. Explain it.
10. How are sil.icon plates prep ared for application of a photoresist layer if they
have been stored for a long time after thermal oxidation?
11. What is the essence af the methods and what is the technique for centrifuging,
spraying, dipping, pouring and rolling? Compare these methods from the goint of
_ view of output cap acity, the possibility of automation and qual3ty of the applied
laye rs ~
12. How does the thickness of the photor~~sist layer depend on the centrifuge rpm
and the vis cosity of the photoresist? Hc~a does the density of the punctures depend
on the thickness of the photoresist 1ayFr? Haw are the centrifuging conditions
selected?
13. What is the mechanism and the fiechnique for the photoresist layer drying
process? What explains the preparation of photoresists on the basis of a combina-
tion of solvents rather tY:an one solvent? What are the basic parameters of the
drying process used on the photoresist layer and how are they selected?
14. Compare convective, infrared and microwave drying.
15. In what atmosphere is the photoresist layer dried?
16. Explain the necessity for the matching oppration.
17. Which methods are used for matching? What is their essence and what is the
technique?
_ l8. What is the role oF the exposure operation? Why are l.ight filters and con-
densers needed for exposure?
i9. Which light sources are used for exposure and why?
20. Name the parameters defining the exposure process. How are they selecte~i?
21. What is the role of the exposure operation? What is the mechanism of the
exposure of negative and positive phbtoresists? What is the technique for per-
_ forming exposure operations?
22. ~Jhat requirements are imposed on the developers?
23. What is indicated by coloring the alkaline developer a raspberry color?
24. What are the characteristic features of tY?e exposure of negative and positive
photoresists?
95
FOR OFFIC[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
25. Name the parameters o� the development process. Haw are they selected?
26. Gampare the purpose and the condi.t'~ons of the operations involved in primary
and secondary drying of a phatoresist. _
27. Compare the methods of obtaining the layout of the elements using photo-
lithography: direct, direct with sublayer, inverse, inverse with sublay~r.

- 28. Present examples of each of the methods of contact photolithography.
29. What is the purpose of the operation of removing the photoresistive mask?
30. What is the essence and the technique for performing the operation of removing
the photoresistive layer by varibus met~iods? Compare these methods.
31. What is a photomask, what is the working zone of the photomask, a module?
What materials are used to make photomasks and why?
32. Explain the three-step production process of making photomasks.
33. Give these characteristics of the two-step and one-step process flow charts -
for making photo~asks.
34. What types of rejects are possible during photolithagraphy and what are the
causes of them? _
35. Wnat measures can be used to i.mprove the quality of photoli.thography?
96
FOR O~~'yCIAI. USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
i
~HAPTER 6. OBTAINING THE CONFIGURATION OF IC FILM ELEMENTS USING FREE MASKS
6- 1. Free Mask Method
~lassification of Methods of Obtaining Film Element Canfigurations. The xequired
s~imensions, configuration and mutual arrangement of thin and tiiick-film elements -
of microcircuits are obtained by using free masks. When using the free masks, the
application of the film and obtaining the element layout are coiubined in a single
production process. These processes, depending on the method of applying the fi1m,
are 3ivided into the free mask method and stenciling. Both methods fail to provide ~
high resolution; therefore when it is necessary to create precision thin-film and,
sometimes, also thiek-film elements, the method of ccm tact photolithography investi-
gated by us is used.
~ssence and Special Features of the Free Mask Method. The free mask method consists
in shielding the required sections of the IC substrate by the free mask from the
f low of particles of the material deposited as the film. This method is used pri-
marily whPn depositing the film elements of IC in vacuum devices. -
The matching of the mask with a f.i1m pattern obtained on the substrate during the
process of preceding deposition is ei.ther done in the air using special devices or
directly in a vacuum chamber. This depends on the structural design of the intra-
chamber fittings of the vacuum devl.ces.
~
~he matching of the maskswith the substrates using special devices in the air
insures higher accuraey of obtaining the film pattern. The matching in an evacuated
space is complicated, and it requires more expensive fi~ctures and the accuracy of
matching is lower. In both cases it is necessary to consider the possibility of
the deformation and expansion of the mask during heating in the deposition process,
_ for it is located under the substrates.
Let us consider the basic features of the fr.ee mask method. The free masks are
made in an independent auxiliary technological process, and when obtaining the
_ film pattern on the substrates they are us~d multiply, ~ust as other tools. Obtain-
ing the pattern by means of a free mask differs from obtaining the pattern ~y con-
tact mask in that the process is less labor-consuming and, consequently, less
economically costly.
rr
The free masks have a thickness app reciably greater than contact masks and, in
- addition, as a res ult of loose fit against the surface, sagging and mutual shift-
ing of the mask in the substrate as a result of various TKLR during the deposition
- 97
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
process there is always a mask-substrate clearance that cannot be eliminated.
This gives rise to lower resolution of the free mask.meChoii than in the case of
photolithography.
The free masks ma.ke it possible to obtain a pattern that is only the inverse of
the mask pattern. It is impossible to use free masks to obtai~.n a closed or spiral _
pattern. The precision of the mask decreases as it is used.
In spite of the indicated deficiencies the free mask m~ethod is basic in the thin-
= film process of manufacturing microcircuits. When it is necessary to obtain a
precision pattern with small dimensions of the film elements, photolithography is
used.
Requirements on Free Masks. The requirement of obtaining a high-quality p attern _
under the film deposition conditions imposes defined requirements on the materials
_ of free masks and their structural design.
The m~terial for making masks must be machined to obtain flat smooth surfaces; it . -
must have sufficient rigidity and elasticity that the mask will~fit tightly against
the surface of the substrate and not be deformed during heating. In addition, the
mask material must clean well to remove buildups from preceding-depositions, it
must have low natural vapor pressure under depositionconditions, minimum gas
release and not 2nter int.o undesirable interactions with the deposited material.
Beryllium bronze, stainless steel, permalloy, molybdenum, tantalum, tungsten,
invar and, among the nonconducting materials, graphite and photopyroceram,
correspond to these requirements to the highest degree. Masks ma.de of tantalimm,
- molyb denum and tungsten are inert with respect to the deposited materials, and
they have high mechanical properties. They clean well without wpar. Invar has low
TKLR and is used in the case of high requi.rements on the accuracy of the transfer
of the element dimensions. Hawever, when making monometal masks it is difficult
to obtain high precision of reproduction of the hole aizes; therefore more accurate
bimetal two and three-layer masks are used.
The mask thickness must not be too sma11 so that the mask will remain sufficiently
rigid and tie deformed less, and it must not be too large so that the pattern will
be transferred exactly. During deformations~ individual sections of the masks
fit loosely ~gainst the si�:'~strate, which leads to a decrease or an increase in the
sizes of the elements as a result of shading of the substrate by the deformed
segment of the mask ("underdeposition") or as a result of the evaporated material
getting under the mask ("overdeposition") (Figure~6-1). ~
D 3 i ae
~ 2 y oM
~ i^i ~ ~ ~ ~ 1 d
�NeBone~n~\ ~ ~ ~ ~ ~,lios b~n" (b)
~ I i , n
. ( a) aK ,
\~v~~ii
MoneKynxpnsiu nomoK (
Figure 6-1. Distortion of the sizes of thin-film elements~-in the
case of loose fitting of the free mask against a substrate.
1-- mask; 2-- substrate; 3-- deposited film
Key:
a.. underdeposition; b., overdeposition; c. molecular flux
9$
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440010055-6
FOR OFFICIAL USE ONLY '
2 3 8r ~-r : I:
_ ,t._
/ ~
1 ~ ~ ~ i . ~
i
~ ~~i~
/i
\~\~~i/,~j~ _
~MoncKynap.~ieiu nomoK (,a~
Figure '6-2. Distortion of the sizes of thin-film elements in the
case of tight fit of the fre~ mask against the substrate.
1-- mask; 2-- substrate; 3-- deposited film; 6T�-- shadow width
Key: ~
a. molecular flux
- The undeformed mask always gives a"shadaw" on the edges of the layout, for not
all of the particles of material arrive in the direction normal to the substrate
(Figure 6-2) . Therefore the thickness of the film deposited through the mask will
be more uniform when using thin-mask. At the present time free masks 50-200
microns thick are used.
The precision of making the free mask layout is determined by the required accuracy
of the rated values of the deposited elements of thz microcircuits. Thus, in
order to obtain resistors with a precision of the resistanCe ratings to +5% and
capacitors with accuraey of the ratings td-+10% requires precision of the hole
sizes in the masks of +5 microns.
6-2. Free Mask Production Technology for Thin-Film IC
_ Mechanical Method~ of Making Free Masks. In order to make monometal foil masks
with simple pattern and relatively large hole sizes (no less than 0.3 mm), it is
possible to use drill~ng, milling, cutouts, stamping an d boring. Hawever, these -
methods do not provide high precision a~1d reproducibility of the results. The
mask billet can be deformed sharply under t~e effect of th,e tools; therefore it is _
impossible to use billets that are too thin. The thickness of the metal form must
be no less than 0.2 mm, and the spacing between holes, 1-2 mm.
Method of Electrosgark (Electroerosion) Machining. Holes in metal billets are
obtained using electospark discharge which is created between two electrodes
placed in a liquid dielectric inedium (Figure 6-3). The f ree mask bi~let or set of
billets made of 20-30 metal foil plates is one of the electrodes. The tool plays
the role of the second electrode. A capacitor is charged fram a voltage source
through a current limiting resistance, 'Jhen the maximum capacitance is re~ached,
'the capacitor discharges, that is, dis~::arge occurs between the electrodes. After
estimation of the electric strength of the electrode gap, the capacitor again -
begins to charge, and the charge-discharge cycle continues. With corresponding
~ selection of the tool mat~rtal and the pulse parameters the so-called polarity
- effect operates in which predominant melting~and evaporation of one of the
electrodes the mask billet take place.
99
FOR OFFICTAL U~E ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
A~ a result of the successive effect of a large number.o~ discharges a hole is
formed in~ the billet which repeats the~ shape of the~ tool. The~ waste is removed
- from the machine zone by forced pumping of the' liquid through the spark gap. An
extraordinary increase in energy of ~ single pulse leads to an increase in rough-
ness of the holes; therefore in order to insure sufficient reproducibility with
optimal pulse energy the pulse repetition frequency is increased.
- ~
, - R.
- . 1 ~
= 2 C
~ U~
_ ~ ~1
Figure 6-3. Diagram of electrospark machining. -
1-- tool electrode; 2-- dielectri c liquid; 3-- free mask billet
electrode
This method also fa31s to provide for o~.taining masks with precisian pattern and
sma11 hole sizes.
Manufacture of Free Maeks by Photolithography. Monometal masks are made using
electrolytic deposition of inetal films or using through local pickling of the metal
billet-foil. In both cases'it is necessary to fnrm a photoresistive mask. The
pattern of the photomask used in the first case must provide for obtaining a
photoresistive mask with layout that is the inverse of the layout of the manufac- _
tured free mask, and in the latter case, the direct pattern of the free mask. When
making bimetal masks the electrolytic local deposition is used in combination with
local chemical pickling.
The method of electro~yCic deposition consists in local deposition of a nickel or
copper layer on the surface of a polished steel plate shielded by a photoresistive
mask. After formation of the photoresistive mask the steel plate is placed on the
cathode of an electrolytic b ath and a metal layer 0.05 to 1 micron thick is
deposited. The nickel and copper films have low adhesion to a polished surface;
therefore the mask obtained can be easily removed from the steel plate. It is
possible to use the steel plate to obtain several free masks. In spite of the
relative simplicity, the method is rarely used, for with small thickness insuring
su�ficient p recision of p attem transfAr, the strength of the �ree maslc'is inade-
quate. Increasing the thickness of the mask lawers the precision and reproduci-
bility of the pattern. It is difficult to obtain masks with complex configuration
of the p attern as a result of the possibility of rupture of the mask when it is
separated from the steel plate. The efficiency of the production p rocess is low
as a result of the electrochemical deposition time.
The method of local chemical pickling is simpl.er and more efficient, and it is
widely used to make monometal and bimetal free masks. _
The flow chart for the manufacture of a monometal molybdenum free mask includes
the formation of a photoresistive mask made.of negative photoresists and 1oca1
pickling of the molybdenum foil through openings in the photoresistive masks. The
- 100
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFF[CIAL USE ONLY
molybden~ is pickied electrolyticallg-.in a compound made up of H3P04:H2SO4:H2~
4:1:4 heated to a temperature of 160�C. ~T~ie' pickling procedds not only in the
direction of the normal to the foil surfacey but also in the lateral directions,
which leads to distortion of the holes and law precision of pattern transfer. On
the average, the width of the side distortion region approaches the depth of
pickling; therefore it is impossible to obtain holes that are wider than the foil
tlitcltness. The deep pickling does not provide reproducible results: the masks
obtained have a divergence in the dimensions of the same elements of about 5-10 -
microns. .
6pvN9a t 1)
NuKenb (2 ~
_ ~
b)
~la6ucawu~uci ( 3)
Kpau HuKenx
d)
Figure 6-4. Diagram of the process of making bimetal, two-layer _
f ree masks .
a-- formation of the photoresistive mask and application of the
protective coating to the back side of the substrate; b-- local
_ elect~+olytic depoaition nickel; c-- removal of the photoresistive
mask and pickling of the mask base; d-- removal of the protective
coating
Key:
- ~ 1. Bronze
2. Nickel
3. Overhanging nitkel edge
~
The process flow diagram for making a bimetal two-layer mask is presented in
~igure 6-4. In the given diagram, in order to decrease the side dlstortions and,
consequently, for more exact transfer of the fmage from the photomaek, photo--
- lithography with a sublayer of nickel is used. At the end of the technological
process the sublayer is not removed, but remains in the free mask and plays the
basic role, being a type of mask, for it determines the ahape and size of the -
elements during deposition. The plate billet made of beryllium bronze foil
100-150 microns thick in the free mask is the structural base .insuring mechanical
strength. A layer of nickel 7-10 microns thick is applied by electrochemical

- deposition, and it is held on the bsonze' base by the forces of adhesion. The holes
in the base are pickled out using a mixture of chromium_anhydride and sulfuric
acid heated to 50-60�C which does not~act on the nickel.
- 101
FOR OFFIC[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
. - : ~ ~
/ y ~ ~ / ~ ;
.
':t: c;:
~ / y .
a~
"i:t~:.~:
; 5�, :i: .
_bl
;;,r.;,.:
~.,~~~4
h
~ �
. ~ . -
Figure 6-5. Flow chart of the manufacture of a three-layer bimetal
free mask.
a-- two-sided exposure; b-- development; c-- electrolytic applica-
tion of nickel; d-- removal of the photoresistive mask; e-- two-
sided pickling of the base
- Bimetal free masks are distinguished by relative simplicity of the manufacturing
- proces~ and quite high precision of transfer of the pat~ern from the photomask.
Therefore the bimetal masks are the most widespread in the manufacture of film
microcircuit~. When depositing the films the substrate is placed on the nickel
layer side; therefore the distortions of the mask base are not transferred to the
pattern of the deposited elements. Nevertheless, they are undesirable, for weak
overhanging edges of the nickel coating are formed which can be deformed, forming
a gap with the substrate or even break off. The basic deficiency of two-layer
bimetal masks is their strong deformation as a result of the differences in TKLR
[thermal coefficient of linear expa~a.ian] of the nickel and the nickel base. In
order to eliminate this deficiency preliminary heat treatment of the billets of the
basE~.arid the finished masks is carried out or three-layer bimetal masks are used.
The plotted process flow diagram for making three-layer bimetal free masks is pre-
sented in Figure 6-5. The decrease in distortion of the base of masks is
achieved by two-sided pickling. The total time for obtaining the holes ia
decreased and the size of the particles of the nickel layer overhanging the holes
is decreased. Obtaining the "mi rror" photomasks.for two-sided exposure of photo-
resistive layers and also matching them before exposure presents special difficulty
in the given technological process.
102
FOR OFFIC[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONY.Y
~
The manufacturing technology of three-layer bimetal masks is more complicated;
therefore they are used when it is.necessary to obtain a more precise and complex
pattern, and when this is econrnnical].y ~ustified.
Manufactu~e of Free Masks by Beam Processing. Precision processing of ~ foil by
a sharply fo cused electron, laser or ion beam permits a numb er of deficieneies of -
the photolithographic methods to be avoided. The pro cessing by an ~~ectron or _
- laser b eam is based on their thermal effect on the billet material. When the
~ corresponding temperature is reached, local evaporation takes place. Lacal removal -
of the material by using an ion beam is connected with the sputtering phenomenon.
The rate of removal of the b illet material can be regulated by varying the energy,
the dosage and the duration of irradiation.
The ion and electron beams have quite high pawer and will permit us to obtain
through holes in the foil up to 100 microns thick or more. The processing of the
foil by laser pulses lasting 5�10'8 Go 10-~ seconde will permit removal of a layer
about 1 micron thick from the surface in one clash and make it possible to obtain
high-quality thiough holes.
The use of electron, laser and ion beams periaits exclusion of the photomasks
and chemical reagents from the technological process. The movement of the beams
over the surface can be programmed and controlled by computer. These methods are
highly prospective, for they make it possible to increase the deficiency of the
technological processes, to improve the percentage yield of usable ~IC as a result
of increasing the precision, and to decrease the sizes and size tolerances of the ~
~ elements. The application of beam processing is stiJ:l being delayed by the
co~lexity and high cost of equipment. _
6-3. Stenciling Method
Essence of the Method. The stenciling method used in microelectronics has ancient
origins. Thousands of years ago~the Egyptians decorated the walls of structures,
tombs, and pottery by stenciling. The stencil is applied to the surface to be
decorated, and paint is forced through the open parts of the stencil by a small =
board. Sten ciling is also used at the present time: for applying patterns to
fabrics in the textile industry (silk screening); in decorative applied art for
reproducing graphics with simple patterns; and to manufacture small series of cards,
_ and so on. Tn IC production stenciling has been borrowed from silk screening and
the printed plates manufacturing technology.
The essence of the method consists in mechanical forcing of special pastes through -
openings in a free mask and subsequent heat treatment to give the film elements
the required properties.
Production Technology of Free Masks for Thick-Film Technology. In order to obtain
- thick-film elements of given configuration, reticular, foil or combination masks _
are used, the manufacture of which is based on the use of the photolithography -
techniques. _
- For reticular masks (Figure 6-6) prima~ily a~tainless steel or nylon screen is
used. The stainless steel s creen is more rigid and resistant to the effect of -
103
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
~
~
I
solvents and other components in the pastes. The .screens are~ uniforml~ stretched I
by a special device and fastenEd to rigid frames made~ of aluminum alloy. The
extra ends of the screen are cut off. T~ien the screen with the frame ia care-
fu11y degreased in hot hydrogen pQroxide, fluahed in water and dried. A negative
photoresist is applied to the working part of the prepared screena (by spraying,
pouring or rolling), so that the screen mesh will be filled. After drying. in the
air the process of applying the photores~st is repeated. As a result of exposure
and subsequent development at the required locatians the photoresist is washed off,
- leaving the open sections of t~ie screen "holes." In order to improve the
resistance of the mask to wear, the photoresist is reinforced by drying at a temp-
erature of 110�C.
Instead of photoresist it is possible to use pigment paper, for example, type
VTU-115-56. The paper impregnated with photosensitive solution and the photomask
prepared in advance are joined by iche emulsion sides, they are i:ightly clamped
~ together, and then the paper base is pealed off the pigment layex which remains on
the photomask. Exposure, development and the pigment layer from the pha~omask are
transferred to the screen. For this purpose the photomask is applied under load to
the screen, it is dried, and after this, the photomask is separated. The masks
obtained using pigment paper have high resolution, for the shielding of the pattern
by the screen is excluded during exposure.
The manufacturing process of foil masks for thick-film microcircuit is the same as
for thin-film microcircuits. The foil masks also can be bimetal and two or three-
layer (with the application of two-sided pickling) based on beryllium bronze and
electrochemically deposited nickel.
The combined masks are divided into composite and all-metal. For the composite
masks, metal foil 25-125 microns thick is used in which the required pattern is
pickled out using the me~hod of photolithography. Then the foil is joined to the
stainless steel screen. The primary difficulty when making compusite masks is
matching the` screen mesh with the pattern lines in the foil.
When making all-metal masks, two-sided ~ickling of the molybdenum foil is used, the
required pattern is etched out on one side, and the screen on the other. The
production of these stencils is quite complex; therefore their cost is high.
3 '
~
2
1
Figure 6-6. Reticular mask.
1-- solid part of the mask; 2-- frame for attaching the screen;
3-- hole for paste to pass through .
104
- FOR OFF'ICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
~ FOR OFFICIAL USE ONLY
Test Questions and Assignments
1. What methods are ~ised to obtain the configuration of IC film elements?
2. What is the essence of the free mask method and in what case is it used?
3. How is the matching done to obtain the configuration of thin-film elements?
- 4. By comparing free aud contact masks, the process for obtaining a layout using
them, analyze the character{ stic features, advantages and disadvantages of the
free mask method.
5. What requirements are imposed on free masks?
6. Why have the methods of machining and electrospaxk processing of free masks not
found broad industrial application for their manufacture? What is the essence of
these methods?
7. Which masks and from what materials are masks made using photolithography?
8. What is the manufacturing process of monometal free masks using a photoresistive
- mask and electrolytic deposition of a metal film? ~ Why is the metal film deposited
on the surface of the photoresistive mask?
- 9. Which method of contact photolithography is used to make monometal masks for
, using local pickling (see ~5-4) ?
10. Which contact photolithographic method is used to make bimetal masks by
- local pickling? What requirements are imposed on the pickling agent of the mask
base?
11. What are the basic deficiencies of two-layer, bi-metal masks and how can they
- be partially eliminated?
12. Compare the manufacturing technology of the Chree-layer bimetal mask with the
manufacturing technology of the two-layer bimetal mask.
13. What is the mechanism for obtaining holes in foil when making free masks using
electron, ion and laser beams?
14. What is the essence and application of the stenciling technique?
15. Which free masks are used in thick-film technology?
, 16. How caaz reti~ular masks be made?
~ 17. What distinguishes composite masks from all-metal masks? Are they used
f req uen!:ly?�
18. Compare all~ the methods of manufacturing free masks from the point of vidw uf
, their advantages, disadvantages and practical application.
105
FOR OFFICIAL USE ONL~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
CHAPTER 7. NEW LITHOGRAPHY TECHNIQUES
7-1. Contactless Photolithography
Limitations of Contact Photolithography. Improvement of the functional complexity -
and, consequently, the degree of integration of IC faces the methods of obtaining
- the images with the problems of improving resolution, the achievement of maxtmum
precision of matching the images and insuring minium defect density in large work-
ing areas. When solving these problems contact photolithography encounters defined
r~stric'tions .
A signiiicant limitation of contact photolithography is the unavoidability of
mechanical damage to the working surfaces of the photomask and substrate, for when
matching tnese surf aces are close to each other (lU- 15 microns), and during exposure
they are tightly pressed against each other. As a result of ineehanical wear of the
film pattern, p artial replacement of the photomasks is required, which makes it
necessary to shut down the equipment and makes automation of the exposure process
_ inexpedient. . .
On contact, the photomask presses dust particles, glass microp articles, and so on
into the photoresistive layer. The photoresist is poured onto~the photomask. In
- addition, any particles that are opaque for ultraviolet radiation getting between
the photomask and photoresistive layer are also the cause:of defects in tihe photo-
resistive mask. .
Obtaining the tight complete cantact between the photomask and the substrate is
in practice an irresolvab le prob lem as a result of bending of the plates (especially
the epitaxial structures) and the substrates, nonidealness ot. the planeness of the
contact surfaces, the presence of foreign particles between them, nonuniformity of
the thickness of the various films and the photoresistive layer, and so on. The
partial air gaps lead to intensification of the diffraction effects and cause addi-
tion al expansion of the lines of the images obtained. However, as a result of the
fact th at the light refraction in the air is approximately half th at in the photo- -
resist, the transfer pattern is distorted still more. The cause of reduced reso-
lution of the contact photolithography is also reflection of radiation fram the
substrate.
