1 University Federico II

Dept of Electronics and Telecommunications

Paolo Spirito
Power Semiconductor Devices
The Silicon Controlled Rectifier (SCR)
The Silicon Controlled Rectifier, also called Thyristor, is one of the oldest power devices, and
it is actually employed as power switch for the largest currents (several kiloamperes) and
voltages (several kilovolts) as presented in the introductory lesson.
Its name indicates that the device behaves a a PIN diode (or rectifier) in his ON state, with a
very low forward drop (if compared with the one of the other devices), but it is normally in
OFF state, and to be brought in the ON state it requires a trigger on a separate control
terminal (called gate). From this effect we can consider the SCR as a silicon device that
behaves as a PIN diode or rectifier - in forward bias if a command is done on the control
terminal.
The main drawback of the SCR is that we can not control it in the opposite way i.e. we can
not act on the control terminal to bring it in the OFF state when it is already conducting (ON
state). Then in order to switch off the SCR we can only rely on the natural current
commutation, in other words we must revert the voltage across it, from forward to reverse
bias (as for the PIN diode) to interrupt the current flowing into it.
The device, like the PIN diode and the BJT, is a bipolar device: its operation relies on both
type of carriers, and the low voltage drop in conduction is actually due to the conductivity
modulation (already discussed for the PIN diode) that takes place at high injection in the thick
and low doped region (needed to sustain high voltages in OFF state).
2 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
P
N
+
N
+
P
+
N
-
anode
cathode gate
w

The cross-section of the structure of an SCR is depicted in fig. (a): the two main terminals
are the anode A (on the bottom of the chip) that is biased at positive voltages in forward
state, and the cathode K(on top of the chip) that is grounded. The gate terminal is the
control electrode that allows to switch on the SCR.
Considering the main portion indicated in dashed line, we see that there are four doping
layers(N
+
P N
-
P
+
) and three P/N junctions from cathode to anode (fig. (b). A simple
analysis of this structure can be done by considering it made by two complementary BJT
(NPN and PNP) closely connected, as indicated in fig. (c,d):
A
G
N
P
N
P
J
1
J
2
J
3
K
N
P
N
P
J
1
J
2
J
3
K
G
A
(a) (b) (c)
G
A
K
N
P
P
P
N
N
(d)
NPN
PNP
3 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
D.C. analysis an trigger condition of the SCR
An SCR approximate analysis, to understand the behavior in forward bias (V
AK
>0) , and to
define the trigger condition that will change its state from OFF to ON, can be done by referring
to the configuration with two BJT seen before.
a) forward blocking (OFF) state
A
G
K
Q1
Q2
IC1
I
B1
=I
C2
I
G
=0
IB2
I
A
=I
E1
I
K
=I
E2
When I
B
= 0, and assuming V
AK
> 0, both BJT are in the active region
and we have: I
A
= I
K
= I = I
C1
+ I
C2
. At high voltage where multiplication
takes place (as seen for BJT eq. (26)) we have:
02 2 02 2 2 2
01 1 01 1 1 1
C C E C
C C E C
MI I M MI I M I
MI I M MI I M I
+ = + =
+ = + =


(1)
) ( ) (
02 01 2 1 2 1 C C C C
I I M I M M I I I + + + = + =
Summing the two currents:
) 1
) (
2 1
02 01
+
+
=
( M
I I M
I
C C
The current I is negligible at low voltages and increase largely only
when M(
1
+
2
) = 1. The max voltage V
MAX
is near to the breakdown
voltage of J
2
because the of the BJT is quite low.
(2)
4 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
In the forward blocking state the P/N junctions J
1
and J
3
are forward biased, while the N/P
junction J
2
is reverse biased, as indicated in figure.
The central N region must be thick and low doped to sustain high reverse voltages with V
AK
positive, needed to keep the SCR in the OFF condition even at high anode voltages if the
SCR is not triggered.
The PNP transistor Q
1
has
1
very low (< 0.5), because the width of the N base region is
large; also the NPN transistor Q
2
has a relatively low
2
because the P layer is not thin nor
low doped. We must consider that both
1
and
2
depend on the V
AK
voltage, because the
effective width of the base for the two BJT (mainly Q1) decreases if V
AK
is increased, due to
the depletion width of region J
2
. The voltage dependence of alphas is important in defining
the IV curves of the SCR in the off-state region.
5 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
A
G
K
Q1
Q2
IC1
I =I C2 B1
IG
IB2
I =I A E1
I =I K E2
PNP
NPN
b) the triggering condition
If a gate current is injected from the gate electrode : I
K
= I
A
+ I
G
and eq. (1) are changed as:
02 2 02 2 2
01 1 1
) (
C G A C K C
C A C
MI I I M MI I M I
MI I M I
+ + = + =
+ =

