1

Note. This monograph about IPRS Băneasa is the extended standalone online version of the

Chapter 3 about IPRS Băneasa which will be included in the book „The Romanian School of

Micro- and Nanoelectronics”, intended to be published by Editura Academiei

IPRS Băneasa

Silicon Technology: Industrial Research and Development

Authors, in alphabetical order:

Viorel Banu, Radu Coteț, Petru Dan, Tudor Dunca, Virgil Gheorghiu, Eugen Popa,

Adrian Veron, Andreas Wild

Memento. The authors want to pay tribute to all

their colleagues who passed away. This

presentation is also meant as a memorial to

their contribution

Contents

List of Abbreviations for Romanian Institutions and Forums .................................................... 4

1. Introduction ............................................................................................................................ 6

1.1 The Decision ..................................................................................................................... 6

1.2 Setup and Milestones ........................................................................................................ 7

2. From Production to Science ................................................................................................... 9

3. Section 2300: The Power Semiconductor Devices Factory ................................................. 12

3.1. Introduction ................................................................................................................... 12

3.2. Historical Perspective .................................................................................................... 12

3.3. Challenges in Power Semiconductors ........................................................................... 14

3.3.1 The Mesa Concept ................................................................................................... 15

3.3.2. The “Core Compromise” ........................................................................................ 16

3.3.3. The Guard Rings ..................................................................................................... 16

3.4. Technologies and Products ............................................................................................ 17

3.4.1. Low and medium power diodes, thyristors and triacs ............................................ 17

3.4.2. High Power Diodes and Thyristors ......................................................................... 18

3.4.3. Specialty Technologies ........................................................................................... 18

3.5. People and their areas of responsibility ......................................................................... 19

3.5.1. Management ............................................................................................................ 19

3.5.2. Product Families ..................................................................................................... 19

3.5.3. Wafer and Chips Processing ................................................................................... 20

2

3.5.4. Assembly and Chemical Protection ........................................................................ 21

3.5.5. Rectifier Bridges and Modules with Diodes and Thyristors ................................... 22

3.5.6. Test, Quality and Reliability Assessment ............................................................... 22

3.6. Partners ...................................................................................................................... 23

3.7. Scientific Contributions ................................................................................................. 24

3.8. Epilogue ......................................................................................................................... 24

3.9. Acknowledgements and Notes About the Authors ........................................................ 25

4. Section 2400 - The Integrated Circuits Factory ................................................................... 26

4.1. Introduction: Integrated Circuits in the 1960s ............................................................... 26

4.2. Starting Integrated Circuits in Romania ........................................................................ 28

4.2.1. Acquiring a License ................................................................................................ 28

4.2.2. Building the Facility ............................................................................................... 29

4.2.3. People ...................................................................................................................... 30

4.2.4. Functioning in the IPRS Framework ...................................................................... 31

4.2.5. Partners ................................................................................................................... 32

4.3. Advancing Silicon Processing Technology (Front End Manufacturing) ...................... 32

4.3.1. Wafer size ............................................................................................................... 33

4.3.2. Mask Making .......................................................................................................... 33

4.3.3. Photolithography ..................................................................................................... 33

4.3.4. Diffusion and Ion Implantation ............................................................................... 34

4.3.5. Metallization ........................................................................................................... 35

4.3.6. Passivation .............................................................................................................. 35

4.3.7. Wafer Probe ............................................................................................................ 35

4.4. Assembly and Test Technology (Back End Manufacturing) ........................................ 35

4.4.1. Assembly ................................................................................................................. 36

4.4.2. Electric Test ............................................................................................................ 37

4.4.3 Reliability and Quality Assurance ........................................................................... 37

4.5. Product Portfolio Expansion .......................................................................................... 37

4.5.1. Digital Integrated Circuits ....................................................................................... 38

4.5.2. General Purpose and Industrial Linear ICs ............................................................. 41

4.5.3. Radio and TV ICs ................................................................................................... 43

4.5.4. Products summary ................................................................................................... 45

4.6. Conclusions and Evolution after 1989 ........................................................................... 45

4.7. Acknowledgements and Notes About the Authors ........................................................ 45

5. Section 2500: The Silicon Transistors and Small-Signal Diodes Factory ........................... 47

5.1. Introduction ................................................................................................................... 47

5.1.1. A brief history ......................................................................................................... 47

3

5.1.2. People ...................................................................................................................... 49

5.2. Technology .................................................................................................................... 51

5.2.1 Wafers ...................................................................................................................... 51

5.2.2 Epitaxy ..................................................................................................................... 51

5.2.3 Thermal oxidation .................................................................................................... 51

5.2.4 Diffusion .................................................................................................................. 52

5.2.5 Ion implantation ....................................................................................................... 53

5.2.6 Photolithography and etching .................................................................................. 53

2.5.7 Chemical vapor deposition....................................................................................... 54

5.2.8 Passivation ............................................................................................................... 55

5.2.9. Gettering ................................................................................................................. 55

5.2.10. Wafer thinning and metallization .......................................................................... 56

5.2.11. Process characterization ........................................................................................ 57

5.2.12. Wafer probe and dicing ......................................................................................... 58

5.2.13. Packaging and final testing ................................................................................... 59

5.3. Product Development .................................................................................................... 61

5.3.1 Diodes ...................................................................................................................... 61

5.3.2 Small and Medium Power Transistors ..................................................................... 63

5.3.3 Power and high voltage transistors .......................................................................... 65

5.3.4 Optoelectronic Devices ............................................................................................ 67

5.3.6 Integrated Circuits .................................................................................................... 68

5.4. Acknowledgments and Notes About the Authors ......................................................... 69

6. Confessions of a Former IPRS Băneasa General Manager .................................................. 70

7. The Twilight of IPRS Băneasa ............................................................................................. 71

7.1. December 1989 .............................................................................................................. 71

7.2. Băneasa SA .................................................................................................................... 71

7.3. The Brain Drain ............................................................................................................. 73

7.4. The Aborted Privatization ............................................................................................. 73

7.5. A Survivor ..................................................................................................................... 74

8. What Happened Since .......................................................................................................... 74

9. References: Published Works Authored or Co-authored by People from IPRS Băneasa .... 77

Figures ...................................................................................................................................... 86

4

List of Abbreviations for Romanian Institutions and Forums

CAS

Conferința anuală de semiconductoare, organizată de ICCE

Annual Semiconductor Conference, organized by ICCE

CCPCE

Centrul de cercetareși proiectare pentru componente electronice, ulterior ICCE

Centre for Research and Design of Electronic Components, later renamed ICCE

DCE

(Catedra de) dispozitive și circuite electronice din Facultatea de electronică și telecomunicații

din IPB, devenită mai târziu Catedra de tehnologie electronică și microelectronică

(Chair for) Electronic Devices and Circuits, in the Faculty of Electronics and

Telecommunications of IPB, which later became Chair of Electronics Technology and

Microelectronics

DCAE

(Catedra de) dispozitive, circuite și arhitecturi electronice din Facultatea de electronică și

telecomunicații și tehnologia informației, din UPB

(Chair for) Electronic Devices, Circuits and Architectures, in the Faculty of Electronics,

Telecommunications and Information Technology of UPB

ICCE

Institutul de cercetări pentru componente electronice, București, anterior CCPCE

Research Institute for Electronic Components, Bucharest, formerly named CCPCE

ICPE

Institutul de cercctare și proiectare pentru electrotehnică, București

Research and Design Institute for Electrotechnics of Bucharest

ICPMS

Institutul de cercetare și producție pentru materiale semiconductoare, București

Institute for Research and Production of Semiconductor Materials, Bucharest

ICRET

Întreprinderea de construcții și reparații pentru echipamente de telecomunicații, București

Enterprise for Construction and Repairs of Telecommunications Equipment of Bucharest