The ma~mum precision of matching in the case of contact photolithography is -
limited by the complexity of creating a system for fixing the transition from
the "gap" position to the "contact" position; therefore making the trsnsition from
106 ~
~ FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFIC[AL US~ ONLY
matching to exposure, a shift of the photomask relative to the substrate is
possible. Errors can also appear as a result of the fact that the operator matches
the patterns of the photomask in the substrate located in different planes.
In connection with the presented limitations of contact photolithography, further ~
improvement of the p rocesses of obtaining the configuration of the IC elements
develops inthe direction of applying contactless methods of exposure and decreasing
the diffraction phenomena.
Photolithography on a Microgap. This method is based on the use of the effect of
a double or multiple radiation source which is created in the exposure systems of
- special design. Ultraviolet beams hit the photomask and substrate inclined at
identical angles to the co~on optical axis of the exposure system. As a result
of inclination of the beams, the diffraction phenomena beyond the transparen~t
sections of the photomask are eliminated or they are reduced to the minimum, the
tn~iformity of irradiation of the substrate edges improves, the precision of the
transfer of the p attern increases. As a result of exclusion of the diffraction
phenomena, high resolution is achieved. Thus, for example, in the case of a layer
of positive photoresist 1.8 microns thick it is possib le to obtain a pattern element
of less than 2 microns (with a photomask-substrate clearance of 10 microns) or less
than 3.5 microns (with ~ clearance of 30 microns). The amount of clearance can be
precisely given by a polyester fi].m or peripheral strips on the photomask, for
example, made of quartz. The industrial eq uipment for expasure on a microgap is
appreciably more complicated th an the contact exposure unit. At the same time the
systems,permit a reduction in exposure time to 2-5 seconds and insure uniformity of
illumination of large-area substrates.
The absence of inechanical contact between the photomask and the substrate makes the
service life of ttie expensive photomask in practice imlimited. Local damage to the
_ photomask pattern by mechanical particles getting between the substrate and the
mask is excluded. This completely eliminates the accumulation of photomask defects
as they are used and rejection of photolithography as a result of the defects; there- ~
fore the percentage yield of usable IC also increases.
Projection Photolithography. Projection photolithography differs from contact
photolithography by the technique used in the matching and exp~sure aperations.
The matching process is simplified, for special objectives are used to project the -
ima~e of the photomask on the plane of the substrate or, vice versa, the substrate .
image is projected on the photomask plane and the operator observes the images of
the substrate and the rhotomask in one plane. Accordingly, the problem of the
depth of field of the objective is excluded along with the problem connected with
it of precise establishment of the small substrate-photomask clearance. The match-
ing time is decreased, and~the precision is improved. After matching, the clearance
between the substrate and the photomask remains, and the ima ~e of th e photomask is
pr.ojected on the substrate by an objective.
Tl~ere are several optical systems for performing the projection photolithography
prc,cess. The system presented in Figure 7-1 is most frequently used. This pro~ec-
tion system provides for optimal illtnnination during matching and exposure. On1y .
the objective and semitransparent mirror are located between the photomask and the
- substrate. The direct path of the beams from the light source through the objective
to the photomask and then from the semitransparent mirror to the substrate comp].etely
- 107
~'OR OFFICIAL USE ONLY ~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFiCIAL USE ONLY ;
coincides with the return path o~ the beam observed uaing the microscope eyepiece.
The operator matches in the plane of the photomask using r~ microscope; then he
replaces the microscope by a source of ultraviolet radiation, condenser and light
filter, and proceeds with e~osure. A deficiency of this system is the necessity
for rearrangement on making the transition from the matching operation to the
exposure operation. In addition, an addition al light source and candenser are
required for matching. _
~
, ~ '
8 Z
3
4
S
7 6
3 t ~
Figur.e 7-1. Optical system for pro~ection photolithography.
1-- light source; 2-- con denser; light filter; 4-- ph oto-
mask; 5-- objective; 6-- semitransparent mirror; 7-- aubstrate;
8 microscope
The resolution of projection photolithography is higher, for diffraction of the
radiation in the gap is excluded. The method of projection photolithography
permits adaptation of the IC production process to a higher degree than the method
of contact photolithography.
The b asic technical difficulty in proj ection photolithography is complexity of
- developing high resolution objects for large image fields; therefore simultaneous
proj ection of the complete pattern of the photomask on the substrate is not always
possible. In the case of successive element-by-element projection of the image
which can be carried out using photocopiers, the efficieney of photolithography
drops sharply.
P rojection photolithography imposes high requirements on the planarity of the
substrate surface, the power and the monochromaticity of the ultraviolet source,
and imiformity of thickness of the photoresi~tive layer.
In spite of the technical difficulties, projection photolithography is the most
prospective optical method of obtaining the configuration of IC elements.
_ 7-2. X-Ray Lithography
Limitations of Photolithography. The theoretical physical factor limiting the res-
olution of optical exposure systems are the light diff raction effects. The
theoretically obtainable minimum line width Q.~n of the pattern as a function of
]_0 8
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
FOR OFFICiAL USE OIdLY
_ the wa~e length a of the irradiation-used for exposure is defined in accordance
with the Rayleigh number by the expression
1 -o,sta (7-1)
?aa+ - a ~
(1) . n sin 2
Key: 1. min
where n is the index of refraction of the medium between the objective and the
image; a is the aperture angle of exit (the angle between the edge rays of a _
conical light beam exiting from the pupil of the ob~ective in the direction of the
image) . ~
In real cases n=1; sin a,/2=0.95; for a=400 nm, Q~n 0.26 microns and for ~=300 nm,
l~n=0.2 micron. This is the theoretical limit for optical lithography. It has
not been achieved in practice. The ob~ectives have significant aberrations (dis-
tortions of the images), the photolayer has finite grain size, the patterns of the
topologic layers of the microcircuits are complex combinations of elements, the
diffraction patterns of which can be partially superposed on each other, and
defects are introduced even in the photolithograph ic process itself.
- In connection with what has been presented, further improvement of the technology
of obtaining the canfiguration of IC elements is develop~ng in the direction of
_ introducing the methods of lithography using irradiation with shorter wave length
by comparison with ultraviolet far exposure.
- Principles of the Method of X-Ray Lithography. The b asis for the method of x-ray -
lithography is interaction of characteristic x-radiation with x-ray r~sists leading
to a change in their properties in the direction ot a decrease or increase in
resistance to the developers.
X-radiation is obtained by irradiation of the target by an accelerated electron -
flux. Depending on the nature of the interaction of the accelerated electrons
with the atoms of the irradiated material, two types of x-radiation can occur:
- white or characteristic. White radiation is caused by breaking of the electrons
on interaction of them with the electrons of the outer shells of the atoms of the
irradiated material and subsequent transitions of the electrons of the external
shelis. Characteristic irradiation is caused by the interaction of accelerated
electrons with the electrons of the internal shells of the atoms of the irradiated
- material, as a result of which the latter go from the internal shells to external
or leave the atom. The ele ctron transitions to th e free internal shells of the
atoms are accompanied by characteristic x-radiation. The difference in kinetic
energi_es of the electrons of different internal shells is appreciably greater than
external; therefore characteristic radiation has significantly shorter wave length
than white radiation. X-rays with a wave length of 0.1-10 nm are used for x-ray -
lithog raphy.
_ X-ray resists, just as photoresists, are divided into positive and negative. The
former are depolymerized, and the latter are crosslinked under the effect of the =
x-rays. In photoresists ultraviolet rays are absorbed by photosensitive components "
109
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFIC[AL USE ONLY
of the molecules. X-rays are observed by the whole molecules, and electrons are
knocked out from the internal. shells of the atoms of the ~x-ray resist. The
released electrons interact with the polymer and play the predominant role in the ~
chemical conversions of depolymerization or crosslinking of the molecules. The
resolution of the resist deFends on the characteristic range of the electrons,
that is, the distances which the free electrons travel in the polymer. Both types
of resists (positive and negative) have identically high resolution.
In practice most frequently a positive resist based on polymethylmethacrylate is
used (RMMA), which is distinguished by high stability of the properties, absence
of sensitivity to ultraviolet radiation, resistance to the effect of acids (except
hydrochloric acid) .
, ~lu Cr
StO~
. n+
_ ,
~ Au
- Cr
Si Oz
_ 9,~~
Figure 7-2. Diagram of the process of making silicon x-ray masks.
a-- epitaxial grawth of the n-Si layer, application of Si02 or
A1203, Cr and Au films; b-- formation of tr,e mask pattern;
c-- local pickling of n+-Si
- The masks for x-ray lithography must be made on a sufficiently transparent base for
the x-radiation used. The material of the film pattern, on,the contrary, must be
opaque. For example, when using characteristic x-radiation of an aluminum target
(a=0.S3 nm) silicon is used for the mask base, and for an opaque pattern, gold. ~
The production technology of patt`erns for x-ray lithography is a quite complex
problem. A slightly alloyed layer of silicon about 3 microns tnick ~is grown on a
highly alloyed silicon plate. Films of silicon dioxide or aluminum oxide, chromium
and gold (Figure 7-2) are applied over this layer. Then the pattern is formed in
the double layer of chromium and gold. Then, using a contact mask made of silicon
dioxide, the highly alloyed silicon is removed by pickling whirh does not affect
the slightly alloyed layer. In order to obtain a flat, unbending mask of large
area, the pickling is done not over the entire surf ace, but locally, removing the
silicon in regi~ns corresponding to the arrangement of the pattern of one micro-
circuit or one semiconductor device. Thus, the initial silicon plate is the
structural base lending rigidity to the mask; the role of the mask itself is played
by a thin layer of slightly alloyed silicon with a pattern of gold applied to it.
X-ray Lithography Process. The order of the production operations for obtaining
images is the same as in ordinary contact photolithography. Let us consider the
basic ones of them.
110
FOit OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400010055-6
FUR UHF'(CiAL U~E ONLY
The application of the layer o~ x-ray resist RMMA {dissolved in methylethylketone)
to the substrate can be done just as the application of the photoresist, for
example, the method of centrifuging. The layer thickness is 0.1-0.5 micron.
- 1 2 4 '
~ 5
B
7
�~q
_ _Tf~~9 _
Figure 7-3. Schematic of x-ray exposure.
1-- electron gun; 2-- electron beam; 3-- target; 4-- x-ray formation
chamber; 5-- x-rays; 6-- exposure chamber; 7-- mask; 8-- x-ray
resist layer; 9 substrate
Matching is a serious technical prob 1em. Several matching methods and devices are
known. For example, matching by special depressions pickled out in the substrate
and the mask, to the bottom of which matching marks made of a layer of gold or
another material that absorbs x-rays well are applied. On the bottom of the sub-
strate, directly under the matching marks, there is an x-ray detector which forms
the mismatch signal. The signal is fed to the device that shifts the substrate
in the plane parallel to the mask to an exact match.
X-ray exposure is done~in the chamber, which in order to prevent attenuation of the
x-rays is filled with helium or evacuated to a pressure of 1.33 Pa (Figure 7-3).
T.he accelerated electron flux is aimed at the target, which serves as the source
of the x-rays. The x-rays pass through a thin bezyllium foil and trana.parent
sections of the mask and they project its image on the layer of x-ray resist. The
beryllium foil shields the x-ray resist from thermal radiation and secondary elec-
trons which can cause polymerization of the x-ray resist. In order to decrease the
erosion of the image projected on the substrate as a result of divergence of the
angle of incidence of the x-rays, the distance from the target to the substrate is
in creased as much as possible. The clearance between the substrate and the mask is
3-10 microns. It is impossib le to make the ex~osure without clearance, for the
~ silicon mask is very brittle.
The exposure time t=D/(~+-u) , where D is the required radiation dosage; ~ is the
~ incldent x-ray flux; u is the coefficient of absorption of the x-rays by the resist
layer. The exposure time in x-ray lithography is from several.seconds to tens of
minutes. The RMMA film is exposed uniformly over the entire thickness, which
ir,sures that a sharp vertical profile of the edge of the image wi11 be obtained.
The RrfMA layer is exposed in a mixture of 40% methylisobutyl ketone and 60% iso-
propyl alcohol.
111
FOR OFFICIAY. USE ONLY -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
FOR OFFICIAL USE ONLY
The x-ray lithography is distinguished by high resolution. A limitation of the
resolution is insufficient absorption of x-rays by the mask gattern film. The
absence of contact of the inask with the~~resist lowers the level of the defects and -
increases the service life of the masks. The contamination transmits x-rays and,
consequently, is not transferred to the resist p attern. The reflectiens and dis- -
persions of the radiation have no influence. on the transfer of the image.
The basic deficiencies of x-ray lithography are large delay during exposure and the
phenom~non of distortion of the image with respect to the pattern field (distortion),
which is explained by the effect of the mechanical stresses occurring in the mask -
when making it. . -
7-3. Electron Lithography
Fundamentals of the Method of Electron Lithography. Electron beam exposure is dane
in vacuum devices, and it is based on the nonthermal interaction of accelerated
- electrons with electronoresist. Various polymers are used as the electronoresists,
including photoresists. However, preference is given to special electronoresists
insensitive to visible and ultraviolet radiation. The absence of~light sensitivity
of the electronoresists facilitates handling them in the process of manufacturing
the microcircuits. During exposure the electronoresists must have no natural vapor _
pressure a.nd must not form chemiCal compounds wh'ich cantaminate the vacuum chamber
of the de~ice.
Electron bombardment causes excitation and ionization of'the elec.tronoresist mole-
cules. The presence of the electric field of the electron increases the n~ber of
methods of rearranging the electronoresist molecules. Having high energy reserve,
the electrons rupture almost all the ~hemical bonds on their p.ath. Transverse
crosslinking of the molecules takes place simultaneously. In each specific example
usually some of these effects predominate. Accordingly, electronoresists, analo-
- gously to photoresists, are divided into negztive and positive.
It has been demonstrated experimentally that the degree of crosslinking of negative
electronoresists and the degree of depolymerization of positive electronoresists are
directly proportional to the radiation dosage, th at is, the magnitude of the elec-
tron charge pex unit irradiated area. The crosslinking or depolymerization take
place completely if the energy of the incident electrons is sufficient for their
free path length to exceed the thickness of the electronoresist layer.
The exposure of the electronoresist can be accompanied by undesirable phenomena:
- contamination of the electronoresist with impurities made of the residual gases of. _
the vacuum chamber of the device; acctmmulation of an electric charge which is the
cause of distortion and worsening of the focus of the beam (this phenomenon is
_ eliminated by the application of a thin transparent film of inetal for the electrons,
fnr example, aluminum); the occurrence of radiation effects in the substrate;
random dispersion of part of the electrons, which has a negative effect ot~ the
- verticalness of the pattern walls.
For electron beam exposure, devices are used with an accelerating voltage of 104
to 4�104 volts, which correspond to an electron w ave length of 100-50 nm. In
- practice electronolithography is used to obtair_ images with element sizes of
~ 112
� FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
- FOR OFF[CIAL USE ONLY _
0.1 to 0.2 micron. The wave length of the electrons and the diameter of the
electron beam decrease with an in crease in the accelerating voltage. Obt aining
electron beams with a wave length of less than 0.1 nm is cansidered technically
achievable, that is, theoreticall~ the resolution of electronolithography can
reach values close to 10-4 micron.
At the present time two methods of e~ectronolithography are used: treatment with a
focused single beam (scanning electronolithography) and electron projection of the
entire image (projection electronolithography).
Scanning Electronolithography. For exposure by a focused beam of electrons, scanning
electron microscopes or specialized devices electron beam accelerators (ELU)
are used. The scanning electron microscopes make it possib le to obtain p attern
lines about 0.1 micron wide in an area to 6 cm2; the line width obtained in the
electron beam accelerators is about 1 micron in an area of 4�10-2 cm2 (the Japanese
_ EBX-2B device).
The shifting, blocking and unblocking of the electron beam are realized using a
remote copying device, a phototelegraphic unit or computer. In the first case
(Figure 7-4) a mask made of glass with enlargement of the pattern is used as the
program for cantrolling the electron beam. The light brightness is varied beyond -
the mask in accordance with its pattern. The photomultiplier locat~d beyond the
mask reads and intensifies this brightness signal. The intensified signal is trans-
mitted to the electron beam accelerator for control of theelectron beaQn in
accordance with the image on the mask, b ut on a correspondingly reduced scale.
When controlling an electron beam from a remote copier the exposure of the electrono- _
resist is realized by raster scanning (Figure 7-5, a), that is, the beam is moved
line by line over the entire surf ace, switching on and off at the required loca- =
tions. During electron beam control using a phototelegraphic set the pattern is =
transferred to the electronoresist from the drawing paper. The beam is controlled
from a computer more efficiently, for vector scanning is used (Figure 7-5, b). In
this case the electron beam is shifted not over the enti.re surface of the electrono-
resist layer, but it scaris only the programmed sections, switching on at the points
of transition from one element to another. Using the focused beam i~ is also possi-
ble to draw ("mill") the programmed outline. This decreases the total exposure time.
In addition to shifting the beam, computer-controlled shifting of the table takes
place on which the substrates are located. This increases t~he e~cposure area. Two
mettiods of using th~e table are used: continuous and by the amount of one crystal
or plate. The ELU-EVM [electron beam accelerator-camputer] system perniits exposure
directly after development of the topology using a computer without making masks,
which greatly reduces the time for the development of the new microcircuits and
insures a rdnimum development-production cycle.
The matching is done using reference marks. When the electron beam falls on the
_ edge of the mark, the signal from the reflected electrons changes and falls on the
detector, tr~nsmitting information about the mismatch in the computer. The computer
chan~es the shift of the beam, matching the created'image with that previously
obtained on the substrate. Usually the beam is deflected by small amounts, about _
2 mm. The matching based on isolation of the signal from the reference mark is -
theoretically different from the photoopt~cal methods of matching requiring micro-
scopes. The precision of this matching is �0.5 micron. _
113
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONY.Y
. .
r
_ _ ~
i-~ ~ -
~ -
~
- ~ -
~ -
~ Om 311Y ~ r
v
~ ( a) r
2 ~ ~
- a
3 �
, , , -
4 =
5
�
6
~
I � -
7 ~ �
K 9ny(b) ~ ~
~
. b)
- Figure 7-4. Diagram of the remote copy- Figure 7-5. Diagram of the raster
ing device for contro lling an electron acanning of a beam (a) and vector
beam. ~ (b) acanning of the beam on e~osure
1-- reflecting system; 2-- kinescope; of electronoresists
_ 3-- light spot on the kinescope screen;
- 4-- objective; 5-- mask; 6-- lens;
- 7 photomultiplier
Key:
a. from the electrQn beam
accelerator
b. to the electron beam
- accelerator
ThQ method of successive treatments by a focused single beam of electrons insures
_ high resolution and precision of matching. Lines 0.1-0.4 micron wide are actually
obtained.
The basic deficiency of the meth od of treatment with a foeused electron beam is
low outp ut cap acity as a result of the long exposure time, for when the be am
. diameter is decreased it is necessazy to reduce it~s current and increase the n~ber
of scanning lines.
114
FOR ()FFiCIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Pro~ection Electronolithography. The entire pattern of the mask is simultaneously
transmitted to the electronoresist (Figure 7-6) . The basic element of the intra-
chamber device of the vacuum unit is a three-layer photocathode which simultaneously
plays the role of a source-of electrons and the role o~ a mask. The photocathode
is a polished quartz plate, on the surface of whict~ the pattem is made on a 1:1
. scale from a layer of titanium diox~de. A palladium~ fi1~ 4 nm thick is applied over
- the titanium dioxide over the entire area of the pattern. The titanium dioxide is
opaque for ultraviolet radiation, and the palladium film has high photoemission
properties. The photocathode on the quartz side is irradiated by ultraviolet, and
- on the ba~ck side from the sections coated only with the palladium film, photo-
electrons are emitted. Then they are accelerated by the electron field, and by
using the focusing system an imagq is pro3ected from the cathode on the layer of
electronoresist in practice without distortion. In the structural design of the
- device provision is made for a deflecting system which permits the projected image
to be shifted in the plane of the substrate and at the same time, matching to be
= provided. The mismatch signal of the matched patterns is picked up from the
= reference marks using detectors. The precision of the matchin~ is +0.25 micron.
12 . 3 4 5 6
~ . .
- Iji ~~i I~1
lil t~~ ~ii
1I 9 ~Yj 8 li~l 7~
. . ~
Figure 7-6. Pro~ection electronolithography diagram.
1-- deflecting system; 2-- focusing system; 3-- ultraviolet
- radiation; 4-- quartz ~base of the photocathode mask; 5-- titaniu:u
dioxide; 6-- palladium film; 7-- substrate; 8-- electron flux;
9 electronoresist layer
The method of electron pro~ection of the entire image has good resolution (lines
1 micron wide are obtained on the working field 25 mm in diameter~, great depth of
field by comparison with the optical methods, reaching up to +50 microns. The
service life of the photocathodes is greater than that of the photomasks. The
efficiency of the process is comparable to the efficiency of photolithography.
- The deficiencies of the method include the complexity of connecting the detectors
for matching and complexity of making the precision photocathodes.
In conclusion let us note that the methods of eontact photolithugraphy and the free
mask method widely used in~ production are not adaptable for fast rearrangement of
the production of systems of a broad nomenclature without significant capital
expenditures. This is ~connected with prolonged processes of m3lcing sets of photo-
masks or free masks. The method of electronolithography p~rmits the expensive
coordinatographs, devices for cutting out the enlarged images and other attachments
115
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
~'OR OFFICIAL USE ONLY I,
for making the photomask masters to be.elimina~ed. During electranolithography, ~
changes in the co~figuration of the~microcircuit structure can.be produced by
corrections in the program if a computer is used as the devlce for assignm~nt and
, reading of the program. The total number of operations in the technological pro-
_ cess of electronolithography is inuch smaller than in the optical processes. For
_ example, the labor consumption of the manufac~t~re of the masks for electrono-
lithography is six times less than the 1Zbor consumption for the manufacture of
photomasks. The technological process using electronolithography has the greatest
advantage when creating special-application nd crocireuit, that is, circuits of
� large nomenclature and small circulation, for it permits significant reduction of '
the production preparation time as a result of the possibility of fast changing of
the electron beam displacement program. Electronolithography is an easily automated
technological process; therefore, in spite of the complexity of the equipment by
comparison with optical and x-ray equipment, it is highly prospective for the manu-
facture of LSI with more th an 10~ elements. ~
Test Questions and Assignments
l. Repeat the basic operations of contact photolithography (Figure 5-5).
2. List the basic prob lems facing the technology of obtaining images of microcircuit
elements. '
3. What are the basic limitations of contact phdtolithography?
4. What methods of obtaining the canfiguration of the elements are included in the
optical sethods?
5. What is photolithography on a microgap and what are its advantages?
6. What is the essence of projection photolithography, what is the t~chnique for
carrying it out?
7. What is the basic technical difficulty~~of pro3ection photolithography?
8. What are the basic advantages of pro~ection photolithography by comparison with
~ other optical methods of obtaining images?
9. What are the theoretical physical limitations of the optical methods of obtain-
- ing images?
10. What distinguishes the characteristic x-radiation from white? Which of them is
used in x-ray diffraction and why? -
11. How is x-radiation obtained for x-ray lithography?
12. What requirements are imposed on the materials and what is the structure of the _
masks for x-ray lithography?
13. c4ake up a flaw chart for manufacturing the mask for x-ray lithography and
check it using Figure 7-2.
_ 116
FOR OFFICIAL USE ONLY _
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
FOR OFFICIAL USE ONLY
14. What aze the requirements on the silicon pickl~ng agent when making the -
mask? Why is pickling self-halting, .leaving a layer of s1igY~tly alloyed silicon?