) ( ) (
02 01 2 1 2 1 C C G A A C C A
I I M I I M I M I I I + + + + = + =
(3)
) 1
) (
2 1
02 01 2

+
+ +
=
( M
I I M I M
I
C C G
A
and we have:
(4)
Eq. (4) is the general equation for the trigger condition of the
SCR, valid also for low voltages with no multiplication (M=1):
in that case we can still have a current increasing indefinitely
(trigger condition) if the condition
1
+
2
(I
K
) = 1 holds.
)) ( 1
) (
2 1
02 01 2
K
C C G
A
I
I I I
I

+
+ +
=
(
6 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices

=
+

+ +
=
T
J
G
T
J
n
T
J
n
G n
n
G p n
n n
V
V
I
V
V
I
V
V
I
I I
I
I I I
I
I
I
2
exp exp
exp
0 0
0
2

Why the emitter current gain is a function the junction current? As seen in the current gain
analysis of BJT (pag. 4), is basically the injection factor of the emitter junction: = I
nE
/I
E
. At
low current we must consider for the current I
E
not only the hole and electron components but
also the generation/recombination current in the depleted layer, that has a dependence
exp(V
J
/2V
T
) from the junction voltage. In the case of SCR we have for the
2
expression:
log I

At low currents I
G
is a substantial part of
the total current I, and is low; if the
current I increases (because V
J
is
increased), the I
G
component becomes
negligible with respect to I
n
and
increases near to unity. If we then
increase the total J
3
current by injecting a
gate current I
G
,
2
will be increased from
<0.5 to about 0.8 - 0.9, and the trigger
condition will be met.
(5)
7 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
When the condition (
1
+
2
)=1 is met, we reach the turn-over point in the IV curve of the SCR,
because dI
A
/dV
AK
= 0. The total current I
A
increases without limits, because both BJTs Q
1
and
Q
2
enter into saturation mode. A positive feedback develops for this configuration, because the
increase of I
C2
current of Q
2
corresponds an increase of the drive base current I
B1
of Q
1
, and
that in turn increases I
C1
and hence the drive base current I
B2
of Q
2
, and so on.
This situation is self-sustaining (latching state), because it can be maintained even if the initial
gate current I
G
is removed, because now the base current I
B2
of Q
2
is generated by the I
C1
current of Q
1
. The I
G
current can be then a short pulse to trigger the switch from OFF to ON
state, but it does not need to be maintained after the triggering (this is an important aspect for
the drive circuitry).
If both Q
1
and Q
2
are in saturation, the N/P collector junction J
2
(for both BJT) must be forward
biased, as indicated in figure, and the whole V
AK
voltage drops to a very low value, being the
(algebric) sum of three forward biased junctions, with J
1
and J
2
summing and J
3
subtracting:
V
AK
= V
1
+ V
2
V
3
.
8 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
A K
J1 J2 J3
P
N P N
V1
V3 V2
p
n
log n,p
x
In the latching condition (ON state) the thick an low doped N
-
region is in conductivity
modulation, but in that case we have a much higher conductivity, because in this state both
the PN junctions J
1
and J
2
are in forward bias and inject holes from both sides of the N
-
region.
As a consequence the hole distribution is more flat than in BJT in saturation, as indicated in
the figure, even with thicker layers, (more than 100 m, as needed to sustain reverse voltage
of several kilovolts), and the voltage drop V across this layer is still low.
9 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
I
v
AK
IG1
IG2
I =0 G
VON VMAX
IH
From the previous analysis, we can now understand the IV characteristics of the SCR.
In reverse bias (V
AK
<0), both J
1
and J
3
are reverse biased, and J
1
will sustain the voltage up
to his breakdown voltage.
In forward bias, if we inject an I
g
pulse the turnover point is reduced and the SCR enters in a
negative resistance region (dashed line) up to the stable ON state (latching state) where the
voltage drop V
ON
is of the order of that of a single P/N diode.
For larger values of I
G
, the voltage V
AK
of the turnover point gets lower, because a lower
voltage is needed to hold the condition
1
+
2
(V,I
G
)=1.
IV characteristics of the SCR
To bring the SCR from ON to
OFF state, we must reduce
the current I
A
below the
holding value I
H
, (a value
quite low respect to the
operating current).
This is normally done by
reverting the V
AK
voltage, as
done for the PIN diode.
10 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
Turn-on dynamics
The lay-out of a typical SCR is reported in fig. (a), together with the electrical symbol for the
SCR device (fig. (b)).
For large current ratings (up to several kiloamp), the device is usually made on a single
wafer, as indicated in the figure for a wafer of 2 inches.
The small gate area with respect to the large cathode one (here the cathode lateral length is
about 2 cm), will make an important issue in the turn-on dynamics, due to the lateral
spreading of the carriers.
A
G
K
(a)
(b)
11 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
P
N
P
+
N
-
IG
IK IK
P
N
P
+
N
-
IG
IK IK
With reference to the cross-section picture
during the turn-on phase, just after the
reaching of the turnover voltage (fig. a), the
high injection region (p=n>>N
D
) in the N
-
region initially starts to develop down the
gate contact area, where the I
G
current inject
holes into the P base.
(a)
(b)
Then the conductivity modulation region
spreads laterally in the large cathode area