IFA

Institutul de fizică atomică București

Institute of Atomic Physics of Bucharest

IFIN

Institutul pentru fizică și inginerie nucleară, București

Institute for Nuclear Physics and Engineering of Bucharest

IFTAR

Institutul pentru fizica și tehnologia aparatelor cu radiații, București

Institute for Physics and Technology of Radiation Appliances of Bucharest

IMT

Institutul național de microtehnologie, București

National R&D Institute for Microtechnology of Bucharest

IPB

Institutul politehnic București, ulterior UPB

Polytechnic Institute of Bucharest, later UPB

IPEE

Întreprinderea de produse electronice Electroargeș, Curtea de Argeș

Enterprise for Passive Electronic Components of Curtea de Argeș

IPRS Băneasa

Întreprinderea de piese radio și semiconductoare Băneasa, București

Enterprise for Radio Components and Semiconductors Băneasa, Bucharest

5

OSIM

Oficiul de stat pentru invenții și mărci

State Office for Inventions and Trademarks of Romania

UPB

Universitatea POLITEHNICA București, anterior IPB

University POLITEHNICA of Bucharest, formerly named UPB

6

IPRS Băneasa

Silicon Technology: Industrial Research and Development

1. Introduction (Andreas Wild, Petru Dan)

The micro- and nanoelectronics emerged and progressed to a large extent through industrial

research. Essential discoveries and inventions happened in industrial laboratories: this is the

case for the transistors (Bell Telephone Laboratories), the thyristor (RCA), the integrated circuit

(Fairchild Semiconductor, Texas Instruments), the microprocessor (Intel), the dynamic random

access memory (IBM) etc. Moore’s Law was formulated by an industrial manager (Intel), and

the scaling theory that opened the way towards the nanoelectronics era was developed in an

industrial company (IBM). Many of these ideas have been considered scientific contributions

recognized with the Nobel Prize. Naturally, Romanian industrial research, in particular the

activity within the company IPRS Băneasa over about 45 years, was an essential element of

progress.

1.1 The Decision

Up to the end of 1989, the communist State in Romania had a monopoly on entrepreneurship:

companies were started only by decisions of the Council of Ministers; the only source for

investments was the State budget; only State Committees could decide upon the manufacturing

volume and the price for every single product; and roadmaps had to be derived from the five-

year plans issued by the Communist Party. The overall objective of the centralized planned

economy of Romania was to make the country self-sufficient by the autarchic development of

the industries.1 In a closed non-competitive market it was not the profitable growth that

mattered, but growth alone, the industry was striving to make quality products available from

domestic suppliers, and sometimes also... taking pride in achieving world class technical

performances. Hampered by a non-convertible currency under tight governmental control,

financing imports was always an uphill battle and import avoidance was an overarching priority.

Communist Romania invested in an electronic industry that by 1989 had 33 units, both R&D

institutes and industrial companies, supported by five University centres and vocational

education. Each company was assigned a market segment, without overlapping: there was no

need for competition in a planned economy.2 Some prominent personalities of those times

recognized the need for a domestic supply of electronic components and actively contributed

to establishing this capability: first and foremost Acad. Prof. Mihai Drăgănescu, the founder

of the Romanian school of microelectronics, Prof. Stere Roman active both in the University

and in the industry, as well as the open minded communist minister Gaston Marin.

The Enterprise for Radio Components and Semiconductors of Bucharest, known as IPRS

Băneasa, was established by the Decision of the Council of Ministers (HCM) Nr. 438 from 12

Mai 1962 intended to create a domestic supply of components for the Romanian electronic

industry, so it could avoid imports. IPRS Băneasa became the main manufacturer of electronic

components in Romania, a major supplier for all industries: consumer electronics, computers,

1 Lucian Boia – „România țara de frontieră a Europei” („Romania the Frontier Country of Europe”), Editura

Humanitas, ediția a 6-a (2016)

2 Andreas Wild – „Rise and Fall of the Romanian State-owned Micro/nanoelectronics”, The 45th ICOHTEC

Symposium, Saint-Étienne, France, 17 – 21 July 2018

7

automation, electrotechnics, electric machines and drives, communications, automotive,

constructions, chemical industry, metallurgy, oil drilling, railway and naval transportation,

agriculture, health etc. This overextended spectrum resulted in an ineffective spreading of

resources, diluting the economic efficiency. The need for imported materials specific for

semiconductor technologies permanently forced the specialists to lower consumption through

novel design ideas, technological innovations, and yield improvements, while using scarce

resources to compete, in a sort of remote race, with performances achieved by the best world-

wide producers. IPRS successfully exported up to 25% of its production; this justified in front

of the State committees its permanent requests for convertible currency.

1.2 Setup and Milestones

In June 1962 began the getting started phase: the enterprise ELECTRONICA seeded the new

company by transferring some of its production lines, enabling IPRS Băneasa to start with two

departments: Section 2200 (Semiconductor Germanium Devices – diodes and transistors,

produced under a license from the French company THOMSON-CSF) and Section 2700

(Passive Components – capacitors, resistors, printed circuit boards). IPRS Băneasa added

technical support functions: Chief Mechanical Workshop, Technical Office, Investment

Department, and Quality Control Department. R&D activities started in 1965. Two years later,

Section 2100 started building in-house tools and equipment items, and in 1969 an internal

Tooling Department was created, and a Psychology Laboratory was added.

A growth phase followed. In the next five years development was explosive. A “Centre for

Research and Design of Electronic Components” (Romanian abbreviation: CCPCE) was

established to perform in-house R&D (1969). Three new sections have been started: Section

2300 on a license from SILEC SEMICONDUCTEURS, France, for silicon diodes and

thyristors (1969); Section 2400 on a license from THOMSON-CSF, France, for bipolar

integrated circuits (1970); and Section 2500 on a license from ITT-INTERMETAL, Germany,

for silicon diodes and transistors (1972). The activities have been focussed on silicon

technology, Section 2700 has been spun off as a new entity, the enterprise specialized in passive

electronic components IPEE Curtea de Argeș manufacturing resistors, capacitors, and

thermistors; later, the printed circuit boards (PCB) production has been moved to another

company, while the germanium products have been discontinued.

In 1975, CCPCE was separated from IPRS Băneasa as an independent “Research Institute for

Electronic Components” (Romanian abbreviation: ICCE), but IPRS Băneasa continued

performing R&D, generating a continuous, sustained flux of innovation, thanks to a surprisingly

modern concept introduced in 1972: „The Integrated Section for Design, Research,

Development, Investments and Production”, a structure anticipating the modern lab-fabs.3,4

IPRS Băneasa was much more vertically integrated that it is customary today. It produced fluids

(hydrogen, oxygen, nitrogen); pumped water from deep wells and deionized it; pulled and

purified first germanium ingots, then silicon 38 mm ingots, then sliced and polished raw wafers

out of them; and it built in-house tools and machinery.

3 Anton Vătășescu – „IPRS Băneasa – 25 de ani de activitate” („IPRS Băneasa – 25 Years of Activity”),

Electrotehnica, Electronica și Automatica EEA, vol. 31 (1987).

4 Doina Didiv – „A Life Devoted to the Electronic Components Branded IPRS Băneasa”, interview published in „Școala românească de micro- și nanoelectronică” („The Romanian School of Micro- and Nanoelectronics”),

Editura Academei Române, București, 2018, p. 59

8

The talent supply has been strengthened after two turning points. First, a specialised section for

“Electronic Components and Devices” was created at the Polytechnic Institute Bucharest

(Romanian abbreviation IPB, as will be referred to from now on), later renamed University

POLITEHNICA of Bucharest (abbreviated UPB). As of 1969 the section curricula were

changed from nuclear engineering to microelectronics. Second, in 1974 a Government decision

forbade graduates to join Universities or R&D institutions. For several years, this put IPRS

Băneasa, a company offering high-tech jobs, in the privileged position to be able to recruit the

best graduates, with appropriate specialized education, and keep them for at least three years,

becoming a top-notch community of professionals.

The context of the planned economy controlled by State Committees and Ministries made

unnecessary a Go-To-Market strategy at the company level. IPRS Băneasa used almost

exclusively the so called “Zero-level Channel” approach, in which the producer delivers

directly to the customer (note that in the planned economy, with non-overlapping specialisation

of the companies, it was not uncomment that there was only one customer for a family of

components). On the positive side, the technologists were directly exposed to customer

demands and could rely on first-hand requirements when engaging in R&D programmes.

Although on a closed market, IPRS Băneasa strived to make its products and capabilities widely

known and increase demand. It printed several Catalogues between 1976 and 1989; published

the technical bulletin “BETA” that had 19 issues between 1979 and 1989 containing more than

110 technical papers and application notes; it even helped the “Alexandru Sahia” Studios

produce two documentaries intended for the public at large entitled “The Transistor” and “The

Integrated Circuit”.5

IPRS Băneasa could not engage in CMOS technology that was reserved for a different factory,

MICROELECTRONICA, established in the 1980s. The structure comprising three fairly

autonomous integrated sections (2300, 2400, and 2500), each one responsible for a product line,

relying on corporate support functions - was operational until 1989, when the world entered a

new era. After 1989 started the last phase of IPRS Băneasa: the twilight.