15. What is the mechanism of effect o~ ~ rays on x-ray resists?
16. How is matching done in x-ray li'thography?
17. Explain the x-ray exposure sysfem (Figure 7-3) .
18. Compare x-ray lithography with the optical methods of obtaining images.
19. What is the mechanism of effect of accelerated electrons on electronoresist?
20. What is scanning electronolithography, and in what devices is it carried out?
21. Haw is electron beam control realized?
22. What distinguishes vector scanning of a beam from raster scanning?
23. What 3s the basic advantage of the ELU-EVM jelectron beam accelerator-computer] -
systems?
24. What is projection electronolithography and how is it carried out?
25. Explain the structural design of a photocathode (Figure 7-6) and the role of
all of its elements. Compile a photocathode manufacturing system. -
26. Why does the photocathode have long service life by comparison with photo-
masks used in contact photolithography?
27. What are the basic features, advantages and disadvantages of the method of
electr.onolithography by comparison with other methods of obtaining images in
microcircuit production?
117
FOR OFFICIAL US1E ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
~
FOR OFFICIAL USE ONLY '
I
~
i
~
~
CHAPTER 8. METHODS OF OBTAINING THIN FIIMS ~
~
8-1. Method of Thermovacuum Depositian '
i
- Principles of the Method. The method is based on creating a di~ectional vapor flow
of the material and subsequent candensation of it on the surfaces of substrates
having a temperature belaw the temperature of the vapor source. During condensa-
t~on a film is formed from individual atoms or molecules of the vapor. The process
. of thermovacuum deposition can be broken down into four steps: the format~.an of
the vapor, application of the vapor from. the source to the substrate, condensation
_ ef *_he vanor on ttaQ suhstrares, formation of nucleating centers and film growth.
The vapor is formed by evaporation or sublimation. The materials are converted to '
vapor at any temperature ab~.~e absolute zero, but in order to increase the intensity
of vapor formation heating is required. With an increase in temperature the m~an
kinetic energy of the atoms rises and, consequently, the probability of rupture of
the interatomic bonds incr-eases. Atoms split off the surface and are propagated in -
free space, forming vapor.
The spe..ific evaporation rate, equal to the number. of grams of material evaporated -
per second from a surface area of 1 cm2 is defined by the expressiori
M
- n~~b,85p:~9 ~8-1) -
(1)
Key : 1. evap
where ps is the saturated vapor pressure of the heated material, Pa; M is the '
molecular mass of the material, g/mo1e; T is the temperature of the evaporator, K.
_ The saturated vapur pressure depends strongly on temperature. The temperature _
increments every 5-10% above the evaporation temperature lead to an increase in
saturated vapor pressure and, consequently, the evaporation rate by one order.
The propagation of the vapor from source to substrate is realized by diffusion and
convective mass transfer which is primarily influenced by the degree of vacuum. -
In order to obtain high-quality films it is necessary that the atoms and molecules
travel to the substrates without collisians with the molecules of the residual _
' gases. This is possible under the condition where the free path length of the
. vapor particles a is greater than the source-substrate distance.
_ iis -
- FOR OFFICIAL USE ONLY -
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFIC[AL USE ONLY
From the kinetic area of gases the mean free path ]ength is determined by the
expression _ _
1
$'ad'N' (8-2)
where a is the mean free path length, cm; d is the effective diameter of the
molecules, em; N is the concentration of the molecules for given pressure and
- te~uperature, 1/cm3.
The concentration of the molecules and the pressure are related to each other by
the expression _ _ _
N (8-3) -
on substitution of which in expreseion (8-2), it is possible to determine the rela-
- tion of the mean free path length to the pressure
~ ~8-4)
_
Beginning with a vacuum of 1.2�(10-2 to 10'3) Pa, the mean free path length (47 cm
to 4.7 meters) wi~l become apgxeciably higher than the source-substrate distance
(10-15 cm), and the probability of collision in the drift space with molecules of
the residual gases is low. It is possible to consider that the atoms of the
- deposited material are propagated rectilinearly by a directional atomic f lux, re-
taining their energy until they encounter the substrate.
The condensation of the vapor on the substrate depends on the temperature of the
- substrate and the density of the atomic flux. According to the modern theory of
condensation, electrically neutral vapor particles, approaching the substrate sur- -
face, Fall into the field of the forces of attraction generated by the instantaneous
dipole moments of the surface atoms and moleeules. At a very close distance the
force of repulsion begins to act on the vapor particle. The vapor atoms, reaching
_ the substrates, can instantaneously be repelled from it (elas~ic collision),
adsorbed and after some time, repelled from the substrate (reevaporation), adsorbed
and after brief migration on the surface, finally remain on it (condensation).
Condensation of the atoms takes place if the energy of the binding of them to the
substrate atoms is greater than the mean energy of the substrate atoms; otherwise
the atoms are repelled from the substrate.
The. temperature above which all the atoms are repelled from the ssbstrate and the
film is not formed is called the critical condensation temperature. The critical
- temperature depends on the nature of the materials of the film and substrate and
the state of the substrate surface. The critical density of the atomic flux for a
given subst rate temperature is the least density on which the atoms condense on
the substrate.
The formation of the nucleating center takes place as a result of the atoms finding
locations corresponding to the minimum fYee energy of the ato~substrate system.
119
- FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY ~
i
The growth of the nucleating centers takes place as a res~lt of connection of new
. atoms migrating over the surface or ~alling an the nucleating center directly from
the source-substrate drift interval. As the vapor condenses, the nucleating centers
graw; connecting bridges are formed between them, the nucleating centers coalesce ;
into large islets. Then the sta ge of coalescence of the 3slets comes with the '
formation of a single grid. The grid beco~es a solid film which begins to grow in
thickness. In the g3ven step the effect of the residual gases on the substrates
and the growing film must be reduced to a minimum. This can be insured by increasing -
- the degree of vacuum or increasing the vapor formation rate. Haaever, the creation .
of supervacuum devices presents significant technological difficulties, and in
- addition, the evacuation time of the operating chambers is high.
- - . ~
A
2
niv~nnnn- 3
_ I � 4.
~
~ ' -
, ; ~ 5 .
. \ I'// 6 .
.
_ ~
~ 8
i
tOmrrq4 ~
~a
Figure 8-1. Diagram of the process of thermovacuum deposition.
1-- vacuum chamber hood; 2-- substrate heater; 3-- substrate
holder; 4-- substrate; 5-- spot valve; 6-- articles of -
~ evaporated material; 7-- evaporator with weighed sample of the
film material; 8-- supporting plate -
Key:
a. evacuation
In microcircuit production frequently metal films are deposited on dielectric and `
semiconductor surfaces. For such combinations of materials of the conc~ensate and
substrate which have no chemical affinity, the reevaporation is high; adsorption
and, consequently, the nucleating center formation and growth of thin films are
complicated significantly.
Thermovacuum Deposition Technique. The diagram of the process of thermovacuum
deposition is p resented in Figure 8- 1. The stationary and removab le equipment of _
the hooded devi ce is periodi cally cleaned to remove encrustations of the preceding
depositions. Weighed samples of the deposited material are degreased, pickled to
remove o:tide fil~ and contamination. Directly before deposition the evaporators -
and the weighed samples are annealed. Then the valve is opened and deposition on '
th e substrate takes place.
The basic pa-rameters of the process of thermovacuimm depositian are th e deposition
rate of the film and the substrate temperature.
120
FOR C2FFICIA~ ~I:~E ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIA~L USE ONLY
The deposition rate of the film is directly proportional to the evaporation rate.
The flux of atoms condensed on the substrate .surface at fixed temperatures of the
evaporator and substrate is defined by the expression
No= ~ i N~' ( 8-5)
Key: 1. evap; 2. flu~c ~2) ~1)
where Nevap is the flow of evaporating particles; R is the distance from the source
to the substrate; A is the coefficient takii~g into account the shape and the molecu-
lar-kinetic characteristics of the evaporator; k is the compensation coefficient
the ratio of the number of molecules condensed on the surface of the substrate
to the number of molecules impacting against it numerically equal to the ratio of
the condensation rate and the evaporation rate.
Evaporation almost always takes place at temperatures exceeding the evaporation
- temperature of the material (the forsed conditions) in order to insure quite high
vaporization rates, and, as a result, to -decrease the effect of the residual gases _
on the properties of the deposited films. At law evaporation rates the formation
of loose, rough films is possible.
The temperature of the substrate is selected as optimal in order to insure condensa-
tian of the vapor and adhesion of the films to the substrates. The heating of the
substrates is required for desorption of harmful materials (gases, moisture, oil
_ from the pumps) which are the basic cause of poor adhesion. The substrate tempera-
_ ture also influences the structure of the deposited film and, consequently, its -
electrophysical parameters.
The uniformity of thickness of the films ~ with respect to the area of the substrates
when using the majority of simplest wire, strip and crucible evaporators is unsat-
= isfactory. The film thickness is maximal at the substrate center, that is, in the
section located directly above the evaporator, and it decreases to the periphery
of the substrate. The uniformity of the thickness of the films can be increased as -
a result of increasing the distance between the evaporator and the substrate, but
in this case the deposition rate decreases and, in addition, the operating chamber
of the device has limited dimensions. At the present time the uniformity of the
thickness of the fi]sns in a large lot of substrates is achieved by application of
~ hooded devices insuring uniform rotation of the substrates fastened vertically to -
the generatrices of the cylinder,around the evaporators located along the central
- axis of the cylinder. The molecular f lux from the evaporators is propagated
through a station;ary diaphragm. The operating principle of such evaporators con-
sists in equalization of the average deposition rate at each point of the sub-
strate surface.
. The structure of the deposited film depends o~ the material, the state of rhe
surface and the temperature of the substrates, the rate of deposition, and it can
be amorphous, polycrystalline, fine-grained, polycxystalline, large-grained, and
monocrystalline. The grain size of the metal films depends on the melting point
- of the metal. Metals with high melting point (ttmgsten, molybdenum, tantalum,
platinum, and so on) form films with sma11 grain sizes. The films become solid
for comparatively small thicknesses. The low-temperature metals (zinc, cadmi.um, -
and so on) form large-grain films.
121
FOR OFFICIAL USE ONLY
~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
~OR OFFtCIAL USE ONLY
At the contact boundaries of the grain structural de~ects-are concentrated in the
solid film. The pxesence o~ the boundaries~itself int~oduces distortions into
the ~rystal lattices of the grains. The eavities b etween the graine are frequently
_ filled by amorphous contamination, oxides, and~so on. Therefore the large-grain
films have smaller cancentration of the structural defects which explains the
large stability of their electrophysical properties.
-Thus, during thermovacuum deposition preference is. given.:to increasing the
deposition rate, and the film grain sizes increase as a~result of an increase in
the substrate temperature durir~g deposition and annealing.
The films are annealed in vacuum devices directly after deposition at substrate
temperatures which exceed;:~he deposition temperatures somewhat. This is done to
accelerate the structure and decrease the internal mechanical stresses of the
films to increase their stability and improve adhesion to the substrate. During
the annealing process the intergrain distances in the films decrease and a~ a con-
sequence, the number of structural defects decreases. The resistance of:the .
resistive and conducting films decreases in this case.
Advantages and Disadvantages of the Method. The method of thermovacuum deposition
has been well mastered; it makes it possible to obtain any passive elements,
metallization of the semiconductor structures; it is used when making photomasks,
and so on. Using thermovacuum deposition it is possib le to obtain films of inetals,
semiconductors and dielectrics. The method provides high grawth rate of the films.
The process is distinguished by a high degree of cleanness and it permits high
quality films to be obtained in a high vacuum with camparatively low substrate
temperatures. Comparatively easy automation of the process permits the creation of
complex vacuum devices and computer-controlled�comp lexes.
The disadvantages of the method include variation of the percentage ratio of the
components with deposition of alloys and complex materials, and sufficient uniform-
ity of thickness of the films on the large substrates, difficulty of obtaining
fil~ of refractory materials, high inertia when using the evaporators (after
disconnecting the heating of the evaporators, the vapor forniation continues;
. therefore the process of film deposition is stopped by using a mechanical s lide
_ valve), comparatively law adhesion of the films, short duration of preparations for
th e pro cess and evacuation, and relative complexity of the equipment.
8--2. Ion Bombardment Sputtering
Principles of the Method. The methods of depositing films by sputtering the
material in a gas discharge plasma, by comparison with thermovacuum deposition,
expand the possibilities of obtaiaing films with given properties. They permit
deposition of films of r-efractory materials, alloys, complex materials and materials
with low vapor pressure. . '
The ion sputtering mechanism was briefly discussed above (�4-5) when describing
ion pickling of substrates.
The material is sputtered in the form of neutral atoms or molecules, and about 1%
of it is ionized by plasma electrons. Sputtering, in contrast to evaporation,
does not depend on the vapor pressure nf the deposited material. This permits
122
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
application o~ films of refractory materials at relativel3~ low temperatures. When
obtaining a11oy films,.initially b asically the component with the larger sputter-
ing coefficient, for eaample, nickel in nichrome,'is sputtered. The influx of
the missing component is realized by diffusion; therefore~equilibrium is quickly
established, and the composition of the sputtered atom flux corresponds to the
composition of the alloy. As a result~, only the first few atomic layers of the
film differ from the basic composition. The application of films of complex
composition can be accomplished also by sputtering several different cathodes.
The particles are propagated in a gas discharge plasma. As a result of collisions
with molecules of the residual gas, the directionalness of the sputtered particle
flux is lost, and the particles go to the surface of the substrates at different
angles. Along with the particles of the sputt~ered material, molecules and ions of
the residual gases also hit the substrates.
The process of formation of nucleating centers and the film formation on the surface
of the substrates during ion sputtering differ significantly from this step in
thermovacuum depositi~n. The sputtered material flux has a number of peculiarities:
greater energy of the atoms reaching the substrate (by 1-2 orders), lower particle
flux densities (by an order), random nature of the directions of impact of the
atoms against the surface of the substrate, to a greater degree presence in the flux
of the ions of the atoms along with the neutral atoms and also molecules of the
residual gas.
A1~ of this causes an increase in the substrate temperature, an increase in the
tangential component of the velocity of the atom, that is, an increase in the
migration rate along the surface, intensive desorption of the deposited molecules
and residual gases.
It has been experimentally established that the fi]~s are deposited for any densi-
ties of the atomic flux and in a wide tem~erature range of the substrate, th at is,
_ during ion sputtering for the formation and growth of nucleating centers there is
no critical temperature of the substrate or critical density of the sputtered
material flux. :~iuii~ wicn the mecY:anism of film growtn from neutra~ p~rticles, ti~e
ch arge me chanism growth operates here, for along with the atoms of the sputtered
source, the ions have the inert gas and a relatively large number of source a~toms
- ionized in the electrode gap hit the substrate. As a result of the additional
electrostatic energy the ch arged p articles increase the s urface migration b etween
the nucleating p articles and accelerate their growth in the p lane. The latter
explains the fact that the films deposited by ion sputtering become solid at less
thickness than the films obtained by thermovacuum deposition. The deposition rates
of the films, as a rule, are lower than during thermovacuum deposition as a result
of low flux densities. The film density is higher as a result of high energy of the
deposited particles.
Cathode (physical and reactive) and ion-plasma sputtering are distinguished.
Physical Cathode Sputtering. In this form of sputtering the ion source is an
independent glow dis charge plasma which is created between the cathode and ~anode
of a diode type vacuum chamber (Figure 8-2). The chamber is evacuated to maximum
vacuum, and then argon is admitted to it to a pressure of 1.2�(10'1 to 10'2) Pa.
123
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFiCIAL USE ONLY ~
On feeding a voltage on the ordex o~ several kilovolts between the cathode and ~
anode with a defined discharge curre~t, glow discharge.sets in which, by the .
nature of the potential distributian; can :be divided into a n~nnber of ~egions in ~
the sp acing between the cathode and.the anode: the dark'cathode space, negativ~ ~
- glaw, dark anode space and positive column. The basic voltage drop applied bet~aeen
the cathode and the anode is iri the dark cathode spaee ad3 acent to the cathode. ;
- In the positive colwmz region of the disch~Irge, -the charged plasma particles,
electrons and ions move diffusely. The ions incident in the vicinity of the
cathode space are accelerated by the electric fi~eld and bombard the cathode. As a
_ result of bombaxdment, the cathode sputtering and secondary electron emission take
_ place. Thus, the cathode is simultaneously the source of material for obtaining
the film and a source of electrons required to support glaw discharge in the '
chamber. During movement of the electrons from the cathode to the anode, their
energy in the catho~e sp ace increases and becomes sufficient for ionization of
the gas molecules. The glaw discharge is called independent, for the discharge
itself insures cathode emission without auxiliaYy means. The positive columms -
performs the functinns of the conductance section between the anode and cathode.
-
i
. 5 _
. i . ~ a~
i
_ i 2 ~-3K~ .
t ~ .
i 3
~ 4 .
Om~avKa ~b ~ -
Figure 8-2. Diagram of the cathode sputtering process
1-- argon ion; 2-- sputtering cathode particles; 3-- substrate;
4 anode; 5 cathode
Key:
a. 3 kev
b. evacuation
The amount of cathode materials sputtered per unit time Qp is proportional to the
voltage between the cathode and the anode U, the discharge current I, and it is
inversely proportional to the product of the pressure in the vacuum chamber p
times the distance R from the cathode to the surface of the substrate:
� .
UI (~6)
= Qp = h p~ ,
where k is a constant. ' ~
- With an increase in pressure ~p, the average free path, length and also ~the dark
cathode space length decrease. The movement of the sputtered particles from the
cathode~to th e substra~e basically is of a diffusion nature; therefore the displace-
ment of the sputtered particles decreases. With an increase in the cathode- -
subs~?rate spacing, the ~probability of collisions of the sputtered particles with
the ~residual gases and; con`sequently, the probability of their return to the
- ~ ~ 124
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OF~ICIAL USE ONLY
cathode, increases; therefore, in order to obtain the maximum grawth rate of the
film, the product p!C must be minimal. ~
LeC us estimate the effect of the discharge current I and the vo~ltage U on the
deposition rate considering the cathode sputtering~ coefficient S. The discharge
current is determined by the n~ber of ions incident on the sputtered cathode and
the charge of the ions I=qN. The sputtering rate VP=NS=(I/q)S. Increasing the .
voltage insignificantly increases the cathode sputtering coefficient, and increas-
ing the discharge current leads to a noticeab le increase in the sputtering rate.
Therefore in order to increase the sputtering rate and, conseq.ugntly, the growth
rate of the films it is expedient to increase the discharge current and not the
sputtering vo ltage.
The amount of material deposited on the substrate Qo is related to the amount of
materials sputtered by the cathode QP by the expression
qr - ~ 8- 7)
Qo = p~,~~ ,
where a=0.06 to 0.2 and 5=0.5 to 2 are the coefficients which depend on the sputter-
ing conditions.
The parameters of the cathode sputtering pro cess are as~follaws: gas pressure in the -
vacuum chamber, voltage between the anode and cathode, discharge current, distance
from the cathode to the surface of the s ubstrates, sputtering time.
Optimal pressure (10-1) Pa. At lower pressures the discharge current, and, conse-
quently, the sputtering rate will be decreased. At higher pressures the increase
in the ntunber of collisions in the gas discharge gap also sharply decreases the
sputtering rate. The voltage is maintained on a level of 1-3 kv. The greater the
distance from the cathode to the substrates, the more frequently the sputtered
atoms collide with the gas molecules and the smaller the n~nber of them reaching
the substrate. A distance between the cathode and anode which is 1.5 to 2 times
greater than the width of the dark cathode sp ace is optimal. `
Z'he advantages of cathode sputtering are as follaws : law substrate temperatures
during the film deposition process; greater film uniformity with respect to area
of the substrates than in the case of thermovacuum deposition, for the surface
area of the cathode is large by comgarison with the area of the vapor source during
deposition in a vacuum; inertialessness (sputtering begins on supplying the voltage
to the electrodes, and it stops instantaneously on taking the voltage away);
absence of the necessity for frequent checking of the source of the particles of _
the growing film (cathode); nonvariability of the stoichiometry of composition of
the film by comparison with the cathode ccmiposition; high adhesion of the film to
the substrates. ~
The basic deficiencies of the method of cathode sputtering are as follows: compara-
tively low deposition rates, contamination of the films by molecules of residual
gases and more complex control of the technological process by comparison with
thermovacuum deposition.
i25
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
In order to eliminate the contamination o~ t~te films, which has for a long time
held up industrial application~ of the method, at the present time sputtering is _
- used with shifting, sputtering of an asymmetric alternat~.ng current of industrial
frequency, getter sputtering and combinations of these methods.
Cathode sputtering is widely used when obtain~:ng dielectric films of Si02, A120g,
T a205 for thin-film capacitors, conducting Cu, A~ films with a sublayer of Cr,
Ti and Cr and NiCr films for thin-film resistors. _
Sputtering of dielectric materials in diode systems''ot~ direct current is impossible,
for the cathode surface is charged with positive ions, and further bombardment of
the cathode is stopped. Therefore sputtering is carried out with a high-frequency
AC voltage. In the case of a negative voltage h alfwave, ordinary Cathode sputter-
ing takes place on the clielectric cathode; in the case of a positive voltage~half-
- wave the positive charge accumulated on the cathode is neutralized by the electrons
extracted from the plasma.
Reactive Cathode Sputtering. In contrast to ordinary physical cathod~ sputtering,
reactive cathode sputtering is realized in a glow discharge of a mixture of inert
and active gases. The particles of the sputtered cathode interact chemically with
the active gas or they form solid solutions with it, an d t he new m a t e r i a l r e a c
h e s
the substrates. In order that the process of formation of the substance of the
applied film not take place on the cathode, which greatly ccmplicates burning of
the discharge, mixtures of argon with active gas content of no more than 10% are
used.
In order to obtaijnoxide films, the sputtering takes place in an ar.gon- oxygen
plasma, nitrides in an argon-nitrogen plasma, carbides in an argon-carbon monoxide
plasma or argon and methane. On introducing different active gases into the
ch amber, it is possib le to obtain films of various compounds which in practice
cannot be obtained by thermovacuum deposition. For exam~le, ferrite magnetic ,
films are obtained on sputtering of a nicke.l alloy with iron in a glow discharge
plasma of argan and oxygen. -
Reactive cathode sputtering makes it possible to ob tain not only films of various
compositions, b ut also it makes it possib le to contxol the properties of the films,
for example, the specific resistance of resistive films. Thus, when sputtering
a tantalum cathode in an argon-nitrogen plasma, thin T a2N films are obtai.ned, the
resistance of whi~h can be varied by the computation of the nitrogen introduced
into the chamber. Reactive sputtering is widely used to obtain high-resistance
resistors.
The basic technical difficulty of reactive cathode sputtering is exact matching of
the active gas introduced into the chamber.
- Ion Plasma Sputtering. Ion plasma sputtering is realized at law temperatures in a
vacuum chamber 1.2. (10-2 to 10-3) Pa. In prder to maintain efficient concentration
of the argon ions at lower pressure, the method of ionizing electrons is increased.
For this purpose, arc discharge~ is used. A~c discharge is independent. To maint ain
arc discharge a thermoemission cathode is needed. In contrast to glow discharge,
_ arc discharge burns at lower voltage between the anode and the cathode (100-300 volts).
The discharge current is several amperes.
126
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
9
2
3 �
--L- ~
-
_
5
6
7 ~OmKa
_ YK4 ('s)
_ Figure 8-3. Diagram of the process of ion-plasma sputtering.
1-- hood of the vacuwn chamber; 2-- heater; 3-- substrate;
4-- anode; 5-- cathode; 6-- target; 7-- needle leak
Key:
a. Evacuation
A three-electrode vacuum system (Figure S-3) is used to implement this method.
An additional electrode-target is the source of particles of the material to be
used for tne deposited film. The substrates are located opposite the targets on
carousels or a drum.
The chamber is evacuated to maximum vacuum, the cathode heating current is switched
on, after heating of the cathode a voltage is applied between the cathode ~nd anode,
and an inert ionized gas is admitted to the~ chamber. Arc discharge is struck
between the anode and the cathode. On feeding a small negative potential to the
target or substrate it is possible to obtain ion cleaning. For sputtering, a
negative potential of 200-1000 volts with respect to the anode is fed to the target.
