IEEE Power Electronics Magazine - June 2015 - 18

challenge [7]. In my disclosure, I projected the following characteristics for
the proposed device:
■■both forward and reverse blocking
capability
■■forward drop similar to a p-i-n
rectifier
■■turn-on and turn-off using a small
gate voltage with low gate current
■■high turn-off gain
■■high dV/dt and dI/dt capability
■■operating at elevated temperatures
■■tolerance to radiation.
I named the device the Gate Enhanced
Rectifier (GERECT) to emphasize its
p-i-n rectifier-like on-state characteristics. These characteristics were projected based on my previous work on
the field-controlled thyristor (FCT)
performed in the 1970s.

a large forward-biased safe operating
area (FBSOA). I eventually published
these observations in 1982 [13]. This led
to my conclusion in 1979 that an ideal
power device could be constructed
using a MOSFET in series with a p-i-n
rectifier. I then began to explore ways
to integrate the MOSFET with the FCT
structure. I found this structure to be
identical to the structures shown in
Figure 1. This second conceptual path
to the IGBT structure, although later is
temporal sequence, was important because it gave me the confidence to project high on-state current density for the
IGBT structure with low on-state voltage drop in my patent disclosure [7]. I
acknowledged the role of my FCT work
on the invention of the IGBT during the
celebration of the 50th anniversary of
the SIT in Japan [14].

Field-Controlled Thyristors
In 1975, GE was blindsided by the anIGBT Proposal
nouncement of a new device concept
My GERECT proposal in September
called the static-induction transistor
1980 was met with skepticism by my
(SIT) and static induction thyristor
GE colleagues. They first pointed out
proposed by Nishizawa in Japan [8]. To
that previous efforts at MOS-gating of
assess the potential for this concept, I
four-layer structures showed latch-up
began investigating these devices in
of the thyristors at low current levels
1976. It was soon apparent that these
[15], [16]. They also pointed out that
devices (called FCTs
my proposed IGBT
at GE) were deplestructure consisted
This would allow rapid
tion-mode structures
of an n-channel MOSthat required a gate
FET driving a wideavailability of IGBTs in
bias to maintain their
base p-n-p bipolar
production quantities
off state. Although I
transistor. Prevailing
without the usual
was successful in imwisdom based on
prolonged development
proving the blocking
decades of work on
phase required for
gain (i.e., the ratio of
power bipolar tranmost innovations due
anode to gate voltsistors recommended
age) using innovative
using a narrow-base
to their unique device
structures [9]-[12],
n-p-n structure to get
structure.
all the power eleca good current gain.
tronics application
Based on this, my
engineers at GE found the normally on
critics said that my proposed device
nature of these devices unacceptable
could be expected to operate at a
from a circuit standpoint due to shootlow on-state current density (below 20
through problems during start-up of
A/cm2). My projections of an on-state
the power circuit.
current density of 100-200 A/cm 2
To address the concerns of the apfor my proposed device based on
plications engineers, in 1979 I investhe p-i-n diode model were considtigated controlling the FCT using a
ered unrealistic.
MOSFET in series with its cathode. I
observed excellent on-state characMOS-Controlled Thyristors
teristics at high current densities and
In September 1980, Victor Temple procurrent saturation characteristics with
posed the MOS-controlled thyristor

IEEE PowEr ElEctronIcs MagazInE

z June 2015

(MCT) structure at GE [17]. His concept was based on a four-layer thyristor operating with latch-up in the on
state. An integrated MOSFET was
used in the MCT to short circuit the
junction between the cathode and the
base region to interrupt the regenerative action and turn off the device.
Temple made convincing arguments
that the MCT would have a lower onstate voltage drop and a larger onstate current density than my proposed IGBT, making it ideal for all
applications. I pointed out that the
IGBT was more suitable for the
replacement of bipolar transistors due
to its good FBSOA and that the MCT
was more suitable for the replacement
of gate turn-off thyristors because
it required snubbers to control the
turn-on and turn-off process. However, Temple's arguments convinced
U.S. funding agencies at the Electric
Power Research Institute and the
Department of Defense to form a consortium to exclusively support the
development of the MCT in the United
States [18]. This decision resulted in
shifting the eventual production of
IGBTs overseas.

The Welch Edict
In October 1980, Tom Brock returned
to the GE Research Center to get a
response to his earlier challenge. During this meeting, I proposed developing and commercializing the IGBT. In
addition to the new device proposal,
my presentation pointed out that this
device could be used for a wide variety of products within GE to impact
the small appliance, large appliance,
drives, and lighting divisions. Most
importantly, I pointed out that I could
engineer the process for my IGBT so
that it could be manufactured in the
existing power MOSFET production
line. This would allow rapid availability of IGBTs in production quantities
without the usual prolonged development phase required for most innovations due to their unique device structure. In contrast, the proposed MCT
required the development of a new
five-layer fabrication process at the
GE Research Center that would have

Table of Contents for the Digital Edition of IEEE Power Electronics Magazine - June 2015