The general managers of IPRS Băneasa in the three main phases of its existence have been:

- Phase 1: Getting started (1962-1968)

Mihai Alexe (for a couple of months in the beginning)

Mihai Oncescu

Grigore Danciu

Ion Chicoș

Lazăr Șandra

Nicolae Cocoș (for about one year)

- Phase 2: Growth and Maturity (1969-1989)

Lazăr Șandra, an engineer coming from the defence industry: a severe, goal oriented but fair general manager. Deputy technical manager: engineer Anton Vătășescu, till 08

Dec. 1979 (Fig. 1).

Anton Vătășescu, (from Dec. 1979 till April 1989), an open minded, highly cultivated, „Western-style” character, a bright professional, excellent manager and great leader.

5 Script by Prof. M. Bodea (IPB) and Andreas Wild (S2400), directed by Alexandru Sîrbu (Sahia Studios)

9

Deputy technical manager: physicist Doina Didiv (former manager of Section 2300).

Deputy commercial manager: engineer Mihai Petrescu (Fig. 2). Anton Vătășescu was

demoted to Section manager because the communist party leaders did not accept

anymore his too liberal behaviour.

Doina Didiv (between 1989 and 1992), a very empathic and supportive leader, always confident that entrusting challenging responsibilities to professionals will be paid off by

their spectacular results. Deputy technical manager: engineer Gheorghe Florea

(formerly acting as manager of Section 2300). Doina Didiv retired in 1992.

- Phase 3: Twilight

Gheorghe Florea (between 1992 and 1999) having three successive technical deputies: Radu Râpeanu, Marian Niculae and Adrian Oanea

Vlad Ciulei and Eugen Popa (co-managers between February 1999 and November 1999)

Gheorghe Florea (November 1999 – March 2002)

Eugen Popa (since 2002 till 2008)

Mihai Stan (for a few months in 2004)

Eugen Popa (till 2008)

Gheorghe Lifu, with Eugen Popa as technical deputy.

2. From Production to Science (Andreas Wild, Petru Dan)

IPRS Băneasa demonstrated over the years of its existence a breadth and depth of technical

achievement unparalleled in the Romanian industry.

IPRS Băneasa collaborated closely with the specialty Chairs in the Faculty of Electronics,

Telecommunications and Information Technology of IPB/UPB, first DCE, then Chair of

Electronic Technology and Microelectronics, now DCAE..

Teaching: the cursus “Theory and Design of Semiconductor Devices and Integrated Circuit” was delivered, from 1973 to 1980, by Constantin Bulucea, the head of ICCE (linear

ICs), and Anton Vătășescu, Technical Manager of IPRS Băneasa (digital ICs); Petru Dan of

IPRS was an invited lecturer for twenty years. There were delegated Diploma thesis co-advisors

(from IPRS Băneasa, Virgil Gheorghiu, Alexandru Hartular, Nicolae Marinescu, Andreas

Wild, Sorin Negru etc) and assistant professors (Andrei Vais and Horia Profeta of ICCE,

Andreas Wild and Rodica Savin from IPRS Băneasa etc.). IPRS Băneasa specialists helped

translating in Romanian reference books and co-authored up-to-date textbooks for the students.

Reciprocally, IPRS Băneasa technologists could pursue a PhD degree while keeping their

regular responsibilities; by 1987, there were four PhD and 12 doctoral students.

Researching: numerous examples will become visible in the description of the work done in all Sections (see following paragraphs). IPRS Băneasa technologists co-authored with

University professors, besides a considerable number of papers and patents, ten volumes

10

published by the Technical Printing House and the Romanian Academy Printing House6

[References 1-7, Books].

The research activities in the sections benefitted from the merging of manufacturing and

research in the „lab-fab” setup; this was one of the main recipes for success: most technical

staff had a double duty, being in charge both of overseeing a workstation in manufacturing and

of developing new products. While managing operators, debugging equipment and fixing

process flaws detracted time from doing research, it also gave access to a solid experimental

infrastructure and eliminated the lengthy and risky technology/product transfer from a research

to an industrial environment. Often, technologists were allowed to take initiatives and launch

R&D programmes. This freedom and the professional passion to obtain significant results was

a real incentive (not a pecuniary one).

A memory of Petru Dan, a co-author. Immediately after joining IPRS, he was presented with

the challenging task to develop a TV scanning device using hybrid integration of a fast thyristor

and a diode. Encouraged to strive for the utmost performance, he proudly presented the results,

only to be told by the customer that, although technically superior, the product was

economically not sustainable. The management did not blame him but was rather appreciative

of the learning effect for the organization. This was typical for the stimulating atmosphere in

IPRS! The young engineer learned the lesson and strived throughout his career for products

both competitive and profitable, as well as for a truly professional working atmosphere,

stimulating the creativity and commitment of the team.

All sections engaged in a rather close cooperation with external research organizations. Besides

DCAE in Academia, numerous collaborative efforts have taken place with research institutes

in semiconductors and related fields, including first and foremost CCPCE/ICCE, but also the

Institute of Atomic Physics (IFA), the Institute for Nuclear Physics and Engineering (IFIN), the

Institute for Research and Production of Semiconductor Materials (ICPMS), the Research and

Design Institute for Electrotechnics (ICPE) and other entities. After 1989, the National R&D

Institute for Microtechnology (IMT) became a major collaboration partner of IPRS Băneasa.

IPRS Băneasa organised a few editions of its own Scientific Session, but the management also

supported publications and participations in conferences such as the Annual Semiconductor

Conference CAS organized by ICCE (international conferences were hardly accessible, given

the severe travel limitations). Articles with authors or co-authors from IPRS have been accepted

6 6.1 Vătășescu, M. Ciobanu, T. Cârcu, I. Rateș, V. Gheorghiu – „Dispozitive semiconductoare – manual de

utilizare” („Semiconductor Devices – Application Handbook”), Editura Tehnică, București, 1975; 6.2 M. Bodea., A. Vătășescu, (Ed.) – „Circuite integrate liniare. Manual de utilizare” („Linear Integrated Circuits.

Application Handbook”) Vol.1 (1979), Vol. 2 (1980), vol. 3 (1984), Vol. 4 (1985), Editura Tehnică, București;

6.3 R. Râpeanu, L. Sârbu – „30 de aplicații ale circuitului integrat βU 1011” („30 Applications of the Integrated

Circuit βU 1011”), Editura Tehnică, București, 1985;

6.4 N. Marinescu – „Radioreceptoare cu circuite integrate” („Radioreceivers with Integrated Circuits”), Editura

Tehnică, București, 1985;

6.5 Dan Dascălu, Gheorghe Brezeanu, Petru Al. Dan – „Contactul metal-semiconductor în microelectronică” („The

Metal-Semiconductor Contact in Microelectronics”), Editura Academiei RSR, 1985;

6.6 Mircea Bodea, Petru Al. Dan, Nicolae Iosif, Andrei Silard, Gheorghe Brezeanu, Eugen Popa, Marian Udrea-

Spenea – „Diode și tiristoare - 1. Performanțe” („Diodes and Tyristors - 1. Performances”), Editura Tehnică,

București, 1989;

6.7 Mircea Bodea, Ioan Teodorescu, Radu Dragomir, Andrei Silard, Sorin Negru, Eugen Popa, Petru Al. Dan,

Marian Udrea-Spenea – „Diode și tiristoare - 2. Aplicații” („Diodes and Tyristors - 2. Applications”), Editura

Tehnică, București, 1990.

11

by prestigious publications, both national like Revue Roumaine de Physique, Revue Roumaine

des Sciences Techniques, Electrotehnica-Electronica-Automatica (EEA), Automatica-

Management-Computers (AMC) and international ones like IEEE Transaction on Electron

Devices, IEEE Journal of Solid-State Circuits, Electron Device Letters, Solid-State Electronics,

Applied Physics Letters, Journal of Electrochemical Society, Microelectronics and Reliability.

The company paid the fees for those who became members of IEEE technical societies. To stay

current with the newest development in the world, Section 2500 reimbursed to cost of mail

orders placed with the library of the Institute of Atomic Physics for western publications

(exceedingly difficult to get in Romania).