The electric field of the target extracts positive ions from the arc discharge
. plasma and accelerates them to energies of hundreds of electron volts. Bombarding
the target surface, the ions sputter it. The target is usually placed at a distance
of 2/3 of the anode-cathode distance from the cathode where the region of the
passive discharge column is located. As a result of the thermoemission cathode it
is possib le independently to control the concentration and energy of the ions in
the plasma, for the gas discharge circuits (anode- cathode) and sputtering circuit
(target-substrate) are electrically decoupled from each other. It is possible to
vary the ion concentration by varying the electron emission current or the accelerat-
_ ing voltage between the cathode and anode, and it is possible to vary the ion
energy by variation of the target potentia:~.
Ion-plasma sputtering by comparison with cathode sputtering is realized at higher
vacuum; therefore the free path length and energy of the sputtered atoms are
greater, Accordingly, the contamination of the films by molecu].es of residual and
inert gases is less, the deposition rate of the films is higher, and it reaches
several hundreds and even thousands of angstroios per minute. The film density is
greater, the adhesion of the .films to the surface of the substrate is better.
127
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
I
FOR OFFIC[AL USE ONLY
The deposition process is more easily.,controlled, the ~ilm thickness is regulated
by the target potential and time. It is possible to carry out a large number of '
. film deposition processes without changing the ~argets, which insures reproducibil-
ity of the properties of the deposited films. T~?e process of fiLm deposition with
_ the given properties corresponds to the requirements of automation and can be used
in the continuous cycle of creating the microcircuits.
The deficiencies of the ion-plasma sputtering ~re as follows: limited possibili-
_ ties of reactive sputtering as a resu~.t--of short service life of the thermflemission
cathode in the presence of active gases, the cathode is an additional source of
contamination; co~parative complexity of the device and operation of the equipment.
~ Ion-plasma sputtering is widely used to make film passive elements, to obtain mask-
ing films on semiconductor plates, for deposition of semiconductor layers and
magnetic films. ~
8-3. Thermal Oxidation
Principles of the Method. The process of oxidation, intensified by heating to
high temperatures, is called thermal oxidation. The method of thermal oxidation is
basic in planar technology for obtaining masking films on silicon and films of the -
gate oxide for 1~IDS-structures.
Silicon has high affinity for oxygen. On a caref~lly cleaned silicon surface at
room temperature a film of silicon dioxld~ 10-15 A thick is formed instantaneo~sly.
This film grows at a rate of about 11-12 A/day to a thickness of about 50-100 A.
Therefore thermal ox3.dation in any case takes place in the presence of a thin oxide
film on the surface. -
_ The process of obtaining a thermal oxide can be broken down into three steps:
delivery of the oxidizing agent to the substrates and adsorption of them by the
surface, diffusion of the o~ddizing agent through the silicon dioxide film to the
silicon surface, chemical interaction of the oxidizing agent with the silicon with -
the formation of oxide. Purified dry or wet oxygen is used as the oxidizing agent.
The oxidati~n rate is determined by the slow phase of diffusion penetration of the
oxidizing agent thxough the growing film to the Si02-Si interface itself. The
- diffusion coefficients depend strongly on temperature. At law temperatures the
diffusion coefficients and, consequently, the grawth rate of the film, are small.
- It is possible to increase the growth rate either by increasing the pressure in the
reaction chamber or increasing the process tempera.ture. The creation of devices
for oxidation at increased pressures is very complicated, and it is economically
inexpedient. In practice the oxidation of the silicon takes place at low pressures,
but at high temperatures (850-1350�C) . The _ diffusion coefficient of water in
silicon dioxide at the same temperature is appreciably greater than the diffusion
coefficient of oxygen. This explains the high growth rates of the oxide in wet
oxygen. However, the growth of films only in water vapor is not used as a result
of poor quality of the oxide. Higher quality. films are obtained in dry oxygen,
but the grawth rate of the films is too small.
128 -
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFF[CIAL USE ONLY
When using dry oxygen the ~ollowing reaction takes place on the surface of the
s ilicon . .
Si-}-O~Si02.
(8-8)
When using wet oxygen an additional reaction takes place
_ _ - - -
Si-}-2H2O-->Si02-}-2H2~. ( 8-9)
The hydrogen formed during the reaction quite rapidly diffuses from the Si-Si02
interface to the oxide surface.
4 '
3, o0000 0 0 0 0 0
~ SL
2
0000000000
1
NZ - - _ ~ ' ,
HZO ~
pZ ~ _ .
T'C ~
Figure 8-4. Diagram of the process of thermal oxidation of silicon.
1-- valves; 2-- rotameters; 3-- quartz tube-reactor; 4-- heating
- furnace
- Silicon atoms are consumed to form the film; therefore the initial substrate surface -
goes into the body of the oxide during the oxidation process. This exp lains the -
absence of the problem of insuring high adhesion-. -
_ At temperatures of more than 1000�C the silicon oxidatfon is subject to a parabolic
law :
z2 kf' (8-10) .
where x is the weight or the thickness of the oxide film; t is the oxidation time; -
k is the growth rate constant which depends on the type of oxidizing agent, its
vapor pressure and the presence of impurities in the silicon plate and in the grow-
ing oxide fi1m.
1
Process Equipment. The layout of the thermal o~dation device is presented..in
Figure 8-4. The reaction chamber of this device~ ~is made from a quartz tube which
passes through the muffle of the heating furnace. The boats of fused q uartz or -
high-resistance silicon filled with silicon plates prepared for oxidation arE
loaded in the reaction chamber heated to the required temperature. Dry or wet
oxygen is passed through the chamber. In order to prov~de for feeding a strictly
129
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
defined amou.~t of water vapor to.the chamber, evaporat~on takes place under
strictly defined conditions. For this purpose a defined amount of water (batch)
is poured into the sealed vessel batcher. The- oxygen coming into the batcher
goea through the water, capturing the water vapo~. This batcher (bubbling type)
is heated, as a rule, to a temperatuxe of no more than 60�C, which provides for
growth of higher quality films.
The control of the film growt~t process is realized by monitoring the substrate temp-
erature, the oxygen flaw rate, the water temperature and time.
- The advantages of thermal oxidation of silicon are its tec~inological nature, high
quality of films obtained. The deficiencies are high ternperature of the substrates
during the film growth process which can lead to worsening- of the properties of the
structures previously obtained in the semiconducting plate, redistribution of tlie =
impurities and a change in properties of the silicon and also 1ow grawth rates.
8-4. Film Deposition from the Vapor-Gas Phase � q
Principles of the Method~. Many processes of film deposition from the vapor-gas
- phase are relatively new and owe their appearance to the intense development of
applied chemistry. The vapor of the initial compounds (halides, hydrides, carbonyls,
= organoelemental compounds) is delivered to the zone of the device where the sub-
= strates heated to the required temperatures are located. As a result of the
- chemical reaction, the material required for constructing the film is isolated in
the so7.id phase, and the gaseous by-products are removed from the zone where the
substrates are located. The entire process can thus be divided into the following
stages: conversion of the ~.nitia.l compounds to the vapor state, transport of the
vapor of the initial campounds to the hot substrates, chemical reaction, the forma-
- tion of nucleating centers and film growth, removal of the gaseous reaction products
and the nonreacting molecules of initial compounds from the reaction zone.
Depending on the chemical composition and the structure of the initial compound,
the chemical reaction can take place both directly on the substrate and near it. -
In the former case first adsorption of the molecules of the initial compound by
the surface of the substrate takes place, then chemical reaction and, finally, de-
sorption of the by-products. In the second case the vapor of the initial compounds
absorbs heat emitted by the substrate; chemical reaction takes place in the vapor
gas phase. The atoms or clusters of the atoms formed near the surface diffuse to
the substrates and are adsorbed by them.
- The basis for the method is the various reactions of synthesis, substitution,
hydrolysis and pyrolysis (thermal decomposition). ~
The processes of film deposition can-be realized in a vacuum, in an inert or active
medium. The application of electric, light, electron beam and other activations
communicates additional energy to the reacting molecules, which permits accelera-
- tion of the chemical reaction with a decrease in the process temperature. For -
example, for activation ,of the processes of ~pyrolysis of organoelemental compounds,
- high-f~�equ'ency plasma, electron and laser beams, ultraviolet 311umination, the
a,ddition o~ oxygen or ozone to the reaction chamber, and so on are used.
130
FOi2 OFFICIAL IJSE ONLY ~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
Process Equipment. ,Tust as in theiinal oxidation o~ silicon, the process takes
place by the open tube method. ~A flow of gas containing a defined concentration
of vapor of the initial compounds is passed thro~gh the.reactian~chamber of the
device. In order to insure the supply of a strictly defined amount of the vapor
the evaporation of the compounds also takes�-place under strictly established condi-
tions. Along with the bubble type batchers, evaporation type batchers are used in
cases where the initial products are easily evaporated. The film growth process _
is regulated by the percentage content of vapor of the initial compound in the
reaction chamber,~ the substrate temperature, and the film deposition time. The
percentage content of the vapor depends on the flow rate of the gas carrier through
the batcher at defined temperature and the dosage of the initial compound. The
regulation of the degree of saturation of the gas carrier by vapor of the initial
compound is realizecl by variation of the gas flaw velocity. The temperature and
dosage are kept constant.
The advantages of the method of deposition from the vapor gas phase ar~ ~as follaws:
~ the possib ility of obtaining a large number of films of different composition and
with different properties which can be alloyed during the growth process; the _
films can be deposited in practice on any substrates and also on complex relief -
s urfaces; the substrate temperatures are comparatively low during the process of
obtaining the films. The method is compatib le with the operatian of preliminary
cleaning of the substrate surfaces by gas pickling. Obtaining thick films if
necessary presents no difficulties. The application of an inert or active gas
environment greatly simplifies the device, and in case of performing the processes
in a vacuum usually a law degree of vacuum is required. The method of providing
for obtaining films of satisfactory quality at comparatively high growth rates and
activation of the process present no great difficulties.
The deficiencies of the method are cansidered to include the following: the
necessity for obtaining~.especially pure initial compounds, the difficulty of con-
trolling the film growth for grocesses with comp lex chemical reactions, the possi-
bility of adsorption of the reaction by~products by the substrate surface (water, -
carbon, and so on) .
Examples of FiZm Deposition from the Vapor-Gas Phase. Deposition from the vapor-gas
phase is most widely used at.the present time to obtain dielectric films and films _
of refractory metals.
The deposition of masking films of silican dioXide differs from thermal oxidation
j by lower temperatures of the substrates during the growth p rocess and the possib il-
ity of ob taining films not only on silican, but also on any other substrates. By
comparison with the vacuum methods, deposition from the vapor-gas phas~ is distin-
guished by simplicity of the pro cesses and simplicity of th e equipment. In the case -
_ of thermal vacuum evanoration of quartz very high temperatures are required and,
consequently, special electron beam evaporators, and, in addition, the~deposited
films consist of a mixture of si-licon dioxide and monoxide and silicon, and they
do not have good masking and other electrophysical properties.
- Dpposition of the gas phase is at the present time finding b road application in the
- finishing stages of silicon planar technology when the application of thermal
oxidation can lead to variations of the parameters of the already created regions
of the semiconductor structures or in planar technology in semiconductors (germaniimm,
1,31
FOR OFFICIAL USE 01`'
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
- FOR OFFICIAL USE ONLY
gallium araenide, and so on) the . thermal oxides of which 3o not have masking
properties.
The method of pyrolyti~ deposition is most widespread after thernoal oxidation. '
~or pyrolysis, organosilicon compounds are used, from which most frequently tetra-
_ ethoxysilane is used jthe e5ter of etT~yl orthosilicic acid Si(OC2H5) The gas -
carrier (argon, nitrogen~ or active- o~rygen) is saturated by tetraeChoxysilane of
vapor by bubbling and goes into the reaction zo~e of the quartz tube where the
_ holder is located with the semiconductor plates.
The folla~aing decomposition reaction takes place in the substrate zone ~
Si (OC=H,)~'~-'~~ SiO, 1-}-2C,H,-{-6H,0. (8-I1) ($-11) '
Silicon dioxide is deposited on plates; the~ ~remaining reaction products are
carried away by the ga.s flow from the tube.. The fi1~ obtained by pyrolysis are
- somewhat inferior to the thermally grown films with respect to their properties.
The best result is provided by pyrolysis in a vacuum. Sometimes the films are ~
compacted, for example, by annealing or introduction of phosphorus anhydride into -
- them.
The oxidation of a gaseous monosilane compoimd proceeds at lawer temperatures
of 300-400�C:
~oo.-+no�c 8-12)
SiH~ 20, -?SiO, j -{-'~2H,0. ~
Films obtained by the silane method are cleaner, for the organic radical and
carbon are not formed as a result of chemical reactions~ .
The deposition of~silicon nitride SigN4 films from the gas phase is realized as a
result of chemical reactions of the interaction be~ween the hydrides or halides of
silicon and a~onia NH3 or hydrazine N2H4:
- � seo.-g~o~c ~ . T ~ 8-13)
~ 3SiH.-}-4NH, -?Si,N4 j.-{- ;2H, j ;
- s~o--fi6o�c ' _ . (8-14)
3SiH, 2N,H~ -i Si,N_~ j 4NH, 4H,;
, _ . - - -
~ 3SiC1,-}-4NH,~~-~~~Si,N~ 1,--~ 12HC1; (8-15)
~
, - -
3SiBr,-{-4NH,~Si;N, 1 12HBr. (8-16)
Silicon nitride fi]ins are the best st~died after silicon dioxide ~ilms; they have
the best di:electric and masking properties and can be used successfully in planar =
technology.
132
FOR OFE'ICIAL USE dNLY . ~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFiC1AI. USE ONLY
The deposition of aluminum oxide ~i1ms by hydrolysis of halide compounds of
aluminiun takes place at temperatures o~ 800�C and it is accompanied by chemical
reactions : _ � -
2A1C1, 3H,O~A1,0, 1-}- 6HC1,
direct hydrolysis; (8-17)
2A1C1, ~CO, 3H,80�~-~?AloO, ; 6HC1 3C0,
indirect hydrolysis. (8-18)
In order to obtain A1~03 films at temperatures of 500�C most frequently the tri-
methylaluminum oxidation reaction is used:
� (8-19)
4 (CH,),Al 30,~600 �?2A1,0, ! -}-12CH~.
The A1203 films are used as the dielectric layer of thin-film capacitors, for the
active elements of IrIDS-microcircuits with high ~adiation resistance as the gate
insulator; they are also prospective for radiation-resistant bipolar structures~
The deposition of refractory metal films fram the vapor gas phase is cart�ied out
most frequently by decomposition of carbonyls (the carbonyl m~thod) or reduction of
chlorides (the chloride method) .
The decomposition of carbonyl takes place in a flow of hydrogen, nitrogen or argon
_ at atmospheric pressure. The substrate temperature is about 250-320�C. The film
deposi~ian process is accompanied b,~ the reactions :
w ~co~,-.w 1-}-sco ~ ; cs-ZO~
Mo ~co~,~ ~o L ~ sco ~ . } .
Chloride reduction is carried out at substrate temperatures of 720-750�C:
WCIe 3f-Is w 1+ Hci i~
2MoC1~ 5H, 2Mo-{- IOHCI t.~ ( 8-21)
Thin tungsten and molybdenum films are used to create rectifying Schottky barriers,
for mett~llization, for the gates of MDS-str~ctures and for cantact masks.
~ The presented examples do not exhaust all of the possibilities of deposition from
- the vapor gas phase. Further development of applied chemistry, and in particulgr,
the chemistry of ~organometal compounds (OMC) perfor.med in the Soviet Union by
Academician G. A. Razuvayem and his students, will permit films to be~ obtained for
thin-film elements more simply than by the traditional vacuum methods. At the
present time the OMC are already being used to obtain chromium, nichrome, copper,
nickel and other films.
133
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
FOYt OFFiCIAL USE ONLY '
8-5. Anodic Electrolytic Oxidation !
Electrolytic oxidation is used to obtain films of semiconductor and metal oxides.
Oxidation is realized on the anode,o~ the el.ectrolytic ba~h. _
The mechanism of electrolytic anodic ogidatioE? differs from the mechanism of ,
thermal oxidation. For electrolytic oxidation-the reaetion of the interaction of
the materi~l with the ions of the oxidizing agent proceeds at the filarelectrolyte
interface.
Thus, for example, in tantalum the growth of the oxide film is accompanied by the ,
reaction _
2Ta+s+ 100H--->Ta206-}~-5Hz0.
wth can be insured by two pro-
- cedures: local stimulation or local complication.
Local stimulation of chemical reactions and also the formation and growth of i -
nucleating centers can be carried out, for eRamp:le, using an electron or laser beam.
The application of this method of localization is limited '~y the comp lexity of the
formation and control of the lbeam displacement (especiall~~ electron) in vapor-gas
media. In addition, nontmiformity of the current density with respect to the cross
section of the electron beam leads to nonuniformity of a local 1y grawing epitaxial
l.ayer.
Local complication of the epitaxial growth i~ carried out using contact masks. In
this case it is necessary to create conditions under which the growth of the layer
takes place only on the open sections of the substrate (selective epitaxtal growth)
or use contact masks which are then removed together with the deposition of a
layer of semiconductor on their surface.
At the present time the selective local epitaxial growth using Si02-cor.tact masks
is most widely used. This is explained by the compatibility of the Si02-masks with
epitaxial-planar technology and energy difference of the formation of the silicon
nu~leating centers on the Si-plates and on the surface of the Si02 promoting growth
of the epitaxial layer in the first stage on the silicon. The selective grow,th in
the chloride method is insured by quite high con centrations of silicon tetrachloride -
in the gas phase and the selection of the temperature range in which the formation
of polycrystalline sil3con on the iuask does not take place. In addition, for compli-
cation of the deposition of the polycYystalline silicon on the surface of the Si02 _
masks, the chloride process is carried out in the presence of hydrogen chloride
vapor. After bringing the temperature of the substrate to the operating temperatures,
_ hydro~en chloride is admitted to the chamber of the device; on comgletion of the
growth process, first the SiClq feed is halted, ther. the HC1: Local growth is
very sensitive to the composition and ratio of the ii~, HC1 and SiC14 in the vapor
gas phase and also the location of the substrate in the reaction chamber of the
' devices. It is also necessary to consider reduced chemical strength of silicon
dioxide in the presence of hydrogen at increased temperatures. At the present time
_ high temperatures are needed to obtair a smooth mirror surface of the local silicon
sections.
= 9-6. A1 loying of Epitaxial Layers
The alloying of layers in the case of epitaxy from the gas phase is realized using
speci al compounds sources of alloying impurity. The source vapors are trans-
ported by the gas carrier to the temperature zone of the device where chemical
153
FOR OFFIC[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY -
reaction takes place with release.of elementary alloying impurity which is
deposited on the plates or substrates together~with the material of the growing
epitaxial layer. The ~ollowing forms o# sources of alloying impurity are distin-
guished. ~
Liquid sources (the halides PC13, POC13, BBr3 and ot~ers) are poured into individual
or cornmon batcher with the sourc~ inaterial~of the layer, The halides of the alloy-
ing eleme.nts dissolve well in silicon tetrachloride; they are easily evaporated
and saturate the passing flaw of hydrogen. At the hot surfaces the vapoxs ~f the
admixture sources are reduced by hydrogen:
2PC1g+3H2-?2P +6HCi. ~q_ g~
Halides have high vapor pressure which depends sharply on temperature; therefore,
, insignificant temperature fluctuations can cause signifi cant variations in concen-
tration of the adnYixture in the growi~g layer. Recently when alloying with boron, -
- high-boiling compounds have-been used, the vapor pressure of which depends only
slightly on the temperature. By using these compounds it is possib le to obtain
silicon layers with a broader range of specific resistances from 0.005 to 5 ohm-cm.
The gaseous sources hydrides, diborane BZH6, arsine AsH3, phosphine PH3 are
distinouished by high toxicity, and therefore they are delivered to the production
facility in tanks in a mixture with hydrogen or inert gases in con centrations of
(5�10-4 to 1)%. The degree of alloying when using liquid and gaseous sources is
regulated by the concentration of the existing compound with alloying gas f low and
speed of the basic gas carrier f low.
When using dibor.ane it is difficult to obtain a mixture of it with a content of
less than 10%; the mixtures are unstable during storage; it is difficult to estab-
lish and exactly measure the small flaw rates of the mixtures; therefore it is
difficult to obtain silicon layers slightly al]:oyed with boron with s~pecific
resistance greater than 1 ohm-cm. Accordingly, high-boiling, liquid boron- contain- -
ing compounds or boron tribromide is used.
Solid sources are used in the relatively new method of gas discharge spark alloying.
The gas discharge chamber using two connecting pipes is connected to the gas dis-
trib ution system of the epitaxial growth device. Pure hydrogen or hydrogen
saturated with silicon tetrachloride vapor is passed thrnugh the chamber. The
electrodes are made of material containing the alloying element, LaP6, B4C, A1B2
or the alloy Sb+(1-?.)%As. When feeding a pulsed voltage between the electrodes,
spark disch arge is excited in the chamber. In the spark discharge chamber the -
_ electrode material is partially evaporated. The vapor is carried away by the
passing hydrogen to the reaction zone of~the device. There it decomposes with
the release of a free alloying admixture. The concentration of the admixture can
' be adjusted by the distance between the ele~trodes, the voltage fed to the
electrodes, the frequency and duration of the pulses and also'the flow rate of the
hydrogen through the gas discharge chamber.
Alloying during vacuum epitaxy is realized by an admixture which enters int the
composition of the semiconductor the source of particles of the growing ~ayar _
of material. The admixture atoms are delivered to the plates or substrates as a
.
154
FOR OFFI~IAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
result of evaporation of the source-or sputtering o~ the target. The admixture
- must be volatile (arsenic, antimony, phosphorus). The concentration of the alloy-
ing admixture in the growing layer can be varied within the broad range, varying
the rate of formation of the vapor or tha sputtering rate. Using somewhat differ-
ently alloyed sources of admixture,'it is possible to grow multilayered epitaxial
structures.
The alloying in liquid phase epitaxy is real~:zed by an elemental admixture. The ad-
mixture is introduced into the melt-solution, from.which-~it is encompassed by the
crystallized material of the growing~layer. For exa~nple, for alloying layers of
gallium arsenide, silicon, germanium, tin, sulfur, zinc and tellurium are used.
Silicon and germanium are amphothermal admixtures. Depending on the alloying con-
ditions, the temperature of the beginning of crystallization, and the concentration
of the alloying admixture in the melt, the silicon and germanium can replace either
the gallium nodes and play the role of donors or the arsenic node and play the role
ot acceptors.
The possibilities of alloying during epitaxial growth are much broader than for
other methods of obtaining alloyed layers of sem~conductor, for example, during
diffusion. A unique characteristic of epitaxy is the possib ility of obtaining
high-resistance layers of semiconductor on law-resistance plate~. Durino epitaxy
it is possible to obtain various distributions of the alloying admixtures, includ-
- ing uniform distribution or with a sharp cancentration gradient at very short
- distance. It is possible to obtain multilayer structures in one growth cycle.
- 9-7. Defects in the Epitaxial Layers
In the epitaxial layers most frequently dislocations, regions of inechanical
stresses, p acking defects and growth defects occur. The defects limit the applica-
tion of epitaxial structures for powerful and high-voltage semiconductor devices
in which large areas of p-n junctions and thick epitaxial layers are used and
specially for LSI and SLSI where rigid requirements are imposed on the defect
density.
The dislocations in the epitaxial layers can occur at the substrate (plate)-layer
boundary in the presence of inechanical disturbances or contamination of the surface
- as a result of crystallographic noncorrespondence of the materials, as a result of
_ ciifferences in t:he degree of alloying of the layer and the plate, as a result of
paint defects, and so on. The dislocations availab le on the surface are traced
- by the growing layer, The presence of various contaminants and noncorrespondence
of the material~ leads to the appearance of inechanical stresses. The plastic
cleformation of the material at high grawth temperatures can remove the mechanical
- stresses with the formation of additional dislocations.