(fig. b), through the carrier diffusion (there is
no large lateral electric field), filling eventually
the whole cathode.
Only at that time the whole N
-
region will
present a low voltage drop, and the current
will grow up to the final transient value,
because it is equally distributed across the
whole cathode area.
12 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
a) di/dt limitation
The I
A
current constriction in the initial part of the turn-on transient poses some limitation on the
current rise time, that must be not less than the time needed for the conductivity region to
spread across the whole cathode area.
If the current rise time is lower than this di/dt|
MAX
value (usually given in the datasheet), the
current forced into the SCR will flow only in the central part of the chip, giving rise to an
excessive current density and to a strong power dissipation in that small area, with an
unacceptable temperature rise (and device failure).
The speed of the lateral carrier propagation is about some hundreds of m/s, so the current
rise time can be of the order of hundreds of s if the lateral cathode dimention is of some cm.
b) dv/dt limitation
The limitation on a dV
AK
/dt|
MAX
is due to the unwanted turn-on that could be caused by an
excessively short rise time of the anode voltage, even if there is no I
G
pulse applied (the device
should stay in OFF state). Recalling the two BJT model introduced, if the V
AK
has a rapid
increase, the current I
C1
in the off state (eq. 3) is the sum of the leakage current I
C01
and the
capacitive current in C
CG1
: I
C1
= I
C01
+ C
CG1
dV
AK
/dt. This induces an extra current I
C1
that has
the same effect in the base of Q
2
as an added I
G
.
The SCR then can be triggered in ON if that displacement current is of the order of the I
G
needed for triggering. The dV/dt limitation indicated the max allowed voltage rise time that does
not generate a critical I
C1
.
13 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
t
t
t*
VAK
IG
ID
tr td
ts
Turn-on waveforms
(a)
(b)
The turn-on dynamics of the SCR is
sketched in these plots. Starting from the
time t* when the I
G
current is applied (fig.
a), we mist consider (fig. b):
the delay time t
d
, that is the time needed
to reach the triggering condition (
1
+
2
)=1
and the turnover point.
the rise time t
r
that is the time needed to
create the high injection region and the
conductivity modulation near the gate
region
the spreading time t
s
that is the time
needed to the spreading of the high
injection region across the cathode area.
The minimum gate pulse duration must
be the sum of these three times to allow
the full switching of the SCR.
14 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
Why the SCR can not be switched OFF by reverting the gate current?
When the gate current is reversed, it will remove only the stored charges underneath the
gate contact, as indicated in figure. Then only the lateral part of junction J
2
will be reverse
biased: the spreading resistance of the thin and long P region under the cathode area will
give rise to a transverse voltage drop that will keep in forward bias most part of the
cathode/gate J
2
junction.
Then the SCR will continue to stay in its latching condition, despite of the reverse gate
current I
G
applied.
The only way to switch off the SCR, as said before, is to reduce the anode current I
A
below
the holding value I
H
.
J
2
J
2
15 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
t
t
VAK
IA
trr
t2 t3 t4 t1
Turn-off waveforms
The turn-off dynamics of the SCR when the
anode voltage (and current) is reverted, is
similar to the one of the PIN diode, because
both devices rely on the natural current
commutation (the anode voltage must be
reverted to switch off the device).
The I
A
current then start to decrease below
the I
H
value and then reverse (fig. a). The
stored charges in the wide N
-
region must
recombine and at time t
3
the I
A
current
reaches its minimum value.
The V
AK
voltage first reduces below the V
ON
(fig. b) and then start to increase in reverse
bias when the current I
A
reaches it minimum
(the N
-
region start to be depleted from
mobile charge).
(a)
(b)
16 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
The Gate Turn-Off Thyristor (GTO)
A
K
G
The GTO is basically an SCR with the capability to being switched off
even with positive anode voltages, with a negative gate current pulse
applied; it is switched on (analogously to the SCR) with a positive gate
current pulse. The electrical symbol of the GTO is reported in figure,
where the bidirectional gate current is indicated.
The ON state characteristic and the turn-on behavior is equal to the one
of the SCR so we will concentrate on the turn-off capability of this
device.
The turn-off can be initially identified in an approximate way by using the two complementary
BJT configuration seen before for the SCR. Recalling the previous analysis done, the latching
state is maintained by the regenerative feedback action of the two transistors Q
1
and Q
2
connected in current loop. To inhibit the regenerative action we must bring the transistor Q
2
out
from saturation using a negative gate current I
G
; in the active region the collector junction (J
1
)
of Q
2
will become reverse biased and the egenerative feedback is blocked.
The condition for Q
2
to exit from saturation is:
2
2
2 2
2
2
1