It is not surprising that, over time, IPRS employees generated a considerable amount of

innovative solutions. By 1987, IPRS had submitted more than 100 patent proposals to the State

Office for Inventions and Trademarks (OSIM), out of which more than 50 were already issued.

Production, research and education cohabited in IPRS Băneasa. The site located in a forest

was nicknamed, with modesty and also pride, the „Silicon Forest of Băneasa”. The Sections

enjoyed a lot of autonomy, but everybody was working to ensure the success of IPRS Băneasa.

Without commercial competitors, collaboration prevailed. It was normal that a section would

help another one „cross border” (or „over-the-fence”) to execute its projects, would share its

technological infrastructure and even its know-how and specialists, to overcome budgetary

shortages and resource scarcity7. This spirit was present in the collaboration with the University

and the external entities: ideas circulated freely, numerous publications and even patents had

authors from different organizations, anticipating in a way the modern “open innovation”

concept.

It was a living, self-standing eco-system that is pertinent even today. Scaled up to the national

level, it underpins contemporary efforts to define in Romania a strategy for the future re-

engagement in micro/nanoelectronics, an essential element in the broader concept of Cyber

Physical Systems.8,9

7 R. Râpeanu, N. Marinescu, S. Negru, S. Puchianu, P. A. Dan, F. Țurțudău, G. Mânduțeanu, T. Dunca, S.

Georgescu, D. Sdrulla – CAS Proceedings, 12 (1989). Re. feasibility of unitary silicon processing workflow.

8 Dan Dascălu – „Transformarea digitală – o regândire a perspectivei” („Digital Transformation – Rethinking the

Perspective”), Market Watch, April 2020

9 Andreas Wild – „Sistemele ciber-fizice – o oportunitate pentru România („Cyber-physical Systems – an

Opportunity for Romania”), Academica, April 2020

12

3. Section 2300: The Power Semiconductor Devices Factory (Petru Dan, Eugen Popa, Viorel Banu)

3.1. Introduction

This is a brief retrospective review of the history, profile, people, growth and fall of the Section

2300 (Silicon Diodes and Thyristors), which later became the Factory for Power Semiconductor

Devices. It is an overview, sometimes from a personal angle, about the unit which in 1969

opened the adventure of silicon microelectronics in IPRS Băneasa, then grew and became an

outstanding success story, finally survived the longest during the sad decline of the enterprise

renamed in 1990 Băneasa SA, and eventually originated the only surviving entity, SC Băneasa

Silicon SRL, one of the top Small and Medium-Sized Enterprises in Romania today.

3.2. Historical Perspective

The power semiconductors evolved from a few seminal inventions. The first semiconductor

diode with germanium was created in 1952 by R. N. Hall of General Electrics, and its

counterpart in silicon was invented by R. Ohl at Bell Laboratories. The p-n-p-n structure, known

as Silicon Controlled Rectifier (SCR) or Thyristor, was proposed by W. Shockley in 1950 as

he was at the Bell Laboratories, was theoretical described by J. L. Moll of General Electrics in

195610, demonstrated experimentally in 1957 and proposed commercially as promoted by F.W.

Gutzwiller of General Electrics in 1958. The “Gate Turn Off” (GTO) thyristor was created by

F.W. Gutzwiller in 1963.

In the world, the established manufacturers of power semiconductor devices were often

vertically integrated electrotechnics companies or were located close to the electrotechnics

industry that was their main market; this is the case of General Electric and Westinghouse in

U.S.A., AEG and Siemens in Germany, Brown-Boveri in Switzerland, but also the case of the

Eastern Block manufacturers like CKD in Czechoslovakia or the factories in the Soviet Union.

In 1970, on the IPRS Băneasa campus, next to the manufacturing lines for passive components

(Building “A”) and germanium transistors (Building “B”), a new facility was erected: Building

“C”, with a new architecture, without any windows. It had four big rooms: one intended for the

newly established CCPCE, one for an assembly line for small signal silicon transistors and two

for a new section to manufacture silicon diodes (later on, the product lines included rectifier

and controlled avalanche diodes, thyristors, triacs, rectifier bridges and modules). It was rather

unusual at that point in time to co-locate power devices with the small signal ones, as it was the

case in IPRS.

Section 2300 was started in 1970 with a French license form SILEC SEMICONDUCTEURS

for manufacturing low and medium power diodes in mesa technology, soldered on metal cases

or connectors and passivated with organic materials. This was followed by the development of

high power diodes in mesa technology, consisting of sandwiches of silicon diode structures

attached to aluminium on molybdenum disks (to improve rigidity, the latter material exhibiting

almost the same thermal expansion coefficient as silicon); they were passivated with an organic

resin, then soldered on a copper base and copper connector. The Section manager, the physicist

Mrs Doina Didiv and her deputy the engineer Nicolae Iosif (Fig. 2 and 3), who led the project

for establishing the section, as well as for acquiring and implementing the license, have been

trained at the facilities of SILEC SEMICONDUCTEURS as part of the license agreement.

10 Moll, J.; Tanenbaum, M.; Goldey, J.; Holonyak, N. – "P-N-P-N Transistor Switches". Proceedings of the IRE.

44 (9) (September 1956), pp: 1174–1182.

13

The first years were dedicated to learning and accumulating experience, but in 1974 the Diodes

and Thyristors Section was already the fast growing „blue chip” of IPRS, so that when the

influx of top-notch graduates, following the Government new regulation, came into IPRS, the

lion‘s share was assigned to the Section 2300. Four graduates from the Electronic Components

and Devices specialty started their careers here: engineers George Mânduțeanu, Marian

Niculae, Eugen Popa and Petru Dan (the valedictorian) were assigned to develop the lines of

low, medium and high power devices. Two graduates from the Applied Electronics specialty,

Adrian Niculiu and Alexandru Zamfirescu, were chartered to design and build electronic test

equipment, as well as perform electronic maintenance. A seventh newcomer was physicist

Florian Țurțudău, graduated from the Physics Department of the Imperial College in London.

Their roles and contributions will be shown in the next paragraphs.

A memory of Petru Dan, a co-author of this chapter. In the beginning of our activity in IPRS,

in September 1974, we were first introduced to the management team of IPRS by engineer Anton

Vătășescu, the technical manager of IPRS at that time. When joining the Section 2300, we had

the great chance to meet there a team of remarkable specialists with various backgrounds and

benefitting of a solid experience accumulated since working in IPRS. We were welcomed by the

Section manager, the physicist Mrs Doina Didiv and her deputy the engineer Nicolae Iosif.

Both were excellent specialists in the field of silicon mesa diodes. They were the ones who

guided us throughout the field of power silicon devices as well as through the „mystery” of

managing related production processes and leading teams of graduates and operators.

The period 1975-1980 is considered as definitory for the personality developed by Section

2300. The staff doubled, the output tripled, and the products started being exported.

Considerable investments brought the equipment set up-to-date, and by 1980 the in-house

research produced a spectacular growth of the product portfolio reaching 19 families of diodes

(normal and fast, Zener, controlled avalanche and suppressor types), 6 families of thyristor and

triacs and a broad range of rectifier bridges. Section 2300 was leading the company in the

number of issued patents.

In 1980, the product range was enhanced with high power diodes and thyristors, in normal and

fast versions, with pressed contacts and flat base, under a German license from AEG, which

won the licensing offers competition with Silec Semiconducteurs (France), Westcode (United

Kingdom), Brown-Bovery (Switzerland). The manufacturing license was granted only for four

thyristor types (two normal and two fast) in four package (housing) versions. This sparked

another creativity explosion, generating a large range of new high-power products developed

in-house, introducing very high-power diodes, thyristors and gate turn-off thyristors (GTO),

entering the markets for railway and machinery driving applications. The power devices were

packaged in flat base and in stud base packages, as well as in disc packages for double side

cooling both in ceramic and plastic housing. The plastic housing for double side cooling

responded to the customers requirement for low-cost products used in common applications.

Finally, power modules including two devices with insulated bases have simplified a lot the

new customers applications.

In order to increase the economic efficiency at customers, as well as to enable the optimal sizing

and correct assembling for the required practical purpose, Section 2300 started manufacturing

some new products with high added value, based on high-power rectifiers and thyristors. A

specialized application workshop was created to ensure both the mounting of power devices on

suitable heat sinks (according to their nominal power) and building up rectifier assemblies or

high-power modules. In addition, the option was offered to include the control circuits on

request. These prototypes workshop was later moved from Section 2300 to Section 2200 that

was transformed into an applications and prototypes section assisting the users of IPRS Băneasa

components.