Ir. order to decrease the density of the dislocafiions in the layers the following
are req uired: selection of the substrate or plates with minimum density of the
dislocations, careful preparation of the surface, sele ction of comp atib le materials
of the layer and plate or substrate. In order to decrease the dislocations
caused by various degrees of alloying, it is not recommended that plates with a
specific resistance of less than 0.01 ohm-em be used during the grawth of high-
resista.-~ce layers, for at the boundary wi~h the~~layer too large mechanical stresses
can arise, and, in addition, it is difficult to manufactura low-resistance, sezni-
conductor ingots without dislocations.
155
- FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY ,
The sections of inechanical stresses around the admixtures are formed during
alloying of epitaxial layers by ac~mixtures with.atom size'differing from the
dimensions of the semiconductor atoms. ~The stress depends on the ratio of the
dimensions of the atoms and the concentrations of the i~urities. The mechaniCal
stresses lead to the appearance of dislocations and_to bending of the structures.
In order to decrease the bending when growing epitaxial layers the method of mixed _
alloying by admixtures causing de~ormation of different sign is used. It is possi-
b le to realize compensation of the deformations of silicon structures by simultaneous
alloying by tin and pl-~osphorus, tin and boron, antimor~y and phosphorus. The vapor
of the impurities is selected so th at the atomic radius of one will be less th an
:
_ the atomic radius of silicon, and the~~other, greater. For ex~mple, the atoinic
radius of tin is 1.40 t~, phosphorus 1.10 A, and silicon 1.17 A. The application of
the method of shifted alloying permits growth of thick epitaxial layers with a
minimum number of noncorrespondence defects.
The packing defects are regions with disturbance of the alternation of the atomic
layers, on the side boundaries of which there is mismatch with the remaining part
of the crystal lattice of the layer. On fitting of the layer during the growth
process, for example, to an oxide islet, the atoms can begin growth of the mis-
matched layer ~f ato~, and as a result ordir.ary alternation of the atom layers can
be disturbed. For example, in a perfect silicon crystal in the (111) direction the
atoms are arranged in layers in a defined sequence ABCABCABC... The layers
designated by different letters (A, B, C) are distinguished by mutual arrangement
of the centers of the atoms. If during the growth process one of the layers.is
skipped, a subtraction packing defect arises, and if any layer is repeated twice,
an interstitial packing defect arises. At the location of the formation the packing
defect is of a point n ature; as growth takes p lace the defect b egins to occupy the
entire region. Thus, points defects growing on the (111) plane are transformed
as the layer grows to regions included inside a right tetrahedron, with crystal
lattice not coinciding with the remaining part of the epitaxial layer. Basically
the packing defects are initiated at. the point-layer interface, and as growth takes ~
place, the~ penetrate the entire epitaxial layer, emerging as the base of a tetra-
hedron, that is, equilateral triangle, on the surface (Figure 9-6). The height of
the tetrahedron equal to the thickness of the epitaxial layer h=a~3=0.816a,;where
- a is the side of the triangle. If the packing defect is initiated not on the plate- _
. layer interface, the height o~ the tetrahedron will be less than the layer thick-
ness. If the packing defects are initiated close to each other, then as growth
takes place, superposition of defects can occur. If three defect layers are
initiated in a row (mi crotwin), then twinrting regions appear in the growth pro cess.
The growth defects appear as follows: at~locations on the surf~ce of th e plate or
growing layer of larger foreign particles, for example, abrasive particles; in
regions of accelerated grawth or encounter of rapidly growing nucleating centers;
in regions with high concentration of impurities when it exceeds the solubility
limit; in the case of high supersaturation wh:en the material is crystallized,by
conglomerates, that is, are clusters of atoms or molecules. During the growth
p rocess these defects are converted to holes, hills, pyramids and polycrystalline
sections. A hole is a formation accurring as a result of failure of the epitaxial _
layer to grow over a section of the plate surface. At these locations the formation
of a polycrystalline deposit is the most prob ab le. A hill is a completely or
partially disoriented monocrystalline formation in bhe form of a protrusion. The
156
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY -
growth pyramid occurs as a result of oriented growth on dislocations or other
defects. Hills and pyramids protruding on both the surface of the layer are
especially dangerous, for they have a.negative ef~ect on the lattice parameters
and lead to re~ects in subsequent technologioal operations.
1~~ .1~ 1jf ~.1' . J
' 7 ,.:.1'I . ` ~,F� t . -
~~1~,,~J r~�^ '�1{ ~ .
1 ~ ~ I ~ ~
: ~
. .v.+,~,,, r~~`~,~', ,
R , �~1
~ ~a~.~ t; .
. ~ +~w~ ~
. M } R .
~ , ~.R ' }�i�
~ '
. ~~~i .
~ y'~ '
� ~ ~ ~ ' * � ��;�le.,�' . . .
. � , . :,t ' f y y,ls~ l
' / . � t y'�~t'~k~?, S ~ -
~ ' ~ - , 1~~ 4
. ' ~ 2
I ,,t ~ � . ~ ~i
L __K~- - - _ ~ ~ . ~~:;I b~) .
Figure 9-6. Packing defect of a silicon epitaxial layer.
a-- microphotography of the defect on the (111) surface of silicon;
b-- three-dimensional diagram of the defect; 1-- epitaxial layer;
2-- (111); Si-plates; 3-- initiation p~int of the defect on the
plate surface; 4-- d~fect region; 5-- emergence of the defect at
the surface of the epitaxial layer
Thus, all of the defects can be c~ivided into those that take off from the sub-
strate crystal and those occurring in the epitaxial layers in the growth process.
The largest number of defects is connected with the quality of preparation of the
~ surface of the substrates or plates, with alloying, with purity of the performance
of the process and with crystallization ~onditions.
- 9-8. Epitaxial Layer Control
During the research and development of the technological processes of epitaxy, a
large number of parameters are controlled, and a large number of various methods
are used. During prodt~ction most frequently the thickness, specific resistance
of the epitaxial layer, the concentration distribution of the admixture with
respect to layer thickness, and defect density are controlled. These layer
- paramete-rs determine the breakdown voltages a~d back currents of the p-n ~unctions, -
the saturation resistan ces of th e transistors, the loss resistances and the volt-
farad cha.racteristics of the structures.
157
= FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOIt OFFiCIAL USE ONLY
. _ _
l
�
_ t
_ a~
. ,
~ l
! . .
Figure 9-7. Oblique (a) and ball (b) microsections for determining
the thickness of an epitaxial laqer
The thickness of the layer very frequently is determined by the methods of coloring
ob lique or ball microsections. For this purpose the control sample is cut at a
small angle of 1-5� or a small hole is cut out on its surface using a ball with a
- diameter D=25 to 150 mm. The coloring of the microsections in special solutions
takes place as a result of the difference in electrode potentials of the p and n
regions which gives rise to selective composition of copper from the solution on
th e p-region or selective oxidation of the~n-region. In order to disccver the _
interface of the concentration homojunctions n-n+, p-p+, selective pickling is used.
It is possible by observation under a microscope to use the colored microsections
to measure not the true thickness of the epitaxial layer h, bu~ an esaentially
large value of Q by the eyepiece scale (Figure 9-7) . The thickness of the epitaxial
layer is calculated by the formulas : h=J~� tg a for 'the oblique microsection,
h=22/4ll, for the b all microsection. The precision of the measurements is on the
average +5-10%. The method of coloring the microsections is applicable for�
determining a broad range of thicknesses of the layers differing from the substrate
b y type of conductivity and degree of alloying.
The cont actless, nondestructive method of infrared interferometry used for layers, ~ -
- the optical constants of which differ sharply from the optical constants of the
s ubstrates insures measurement precision of +5%.
The infrared beams are partially reflected from the air-layer interface, they pass
through the epitaxial layer and are ref lected from the layer-substrate interf ace.
As a result of the application of two reflected be,ams, the interferenee pattern -
of alternation of dark (minima) and light (maxima) strips is observed. The thick-
- ness of the epitaxial layer is defined by the formula -
oN _ _ (9-9 )
h - 2n~k ,
eaks or minima in the measured interval; n is the /
w~here ~N is the number of p
coefficient of refraction. of the layer; Ok=1/ai-1/a2 is the dif.ference in values
. inverse to the wave lengths of the observed maxima or minima.
158
FOR OFFICIAL USE ONLY
I
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR UFFICIAL USE ONLY
Yr~
_ 'rhe method is distinguished by rapidity of ineasurement, but is applicable only
for epitaxial layers from 2 to several tens of microns thick grown on highly
alloyed substrates.
The specific resistance is measured by the sonde and volt-farad methods. For
epitaxial layers, the ordinary four-sonde metfiod is used on high-resistance sub-
strates of opposite type conductivity. The specific resistance is calculated by
the formcla
P = 4,~3 ~ !t, ~9-10)
= where U is the voltage drop between adjacent sondes, volts; I is the current, pass-
ing through the epit~ial layer between the edge sondes, amps; h is the thickn'ess of
the layer, cm.
- The snecific resistance of the epitaxial layers on the high resistance substrates I
(pe/p5i.50) of the same type of conductivlty is measured by the free-sonde me~thod
(Figure 9-8). The specific resistance is determined from the equation
U~~~~ 68,39po,6360,.~..21P0,e~7s~ ' (9-11)
Key: 1. breakdown
where Ubreakdown is the breakdown voltage between the sondes 2 and 3.
The breakdown voltage is measured using an oscillograph which has a large input
imp edan ce .
The volt-farad method of ineasuring the specific resistance is based on using the
re].ation that relates the capacitance of the p-n junction to the voltage at, the
junction and� the concentration of the admixture. '
The concentration distribution of the admixture with respect to thickness of the -
~ epitaxial layer is determined by the method of differential electrical conductivity,
that is, measurement of the electrical canductivity after successive removal of
an insignificant part of the thickness of the epitaxial layer. On the surf 3ce of
the epitaxial layer, the electrical canductivity al is measured. Then part of the
epitaxial layer with a thickness ~ is removed, and the surface electrical co;~ductiv-
ity a~ is again measured. The mean concentraticn of the admixture in the rem:~ved
part of the layer is calculated:
N- -(Q'_~z)~ (9-12)
qr~o
where q is the electron charge; is the iuobility of the current carriers.
Then part of the epitaxial Iayer is again removed, and Q3 is measured, and so on.
The thick epitaxial layers can be gradually ground off. The removal of thin
epitaxial layers is by cathode pickling or anode oxtdation in an electrolyte with
159
FOR OFFICIAL USE ONLY
i
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLV -
subsequent acid pickling o~ silicon dioxide. The prec~sion of these methods of
_ removing the silicon is 20-50 and ~~50-300~ angatroms, respectively.
. ~
_ O _
- yn x
_ ~ 1) + RN
~ _ - 1 1 3 :
n
i
Si, , n+ .
Figure 9-8. Diagram of ineasuring the specific resistance of an -
~ epitaxial layer by the three-sonde method
~ Key:
_ l. IP
The structural defects of the epitaxial layers ars 3nvestigated using the trans-
parent x-ray topography (the Long method) , electron microscopy and visual observa-
tionunder a microscope of tne surface of the layer after discovery of the defects
by selective pickling. The l~st method is the simplest and takes the least time.
After pickling at the points of emergency of the packing defects at the surface,
_ depending on the crystallographic orientation of the substrate, various pickling
configurations are formed: triangles, tetrahedrons, trapezoids, Identical or
V-type lines and also combinations of pickling figures if the defects are superposed
on each other. At the points where the dislocations emerge on the surface, ;disloca- _
tion pickling holes appear. The number of holes or pickling figures and also their
~ size can be determined usin~ t:~e rill~'~7 (200 paw~r) and MMU-1 (190 pc~wer) metallo-
graphic microscopes. The compositions of th2 pickling agents for discovery of
defects are selected as a fim ction of the material of the epitaxial layer, the
crystallographic orientation of the substrates, and ~o on. The defect density is
determined by the number of defects per unit area of layer and, in accordance with
the techni cal s~.ecifications, can vary within the limits f rom 0 to 5�104 cm 2.
Test Questior.s and Assignments
1. What is epitaxy and what is its purpose in IC production? Explain autoepitaxy,
" heteroepit axy and chemoepitaxy.
2. Which epitaxial structures are used in IC production?
3. Ho~a are the types of epitaxial_structures deciphered? Check yourself on the ,
' markin g examp les presented in �9-1.
4. What is the mechanism of epitaxial growth? What factors influence the g'rowth
rate?
. 160
FOR OFFI~IAL USE ONLY ~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFIC[AL USE ONLY
5. Why is epitaxial growth carried out on sli~htly disoriented silicon'plates?
6. What is the essence of the chloride method of growing silican layers? ,
7. Using Figure 9-2 explain the order of the~ epitaxy process.
8. At what tempe ratures are monocrystalline layers grown? Poly crystalline layers?
Why? At what temperatures do the layers not grow? What are the temperatures
called at which the layers are not formed?
_ 9. What chemical reaction provides the basis for the silane method of growing -
silicon layers? Compare it with reaction 9-1.
10. What distin guishes the techniques of epitaxial growth of silicon by the silane
- method from grawth by the chloride method? -
11. Compare the ~hl~ride and silane methods of growing epitaxial silicon?
_ 12. What are the characteristic features, advantages and disadvantages of thermo-
vacuum depositian of silicon? -
13, What is the basic application of liquid phase epitaxy (ZhFF) ?
- 14. What is the essence of the process of growing epitaxial layers from solutions -
in molten metals?
15. What is the technique for growing gallium arsenide layers from the liquid ~
phase using Si, Ge, Sn-alloying admixtures? When using Zn, S, Te-alloying ~~+~aixtures?
Explain the structure of the holder and the temperature-time chart of liquid-phase
epitaxy by Figure 9-5. `
16, What are the advantages and disadvantages of liquid phase epitaxy? -
17. What is the role of heteroepitaxy in the production of microelectronic devices?
In what two b asic directions is modern heteroepitaxy developing?
18. Give examples of heteroepitaxial semiconductor junctions and explain the
- essence and technique for obtaining them.
19. What are the peculiarities of silicon on sapphire epitaxy? What method is
most frequently used to gro~r silicon on sapphire?
20. E xplain why the p-Si-layers are obtained when growing on sapphire without
; alloying. ,
21. What are the advantages of epi~axy when growing silicon on spinel?
22. What is Iocal stimulation and local complication of the processes of epitaxi al
- growth? What method of localization is used in practice? Why?
?.3. How is alloying of the epita~cial layers realized?
161
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
,
FOR OFFICIAL USE ONLY
24. How is the degree o~ alloying of epitaxial layers controlled?
25. What distributions of the admixture in the epita~dal layers can be obtained
in practice?
- 26. What is the unique feature of epita~ry by camparison with diffusinn and ion
alloying:
27. List the possible structural defects of epitaxial layers.
28. What are the causes of defectiveness of epitaxial layers and how can they be
de cre as ed?
- 29. What are the parameters of the epitaxial layers and why are they controlled .
during the production process?
30. How is the thickness of the epitaxial layers determined?
3I. How is the specific resistance of the epitaxial layers determined?
32. How are the concentration distributions of the alloying impurities in epita~d al
layers determined?
33. What defects of the epitaxial layers ar~ controlled visually?
~
J
0 .
162
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY _
' CHAPTER 10, HIGH-TEMPERATURE DIFF[1SION
10-1. Fundamentals of the Method of High--T@mperature Diffusion
Application of the Method. Diffusion as an independent technological process is -
the most widely used method at the present time for introducing alloying impurities
- into semiconductor plates or into the epitaxial layers grown on them in order to ~
obtain regions of opposite conductivity by co~arison w~th the initial semiconductor
or with lawer electrical resistance. In the first case, for example, emitters,
bases and insulating regions of the transistor are obtained; in the second case,
n+, p+-latent regions decreasing the resistan'ce of the collector body or tfie contacC
- regions decreasing the in3ection of the minority carriers from the ohmic contacts
and improving their quality.
- When manufacturing high-speed structures with good pulse properties diffusion is
used to introduce the admixtures forming deep levela in the forbidden zone of the
semiconductor and decreasing the lifetime.of the minority carriers of the current.
- For silicon gold and nickel arg~such admixtures.
Motive Force of Diffusion. Various nonceasing mechanical movements of the atoms
are hidden behind the apparent external calm of solid states: vibrations,~;rotations,
random translational movements~ The translational movements are�~.random as long as -
the solid atate is an equilibrium corresponding to minimum free energy. On dis- -
- turbance of th~ equilibrium, that is, on the appearance of any nonuniformities the
translational movements of the atoms become directional.
- Diffusion is the phenomenon of directional displacement of the particles of matter _
in the direction of their decreasing cancentration. Diffusion is aimed at having
the solid state avoid nonuniformity, gi-ving it the possibility of releasing excess
free energy and conversion to the equilibr~ im state. The d~tffusion takes Ylace
while various sections of the solid state tiave different concentration of particles.
The motive force of diffusion is the cancen~ration gradient of - the atoms or mole-
cules of the rnaterial. The sreater the concentration gradient, the more intense
the diffusion. ~
When manufacturing diffusion structures, increased~ concentrations of the alloying
admixture are created on ~the surface o~ the .semiconductor plates. The admixture
- b~gins to diffuse deeply.
163
FOR OFFICIAL USE ONLY .
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Diffuaion Mechanism. In real se~dconductore diffueion can be realized by three
methods (Figure 10-1) . In the vollaaetric method a eimple exchange of places by
_ two atoms or circular exchange with the partfcipation of several~.atoms takes place.
In the vacancy mechanism the di~~usion is realized by�successive ~umps of the
admixture atams ot the substitution ~from the~ nodes~ tfiemselves to~ the vacancies
(free nodes). The~diffus3on with intemodal mechaM~sm is realized as a result of
successive transitions of the interstitial admi~cture ftom one interstitial node to
another. The atioms of the a~mixture somehow~ are "forced" between the atoms at the
nodes of the crystal lattice. The coc~cepts of vacancy and internodal mechanisms of
diffusion were introduced into physics by the Soviet scientist Ya. I. Frenkel'.
_ _ _ _ - -
- - - .
- - - - - � � r-
_ ~ :I
.
~ ~ ~ 0 ~f~ ' ~ ~ �
_ s i
� � ~~..,,o. e: � �
~ ~ ~ ~ ~ � s � e � .
� � � . � �I ~ � � � � , i
_ a) � b ~ - .
_ , ~
Figure 1Q-1. Diffusion mechanism in semiconductors.
a-- circular; b-- vacaney; c-- internodal
The diffusion of the basic alloying admixturea in germanium and silicon can be
realized by all three mechanisms, but the most probable for them is the vacancy
methQd determined by the presence in fihe crystal of Schottky and Frenkel' defects,
that is, the presence of vacancies or paired vacancy-atom defects at the internode.
The formation of such defects is connected with the thermal vibrations of th~
lattice. With an increase in temperature, the number of vacancies increases~ in
accordance with the expression
~ rt~~'=n,e-B~ckr~,
Key: 1. vacancy
- where nyacanc is the number or vacancies, cm 3; e is the base of the natLral
logarithm; E~s energy required for the formation of the vacancy; k is the Boltz;~~.n
constant; T is the absolute temperature; n is the number of atoms of semiconductor
at the nodes of the cYystal Iattice.
For the formation of one vacancy, an energy equal to several electron volts is re-
quired. At ordinary room temperatures the~number of vacancies per unit volume of
semiconductor is small; for 1015 to l0ia semiconductor atoms there is one vacancy.
With an increase in temperature to 1000-1200�C ~the nwnber of vacancies becomes
comparable to the number o~ semiconductor atoms. Under the effect of the thermal
vib rations the admixture atoms can occupy the location of the next vacancy and thus
mov~. The probability of transition of the admixture from the node occupied to it
_ 164
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFIC[AL USE ONLY
to the vacaney node depends on the probability~ of the presence of adjacent
vacancies and the probability of overcoming the~potential barrier when the atom
moves to the vacancy location:
P ~ e es/ckrl
. (10-1)
Here the value of aE is the so--cal~ed'~ energy of activation of the diffusion process;
it is made up of the energy required for formation uf the vacancies and the energy
of transition of the admizture atcnm from its position to the vacancy node. For
vacancy diffusion in silicon aE=3.~ to 4.3 electron volts.
Alang with the process of diffusion of the alloying impurities, there is also
diffusion of the atoms of the semieanductor itself self-diffusion. Iiowever, on
the basis of the large chemical bond of the atams themselves the self-diffusion
process is insignificant by comparisa.i with the diffusion of the alloying admixtures.
Many admixture atoms of the first, second, sixth, seventh and eighth groups of the
periodic table in silicon occupy locations at the internodes, that is, they form
solid interstitial solutions. The diffusion of these admixtures is realized by
the internodal mechanism. The probability of internodal transitians of the atoms
is appreciably higher than the probability of transitions from node to adj acent node.
Therefore the diffusion of the interstitial admixtures takes place appreciably
faster than the diffusion of the substitution admixtures.
Solubility of Admixtures. With an increase in the temperature, the solubility of
- the admixtures in the solid semiconductor increases, and after reaching the
maximum solubility, it begins to decrease. The ma~.mum solubility of the impurity
in the sPmiconductor is the maximiun possible amount of defined admixture per unit
volume of the given semicanductor at a given temperature. The maximum solubilities
of the alloying admixtures in the silicon are preaented in Table 10-1.
Table 10-1. Maximum Solubility of Admixtures in Silicon
. _ . . Maximinn solubility
Admixture ~ Maximum'solutiility;~cm'3'~~~~~ ~ " 'femperature; �C
Aluminum 1019-1020 1150
Boron 5�1020 1200
Phosphorus 1.3�1021 1150
Galltum 4�1019 1250
Indium 1019 1300
Antimony 6�1019 I300
Arsenic 2�1021 1150
Go ld 1Q1~ 1300
Using the di~fusion method it is possib le to introduce an admixture into the semi-
- conductor to concentrarions no greater than the ma~dmum solubility ~or the given
temperature or the maximum solubility for the temperature corresponding to it.
First Diffusion Law. The processes of diffusion transfer of the material in the
semiconductors are described by the two ~ick equations (laws). The thickness of
165
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USF ONLY
the dif fusion layer usually is apprec~ably . less tha~~ the glate area. If the ,
dif fusion proceeds with respect tc~ . the~ entire surface o~ the~ plate . it is possib le
to consider that the diffusion is unif4rm, for the admixture diffuses primarily
_ in the direction normal to the sur~ace.
The first equati~n of ane-dimensional di~fusion.determines the flaw of impurity '
atoms fr~m the region with fncreased~concentration to the region with decreased
concentration. ~
,.r .
_ ~F_-D~� � (10~2)
Here D is the diffusion coefficient numerically equal to the number of adm~.xture
atoms per unit time passing through-a unit area normal to the diffusion direction
with a concentration gradient of the admixture equal to on~. The minus sign in
. the righthand side of the equation indica~es that the diffusion ahift of the atoms
~ takes place from the region with higher concentration to the region with lower
concentration.
Second Diffusion Law. The second equation of diffusion is derived from the first
under the assumption th at the diffusion coefficient does no t depend on the concen-
_ tration:
dN _ d'N (10-3)
~ - D x' '
The second Fick law is the basic diffusion law. It defines the concentration of
the admixture introduced into the semiconductor at any point in time at any
distance from the surface for the given diffus~an temperature. The temperature
enters into the secoi~d equation not explicitly, but through the dif.fusion coeffi-
cient : _
D = D~e-eE/(k7~ (10-4)
Here D~ is a constant numerically equal to the diffusion co efficient at infinitely
high temperatures; DE is the energy of activation of the dif fusion process of
the given admixture, that is, the ener~y required for the admixture atom to ~ump
to the vacant node of the lattice. At ordinary room temperatures the diffusion
in the solid states is not observed in practice., Bif.fusion processes in semi-
conductors take place at high temperatures of 800-900�C for germanium and 1000-1350�C
for. silicon.
Distribution of the Alloying Admixture with Respect to Depth of the Diffusion Layer.