= > where
B
C
I
I
(6)
17 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
A
G
K
Q1
Q2
IC1
I =I C2 B1
IG
IB2
I =I A E1
I =I K E2
PNP
NPN
G A G C B
I I I I I ' '
1 1 2
= =
A C
I I ) 1 (
1 2
=
Recalling the schematic for the two transistors equivalent, we have:
Substituting (7) in (6) we have:
(7)
2
1
1
) 1 (
'

A
A G
I
I I

>
OFF
A
G A A G
I
I I I I

>

+
=

> '
1 ) 1 )( 1 (
'
2
2 1
2
2 1
1

and from the value of
2
:
1 '
2 1
2
+
= =

G
A
OFF
I
I
where a turn-off current gain: can be defined.
18 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
From the simplified one-dimensional analysis made with the two transistor equivalent, in
principle it will be possible to turn-off the SCR, but with the usual values of
1
and
2
the turn-
off is about 1, and we need to switch off the device a gate current equal to the anode one.
Moreover, this analysis neglect the transverse voltage drop due to the base spreading
resistance that inibit the SCR turn-off.
We need to change the gate and cathode structure in order to a) increase the
2
value of the
NPN transistor, so to have a turn-off 10, and b) reduce the base spreading resistance.
The main change is to resort to an interdigitated structure for the gate and cathode contacts, as
schematically indicated in the GTO cross section. The cathode is raised above the gate, and
its lateral width W
C
is contained, and the gate is contacted by an inner metallization plane. As a
result the GTO is made of many elementary cells as indicated.
P
P+ P+ P+ P+ P+ P+ P+ P+ P+
N-
N+ N+ N+ N+
N+ N+ N+ N+ N+ N+ N+ N+
GTO cross-section
W
C
19 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
P
P
+
P
+
P
+
N
-
anode
cathode
gate gate
w

N
+
N
+
N
+
With reference to the elementary cell structure of
the GTO, we can do the following comments:
a) the gate contact is obtained by etching the
top surface after P and N
+
doping on the
whole wafer; in this way the cathode has an
high periphery/area ratio and the charge
removal from the cathode region is more
effective
b) the doping and thickness of the gate layer are
chosen to obtain a quite high
2
value and a
turn-off of about 10.
c) the anode region is made of alternate P
+
and
N
+
regions (the N
+
with a lower area than the
P
+
one): the N
+
layers are named anode
shorts because they act as localized short
circuits across the anode junction J
3
. Their
role is to allow a faster removal of the stored
charge in the N
-
region and to reduce the
turn-off time.
20 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
GTO turn-off
At the beginning of the turn-off transient, the negative gate current removes the stored
charge from the gate area far from the cathode region. The high injection region with p
and n carriers then concentrates below the cathode area, and subsequently it shrinks
down due to the carrier removal through the gate lateral region, until the P/N
-
junction
(collector junction of the NPN Q
2
) becomes reverse biased, and the regenerative
feedback between Q
1
and Q
2
is blocked.
The turn-of dynamics is sketched in
this figure, with reference to the
elementary cell of the GTO.
21 University Federico II
Dept of Electronics and Telecommunications
Paolo Spirito
Power Semiconductor Devices
The anode shorts N
-
in the anode layer help in removing the stored charge accumulated
near the anode junction that would be otherwise blocked by the N/P junction. Then the
recovery time is reduced and the switching speed is increased.
As for the PIN diode, the dynamics of the turn-off is dependent also from the lifetime in the
low doped region: the combined effects of the lifetime control and of the anode shorts are
beneficial in the recovery time reduction, but are negative on the on-state voltage V
ON
in
forward conduction: as a result a trade-off relation between the reverse recovery losses and
forward steady-state losses can be defined for any GTO (and SCR) device, linking the best
choice of the relevant parametersa to the specific application (high frequency operation or
low frequency/high current operation).
It must be noted that the anode shorts will pose a significant limitation in the GTO ratings
with respect to the SCR ones: the GTO can not withstand reverse voltages (contrarily to the
SCR) because the J
1
junction is shorted by the N
+
shorts, and can not sustain reverse bias.
Asd a consequence, the GTO can be used only as a controlled switch for positive anode
voltages.