14

The portfolio diversification of high-power semiconductor devices was almost immediately

followed by a national increase in high-power product demand. Imports of high-power

semiconductors from COMECON countries (Czechoslovakia, Poland, USSR) were soon

abandoned. The pressure on the high-power division was huge, given the limited production

capacity of the manufacturing line delivered under the AEG license. There was no legal

limitation in expanding the AEG technology to other similar products, but the license was only

for the domestic market.

Unfortunately, the Romanian state policy almost completely banned new imports of machinery

and equipment, which were necessary to increase the capacity. The only chance was to utilize

the local expertise and production resources. In this respect the collaboration with other

enterprises or research institutes became essential. It was a tremendous effort made by a small

team which, apart from the current production tasks, had to ensure the assimilation of materials,

technologies and new equipment aimed to increase not only the production capacity but also

the product quality. For instance, shortly after production began, a laser machine was needed

to cut dice from the silicon wafers. Because it was forbidden to acquire imported lasers, a

research team from the Lasers department of IFTAR was identified and accepted to collaborate

in the design and construction of a YAG:Nd laser for this purpose. This was the first laser built

in Romania for the circular cutting of silicon wafers in industrial regime, which enabled to

extend the diameter range of silicon wafers up to 3 inches. It has also been applied to cutting

ceramic tiles used as internal electrical insulators for power modules.

The maximum production capacity reached over 250,000 high-power devices per year, more

than three times the licensed capacity.

A crucial point for increasing the economic efficiency was the spare parts recovery and

materials saving. Metals such as gold, silver, molybdenum, copper or even ceramic housings

were the main target of this policy. The two electron beam devices that were built in cooperation

with a section of Romanian Academy in Cluj-Napoca allowed Section 2300 to completely

replace the use of gold for pellet fabrication and to recover the molybdenum discs from the

failed devices. It is worth mentioning that the gold-plated molybdenum disc was one of the

most expensive parts in the fabrication of the pellet devices (diodes or thyristors).

A memory from Eugen Popa, a co-author of this chapter. The last major development phase of

Section 2300 took place in 1980-1981 when the license for high power diodes and thyristors

from AEG Germany was acquired and put into operation. I was part of this project and …it

was also the moment when my career took a different turn. I should have been part of the license

implementation team but, as I did not have a “clean” political file from the viewpoint of the

communist rulers, I was not eligible. So I embarked into the new task as head of the team for

the design and manufacturing of the rectifier bridges for Oltcit cars: new challenges and

opportunities that helped heal the wound of not being sent to the training stage in Germany.

The level achieved by Section 2300 in the early eighties allowed it to navigate the much more

challenging period that followed, when the political priorities resulted in massive reductions

both in the capital investments and in the access to imported materials and piece parts.

3.3. Challenges in Power Semiconductors

The power semiconductors have a few essential characteristics that differentiate them from the

rest of the industry. Among them:

15

- While many products strive for miniaturization, the power devices are getting bigger and bigger, to be able to handle increasing amounts of energy.

- The electronic stress, in terms of levels of injection in the conductive state, respectively size of the depleted regions and the electric field intensity across them in the blocking

state impose unparalleled levels of purity for the silicon crystal that must be free of

defects and maintain its quality throughout all manufacturing steps.

- The intense thermal and mechanical stress they must sustain in silicon, in the package and in the whole built of the application system is without comparison.

- The unit cost and therefore the product price are determined primarily by the high material costs in manufacturing that cannot be eroded as for miniaturized electron

devices.

It is useful to start with a brief overview of the basics of the technologies used for power silicon

devices in Section 2300.

The manufactured devices covered a large range of power values: forward currents between 1-

5000 A, reverse blocking voltages up to 4000 V (the most performant being a diode of 5000 A

/ 200 V for welding resistive steel pipes and a diode of 4000 A / 5000 V for a special order, in

the biggest case used in IPRS at that point in time, T70). It is important to note that the

designations high-, respectively low-power are context dependent: the technologies and

products at the lower range of currents and voltages shown above are referred in this chapter as

„low power”, but they actually can operate at levels one or more orders of magnitude higher

that the integrated circuits or small signal devices designated as „high power”.

3.3.1 The Mesa Concept

The mesa structure is the specific common feature of the products manufactured in Section

2300. The shape of the silicon dice remembers the mesa flat-topped hills bounded from all sides

by steep slopes, normally slightly bevelled.

A rectifier device is supposed to work in two basic modes: (1) allowing a current flow through

it when directly biased, and (2) blocking the current flow when reversely biased, up to certain

voltage values. The pn junctions exhibit a low barrier for the direct current flow, and a high

barrier in the opposite direction, sustaining the reverse voltage across a space charge region

depleted of mobile charge carriers. But the reverse voltage must remain below the value at

which the electric field it generates across the depletion layer will cause a breakdown, i.e. an

uncontrolled current flow. The volume breakdown depends on the doping values and profile.

The breakdown may also be provoked by crystal imperfections (defects). It occurs differently

at the device borders and worsens at the surface of the bevel.

If the lateral bevel intersects the junction at a positive angle (i.e., when the junction area is

decreasing from the heavier to the lighter doped side), the depletion region bends towards the

bevel, its thickness increases, the electric field intensity underneath the bevel is diminished and

the breakdown voltage here becomes higher than in the volume. Obviously, a negative angle

will have the opposite effect, diminishing the breakdown voltage. The bevel surface is actually

protected by a passivation layer; it creates a complex bevel interface between the silicon crystal

and the passivation coating material, that must exhibit very good and stable dielectric properties

as well as a strong chemical compatibility with and adherence to the semiconductor. Local non-

ideal behaviour like lateral current leakage or even lateral spot breakdown may occur,

depending on the dielectric properties of the coating as well as on the electric charges

accumulated at the semiconductor surface. The mesa profile of the edge secures a high reverse

voltage, closer to but always different than in the volume.

16

A memory of Viorel Banu, a co-author of this chapter. In October 1978, when I was hired, the

800 V thyristor was considered a great success although it exhibited uneven blocking voltages

because of the asymmetrically bevelling of the anode and cathode junctions: if one angle was

positive, the other one had to be negative. I came up with a new, rather simple, high-

productivity bevelling procedure that produced positive angles for both junctions. Once, when

the Section and IPRS technical management were abroad to negotiate licenses, being “home

alone”, I decided to apply the “double positive bevel angle” procedure to all thyristors families.

Upon their return, the management was surprised to see perfectly symmetrical characteristics

sustaining regularly 1600, 1800 and even 2000V, never attained before. It was like a mini

revolution.

3.3.2. The “Core Compromise”

When speaking about power devices (such as rectifying diodes, thyristors, triacs), in order to

ensure a maximum power for a given device size, a careful optimization, which can be named

the core compromise, is necessary to achieve simultaneously the specified values for the

forward current (crossing the whole volume between the top and bottom of the silicon

structure), the reverse blocking voltage (sustained by the deep pn junctions), the maximum

allowed temperature, the desired reliability under thermal cycling, the surge current and

voltage values, and the switching speed.

The lower the doping and the thicker the silicon wafer, the higher the sustainable reverse

blocking voltage (of course, taking also into account the edge effects). On the other hand, a

lower doping and a thicker silicon wafer result in a higher electrical resistance and a higher

power dissipation, hence limiting the forward current the structure can lead without exceeding

the maximum allowable temperature. Good electrical and thermal contacts to the metal parts of

the case may substantially improve the situation by effectively removing heat, so that a higher

power dissipation would be tolerable, enabling higher maximum forward currents. Next,

structural defects in the semiconductor and at its edge may impose limitations on the

withstandable surge currents and voltages. The monocrystal quality is therefore especially

important, the careful limitation of monocrystal damaging during high-temperature treatments,

as well as the accuracy of the mesa etching and passivation are crucial. For fast switching

devices, the lifetime of the charge carriers has to be lowered, which in turn results in a higher

semiconductor resistivity, hence a lower withstandable forward current; this can be mitigated

by thinning the wafer. Finally, larger area devices are subject to substantial expansion and

contraction during thermal cycling, which may cause microstructural damages of the silicon if

it is tightly attached (e.g. soldered) to other materials like copper piece parts of the case, which

exhibit very different thermal expansion coefficients. A molybdenum interlayer between silicon

and copper is used to attenuate the mismatch of the thermal coefficients, while pressure contacts

give silicon the flexibility to expand/contract independently, being mechanically disconnected

from the other metal surfaces (yet maintaining good electrical and thermal contacts).