The solution of the basic equation for specific diffusion canditions determines
the concentration of the admixtures at different depths f or different duration of
the process, and the function N~f(x) is thus found for the given di~fusion tempera-
ture.
' The distribution of the impurity for uni~orm distribution from an infinite and
constant source w~ich insures constant replenishment of the admixture going into
the semi conductor, is described by the equation of the comp lex function of the
one's complement of the error integral:
_ 166
- FOR OFFICIAL USE ONL4l
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
~
FOR OFFICIAL USE ONLY
- -
N~x, t) ~ N, erfc y"o~.,: ;
(10-5)
where N~ is the admixture concentration-on tIie~surface o~ the plate; x is the
depth of dif~usion; D is th.e di�~usion coefficient; t is the time for conducting
the process; erfc is the provisionai notation for t~ie fvnction of the one's
complement of the error integial. '
The concentration distribution graphs of t~ie a~ai~cture are presented in
Figure 10-2. The amount of admixture going to~~the surface�is equal to the amount
of admixture leavtng the s~rface for the body~of the plane. In the case of
- practical calculations tIie admixture distribution is described by the equation
- (10-5), the va~ue of x/2~ is-defined for'defined temperatures and diffusion
time, and then, using the table, erfc (x/2~) is found.
- - -
; - -
-
~
:-.~f-- ' ~
� , ~ ' '
~ , . - ~ NOf
N(x) ' . , . . ~ . ~ . ~ t~ ;
No , - ~ ~ . ~ ; .
. , . Nae t,'t2 > t~ ' ~ ~
E~~t2~tt� . . ' , '
� . . , ' tz
� t t2 t3 . Mp~ f3 '
~ / , ,X '1
' ' L X
.0
Figure 10-2. Admixture distribution Figure 10-3. Admixture distribution
during diffusion from an infinite during diffusion from a limited
constant source source
The admixture distribution for uniform diffusion from a limited source, for
- example, created�in a thin surface layer of the plate and protected by the masking
film from diffusion of the adnnixture into the surrounding space is described by
the normal (gaussian) equation
-N ~Z' t~
r . Q
. 8_~l4 ~1~-6~
- � aDt
where Q is the total amount o~ the admixture in the semiconductor per cm2 of
surface.
The distribution graphs o~ the admi~tures for this case are.pnesented in
Figure 10-3. As the admixture goes~into the Body of the plate the concentration
of the admixture in tfie source is depleted, and influ-~c from the outside is
absent.
A characteristic feature of the presented distributions of the admixture
corresponding to the complementary erior.function and the gaussian function is
~ 167
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOit OFFICIAL USE ONLY
a monotonic decrease in the concentration .o~ the ~admiacture from the surface and ~
to the semiconductor. ~The inaacimwn concentration of the admixtures corresponds to
~ the plate surface. ~The surface concentration with diffusion from an unlimited
ebu~~~ce is constant~�independently .of the diffusion time~; and with diffusion from
the limited souxce, it decreases as the time tncreases.
The practical distributions of~~the~a~mixtures, as a rule, do not c:orrespond to
the simplest theoretical cases. ~Verq f~equently the boundary conditions are dis-
tinguished or they are not~maintained~during~the~diffvs~on process and, in
addition, the influence of variaus factors not��coneidered by the calculation ie
felt on the diffusion proeesses. First o~ a11, this is inco~nstancy of the
diffusion coefficient which depends on the concentration of the introduced impurity,
the presence in the plate of initial admi~ctures-and structural defects. The
dependence of diffusion coefEicient on the concentration in practice ig known
very approximately, and the solution of the basic equation even in the case of
uniform diffusion can be obtained in axceptional cases.
More complex cases of diffusion inelude multiple successive diffusion of a number
of admixtures at various temperaturea, for example, when creating a transistorized
structure; the diffusion accompanied by evaporation; diffusion into the plate in
the presence of a surface oxide; diffusion in the presence of a counterflow of
impurities diffusing from ~~he plate volu~e to-the outside, and so on. :Finding the -
solutions in these and other intermediate cases is a complicated problem.
From what has been indicated it is clear t~at the actual distributions of the
admixtures differ from the calculated ones. At the same time when estimating the
diffusion process ~n practice frequently the simplest solutions of the Fick
equations are used, and t~e calculation results are checked and more precisely
- determined experimentally.
10-2. Characteristic Features of Diffusion in Planar Technology
Basic Equation for Local Diffusion. In the~planar structure the dimensions of the
openings in the masking film detezmining the dimensione of the alloyed regions are
sma11 and are compa~rable with the depth of_diffusion; therefore in order to find
the concentration distribution of the admixtures in the diffuaion layer it is
impossible to use the solution of the one-dimensional diffusianequation (10-3).
The basic equations for local diffusion of the admixtures in the general case is
three-dimensional: ~ J
dN d'N d'N~ d'N
a~ = D ~az, -I-ay~ -F- ax~
j � (~o-~)
The solutions of the three-d3mensianal Fick equation are complex and highly awkward.
In practice these complex and awlcward solutions can only be used on applying
a computer. ,
Planar p-n Junction ~'ront. In the case of local diffusion, the admixture pene-
trates into the seffi:conductor not ~iy a planar front, but a front distorted at the
edge of the opening. The distortion of~the ~unction front at the edge of the
opening in tfie ~masking film has great practical significance �or the parameters
of the planar stzuctures. First, as a~result of the effect of the edge distortion
the density of the electric current with respect to the area of the p-n ~unction
168
FOR OFF'ICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
_ FOR OFFIC[AL USE ONLY
is nonuniform, and breakdown is possible.for~.-voltages t~~tat are less than in the
case of the planar -~unc~ion of ths~ same area: :5econdly, the actual ~area of the
~unction increases, which leads to~a~ tncrease in tlie capacitance, variation
in tlie resistance o~ the di$fus.ion laye~ and ~to th,e~ pc~ssibility of electrical
short-cir-cuiting of the ~~cttons=in tTie~nearby openings. The latter fact imposes
a restriction on an increase in tfie degree of integration of~the microcircuits.
Influence of the Masking Fi1m. Local diffusion of the a~xtures takes place _
~der somewhat different conditions than~ordinary o~e-dime~sfonal diffusion~:~which _
proceeds over tfie e~tire surface of the sr~mi.eonductor~ plate. The presence of
a masking film, as a resu3.t of the~ differeaEe� in valuea of the thermal coefficient -
of linear expansion~of it and the semiconductor plates leads to the occurrence of
mechanical stresses. In the masking ~~lm these stresses are tensile stresses.
Mechanical stress causes the appea~ance of-additional dislocations at the edge of
the opening and bending with respect to the direction -~of :t~e silicon surface.
The intensified diffusion of the~admi.xtures along these dislocations can lead to
rejection of the planar structures.
It is very important t~at the p-~n ,junction reaches the surface of the plates
under the masking film, that is,-it turns out to be protected from further produc-
tion processing anxi external effects ~til its formation. This explains the
high stability of the parameters of the planar structures.
Effect of Thermal Oxidation. It is necessary to consider that at high temperatures
thermal oxidation of silicon takes place simultaneously with diffusion. The
diffusion coefficients of the alloying admixtures in the oxide film are appreciably -
smaller than in silicon. Therefore the p-n 3unetion formed in the presence of
a s urface oxide film has significantly less depth by comparison with the theoret-
icall.y calculated film. In addition, as-a result of differences in solubility in
the oxide and in the silicon, redistribution of the admixture takes place in
accordance with the distribution coefficient of the silicon oxide. For the most
standard admixtures of boron and phosphorus the distribution coefficients are :
- opposite, the surface layer of the silicon is impoverished with respect to boron -
(the distribution coeff~cient of the boron is greater than one), and in the case
of phosphorus diffusion, on the contrary, it is enriched (the distribution
coefficient of the phosphorus is less than one). All of this leads to complication _
of the diffusion process and to anomalous distribution of the admixture. In the
case of diffusion of the admixture with the silicon oxide distribution coefficient
- less than one the maximum concentration shifts into the depths of the diffusion
layer, Decreasing the surface concentration has a negative effect on the quality
of the resistance contacts to the local regions.
10-3. Methods of Achieving Diffusion
Two-Stage Di~fusion. In order to obtain reproducible par~neters of the diffusion
- layers and comparativel}~ sma11 sux~ace concentrationswhich are almost always
- required when ~aking IC, the ma~ ority of the difi~usion processes are conduc~ed
in two stages.
In the first stage, a defined amovnt,of alloying admixture from an imlimited
- sour.ce is introduced into tfie thin sur~ace layer of a semiconducting plate. At
the same time, in the first, f~equently called "buildup," the surface layer of _
.�.r~c~�:.~eued concentration is created --_a source of admixture for the second step.
�~r`ace concentration of the a~mixture is large after the "buildup" phase.
169 _
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
The first stage usually is carried out~qui,ckly and~at crnnparative~y low tempera-
tures. A layer of admixture eilicate~.glass is.formed on the surfac~ of the
ailicon plates.
The second stage,11) .B .
(aa) ~
li,~~cKOCme � D ~ ~
(�o) I ~2f~
_ , ,
B - ~ ~1 ~
- I ~ C Xo~nocmod j Padovud
A xo9 ~ ,roB
1 i ~hF)pe~ga~ ~ peaKw ~e~)
I (~8) aocxocm~ (111)~
2 r ~ /
~f ;
i
. a~ ~8a~laocKOCms (110) ~'0�Sd~
- Figure 17-4. Influence of the cryatallographic or.ientation of the
. plate (a) and direction of the cuttitig (b) on the shape of the
- crystal (c). 1-- sewiconductor plate; 2~-~- ingot,
Key: a~. plane
bb. idle of the cu~tting tool
c~e. working atroke of the cutting tool
The basic de�iciencies are as f~]_1ows; low accuracy of the geometric dimenaiona
of the crystals obtained; depende;~c~~ of the quality of separation on the ratio of
the crystal dimenaions and thicknesa of the separated pl.ate. The minimum size of
crystals a and thickness of the plate h are related by the expresaion a= k.h. For
silicon k= 4; for germanium k= 32.
Laser Separation of Plates and Substratea. Separation by laser radiation is among the
contactless r~ethods in which there is no mechanical effect on the machined material.
- Separation can be carried out e.~ther with preliminary obtaining of lines ~iaser
- scribing), or by passage thraugh the entire thicknesa of the material (laser cut-
ting).
The formation of the lines tak:_ place as a result of evaporation of material by
the high-power focused laser beam. In through cutting there is also fusion.
The application of laser ecribing permits a fourfold to fivefold increase in effi-
ciency of the proceas by comparieon with diamond scribing. Aa a result of the great
depth of the lines (40-50 microne) the percentage yield of good structures after
breaking increases. Using a laser beam it ia poasible to cut through the oxide and
. the metal layer. As a result of absence of inechanical effect there are no micro-
chips or microcracks. The fusion of the material along the edgea of the line de-
creases the probability of peeling of the film coatinga. The�cryataldgraphic orien-
tation has no influence on the quality of separation or ehape of the crystals. The
crystals obtained have in practice vertical lateral aurfaces which greatly facili-
tates automatic assembly.
The line dimeneions are determined by the diameter of the.laser beam, the apeed of
displacement of the plate or the substrate relative to the laser beam, power, fre-
quency and duration of the radiation pulses.
295
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
= FOR OFFICIAL USE ONLY
The defici,enciesoflasex separati.on i,nclude the follo~ring; high cost and complexity
of the equipment, contamination of the surface of the structures by products of
evaporation and melting, the occurrence of a deformed zone in the silicon with
structural disturbances and a defective zone, in which changes in the properties of
t1~e material take place and under the effect of laser treatmenic.
Chemical Separation. Separation by through chemical pickling is used comparatively
rarely f or semiconductor structures and provides the corresponding preliminary ma.sk-
ing of the surface for which the photolithography process is undertaken.
17-2. Basic Assembly Methods
Soldering. Three basic methods of joining parts are used for assembly: soldering,
welding and bonding.
Soldering is the process of joining two parts in the fiolid state using the molten
material solder. The solder must wet the ~oined surf aces well, spread, f illing
the entire space between them.
When heating the solder to the melting point, the following take place simultaneously
between the solder and the j oined parts: solution of the joined materials in the
liquid solder, diffusion of the solder into the joined materials with the formation
of a solid solution, chemical interaction of the solder with the materials of the
joined parts with the formation of intermetallic compounds. During cooling the
solder crystallizes on the surfacesJof the joined parts, entering into a strong
metallic bond with them.
The solder must not change its properties at the operating temperatures of the IC _
- (:.25� C for silicon and 85� C for germanium). The temperature of obtaining the
so:LderPd joint must be as low as possible in order not to have negative effects on
the parameters of the f inished products.
Depending on the melting point of the solders, Iow temperature aolder (to 450� C)
= and high-temperature solder (above 450� C) are distinguished.
So�t or low-temperatu~re solders include tin alloys with lead (POS-40; 40% Sn+60% ,
Pb; POS-61: 61% Sn+38.2% Pb+0.8% Sb), tin with bismuth (POVi-05: 99,6 to 99.4%
Sn+0.4 to 0.6% Bi). The solid or high-temperature aolders include alloys based on
silver (PSr-45: 45y Ag+30~ Cu+25% Zn; PSr-72: 72% Ag+28% Cu). Basically low-tem-
perature solders are used for microcircuits.
The quality of ,joining by solder is determined by the quality of cleaning the sur-
face of the joined parts and the solder to remove contamination and oxides and also
the choice of the design of the j oint.
- In order to improve the wetting f~.uxes are used which remove the surface oxides -
during the soldering process, prevent new oxidation and lower the surface tension -
of the solder. When manufacturing microcircuits, acid-free anticorrosive fluxes
base3 on colophony, types FKSp, FPEt, FKTS and based on zinc chloride FKhTs, are
used. However, during soldering the flwces introduce contamination; therefore an _
effort i,s made to use them in rare cases. In the majority of cases the soldering is _
' done without flux, but in a reducing (hydxogen~ formisgas) or inert environment
(argon, krypton and helium). Dur in~ soldering hydrogen forces other gases out of
296
FOR OFFICIAL USE ONLY ~
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
the soldered ~o~nt, and then is easi.ly~ removed ~tself during de~assing of the micro-
circuits. Sydrogen must be puri~ied of moisture (dew point no more than -50 to -60�
C) and oxygen (no more than O.UU3 to Q.OOSY). Hydrogen is explosion hazardous;
therefore when soldering in large spaces furnaces fnrmirgas is u~zd (a mix-
~ ture of 85% nitrogen and 15% hydrogen).
The parts are butt and lap joined and also ~oined by combination methods. The me-
chanical strength of the ~oints determined by the method and the conditions of sol-
dering, the pr~paratio~ of the surfaces, the strength of ;_:ie solder in the joint,
the strength of hinding of the solder to the matErial of the j oined parts, the
presence of intermetallic compounds in the joint, the strength of the joined ma-
- terials in the joining zone after cooling. The strength of the butt joints is lo-
wer than lap ~oints.
The advantages o~ soldering are simpl~.city of the proceas, the absence of fusion,
relatively low heating of the ~oined parts, the possibility of ~oining parts of
complex conf iguration.
_ In order to perform the operatiAns of assembly of the microcircuits, the soldering
- is done in hydrogen f urnaces; by an electric soldering iron; by heat transfer from
a tool that is pulse-heated by an electric current; ultrasound; electrical resis-
tance as a result of joule heat released at the point of ~ oining the parts; sub-
mersion in solder; radiation.
Welding. Welding is gradually displacing soldering for the assembly of microcir-
_ cuits. This is explained by the high quality of welded ~ointe and the slight
effect on the parameters of the structures. Welding is the process of ~oining two
parts without the participation of solder as a result of bringing them to the dis-
tance of atomic effect. During welding the following states of the surface materials
of the joined parts are possible: plastic deformation, fusion and plastic deforma-
tion, fusion and subsequent crystallization.
- Welding can be done with heating or without heating, in the presence or absence of
a compressive force and also with simultaneoua effect of heating and compressive
force.
In order to perf orm the operationa of assembly of the microcircuits moat frequently
_ the following welding techniques are used: thermocompreseion, indirect pulse heat-
ing, ultrasonic, double electrode, la.ser spot, electron beam.
Bonding. Bonded joints do not require complex equipment9 they are easy to make,
but they do not always provide good contact quality. This explains their applica-
_ tion primarily for microcircuits operating under nonharsh operating conditions.
~ Current nonconducting ar~d current conducting (contactols) adheaives are diatingui-
shed. In order to obtain the ~oints, the adhesive is applied in batches to the
surfaces, they are bonded to contact and in the absence (or without) compressive
, forces the glue hardens. During hardening the glue shrinks. The mechanical
stresses accurring in this case provide for drawing the parta together and a tight
mechanical joint of the parts. The current nonconducting glues insure high mecha-
nical strength o� the ~oints, whj.ch incr~ases in the presence of an external com-
pressive �orce. However, in this case the probability arises of transmisaion of
- an electric current through the metal contact spots of the parts, by tunnel
' 297
FOR OFFICIAL USE ONY.Y
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
r OR OFFICIAL LJSE ONLY
in�iltration of the electrons thxaugh tI~ tbin interlayex o~ glue and also through
the conducting channels obtained by d~�~usion of the metal through the de ~ective
sections. The joints using contactols have high electrical conductivity but less
mechanical strength. An increa.se ixi mechanical strength is achieved by decreasin.g
~ the metal f iller content.
- 17-3. Installation of Crystals and Plates
Direct MoLnting Method. The direct method, that i~, with the working surface up,
is used to mount structures of the microcircuits based on cases, the seats of the
lead frames (strips) or for additional contact substrates. The mounting of the
mounted active elements on the passive part of hybrid micror.ircuits can also be
carried out directly.
The Lasic requirements on the mounting operations are as follows; insurance of high
mechanical strength of the joir.ts, good heat removal from the structure and in a
number of cases, good electrical conductivity. The temperatures and the compressive
forc~s when doing the mounting must not be too high so as not to disturb the pre-
viously obtained joints, not have a negative inf luence on the structure parameters,
and not destroy their mechanical integralness. At the same time they must be suffi-
- cient for strong joining. The elements of the microcircuits take up only the
surf ~ce part of the crystals or are located on the surface; therefore the operations
of direct mounting are not critical to the depth of penetration of the connecting
~oint.
Electrical insulating connections are made using current nonconducting adhesives,
glass or special compounds.
The joints by adhesives and conpounds are quite broadly used for mounting, for they
are distinguished by simplicity of the process, low ha.rdening temperatures and
suff icient mechanical strength and reliability. By bonding it is possible to join
various ma.terials of different thickness. The bonded ~oints simplify the struc-
tural design, they increase the weight, and they save with respect to the consump-
tion of expensive metals.
The bonded surfaces must be carefully degreased and dried well to complete removal
- of the solvents, for later when the glued joint hardens the remains of the solvent
will lead to the appe.~rance of porosity and mec~anical stresses that lawer the
- strength of the ~oint. Sometimes vacuum annealing is used to clean the surface of
the bases of the case or the lead frames.
The thickness of the layer of glue applied ta thejoined parts must be small, for
with an increase in thickness the strength of the joint decreases. The glued
mounting is done in holders when the spreading of the glue is small and an exter-
- nal compressive force is required or it is done without a holder when the glue
spreads well over the surface and external pressure is not required. The heat
treatment of the glue, as a rule, is carried out in two steps: f irst for complete
remov~l of the solvent, then at a higher temperature, for hardening. The methods
and conditions of heat treatment are determined by the composition of the glue and
the structuxal design of the joined parts. Epoxy resin ED-S= glue VK-2 (a solution
of organos~,licon res~.n in an organic solvent with finely disperse asbestos), the
glues K~4QQ, KT354--61, VK~4, VK~8, VK
32~2~0~ and so on are used at the present
time for mounting crystals and plates. These adhesives provide good strength at
298
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFl~'ICIAL USE ONLY
temperatures to 300� C, which pexm~,tst subetec~uent j oining of the leads without rup-
turing the glued joi,nt.
The deficiencies of glued ~oints includ~ complexity of repaii (replaca~ment of
' structures) and poor heat removal.
Joining by glass provides good matching of the TKLR, it permi~s simultaneous ~oin-
ing of the structure and leads to the frame. The baeic deficien.,tes of joining by
glass are the high temperatures of the process ~about 5~0� C) and poor thermal
~ conductivity of the glase. Tt ~is recommended that the ~.ounting with glass be used
- for the structures of small IC.
The electrically conducting ~oints are basically used for mounting erystals on the
base of inetal cases. The inaulation of tlie crystal ~is not required, for it always
_ has low potential during the operati,on of the microcircuit.
_ The mounting of the silicon cryatals can be done by electric aoldering and in some
casea by welding or 3oining by contactols. ~
Eutectic soldering is the most wi.despread methad of mounting silicon crystals. Moat
frequently eutectic soldering is done using gold-silicon or gold-germanium aolders
having a melting point of 370 and 356� C, respectively. Let us remember that alloys
are called eutectic in which crystalliaatioa of 'the components takes place simul-
taneously over the entire volume at the lowest eutectic temperature for the given
system. During cooling of the solder at a time correspoadiag to the eutectic tem-
perature, a complete discontinuoua transitioa tab,ea place in both alloy components
from the liquid to the solid phase. The hardeni:~g ie accompanied by the release of
fine crystals well bound to each other and this providea for a high-quality soldered
3oint. �
Eutectic soldering can be done using solder inaerta (or balls) or by direct contact
joining of a silicon crystal to the seat to which a layer of gold 6-9 microns thick
_ has been applied in advance. At the present time contact soldering is most widely
used. The nonoperating side of the semiconductor subatrate is also coated with
a gold film or gold with alloying admixtures correaponding to the electrical con-
ductivity of the crystal in order to improve the wetting. The gold film is melted,
J and then the plate is divided into crystals. By uaing a vacuum capturing capillary
and a matching system the crystal is approximately installed in the seat. Heated
- inert gas (nitrogen, argon) or formirgas ie fed to the ~oining point. The optimal
conditione of eutectic soldering of a silicon crystal are as followa: temperature
390-420� C, time 3-5 aeconda, pressure (3-5)�10~ newtons/m2. For intensif ication
of the soldering process, a compressive vertical force and horizoatal ultraaonic
vibrations, vibrations of industrial frequency or mechanical vibrationa with a
frequency of 4-8 hertz are used. This provides reliable contact of the ~oined
parts, the eutectic forms quickly and uniformly over the entire area. '
Eutectic soldering is widely used for automated mounting of cryatals oa metal strips
or cranea, the seats of which are ccated with gold. Here group placement of the
crystals in the stenciling hol~der, the holes in which correspond to the location
- of the seats of the strip or fzame will be used. The cryatals are brought into
_ tight contact with.the strip ox �rame hy special clamps of tbe upper cover of the
holder. Then the holder is put in the fuxnace and the soldering takes place.
299
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
. FOR OFF[CIAL USE ON~.Y
For mounti,ng aili,con cxystals somet~mes goldTti,n solder (80% Au + 2~% Sn) is used
with a eutectic temperature of 280� C or POS--61 is used with a eutectic temperature
of 185� C. Th~e advantage of these soft solders low soldering temperature can
turn out to he insuf�icient if further apsrations are carried out at a higher tem-
- perature. The basic advantage of ~nounting by low-temperature aoft solders is ease
of dismantling the rej ected cryatals.
The mounting using contactols, as was already noted, is distinguished by simplicity
- of the process. For connecting the mounted active elements to the contact areas of
the passive part of the hybrid IC, the contactols K-I, K-II, K-III, VK-20T and so
on are used.
