1/19/10 C. Nguyen Lecture 1: Administration & Overview; History of IC’s Copyright @ 2009 Regents of the University of California 1 EE143: Microfabrication Technology This Lecture: Administration & Overview Reading: Handouts Lecture Topics: – Course information – Syllabus ------------------------------------------- Welcome to EE 143: “Microfabrication Technology” This is our course on wafer-level fabrication of transistor integrated circuits and other micro-devices, such as MEMS → Pass out course info sheet → Pass out course syllabus → Lab juggling; get info on the order of people signing up for labs, then only make those who were last in a section that is full move; must show data on lab section counts on first lecture sheet → Show calendar and settle the office hours ------------------------------------------- Goals of the course: – Teach the skills needed to design and fabricate micro- and nano-devices, including integrated circuits and micro electromechanical systems (MEMS) – Design emphasis: This is NOT a survey course; you will be expected to design and layout physical MOS devices (and MEMS devices, if there’s time) – Hands-on emphasis: Give you actual hands-on experience fabricating micro-devices using a wafer-level process in a cleanroom The mechanics of the course are summarized in the course handouts, given out in lecture today – Course Information Sheet → Course description → Course mechanics → Textbooks → Grading policy – Syllabus → Lecture by lecture timeline w/ associated reading sections → Midterm Exam: Thursday, March 18 (tentative) → Final Exam: Monday, May 10 Reading: Jaeger, Chpt. 1 Lecture Topics: – History of IC’s – Devices of Interest → MOS transistor → Micromechanical structure ------------------------------------------- History of IC’s: – 1834: Difference Engines (mechanical computers) → Gears, cranks, levers, decimal, pipelining! – 1904: Vacuum tube invented → Yielded the ENIAC vacuum tube computer – 1925: J. Lilenfield proposed the MOSFET transistor → Problem: knowledge of materials not sufficient to get this to work – (instead) – 1947: Invention of the transistor (Bardeen, Brattein, Shockley) – 1949: Invention of the Bipolar Xsistor (Shockley) – 1956: First digital logic gates (Harris) – 1959: Invention of planar silicon processing (Kilby, Noyce) – Then a slew of bipolar technologies Administration & Overview IC History & Review of Devices1/19/10 C. Nguyen Lecture 1: Administration & Overview; History of IC’s Copyright @ 2009 Regents of the University of California 2 EE143: Microfabrication Technology → TTL (1965) → ECL (1967) → MTL/I2L (1972) → SiGe heterostructures (1990’s) – Bipolar ruled during the 60’s and 70’s, because it was faster than anything else, incl. MOS – But soon, its excessive power consumption caught up, and MOS began to come into favor as small channel lengths boosted the speed of MOS – Fairly simple process: only 5 masks; note that this is much smaller than today’s proess, which might have more than 28 masking steps – The rise of MOS occurred in steps: – 1965: PMOS w/ Al gate → Used <111> wafers because bipolar used them → This forced the use of PMOS, since oxide charge was dense in <111>-Si to oxide interfaces → Oxide charge made it difficult to isolate NMOS devices – 1967-70: Al gate NMOS → Use of <100>-Si together with sintering reduced oxide charges → Speed faster than PMOS and path to matching bipolar speed could be seen1/19/10 C. Nguyen Lecture 1: Administration & Overview; History of IC’s Copyright @ 2009 Regents of the University of California 3 EE143: Microfabrication Technology – 1970: Si-gate NMOS → Advantage: self-alignment of source & drain → Problem: power consumption (similar to bipolar) – To reduce power consumption, a complementary device was needed – This is where CMOS looked advantageous – 1963: pwell CMOS → CMOS gate actually came before NMOS or PMOS, but poor control of oxide quality at the time prevented it from thriving → Why didn’t CMOS thrive in 1963? 1. Higher fabrication cost. 2. Latch-up problems. 3. Lower packing density due to need for wells. 4. CMOS slower than NMOS due to larger gate capacitance. – But soon power became an issue: – 1971: Intel 4004 4-bit microprocessor → 2,300 devices (PMOS) – 1978: Intel 8086 16-bit microprocessor → 29,000 devices (NMOS); power dissipation beginning to get up there: 1.5W @ 8MHz – 1985: Intel 80386 → 275,000 devices Æ NMOS light bulb! → A low power technology was needed – Result: CMOS takes over → Intel 80C86 (CMOS version of 80386) → Intel 80486: 1.2 million Xsistors → Intel Pentium (P5): 3 million Xsistors → Intel P6: 5.5 million Xsistors in core, 15 million more in secondary cache → And of course it keeps going to today … → Intel Core 2 Duo: 820 million Xsistors
View Full Document