3.3.3. The Guard Rings

As far as Zener, controlled avalanche and suppressor diodes are concerned, their precise reverse

breakdown voltage should depend primarily upon the value of the semiconductor doping level.

The edge breakdown or leakage remain a vulnerability, that is mitigated by different means for

the various voltage ranges. The very low Zener voltages up to 6.8 V were obtained by simply

using shallow planar junctions (formerly there were alloyed pn junctions instead of the diffused

ones); for the next Zener voltage range up to 10 V a guard ring was used around the planar

diffusion; finally, the highest voltage Zener diodes were realized in mesa technology with

17

deeper diffusion for the junction. The Schottky diodes were realized also in planar technology

with guard ring lateral protection.

3.4. Technologies and Products

The power silicon devices were manufactured by Section 2300 on high resistivity (low doping)

silicon wafers, with high temperature deep diffusion of impurities at lower concentrations to

form the high voltage junctions, and shallow, higher concentration diffusions for ensuring a

metal-semiconductor contact with very low resistance. The largest wafers in use in Section

2300 had a diameter of 3 inches. The wafer thicknesses ranged from 150 microns (for fast

rectifying diodes) up to 700 microns for 2400 V thyristors and could go up to one millimetre

for 4-5 kV rectifying diodes.

The main manufacturing steps – also called “unit processes” - are quite similar for all products

manufactured in Section 2300, although it will depend upon the product which unit processes,

with which recipe and parameters, and in which order will be integrated in the manufacturing

flow.

In generic terms, for all products there will be a manufacturing sequence that will be applied to

the wafer, and may include: oxidation and possibly lithography to open windows in the oxide

and define localized area on the chip; doping either selectively (i.e. in the windows defined by

photolithography) or non-selectively; metal deposition; and the definition of the mesa bevel.

Then the chips are separated from the wafer (dicing), the dice are attached and contacted to the

metallic parts of the future package, then the package is formed and protected. A crucial step is

the etching and passivation of the mesa bevel, which may be performed either on wafers before

separation, or after soldering the dice to the case parts. This is of course the case for smaller

dice, when many of them would fit on the same wafer. For ultra-high voltages it may be

necessary to use several wafers, stacked one atop of another and separated by aluminium disks

to assure the mechanical strength. When the product must handle extremely high currents, it

may be necessary to use the entire wafer for one or very few silicon structures. The silicon

structure is alloyed with aluminium on a disc of molybdenum as interposer between the silicon

and the copper package parts to ensure the mechanical rigidity and to attenuate the mismatch

in the thermal expansion coefficients, in order to avoid the silicon micro-cracking. It is this

sandwich that will go in the final package. In the following we will indicate the typical

operations for the main product families.

3.4.1. Low and medium power diodes, thyristors and triacs

Uniformly low-doped wafers, with appropriate thicknesses are used.

For diodes, the junction is defined by high temperature deep diffusion at low concentrations,

while shallow, high concentration diffusions will be performed in the contact areas to ensure

high conductivity, ohmic contacts. The wafers are then electrochemically plated with a nickel-

gold metal layer, then the dice are separated from the wafer into square, hexagonal or round

chips, from 1mm to 1 cm on a size, by a chemical etching process which also defines the lateral

mesa profile. In a subsequent processing step, the bevel is passivated with organic resins or

thermally sintered glass. As an exception, the very small chips of the 1 A diodes are separated

by wafer sawing with diamond edge disk and are fixed between two silver-coated copper disks

to form a sandwich for safer mechanical handling; next they are chemically etched to define

and smoothen the lateral mesa profile; after being further soldered to the copper piston-shaped

terminals, the sandwiches are passivated with organic resins.

For thyristors and triacs, the junctions are defined by high temperature deep diffusion at low

concentrations, while shallow, high concentration diffusions will be performed under the

18

contacts. The grooves of the mesa profile are chemically etched, then passivated with a

thermally sintered glass layer, or in some cases with organic resin. The chips are selectively

metalized by chemical deposition of a nickel layer. The selective diffusion, etching and

metallization processes involve single and even double face aligned photolithography (e.g. for

low/medium power thyristors or for triacs) making it a mesa-planar technology. The double

face alignment process was the result of the successful in-house research. The capability for

double face photolithography developed by the Section 2300 was unique in Romania! The

deposited metal is later covered with soft solder alloys. The dice are separated by wafer sawing

with a diamond edge disk.

The dice are soft-soldered with lead-tin-(sometimes)silver on copper case parts, while the gate

for thyristors and triacs is contacted by thermosonic wire bonding. They are packaged in both

electrically welded metal-glass or metal-ceramic cases, as well as in moulded plastic cases.

From an organizational perspective, these products were consolidated in an integrated module

for medium power diodes and thyristors.

3.4.2. High Power Diodes and Thyristors

Thicker silicon wafers with lower doping will be used.

Deep non-selective impurities diffusion at high temperatures are used to define the high voltage

diode junctions. For thyristors and triacs, in addition to the deep non-selective and selective

diffusions there is an additional diffusion for the control gate. Moreover, in order to be able to

uniformly trigger the thyristor/triac anode current, the anode presents holes (dots) through

which the gate-to-anode contacts are distributed across the whole area. Photolithography must

be used to define these localised features; hence this is a mesa-planar technology. Of course,

shallow high concentrations layers are also diffused on all devices, to ensure good ohmic metal-

semiconductor contacts. The wafers are cut into round structures by sandblasting, or by laser

cutting for larger diameters, laser cutting being also capable to separate structures with different

diameters from the same wafer, useful to optimize the material usage. The silicon disks are

vacuum plated with thin metal layers; to assure rigidity, they are pre-assembled in sandwiches

by high temperature alloying of the silicon structures with aluminium on a molybdenum

interposer matching the expansion coefficient to that of silicon. The mesa lateral profile is

defined by mechanical wet-powder grinding, then refined by chemical etching, followed by

passivation with organic resins. The sandwiches are contacted to the copper parts of the cases

either through soft soldering, or by pressed contacts11 (on one or on both faces) conceived for

avoiding mechanical tensions between the sandwich and the case parts, making sure that the

product will reliably operate during thermal cycling. They are packaged in metal-glass or metal-

ceramic welded cases, as well as in moulded plastic cases.

From an organizational perspective, these products were consolidated in an integrated module

for high power diodes and thyristors.

3.4.3. Specialty Technologies

In the early phases of the Section 2300 the silicon wafers for fast switching rectifying devices

were doped with gold through high temperature diffusion. Later, the expensive gold doping

process was replaced by a revolutionary nuclear technology, not only cheaper but also much

more effective; it has been protected by patents and was applied across the board, with the

exception of the planar ultra-fast diodes that remain gold-doped.

11 V. Banu – CAS Proceedings, 8 (1985). Re. power devices with pressed contacts.

19

3.5. People and their areas of responsibility

The intention of this chapters is to mention the names of the technologists as well as their main

contributions to the growth of the Section 2300, the „heroes” of this success story.

3.5.1. Management

The Section 2300, later changed into the Factory for Power Semiconductor Devices, was led

by eight successive management teams:

(1) phys. Doina Didiv (manager) with dr. eng. Nicolae Iosif (deputy);

(2) dr. eng. Nicolae Iosif with eng. Gheorghe Florea;

(3) eng. Gheorghe Florea with dr. eng.Petru Dan;

(4) dr. eng. Petru Dan with dr. eng. George Mânduțeanu;

(5) dr. eng. Petru Dan with eng. Eugen Popa;

(6) eng. Eugen Popa;

(7) eng. Marian Udrea-Spenea;

(8) phys. Marius Jidveian was the last manager of the Factory.

3.5.2. Product Families

It was already mentioned that the products of the Section 2300 have been basically split in two

big families mainly as a function of the power they could handle (plus a few other

characteristics). For a better coordination and management of the design and manufacturing

chains, a concept of integrated modules for designing / wafer processing / assembling /

measuring and testing was established and implemented. and the technology portfolio of the

two modules has been described before, here we will also list their main products. In addition,

these two modules did not cover the entirety of the activities in the Section 2300, two more

specialty chains emerged, as shown below:

1) Integrated module for medium power diodes and thyristors under the leadership of Petru

Dan produced mesa diodes between 1-80 A and mesa thyristors and triacs between 1-40 A (also

the above-mentioned planar Zener and ultra-fast diodes). Other products, more or less similar

with these, included monolithic rectifier bridges or rectifying stacks, as well as monolithic

power modules with diodes and thyristors.