"Rotated Crystal" Method. Crystals with rigid three-dimensional leads entering into
the composition of the hybxid microcircuits or microassemblies are mounted on the
- passive part. of th,e working side of the crystal at the bottom, Thus, it is possible
to mount the crystals with expanded contacts. In this case the joining protrusions
are executed on the contact sites of the substrates. Using volumetric leads or
protrusions on the substrates, both the crystal and all the leads are simultaneously
connected. The basic diff iculties of the installation by the rilcated crystal
method are reduction of the difference in height of the leads protruding above the
crystal or substrate to a minimum and matching them with the contact sites. The
difference in height leads to the necessity for creating sufficient deformations
for the most protruding balls or columns in order to provide for contact with the
lowest protrusions. Here the deformations must not escceed the admissible norms,
for the mechanical stresses occurring after installation lead to potential failures
of the microcircuits. The matching operation is performed using q~ite comp~ex
optical devices with stereomicroscopes or infrared microscopes with optoelectronic
- image converters that shift the radiation to tlie visible part of the spectrum. The
majority of semiconductors transmit infrared beams, and metals are nontransprent
for these beams; therefore against the substrate background the metal sections are
obvi~us in the form of black spots. In some devices the image is projected on a
~ screen to facilitate ma.tching.
I
! The matching of the rotating crystals with the bar leads is simpler to do, for they
; go beyond the crystal limita.
~ The process of ~oining the volumetric leads will be investigated in ~ 17-5.
~ 17-4. Wiring
i Final Installation Operations. After ~oining the plates and crystals we have the
; operations that complete all of the electrical connections between the elements and
I also between the structures and the external leads of the finished microcircuits.
These include the connections of the ohmic contacts of the active suspended elements
with the film contact sites of the passive part of the hybrid IC, the contact sites
of the IC structures with external leads of the cases, contact sites with contact
sites (or volumetric leads) of the monocrystalline IC. All of these ~oints are
made with soldering, welding and bonding 17-2).
_ Depending on the means used for making the ~oints, wire and wireless mounting are
distinguished. In the case of ~i,re mounti,ng~ bas�cally gold and aluminum wire are
used. The wireless mounting includes the following: ~oining the crystals to the
volumetric leads, aseembly on frame, a strip or f lexible carrier (the "spider" lead
method).
, 300
- FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Special ~eatures o~ Wixing. A basi.c ~eature of the jo3nts made using gold or
aluminum circular wi,res is a great dif~erence 3.n thickness of the joined parts,
~ As a rule, th,e wire diameter is app~ecia~ly gxeater than the film thickness of the
contact areas and apprecia.bly less than the diameter of the case leads. The micro-
contacting procesaes are therefore highly critical, for the formation of a connec-
ting weld with a thickness comparable to the f ilm tlzickness of the contact site is
possibl.e. The wiring is individual, each connection is made separately. Wiring is
one of the weekest points in microcircuit production. Unreliability of the indi-
vidual wiring assembly is the cause for the greater part ui failures of microcir-
cuits. Nevertheless, up to the present time the majority of connections have been
- made using wires. The greatest reliability of the wiring connections is insured
by welding and soldering. Let us consider the varieties of wire connections using
these methods.
Thermocompression Welding. Microcontacting by thermocompression welding is done
- with simultaneous effect of temperatuze and preaeure on the joined parta for a de-
= fined time interval. A necessary conditioa of the formation of a strong joint is
- plastic deformation in the contact zon2. For this purpose the ~oined wires are
selected from the soft plastic metala: gold, aluminum, eilver, and so on.
The connection temperature during thermal compression must not exceed the tempera-
, ture of f ormation of the eutectic of the 3oined materials, and it is usually close
to the annealing point of the more plastic metal. It ia most expedient to heat by
a hot working tool, for localization of the heat release in the welding zone is
insured.
The pressure is transmitted to the ~oined materials through the welding torch
. which is in the shape of a wedge (needle), capillary or "bird's beak" (Figure 17-5).
The pressure must provide no less than 30,�6 deformation of the ~oined material, but
no more than 60%. On application of pressure the contacting tt~kea place first at
individual protrusion points of the ~oined surfaces. The deformation of the pro-
trusions promotes an increase in the contact zones and approach of the joined sur-
faces. Here forcing of the adaorbed gases and surface contamination out of the
- welding zone begins. In the case of sufficient approach, the phyaical interaction
of the atoms of the ~oined materials begine as a reault of Van der Waals forces.
= The stronger chemical bonds occur as a reault of heating and plastic deformation.
Plastic deformations lead to effective rupture of the surface oxide filme, which
are the basic obstacle to chemical interaction. On rupture of the oxides, sectiona
of the clean surface are denuded, on which the strongest chemical bonds occur.
The temperature and pressure are interrelated; they must be selected so as to in-
aure optimal deformations, closeness of the ~oined parta and activation of the
surface atoms to the formation of chemical bonda.
The optimal conditions of thermal compression for each pais of welded parts are
selected experimentally beginning with the requirement of obtaining maximum strength
of the ~oint. Far this purpose the relations are defined for the strengrh as a
- function of temperature and pressure, respectively, f or pressure and time constants
or temperature and time and also dependence of the strength on time for constant
tewperature and pressure.
The duration of the thermal compression process usually is fractions to tens of
seconds depending on the quality o~ pxepaxing the ~oined parts, the properties of
301
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
~
1 I ~ '1 .
5432 5432 6438
_ a) b~~)
Figure 17-5. Thermal compxession tool. a M wedge; b-~ capillary;
c-- "bird`s beak"; 1-- tool; 2~-- joined wire; 3-- joined area;
4-- aubstrate; 5-- welder table.
the juined materials, temperature and presaure. When the opti.mal time is exceeded,
a lowering of the strength of the thermal compression joint is observed. This is
explained by the f act that f or the creation of a strong contact "fresh" bonds are
needed; if the pressure is applied longer, then the process of rupture of the
formed bonds begins.
The thermal compression conditions recommended for certain pairs of 3oined parts
are presented in Table 17--2.
- Tab1e 17-2. Thermal compression conditions
' ~Tempe- Pressure ~Lead de- Time,
Welded parts rature, x10~~ n~w- f ormation, seconds
�C tons/m ~
- Silicon and gold wire 350 14-15 60 5-10
Aluminum f ilm and gold wire 350 10-11 50 0.5-3
Gold f ilm and gold wire 320-34U 7-10 50 1-5
Silicon and aluminum wire 450 7 60 10
Aluminum f ilm and alwninum wire 400 6-7 6Q 1-3
Gold f ilm and aluminum wire 320 6-7 60 1-3
Aluminum f ilm and .ailver wire 400 ~8-19 5-7
G.old..f.ilm and silver .wire,. . 350 18-19 5-7
In order to carry out thermal compression welding the IC structure or holder with -
IC structures is .fastened to the work table of the device. Before welding using
the MBS-2 microscope or a special projector and manipulators for displacements,
the joined parts are matched. The welding is carried out in the sir, but in some
cases to protect the structures from oxidation nitrogen, f ormirgas or argon is
fed to the welding zone through the work table. The procedure for dieplacement
and separation of the joined wire is determined by the structural design of the
welding tool and unit.
It is possible to have butt and lap thermal ~ompreaeion. Butt thermal compreasion
is carried out by a capillary (Tigure 17~-S,b) with preliminary formation of a ball
on the end of the ~oined wire uaing.a hydrogen buxner. Thus, only the gold wires
are co~ected, for the formation of a ball on aluminum aad other materials is
dif f icult as a xesult o� their ox3,dation. ~Ihen transferring the pressure and
302
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400010055-6
- FOR OFF'iCIAl. USE ONLY
heating the gold ie split to size~ Appxoxi,ntately twice the di.ameter. After the
- formation of the ~oint cap~llary is raised, it is taken to the case lead (or
another part) r~nd tlie second jaint is made lap (without a ba'Ll) . The wire can be
- broken by cutting, pulling after making the second loop connection or using a
hydrogen burner with subsequent cutting off of the "tail."
Lap thermal compreasion is carried out hy means of a needle o~c a tool in the form
of a"bird's beak" (Figure 17-5,a,c). During thermal comr ~ssion the wire is fed
by a needle using an auxiliary capillaxy.
_ . ~ -
r.._~p ~ - _ _ - ~ -
1 Oe~ne
~ BpeMeHU (a)
i
I
_ L 1 I J Ilpeod aea
~ BamQ
b ~b~
! 1
.Z
~ ^~2208
~ (c)
Figure 17-6. Diagram of microwelding by the indirect pulae heating.
1-- welding tool; 2-- connected wiring; 3-~- coatact site; 4--
aubatrate.
Key; a. time relay b. converter c. ~220 volts
The quality of welding is detezmined to a significant degree by the quality of the
welding toal; the end o� the tool must be flat and parallel to the contact eurface
of the site. When wozking on thermocompression units it ie necessary to preserve the
tool from pollution and contact w~,th harder materials.
The advantages of microcontacting using thermal compresaion includes simplicity of
the welders, long service life of the tool (to aeveral hundreds of thousands of
welds), easy control of the process, stability and low senaitivity to sma11 devia-
tions of the welding conditions. The output capacity of the Soviet unita is
180-800 welds/hour.
- Indirect Pulse Heating Welding. This method ie distinguished from thermal campres-
sion welding by th: fact that the heating takes place by the direct passage of an
electric current through the working tool and only at the time of ~oining of
the parts (Figure 17-6). The released heat, as a result of the atructural design
of the tool, ia concentrated in ita lower working part. This permits more exact
regulation of the magnitude and the duration of heating of the compared parts.
_ The heating temperature o~ the tool dependa on the fed voltage and the duration of
the welding pulse. The conditions of welding by indirect pulse heating are
characterized t~;y a primary voltage, pulse duration, the free tool pressure. The
duration of the welding process by indirect pulse heating is less than the thermal
compressi~n process, for the concentration of h~at at the tip of the tool insures
more effective heating of the welding spot. This permits an increase in the weld- _
ing eff iciency. The preasure of the welding tool ie selected as a function of the
diameter or thickness of the welded wire and the ductility of the joined materials.
The recommended welding conditions are presented in Tatale 17--3.
303
FOR OFFdCIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
~'OR OFFICIAL USE ONL~f
. Tahle 17--3. TrIeld~ig cond~,ti.ons by~ i.ndixect pulse heating
W3,'~e Pximary Pulse Compres-
diame~ voltage, duration, sive force
Welded parts ter,, volts sec x 102, new-
microns tons
Aluminum wire and gold film 100 110 0.2 150
on pyroceram
Aluminum wire and gold kovar 100 120 0.2 120
_ Aluminum wire and metalized 30 120 0.2 60
silicon
Gold wire and gold f ilm on 50 105 0.7 80
pyroceram
At the present time the~e are a large number of microwelding units that use indirect
pulse f eeding: "Kontakt-3A,'~ SKIN-1, MKS~02 ~ USP-~Ul, E1~~4a8A, EM,-425A, EM~-440,
EM-441, MS--3R2-2. The USP-O1, EM-~425A and EM-441 welders with semiautomatic con-
trol have an output capacity of 900, 800 and 1600 welds/hour, respectively. The
automated EM-440 welder has a capacity of 2000 welds/hour.
Ultrasonic Welding. As a result of the displacement of oxide films and contamina-
tion from the weld-affected zone and activatior~ of the surface atoms, longitudinal
ultrasonic vibrations intensify the process of 3oining and essentially improve its
- quality and reproducibility.
The weld is made using normal pressure and longitudinzl vibrations of the tool with
an ultrasonic frequency of 20-60 kilohertz. The ~oined surfaces are sub3ected to
harsh shearing forces, as a result of which the microirregularities are sheared off,
the oxide f ilms are rupture.., and they are forced into the gaps between the micro-
protrusions. The heating as a result of friction not exceeding 30-50% of the
- melting point of the ~oined materials, leads to plastic deformation and promotes
the occurrence of a direct metallic b.ond and a strong ~oint without structural
changes in the material in the weld-affected zone. .
The welding procesa parameters for some of the,Soviet welders are pre3ented in
Table 17-4.
Table 17-4. Characteristics of ultrasonic welders
Welder Output Lead di- Operating Welding Force of compres-
capacity, ameter, frequency, time, si~n of parts x
welds/hr microns kilohertz secondg lU , newtons
UZP-03 600 20-5C' 75+2 ~.OS-0.5 10-100
EM-424A 800 25~60 66+6.6 0.08-3.6 10-120
MS-41P3-3 1000 20-50 66 0.05-1.5 20-150
UZP-02 25U0 24-40 70+5 10-20
NPV'-2 240U 24-,40 :74+5 10--20
Ultrasonic welding tnakes it possible to weld parts from different materials, includ-
ing dielectrics that differ sharply with xespect to thickness for low requirements
304
' FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
on the qual~,ty o~ pxeparing the sur~ace~�. In pxactice ultrasonic welding is car-
ried out most �requently ~,n combina.t~,on with indirect pulse heati.ng.
Double Electrode Welding. This procedure is a version of resistance welding. The
joint is made by passing a high-density electric current pulse through the welding
zone and simultaneously chemical compression of the welded parts, The electrodes
through which the electric current is fed are located on one side of the welded
parts (Figure 17-7). On pass~ge of the electric current the basic part~ of thc
joule heat is released in the region of maximum resistancE. - at the point of con-
tact of the joined parts. The contacted protrusions of the joined surfaces heat
rapidly, they are brought together under the effect of compression, and the contact
2one of the parts begins to expand. The contact resistance and current density
diminish in this case. The welded surfaces are sub~ect to plastic deformation and
melt. The current feed is atopped, and the melt region, not extending to the near-
est sections, begins to be cooled, forming the weld. With careful cleaning of the
- surfaces of the welded parts and the electrodes, a high-quality weld can be
_ achieved which is not inf erior to the basic material with reapect to strength
characteriatics.
The welding conditions are determined by the energy and duration of the welding
_ pulse, the pressure on the electrodes and time. For example, the Soviet USDYe
welder is characterized by a pulse energy of 4.8~-5.2 watt-seconds. The pulae dura-
tion is 0.02 to 1 second, the load on the instrument is 0.5-20 newtona with weldable
wire diameters from 30 to 150 microns.
P~- _ _ .
9
1 II
' II
I
II '
li Z'~ 9
Figure 17-7. Microwelding with dual electrode.
1-- tool; 2-- wire; 3-- contact site; 4--
substrate.
- The heating tirae and presaure on the ~oined parte during double electron welding is
less than for indirect pulse heating welding. Thia decreaaes the harmful eff ect of
the tool on the adjacent regions of the atructures.
The electrode material must have high electrical and thermal conductivity with high
mechanical strength. For the uncombined electrodes, bonds9 tungsten and special
alloys are us~d; for the combined electrodes, the work,tng part ie made of copper
alloy with tungsten. The roughness of the work.ing aurfaces of the electrodes must
be no less than class 9-10.
The basic advantages of the mWthod are ae �allows: a small zone and short time for
the thermal e.�f ect, high output capac~ty ~to 1200 welda~hc>ux) .
305
FOR OFF[CIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
The deficienciea of the dual electxode weldi.ng are as follows; the poseibility of
current leakage and thermal itnpact.
Laser Spot Welding. This method reduces to welding by fusion with subsequent
crystallization of the fused metal. Th,e welding is done uaing a laser beam fdcused
to a diameter of 0.25-1 mm with a specif ic power of 105 to 106 watts/cm2. As a
result of nonuniform temperature distribution inside the welding zone the metal
boils, a high pressure region is created, and a thermal explosion can occur. In
order to prevent spattering of the metal from the welding zone, it is necessary to
select radiation that is not too higll
powered. In practice, the weldiag is fre-
quently done by a diverging lasex beam, that is, the weld-aff ected zone is located
below the focal plane. The elimination of inetal splashes is one of the basic pro-
duction problems during laser beam welding. The pulse duration of the laser irradi-
ation T must be less than the time required for the beginning of fusion of the film
material and more than the time xequired for fusion of the wire connected to the
film contact site: tmelt ~ T~ twire~
By using a laser beam it is possible to do the welding near the elements of the
microcircuits, glass or ceramic insulators. By a laser beam it is possible to
- perform remote welding in a vacuum, in high pressure chambers and in other environ-
ments. The weided connections obtained using a laser beam have smaller (by approxi-
mately an order) transient resistance by comparison with the ~oints obtained by
- pressure welding.
Table 17-5. Laser ~pot welders
= Welder Welded Dura- Maximum Pulse Pulse
_ depth o~ spot di- tion, radiation duration, repetition
fusion, ameter, part/ energy, milli- f requency,
utm mm minute 'oules seconds hertz �
SLS-10--1 0.3 0.4-1.5 30 8 2-4 0.1-0.5
"Kvant" 0.5 0.4-1.5 60 15(30) 4 0.1-1
"Kvant-16" 0.7 0.4-1.5 30 30 6-7 0.1=0.3
The application of laser welding has been held up for a long time as a result of
higk~ cost and def iciencies of the lasers. The characteristics of aome Soviet glass
laser welders are presented in Table 17-5.
Cathode Ray Welding. Analogously to laser welding, cathode ray welding is accom-
panied by fusion and subsequent crystallization during cooling after the pulse
eff ect of the beam of a~celerated~~.electrons. Welding is realized in vacuum units by
an electron beam focused to a diameter of 0.5 mm with a specif ic power to 2000 kilo-
watts/cm2. The fusion of the contact region under the cathode ray spot takes place
as a result of conversion of the kinetic energy of the electrons to thermal energy.
The basic advantages of the method are the f ollowing: the possibility of welding
many metals at?d $lloys (nickel-nickel, nickel-copper, nickel-steel, nickel-kovar,
copper-rmolybdenum, steel-kovar, kovar-steel, k~var-kovar, and so on); the small
thermal effect zone permitting welding during glass insulators and a metal-ceramic
~unction; the p~ssibi.lity of contxolling the beam displacement aad automation of
the assembly processes; degassiag of the weld izz a vacuum.
306
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
FOR OFFICIAL USE ONLY
The hi,gh. coat o~ equip~ent and diirat~,on of ttie evacuation pracese limit the appli-
cation of the methad.
Wire Joints Using Microsolder. The process for connecting wires with solder ie
highly critical with respect to the cho3.ce of the solder material, temperature and
time of soldering. The latter must be as emall as possible, and the solder must not .
noticeably dissolve the film material and form brittle intermetallic compounds.
Soldering, more rarely than welding, is used for wire assembly of semicouductor
microcircuits, for it dops not provid~ suff icient stabilicy of the results, and
reliability, and it has comparatively low output capacity. On the contrary, solder-
ing is more frequently used for the assembly of the hybrid IC.
In order to increase the strength of the j oints, the area of the soldered contact is -
increased, the wires are fastened by special clamps or the wire ~oints are made -
paesing through the substratea (Figure 17-8).
The leads are f irst tinned, and then for soldering to the contact aites, a"split
electrode" is used along with a 3et of hnt inert gas, and aa "infrared soldering
iron" (soldering by f ocused infrared radiat ion). Soldering "by infrared soldering
gun" which ie among the contactless methods, insurea relatively preciae localiza-
tion of the ~oining point and comparatively high output capacity. For welding the
wires to the thick-film contact sites the latter are tinaed by a eolder wave or the
stenciling method.. In the former case the liquid sold~r continuously fed by a pump
to a nozzle set at an angle to the horizon, forms water, through the crest of which �
the thick-f ilm contact sites pass. The tinning by the solder wave is distinguiahed
bp high output capacity, it prPventg contaminat~on with slag and flux residue, for
the process is performed in a moving jet with continuoua renewal of the aolder. The
soldering technology of wires to thick-film contact sitea is less critical and pro-
vides for obtaining contacts with ~igh atrength.
1~ 1 ~ 1 2 ,i 1� 2 , -
~ J ~ '
_ _J
~M S ~F 4 ~ ~
r~ v~ , ~ s~ ~
a) b~) c )
~igure 17-8. I~i.cxos~ldering o� ~r~,xe lead~ to tb.i,n~~i,lm contact si.~es.
a--- suz�ace connecti,on; b--- connecti,on with clamp; c~-- coanection of
- a througi~ lead; 1-- lead; 2~-- solder; 3-- contact site; 4-~-~ sub--
atrate; 5 clamp .
17-5. WireleQa Melting
Connectizlg the Cxystals to the Bulk Leads. The hulk leada of tLe mounted active
elements or semiconductox mounted m~,cxoc'~ccuite combine both the ohiaic contact and
the lead ~rom ~t. In g 13~3 ~.t was nQted that structurea with leads in. tbe form
of balls or columns are mounted hy the rotated crystal method. The structures
. 307
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007102/49: CIA-RDP82-00850R000400014455-6
~
FOR OFFICIAL USE ONLY -
with bax ieads can also be mounted ~t~,th Korki,ng sur~ace up. The developmeat of
the structures w~.th bulk leads made ~,t poss~.ble sharply to reduce the number of
unreliable w:Lre connections, to improve the output capacity and automa~re the
assembly. '
Af ter matching with previously grayed contact sites, th~e bulk leads in the f orm of
balls or columns coated with soft solder are joined hy soldering. The surface
tension forces of the molten solder attract the crystal exactly to the location.
_ The short circuiting of the leads by the solder does not occur, for the solder
does not wet the passivated surface of the crystal and the unt inned sections of -
the substrate on which the crystal is mounted. The solder f iZls all of the cavi-
ties, it eliminates the irregularities, forming a soldered joint on cooling. The
crystals connected by the soft solder are easily demounted, large mechanical ;
stresses do not occur in the soldered joint. For connecting the solid solder leads
usually thermal compression, welding by indirect pulse heating, ultrasonic welding
or combined methods are used. The thermal compression is used primarily for making
gold-gold joints. The ~oining by the rotated cryatal method is possible also when
using current-conducting adhesives. It is expedient to join the crystals to the
bulk leads using the holder method of assembly. Thus, for example, the holder
method of assembly used in the EM-432 ultrasonic welder provides an output capacity
of 4000 welds/hour.
The bar leads are basically connected by microwelding with ultrasonic iaitensif ica-
tic,n. Sometimes soldering is used. In contrast to the ball and column leads, the -
bar leads remove the heat better from the crystal, they are more eaeily matched with
the centact sites. The compressive force during welding can be applied directly to
the leads and not to the crystal. However, crystals with bar leads occupy a large
area of the substrate on which they are mounted. In order to decrease the substrate _
_ area used by the bar leads, they are arranged in a comb (Figure 17-9).
"Spider" Lead Method. The "spider" leads, that is, planar leads made of m~tal foil
arranged in the radial direction in accordance with the location of the contact
sites to which they will be connected, can make up a united wholewith the metal
frame or strip, or they can be manufactured on a flexible dielectric carrier.
The lead frames or strips are made of thin (25-75 microns) copper, Kovar or aluminum
foil using photolithography and through local pickling. The metal frames or strips
can be used to obtain the external leads of the microcircuit case from the contact -
sites of the structures and also for installafiion of the mounted elements. In the
former case (Figure 17-10) after simultaneous connection of all of the leads to all
of the contact sites it is possible zo encapsulate the structure and cut off the
excessive part of the frame. The end of the "spider" lead f ree a�ter it is cut off -
is the finished microcircuit lead. In order to lend rigidity to the structural
element, the structure is first mounted on an auxiliary dielectric substrate. -
In the latter case (Figure 17-11) the free en~is of the "spider" leads after ma~ing
the joint with the contact sites ot the mounted element snd cutting the frarne are _
- shaped and connected to the contact sites o~ the passive part of the microcircuit.
The flexihle dielectxic carrier ~txip with edge perforation (for automa.tic f eed)
is made o~ polyiwide, Iavsan, polyeatex or another polymez 80~120 microns thick.
Metal foil, ~or example, 'alwuinum ~o~.l, type A--7 ~ 40 miczons thick~ is rolled onto
- 308
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL U~E ~+AiL:Y
1
2
~ . _
Figure 17-9. Comb location of the bar leads -
1--bar lead, 2--crystal; 3--free part of the substrate
~ -
_
i
. 2
~ , ~
V
Figure 17-10. Wireless inst~ll..ation using "spiderless"
leada
1--lead frame (tape); '
- 2--"spider" leads
3--contact area -
4 --microcircuit structure
this strip uaing a nondrying glue or 8pecial adheaide~ Then by means of photo-
lithography with continuoua rewinding of the etrip frames are formed with the
"spider" lead pattern. The group method ia used to ~oin�.~ the contact aites of the
crystal to the internal leads of each frame, in the center of which there ie a hole.
In order to prevent electric short circuite between the leads and the crystals it is
possible to use A1203-inaulating ringa which are obtained by local electrolytic
oxidation of aluminum (Figure 17-12). The leade located on the flexible carrier are
insulated from each other; therefore the parameters of the crystals can be con-
trolled directly in the technological process. After installation of the crystals,
309
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
I
FOR OFFICiAL USE ONLY -
the outer ends of the spider leads are simultaneously connected to all of the con-
- tact sites of the passive part of the hybrid IC. -
~ Z s ;
~ ~ s
y; t
Figure 17-11. Installation of a mounted crystal
using a lead frame. 1-- frame with leads; 2--
contact site of cryatal; 3-- contact siCe of the
passive part of the hybrid IC.
' 1 2 ~ ' " -
5 4
Figure 17-12. Mounting on a flexible carrier.
1-- A1-lead-"spider"; 2-- crystal; 3-- con-
tact site of the crystal; 4-- A1203 insulating
_ ring; 5 carrier dielectric.
The "spider" leads are ~oined by the invea~tigated methods of microwelding. The
difference from connecting the wire leads consists only in the structure of the
- working part of the welder.
_ The wireless methods of installation are r~latively new, and they are not yet fully
developed with respect to structural engineering. In spite of the temporary diff i-
culties wireless mounting is atill the only patih to complete automation of the
assembly and encapsulation processes.
17-6. QuaZity Control in the Welding Procesa
The most effective methods of quality control of ~ oints are metallographic ana~.ysis
- and mechanical strength testing.
In order to check the mechanical strength of the ~ oints there are many attachments
and devices and also test methods. For example, when shear testing, the structure
with the connected leads is sub~ ected to stretching by a force acting parallel to
the surface of the subatrate. If the etrength of the ~oint is no less than 70~ of
the strength of the wire used, the ~ oint is considered to have high quality. The _
~oints are rupture tested by multiple bending of the lead at an angle of 30, 45 and
90� with respect to the substrate surface (the UKPM-1 device).
310
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
- FOR OFFICIAL U~E ONLY
The strength of bonded joints is determined by rupture teste. The rupture strength
of e bonded j oint must be no less than (125--150) � i05 newtona/m2.
The metallographic analysis coneists in examination of transverse or oblique macro-
sections of the welds, and it permits discoverp of their internal structure and
, detection of sections n~t wet during soldering, through fusion, microcracks, pits,
pores, intermetall.ic inclusiona, and traces of fusion of the solder with respect to
the grain toundaries.
X-ray defectoscopy using a diverging beam permits detection of internal defects and
offers suff icient information about the reliability of the joints. In contrast to
the metallographic analysis this method is noadestructive.
Visual monitoring makes it possible to detect breaks, short circuits, shif te of the
welding or soldering zone, pores, microcracks and deformations. The relative de-
formation of the connected lead ie determined by the widtfi of the welding spot.
With good quality of the soldered ~oint the solder flows out of the clearance be-
tween thefoi.ued parts, forming a good ap~earing solid fillet.
Cotitrolling the electrical parameters of the microcircuits permits e~timation of the
correctnesa of the selection of the assembly conditioas.
Test Questions and Assignments ~
1. What technological operations pertaia to the asaembly of IC and what are their
peculiarities?
2. Repeat ~ 3-2.
- 3. What are the advantages and the deffciencies of abrasive cutting of platea and
substrates with f inished structures?
4. What is the essence of scribing and what is the technique for applying lines?
5. Compare the methods of separatioA of the atructures af ter applying lines. How
is it possible to maintain orientation of the cryatals and plates after separation?
~ 6. What determines the scribing ~ff iciency?
7. What is the characteristic feature of aeparation of monoc.ryetalline plates and
substrates? What are the peculiarities of the scribing of planar structurr~s?
8. What are the advantages and dieadvantagea of acribing by a diamond cutting tool?
9. Compare laser acribing wi~h diamond scribing.
10. What is chemical separation, what are the d ifficulties with it?
11. What is soldering and what are its mechanisms, advantages and deficiencies?
12. How is it possible to improve the quality of a soldered ~oint when aASembling
~ IC?
311
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
EOR OFFIC[AL USE ONLY
13. What methods are used f or soldering when assembling IC?
14. What do welding and soldering have in common and how do they differ which can
occur on the surfaces of welded parts?
_ 15. What ~telding procedures are used for IC assembly?
16. Compare the properties of the ~oints obtained using current conducting and
current nonconducting joints. What are the peculiarities of the glued ~oints by
comparison with soldered and welded ones?
~ 17. What requirements are imposed on the installation operations?
18. What is the essence and ~hat is the te~hnique for direct mounting using the
ele~trical insulating compounds?
- 19. What is the essence and what is the technique for eutectic soldering of sili-
con crystals?
20. What is installation by the "rotated crystal" method and what are its basic
dif f iculties?
- 21. What are the characteristic features of wire mounting?
22. What is the thermal compression procedure, what technique implements it and
how is the optimal condition� selected?
23. Compare welding by an indirect pulse heating with the.r.weS.compression welding.
24. What are the basic advantages of ultrasonic welding?
25. What is dual electrode welding, what is its essence and technique? What are
the basic advantages of this welding over w~elding by indirect pulse heating and
thermal compression?
26. What microwelding procedures are contactless? What do laser and electron beam
have in common and what are their differences?
~ 27. Give a brief description of microsoldering during wire installation.
28. What methoda of wireless installation are used when assembling IC and what is
their essence?
29. What is the process of joining the crystals to the bulk leads by the "rotated
crystal'1 method?
- 30. What is the nethod of "apider'~ leads and wha.t are its advantages? Compare the
wiring on a f lexible carrier and on metal lead frame or strip.
31. What is the basic advantage of wireless inetallation?
32. How is the quality of the ~oints checked after assembly?
312
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFICiAL USE ONLY
CHAPTER 18. ENCAPSULATION OF MICROCIRCUIT~
18-1. Microcircuit Cases
General Information on Encapsulaton. The problems of encapaulation are ae follows:
insurance of reliable connectioa of th,e micro~ircuits to the equipment, protection
of the structures from all types of external ~aperating loads and alsa external
aesthetic appearar~ce.
With respect to atructural engineerinR attributes it is possible to divide encapsu-
lation into three types: case, caseless and combination.
Case encapsulation provides for preliminary manufacture of the part (cover, insula-
tors, leads, flange, aolder inserts) and subassemblies of the aases (the case base).
In the case of caseless encapsulation the process of manufacturing the case is
matched with the encapsulation process. The s~ructure of the microcircuit turns
out to be included in the shell material and repreaents a united whole with it.
Combination encapsulatioa is done by maaufacturing the cover, in the free space of
which the structure connected to the lead holder ia placed, and then the sealing
eompound is poured over the structure through the free entrance of the cover.
Types of Cases. Standardized cases are used to encap~sulate microcircuits. This
eimplifies the manufacture, it permits mechaniza~tion arid automation of the proceases
of assembly and encapsulation, it lowers the cost of the microcircuits and also
simplifies the construction of the equipment.
The classif ication of cases by external structural appearance ia preaeated irc Table
18-1. The overall and connecting dimensiona of each type of case are strictly
standardized.
Dependirig on the materials used to malce the cases, the latt~r are divided into glasa,
_ ceramic, cermet, metal-glass and plastic. The glass and ceramic cases differ from
metal-glass and cermet cases in that only the leada enter into their structure made
of inetal parts.
- Manuf~cture of the Parts and Subassemblies of the Casea. In the production sections
the following are realized in a defined se4uence (Figure 18-1): entrance control of
the materials, obtaining the billets, stamping th.e parts, obtaining the ~unctions,
application of the galvanic coatin.ga, output control.
The metal parts of the cases are basically made by the method of cold atamping.
Before etamping the strips, tapes, bars, tubes and wire made of kovar, copper,
313
FOR OFF[CIAL USE ~NLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
Table 18-1. Types of cases
~ Form of pro3 ection Location of the projec- Location of the
Type of the case body on tio~.of the leads on leads with respect
the base plane the base plane to the base plane
~ 1 Rectangular Within the limits of Perpendicular
. projection of the case
body
2 Rectangular Beyond the limits of The same
pro~ection of the case
body
3 Circular Within the limits of The same
pro~ ection of the case
_ body with respect to a
_ circle
4 Rectangular Beyond the limits of Parallel
pro~ection of the case
body
( j ) BxonxoN Koerpoas rerepNenos .
~2~ 8erosolesai~s~q yvaczoK
~3~ Yexeeo~serAnsoawll yvac:os
~4) Cx~qt icornaaxreaix aero:osor ~
(5) YwoioK oaees
(6) Pa~~~wxr~eosxp yvtcsoK
(7) ;lonus~tne ~ wxoxeoM roarpoaf
Figure 18-1. Basic production sections for
the manufacture of case parts and subassemblies. -
- Key: 1. input control of materials
2. billeting section
3. mechanical stampin$ section
4. billet make-up storage area
5. junction section
6. galvanizing section
- 7. teating and output control
steel and nickel are laid out, degreased, annealed in hydrogen furnaces and pickled.
The tapes and strips are laid out by roller ahears into billets of the required
sizes. From the obtained billets the parts are made on mechanical stamps, the
operating tool of which is a punch and die. The abtained part is forced through the
314
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICiAL USE ONLY
= opening in the die into the receiver for the finiahed parts. Cold atamping per-
mite manuf acture of parts of complex shape, it is distinguished by high output
capacity and is easily automated. It is poasible to perform the following opera-
tinns by cold stamping: cutting out, trimming, punching holes, d~ ~ing, die
forging, upsetting. Cutting out is a comglete aeparation of part of the material
with respect to a closed outline from the total mass. The less the punch-die
clearance, the less rough the cutout surface and the fewer burrs on it. Trimming
and punching are operations analogous to cutting out. Drawing is an operation af
making hollow parts of closed outline from a two-dimensior~al billet open on one
end (for example, the case cover) . In order to prevent the formation of wrinkles
_ when drawing the billet by the punch into the die, the billet is clamped to the
aurface of the die. The thickness of the billet is maintained only at the center
- of the bottom of the part; the material thins out at the points of transition to
the walls, and at the open edge it gets somewhat thicker. IDie. ~:farging is
used to make flanges and bases for the cases. During volumetric stamping redis-
tribution of the metal takes place with a decrease or preservation of the height
of the billet.
- The cleaning of the parts and removal of burra after cold stamping is done by tumb- ~
ling in drums with a tumbling mix (metal needles with nonreaonance wood shavinga),
chemical pickling or abraeive grinding.
The metal-glass subassemblies which provide reliable electrical insulation are ob-
tained by soldering. Before soldering the glass insulators (beads and tablets)
are pickled, they are washed in running water and dried.
Good adhesion of the glass to the metal parts ia achieved if the machining of the
latter corresponds to roughness class 5-7. Before soldering the metal parts also
go through a preparation cycle: degreasing, pickling, washing, drying, and the
degassing in hydrogen. -
- After mechanized assembly the joined parts in special graphite holders are trans-
ferred to the traveling oven. The approximate time-temperature conditions of
joining the kovar to 549-2 glass are preaented in Figure 18-2. The parts are heated
in section I, the glass ia sof tened and the kovar ia oxid~zed by oxygen added to
the nitrogen flow. In section II at maximum temperature of 900-1000� C the glass
is soldered to the kovar in a nitrogen atmoaphere. In aection III the oxide film
is reduced on kovar surfaces free of glasa in a hydrogen atmosphere. At a tempera-
ture of 575� C, annealing of the obtained ~unction begins to remove the mechanical
stresses (section IV) . Then the holders are cooled to room temperature (section
_ V), and the finished subassemblies are taken out of the holders. During soldering,
stability of the gas regime ia important. Hydrogen ahould not get into the solder-
ing zone (section II), for the reduction of the oxides on the surface nf the metal
parts h~s a negative effect on the quality o� the ~oint. With an oxide ~oint as _
a result of diffusion of the surface of the oxide layer of inetal into the sof tened
glass, an intermediate layer is formed which lowera the mechanical stresses and -
increases the strength of the j unction. It is important to ineure optimal thick-
ness of the oxide, for with small thickness the atrength of the junction ia less,
and with great thickness, the sea1.
The case parts made of vacuum-type ceramic are primarily made by the method of hot
pressure casting of paraffin ceramic masa (slip) in metal molds on special casting .
31.5 -
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
~00 r~ ~ -
800 - ~
i i ~
I
600 , , - - - '
~ - - - -
' 400 - - - - I- i � ~ - -
- I � ; II IQ { IP -y-Y-
200 I ~ i I I
_ t
0 8 12 18 2~ 30 ~6 42 48 54 60 60 ~r~ (a)
Figure 18-2. Time-temperature conditions of joining
kovar to S49-2 glass.
Key: a. minutes
machines. The paraffin is added to lend the ceramic mass plasticity during the
hot pouring process. The pressed ceramic parts are annealed in a bank of adsorbed
powder (alumina). After cleaning off the alumina using a~et of compressed air,
final annealing of the parts takes place to lend the ceramic the required physical-
_ chemical properties. The annealed parts are checked for porosity and correspondence
to the given dimensions, they are ground with diamond discs, washed in hot water,
dried, heated in the muffle furnaces and again checked for the presence of chips,
cracks and dark spots.
For joining the ceramic parts to metal parta, for example, leads with a frame for
- a flat ceramic case, the ceramic i.s metallized using a molybdeaum-~manganese pa~te
or metal foil 30-50 microns thick. The metallization is carried out by brushing on
a paste, spraying from an a3r gun through a free mask, stenciling, cutting out from
foil with subsequent gluing. A layer of nickel or copper 3-5 microns thick is
electrolytically applied for metallization, and then it is burned in. The second =
method of ~oining the metal to ceramic is through a layer of soldering glass which
is applied in the form of a suspension to the 3oined surfaces and then siutered at
a temperature of 400-500� C.
The galvanic coatings of the case parts are needed to obtain smooth surfaces, for
, protection againat corrosion, to insure high quality of their joints during encapsu-
lation. Before the application of the galvanic coatings the parts are carefully -
degreased, pickled and washed. The surface of the parts after picicling is again
quickly covered with oxide; therefore the pickling operation is carried out directly
before putting the parts in the electrolytic bath. The following operations are
performed in the galvanizing section: chemical nickel plating of the kovar parts,
electrolytic nickel plating of copper and steel parts, copper plating of insulator�s,
gold plating of knives, and so on.
18-2. Methoda of Sealing in a Case [Encapsulation]
Methods and Procedures for Encapsulating Microcircuits. The same methods are used
to encapsulate microcircuits as to assemble them: soldering, welding and bonding _
(Figure 18-3). The methods of solderin~ and welding are most widely used, for they
make it possible to obtain vacuum-tight sealed ~oints. Bonding is one of the
316
~"~~lt OFFQ~'[AL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6
FOR OFFICIAL USE ONLY
simplest and most economical methods, but it does not allow for obtaining sealed
joints. In many cases preference is given to soldering when encapsulating.
I
T~s~aW~ uxseoc~t
- ~ xopayo~
; Sery~-nxol~ua ~2~ ~3~ Nerepre:rawa
Ilesennoosex-~ MoreuoKepe- Xe rrvecxue CseK exwe Ilteorraoco~Ye
YOp1~y08l4~ ~toP0.Yao 5 ~P4Yaa so~yoe xopayoe
~
~ '
Ci~Pif? 9 n~wu ~ CKaBND~lYB .
(12 (13 ( 4'~ ( 5 (16) 1 ) ~18) ~ (1 )
~ ~ ~ ~
~ ~ , ~ ~
I ~o ~ � ~ m ' o~ � ~ti
aa
c.~
Figure 18-3. Classification of inethods and meana of encapsulating IC.
_ Key: 1. encapsulation of microcircuits
- 2. vacuum-tight 11. bonding
3. unsealed 12. cold -
4. metal-glass cases 13. resiatance
5. cermet cases 14. ultrasonic
6. ceramic cases 15. argon-arc
7. glass cases 16. laser
8. plastic cases 17. electron beam
9. welding 18. convective in furnaces
10. soldering 19. hot gas ~et
- The structural designa of many of the widely used microcircuit cases with location
of the projection of the leads beyond the limits of the pro~ ection of the case body,
including with parallel arrangement with reapect to the base plane (see Table 18-1),
and with the presence of glass or ceramic insulators directly under the sealing
zone, the application of welding and pressure are made imposaible. In addition,
wide use is made of ceramic cases, the ~oints are made by metallization when seal-
ing. During soldering the metallic, metallized or glass-coated surfacea of the
case parts are connected to the sealing system with the help of solder and flux, -
the role of the flux can be replaced by hydrogen, inert gas or special additives
to the solder. During soldering the entire microcircuit is heated to temperatures
of 200-350� C, and the presence of a flux can have a negative effect on its elec-
trical parametera. The advantages of the aolder include the absencs of signif icant
- compressive forces and apecial tools. Soldering is done by a hot gas ~ et or con-
vective heating of the holder with the assembled parts in f urnaces. The soldering
procedures used for encapsulation in practice do not diff er from the soldering tech-
niques during assembly; we shall consider them in ~ 18-3, in examples of sealing
specific cases. In this section we shall conaider the soldering procedures used
for encapsulation.
317
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
Cold Welding. This is pressure welding without heating realized by joint plastic
deformation of the surfaces of the ~oined parts. Careful preliminary clea.ning of
t~ie parts ie inaufficient to inaure strong metal bonds. A necessary condition of
obtaining a sealed welded ~oint is the presence of an oxide film or galvanic coat-
ing on the juined surfaces having great hardnese and brittleness by comparison
with the material of the parts. In the majority of cases after careful cleaning
and annealing the parts are nickel plated or chrome plated. The nickel or chrome
coatings 3-9 microns thick protect the clean surfaces of the parts reliably before
welding.
When encapsulating microcircuits one-way cold lap welding around the perimeter is
used. The base of the case is placed in a lower hollow punch to the protrusion and
it is covered with the case cover, aligned by means of the upper punch (Figure 18-
4). Under the eff ect of the compressive f orces of the punches the surface f ilm of
nickel or chromium cracks and it is forced out of the welding zone; the base ma-
terial of each part is bared. As a result of bringiag the welded surf aces together
to the distance of effect of interatomic forces, a common electron cloud is formed,
a metal bond arises between the surface atoms. On increasing the pressure, the
bond zone grows, and a strong sealed joint is formed.
ThP force required to execute cold welding depends on many factors and can vary
within broad limits. The pressure mu~t be no less than 3 X 109 N/cm2. The cold
welding conditions usually are deteimined by~:the degree of deformation:
' 1~'/a (18-1)
_ k = 2 2H t _
where 2 H is the total thickness of the welded parts; t is the thickness of the
welded j oint . . _ _
_ P~ ~ ~ 2
3~.
~ P~
Figure 18-4. Diagram of two-way cold welding of a IC case. 1--
upper flat punch; 2-- cover made of sof t material (copper); 3--
base of the case made of harder material (kovar); 4-- lower
trapezoidal punch; P-- compressive force.
Thus, for high-quality welding it is necessary to inaure the following: cleanness
of the surfaces and the presence of hard, brittle, sufficiently thick films of
- nickel, chromium or other f ilm; pre;,iaion assembly of the parts; suf f icient def or-
mation of the welded materials (75-80~6); selection of plastic materials for welding.
The punches are made of hard, high-quality aZloy steele Kh12M, Kh12Fl. The rough-
- ness of the working surfaces of the punch must correapond to classes 9-11. During
operation it is necessary to see that there are no dents, chips or burrs on the
punches. The punches should be wiped with a dry coarse calico regularly.
318
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400014455-6
FOR OFFICIAL USE ONLY
The required degree of deformatioa of the welded parts ia given by the restriction
of the travel of the punches by the ends of a epecial sleeve.
The advantage of cold welding ie the following: abaence of heating, gas releases
and splattering of the metal and also simplicity of the welding equipment. The
defici.encies include an increase in the perimeter of the outaide contour of the ca~e
and significant deformations of the weld-affected zone, diff iculty in reliable
connection of the thin-walled parts, nonunif ormity of plastic deformations with
respect to the perimeter of the rectangular case, the pos~ibility of disturbance
of vacuum tightness of the weld as a result of incomplete fuaion or undercutting
on the part of the sof ter material, a limited selection of materiala both with re-
spect to thicknesa and with respect to properties. It is recommended that the
parts made of 29NK alloy or 47rID alloy with MB copper or M-1 copper be ~oined.
Electrocontact Welding. When encapaulating microcircuita most frequently capacitor
spot or roll welding is used. The welding is done at the time of diacharge of the
capacitor b~nk to the primary winding of the welding transformer. The secondary
winding of the transformer ia connected to the electrodes that play the role of the
welding tool. Sealing by spot welding ia done using electrodes, the shape of
which repeats the shape of the case perimeter (Figure 18-5). The base of the sealed
case is inserted into an opening in the lower electrode, and the cover of the case
- is placed on top. When pressing on the pedal of the welder, the upper ele~~trode
is lowered, and in one current pulae welding is accomplished with respect to the
entire perimeter of the case. A epecial sleeve is used to align the welding rods.
_ . In the intervals between welding, the capacitors are charged from an ac network
through a rectif ier .
_ . ~ ' ~ ~p .
. ~
~ Tv z .
Cema ~
_ B 1 ~ I 3
1 NUBU
~ 4 P .
Figure 18-5. Syetem for apot capacitor welding. 1-- case;
2-- electrodes; 3-- sleeve; 4-- electric network of the
welder; P compreasive force.
The optimal welding conditions are selected by regulating the capacitance of the
capacitor bank, the tranaformation coefficieat of the welding transformer and the
force of compresaion of the electrodes. The quality of the weld depends on the
quality of preparing the ~oined surfaces, the ahape and area of the contact, the
proper choice of the pair of joined materials, cleannesa of the machining of the
, joined surfeces which muat be no less than class 5 and the working aurfaces of the
electrodes which must be no less than clase 9-10.
= The me~hod is widely used to seal smAll,: round metal glaas�hingea. The applica-
tion of the given procedure for sealing largs-perimeter casea has been dela.yed for
some time as a result of absence of powerful capacitor banks. At the present time
319
FOR OFFICIAL USE ONLY
APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400010055-6

APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400014455-6
FOR OFFiCIAL USE ONLY
Soviet industry has developed powerful welders. For example, the ShchYaM 1.124.001
makes it poasible to weld rectangular cases from 55 to 100 mm arauad the perimeter.
Roll welding, in contrast to spot annular welding, is welding by moving electrodea.
The electrodes supplying the current have the shape of rolls, and during welding,
they are rolled over the perimeter of the case at a short distance from each other
on the same aurface. The welding is argon shielded. During welding it is necessary
to see that there are no defects on the electrodes: dents, chips, erosion, and so
on. The electrodes are periodically cleaned to remove contamination. The method
provides for obtaining a tight weld, it is used to seal metal-glass and cermet
rectangular cases up to 75 mm around the perimeter.
Sealing by resistance welding is used to ~oin parts made of 29NK kovar or 47ND alloy
to parts made of NP-2 nickel, 0.8 KP, 10 or Kh18N9T steel. Parts made of nickel and
stainless steel are welded without a coating. The parts made of kovar are
shif ted on the carriage along a guide rail and the cases assembled into holders are .
welded. In the Soviet welders USKM-2, USKM-3, USKM-4, a lock of about 100 cases is
welded simultaneously.
The basic disadvantage of the method is high thermal effect on the welded parts and
the necessity for increasing the welding bead connected with this.
Laser Ion Welding. A new type of welding has become possible as a reault of the
- development of YAG lasers, which in contrast to the ruby and neodymium glass lasers
previously used for spot welding, have higher pulse repetition frequency and radia-
tion power. During pulsed laser welding, a long weld is formed by superposition of
the points on each other with some overlap k= k/ds where R, is the overlap; d is the
diameter of the welded spot. The welding speed is def ined by the .formula
V-fd(1-k)~, (18-2)
where f is the pulse repetition frequency.
L The laser welding of nickel, kovar, stainless steel, titanium and other materials
is done on the "Kvant-12" device which provides aispeed of not