2) Integrated module for high power diodes and thyristors led by George Mânduțeanu

delivered diodes between 100-3000 (exceptionally 5000 A) and thyristors between 50-1000 A

(including related versions like gate turn-off thyristors GTO).

3) Three-phases rectifier bridges for automotive applications was a third distinct family that

emerged due to the increasing demand for automotive parts. It was led and developed by Eugen

Popa. He took over these products form engineer Traian Cârcu, future manager of Section

2500, and engineer Dănuț Bodea, future manager of the export department, who initiated the

production of the first type of such bridges responding to a customer demand.

4) The ready-to-use modules with power devices was a fourth, future-oriented avenue opened

thanks to the technical skills and innovative creativity of some application-oriented people.

Most of the products were custom-designed, generating a considerable added-value; some of

these products were designated in the world as „mechatronics”. The leader of this department

was eng. Mihai Chiș, a brilliant professional coming from one of the main customers,

ELECTROTEHNICA SA.

20

3.5.3. Wafer and Chips Processing

The Silicon wafer processing, including oxidation, doping by diffusion, heat treatment,

chemical deposition of nickel and gold, followed by acid mesa defining and etching, for low

and medium power diodes, were initially handled by phys. Elisabeta Tsois and eng. Iosif

Lingvay, and later by phys. Eugen Lakatoș (high temperature processes) and eng. Elena

Seremeta (chemical processes).

The silicon wafer processing (same kind of processes as above, plus photolithography) for high

power devices with mechanically defined mesa structure, as well as for the whole range of

thyristors and triacs, including the double face aligned photolithography were led by physicists

Carmen Liiceanu, Florian Țurțudău who were later joined by the younger engineers Gabriel

Dumitrescu and Mihai Apostolescu. The planar processing of the wafers for low voltage Zener

diodes, ultra-fast diodes and Schottky diodes was entirely the responsibility of eng. Mihai

Bucur, a colleague „borrowed” from the staff of the Section 2500, where performed the related

processes. Phys.Mihai Răuță contributed here with an innovative design procedure for guard-

ring Zener diodes.

The contact between the semiconductor and its metal electrodes may dramatically influence the

electrical and thermal behavior of the device, as well as its reliability. The real metal-

semiconductor interface is very different from the ideal model of the metal-semiconductor

contact. Obtaining a very good ohmic contacts or a high performance Schottky contact is a

matter of technological art. That is why an extensive research on this topic was developed by a

joint team of academics and technologists12,13,14,15,16.

The neutron irradiation technology17 mentioned above for fast semiconductor devices was

invented, developed, and implemented by phys. Eugenia Hălmăgean. She also developed a

technique of doping silicon ingots by nuclear transmutation of the silicon into phosphorus,

capable to considerably improve doping uniformity.

Eng. Viorel Banu was running the vacuum metal deposition, such as titanium-nickel-silver or

chromium-nickel-silver for low and medium power devices; chromium-gold, chromium-gold-

chromium and nickel-silver for high power devices; as well as aluminium for alloyed Zener

diodes and high voltage rectifying stacks. He replaced the previously expensive gold plating of

the molybdenum disks with silver. By using an equipment for plasma enhanced chemical vapor

deposition designed and built by IPRS Băneasa in collaboration with the enterprise ICRET

Bucharest he obtained super-low forward voltage Schottky diodes with a maximum voltage

drop of 0.25 V at a current of 10 A. He also realized an innovative technique for ultra-fast

12 P.Al. Dan - Contributions to the study of the metal-semiconductor contact (in Romanian), Ph.D. Thesis, Faculty

of Electronics and Telecommunications, IPB, , Romania, 1988.

13 D. Dascălu, G. Brezeanu, M. Suciu, P. A. Dan – Solid State Electronics, 27, 359 (1984). Re. effect of geometry

and heat treatment on non-ideal aluminium-silicon contact.

14 P. A. Dan, G. Popovici, D. Dascălu, G. Brezeanu, A. Popa – Journal of Electrochemical Society, 130, 2472

(1983). Re. chemically deposited nickel-silicon contacts.

15 G. Brezeanu, C. Căbuz, D. Dascălu, P. A. Dan – Solid State Electronics, 30, 527 (1987). Re. silicide-silicon

contacts of vacuum deposited platinum and chromium.

16 G. Brezeanu, D. Dascălu, P. A. Dan, S. Negru, V. Trăistaru – Microelectronics and Reliability, 28, 205 (1988).

Re. aluminium-titanium-silicon contacts.

17 E. Hălmăgean – CAS Proceedings, 2 (1979). Re. lifetime reduction through irradiation with fast neutrons.

21

power devices18,19, which consisted of combining gold diffusion with electron irradiation for

improving the mitigation between forward conduction and reverse blocking.

The classical passivation of low and medium power devices was based on organic silicone resins

polymerized at moderate oven temperatures. It was generally performed on chips already

attached to the copper base of their case. A genuine alternative solution was the chemical

etching and passivation with organic resin of separated dice tightly pressed between acid-

resistant foils, a patent developed by eng. Vasile Obreja from ICCE. A substantial

improvement occurred for devices manufactured with glass-passivation20,21 (instead of organic

passivation), with nickel metal plating covered by soldering alloy, based on an original idea by

eng. Sever Grigorescu, further developed by eng. Anca Nichita. The glass passivation also

enabled a major improvement in manufacturing controlled-avalanche devices22. Another

original passivation alternative for the nickel-gold plated mesa chips invented and patented by

eng. Mircea Romanescu consisted of passivating the chemically etched grooves on the wafer

with a thixotropic organic resin (used before only in aeronautic technologies) having very good

chemical and dielectric properties and a short hardening time by thermal treatment, then

mechanically dicing the wafer by sawing with diamond edge disk. This technology was

particularly successful for mesa Zener diodes over 10 V, where the planar technology was no

more appropriate.

3.5.4. Assembly and Chemical Protection

The low and medium power devices chips (either already passivated or not) were assembled by

soldering the chips with special alloys containing tin, lead and silver on metal parts (bases) of

metal-glass or plastic cases, through a heat treatment in inert (nitrogen), then reducing

(hydrogen) atmosphere in belt furnaces, followed by sealing (electric welding of metal cases)

or moulding (plastic cases). In the case of non-passivated chips, a chemical etching was used

after soldering for refining the mesa edge and before applying the resin passivation on the bevel.

For high power devices the chemical mesa processing and the organic passivation were always

performed on sandwiches. The alloying of high-power silicon-molybdenum sandwiches, as

well as the soldering of these sandwiches to the copper package parts were performed in similar

kind of furnaces. The sandwiches were then either sealed in metal-ceramic packages by

electrical welding or moulded in plastic packages.

The assembly and test production line for low power diodes was led by phys. Carmen Nan,

and later by eng.r Gheorghe Lazăr. Eng. Cătălin Georgian succeeded to implement here the

more economic use of glass-passivated chips, instead of the organic-passivated sandwiches, for

1 A diodes. Some similar technology for assembling diode chips in single-phase moulded or

resin-filled rectifier bridges was used and diversified by eng. Doru Liiceanu for general

applications.

18 V. Banu, G. Dinoiu, E. Iliescu, E. Lakatoș, C. Liiceanu, F. Țurțudău – CAS Proceedings, 12 (1989). Re.

irradiation with electrons versus gold and platinum diffusion.

19 V. Banu, E. Iliescu, Anastase Niculescu – în Electrotehnica, Electronica și Automatica, 33 (1989). Re.

manufacturing process for fast high power thyristors.

20 M. Udrea-Spenea, S. Grigorescu, A. Nichita – CAS Proceedings, 6 (1983). Re. glass passivation.

21 M. Niculae, A. Nichita, M. Udrea-Spenea, S. Grigorescu, V. Marinescu – CAS Proceedings, 8 (1985). Re.

breakdown of glass passivated versus organic passivated chips.

22 E. Popa, A. Stan, M. Udrea-Spenea, A. Nichita – Electrotehnica, Electronica și Automatica, 33 (1989). Re.

medium power suppressor diodes.

22

Petru Dan was the manager of the assembly and test line for medium power devices, as part of

the module he was leading. The team was great; it included engineers Viorel Marinescu,

Mircea Romanescu, Grigore Popovici, Adrian Albu, Nicolae Popescu and Any Ichim. The

latter two invented, registered and implemented a technology for alkaline etching of nickel

plated mesa chips (thus avoiding the additional expensive gold plating), soldered with lead on

the copper case base, followed by organic passivation.

The assembly and test of high-power devices were led in the early days by eng. Marian

Niculae, and later by phys. Marius Jidveian. It included the sandwich processing, assembling

with case parts, sealing or moulding, and finally testing and sorting. Eng. Mihai Luca worked

there for a while, for measurement and testing. The electronic maintenance of the whole

production line for high-power devices was taken care of by eng. Dan Halip.

The sealed devices (either in metal or plastic case) were protected by electro-chemically

deposited metal cover (like nickel or tin), the related processes being led by eng. Ștefan

Armeanu. He was also in charge with supplying the high-grade de-ionized water, whose high

purity was crucial for the quality of mesa technologies.

3.5.5. Rectifier Bridges and Modules with Diodes and Thyristors

Part of the medium power diodes and even chips where further used to build three-phase

rectifying bridges for automotive alternators, with average currents up to 100 A. Before 1990

it was a substantial demand mainly from the Romanian car manufacturers, established as state-

owned enterprises, as well as from the famous motorcycles manufacturer Java in

Czechoslovakia. The latter was the biggest export business of IPRS Băneasa. After 1990 the

international market demanded an unexpectedly large spectrum of design versions, with diodes

either soldered, or press-fitted on heatsinks, or even with passivated chips directly soldered on

heatsinks. The creativity of Eugen Popa and his team, where Florian Țurțudău played a major

role, led to ingenious custom-designed solutions for all three kinds of designs, part of them

protected by patents, enabling the successful entry and growth on the worldwide automotive

market. This development required extensive in-house design efforts for the mechanical parts

and corresponding tooling, professionally assumed by eng. Sorin Dobrinescu.

A particular interest emerged for modular assemblies of power diodes and thyristors together

with other components, exhibiting significant added value, in discrete version fixed/pressed on

aluminium heatsinks like the so-called „application kits” conceived and developed by eng.

Mihai Chiș, as well as in monolithic versions, moulded or filled with epoxy resins, designed

by eng. Milan Peleanu. Due to the impetus of these developments and the additional space

requirements, the production of the heatsink-mounted assemblies led by Mihai Chiș was located

in the older germanium unit, Section 2200.

3.5.6. Test, Quality and Reliability Assessment

The resulted final devices and assemblies were submitted to complete electrical testing and

sorting. This was a 100% screening and sorting process done in the assembly and test

department. In the beginning of Section 2300, when SILEC diodes were manufactured under

license, the responsibility was with eng. Gheorghe Florea, then it was assumed by Eugen Popa

together with eng. Eva Rado after thyristors have been introduced.

At the same time, the quality assurance pulled random samples from all batches and performed

checks and assessments, as well as reliability evaluations. The quality assurance methodology,

formerly dealt with by eng. Radu Dragomir, became rather obsolete after two decades of

classical approach and was upgraded when Florian Țurțudău got involved in implementing

the ISO standardization and in establishing an up-to-date quality system, which strengthened

23

the competitive advantages of the products, mainly after expanding the presence on the

international market after 1990.

The reliability sector had a spectacular development, in line with the development of the

company, and allowed the manufacturing of high reliability components, which were classified

according to the specific operation conditions: (a) „yellow program” with reference to

components for industrial applications requiring guaranteed reliability; (b) „red program” for

professional components with a minimum reliability index λ 10-6 h-1; (c) „blue program” for

components with special protection for naval and aviation applications; (d) components for the

Romanian Railway Company; (e) components for military applications. The annual reliability

report became one of the most effective instruments for the continuous quality improvement.

The reliability evaluation had been started by phys. Barbu Constantinescu, but later it was

broadly developed, upgraded and extended by eng. Marian Udrea-Spenea23 to cover a wide

range of power domains (i.e. up to very high voltages and currents, including unusually high

surge values to be checked), for almost all device types (rectifier and controlled avalanche

diodes, thyristors, triacs, rectifier bridges, modules), and for an extended span of controlled

performances (electrical, thermal, mechanical, environmental, aging under full charge

conditions). Marian Udrea-Spenea had contributed also to the progress of Section 2300 with

several new products and processes. Eng.Augustin Stan added valuable knowledge about

failure mechanisms in power devices24, based on his previous work for the assembling of low

power diodes and automotive products.

3.6. Partners

A major support role in the research, design and development of the specific silicon devices

and the related technologies was played by the specialized research institute ICCE, as well as

by other research units such as ICPMS, IFA, IFIN, and by universities, mainly by the Faculty

of Electronics and Telecommunications of IPB (later UPB). This support was constantly

provided not only to Section 2300 but to all semiconductor production units of IPRS.

In the beginning of the operations of Section 2300, the early version of ICCE (at that time

CCPCE) was located in the facilities of sections 2200 and 2300, than moved in its own tall

building, over the fence of IPRS Băneasa. In spite of the large span of concepts, products and

technologies developed by ICCE, only very little interest was actually devoted to the power

semiconductor devices dealt with by Section 2300. It is worthwhile to mention here the

passivation technology with organic resin for nickel-gold plated chips for medium power

devices, as well as some technologies for fast and Zener diodes.

A much more substantial support was given by the Faculty of Electronics and

Telecommunications of IPB (later the Faculty of Electronics, Telecommunications and

Information Technology of UPB), mainly by the Chair CDE, which was renamed Chair of

Electronics Technology and Microelectronics (the later DCAE). A special emphasis is deserved

by the long lasting co-operation with Prof. dr. Dan Dascălu (nowadays member of the

Romanian Academy) and Prof. dr. Gheorghe Brezeanu in the fields of metal-semiconductor

contacts and Schottly diodes, Prof. dr. Adrian Rusu in the fields of Schottky diodes and

semiconductor device modelling, Prof. dr. Andrei Silard in the field of power thyristors, Prof.

dr. Mircea Bodea on various semiconductor theoretical and practical issues.

23 M. Udrea-Spenea, T. Mochi, A. Stan, R. Giurconiu, M. Dabija – CAS Proceedings, 14 (1991). Re. failure rate

of semiconductor devices.

24 E. Popa, A. Stan, A. Nichita – CAS Proceedings, 9 (1986). Re. thermal fatigue of medium power devices.

24

3.7. Scientific Contributions

The high performance obtained in designing products, imagining new device concepts25,26,27

and managing production processes was per se rewarding for professionals working in IPRS

Băneasa. However, the production challenges led them to more ambitious goals, such as sharing

the acquired experience within the scientific community, or even to protect the results by

registered patents Thus, a lot of valuable results were presented at professional conferences like

the Annual Conference for Semiconductors CAS, or published in local and international

scientific journals, as well as in comprehensive books (Fig. 4, 6). Most of such contributions

were the result of joint activities with partners from universities or research institutes.

To provide a systematic view of such works authored or co-authored by people of Section

2300, it is worthwhile to classify them according to the approached topics: silicon processing

[1-6], power diodes and thyristors [7-25], fast switching devices [26-34], Zener, controlled

avalanche and suppressor diodes [35-42], Schottky diodes [43-45], metal-semiconductor

contact [46-74], mesa passivation [75-79], reliability of power devices [80-87], modelling and

characterization [88-97]. In the detailed reference list of this chapter a brief abstract in English

is added at the end of each reference, instead of the titles in Romanian.

The universities offered to IPRS specialists the opportunity to obtain advances degrees. Thus,

four people employed in Section 2300 engaged in PhD research and successfully defended their

thesis, while keeping their regular production responsibilities: Eugenia Hălmăgean got a PhD

in physics with the thesis on neutron irradiation of semiconductors; George Mânduțeanu

became a PhD in electronics with a thesis on modelling power silicon devices; Petru Dan got

a PhD in electronics with a thesis on metal-semiconductor contacts; Nicolae Iosif became a

PhD in power electronics.

3.8. Epilogue

The story of Section 2300, later the Factory for Power Semiconductor Devices, is not only about

a successful production unit belonging to the elite industrial company IPRS Băneasa, but it

offers also a panoramic view over the life cycle of one of the most advanced production units

of the Romanian industry. Its growth lasted for more than a decade, really impacting the

progress of many industrial sectors dependent on such devices. The maturity decade gave it the

chance to consolidate its strengths, to diversify its offer, to capture the local market and to open

doo