Professional Documents
Culture Documents
1. INTRODUCTION
Power semiconductor diode is the power level counter part of the low power
signal diodes with which most of us have some degree of familiarity. These power
devices, however, are required to carry up to several KA of current under forward bias
condition and block up to several KV under reverse biased condition. These extreme
requirements call for important structural changes in a power diode which significantly
affect their operating characteristics. These structural modifications are generic in the
sense that the same basic modifications are applied to all other low power
semiconductor devices (all of which have one or more p-n junctions) to scale up their
power capabilities. It is, therefore, important to understand the nature and implication
of these modifications in relation to the simplest of the power devices, i.e., a power
semiconductor diode.
2. Review of Basic p-n Diode Characteristics
A p-n junction diode is formed by placing p and n type semiconductor materials in
intimate contact on an atomic scale. This may be achieved by diffusing acceptor
impurities in to an n type silicon crystal or by the opposite sequence.
In an open circuit p-n junction diode, majority carriers from either side will defuse
across the junction to the opposite side where they are in minority. These diffusing
carriers will leave behind a region of ionized atoms at the immediate vicinity of the
metallurgical junction. This region of immobile ionized atoms is called the space charge
region. This process continues till the resultant electric field (created by the space
charge density) and the potential barrier at the junction builds up to sufficient level to
prevent any further migration of carriers. At this point the p-n junction is said to be in
thermal equilibrium condition. Variation of the space charge density, the electric field
and the potential along the device is shown in Fig 1.1 (a).
Fig 1.1: Space change density the electric field and the electric
potential in side a p-n junction under (a) thermal equilibrium
condition, (b) reverse biased condition, (c) forward biased
condition.
When an external voltage is applied with p side move negative then the n side
the junction is said to be under reverse bias condition. This reverse bias adds to the
height of the potential barrier. The electric field strength at the junction and the width
of the space change region (also called the depletion region because of the absence
of free carriers) also increases. On the other hand, free minority carrier densities (np in
the p side and pn in the n side) will be zero at the edge of the depletion region on
either side (Fig 1.1 (b)). This gradient in minority carrier density causes a small flux of
minority carriers to defuse towards the deletion layer where they are swept
immediately by the large electric field into the electrical neutral region of the opposite
side. This will constitute a small leakage current across the junction from the n side to
the p side. There will also be a contribution to the leakage current by the electron hole
pairs generated in the space change layer by the thermal ionization process. These two
components of current together is called the reverse saturation current Is of the
diode. Value of Is is independent of the reverse voltage magnitude (up to a certain
level) but extremely sensitive to temperature variation. When the applied reverse
voltage exceeds some threshold value (for a given diode) the reverse current increases
rapidly. The diode is said to have undergone reverse break down. Reverse break down
is caused by "impact ionization" as explained below. Electrons accelerated by the large
depletion layer electric field due to the applied reverse voltage may attain sufficient
knick energy to liberate another electron from the covalent bonds when it strikes a
silicon atom. The liberated electron in turn may repeat the process. This cascading
effect (avalanche) may produce a large number of free electrons very quickly resulting
in a large reverse current. The power dissipated in the device increases manifold and
may cause its destruction. Therefore, operation of a diode in the reverse breakdown
region must be avoided.
When the diode is forward biased (i.e., p side more positive than n side) the
potential barrier is lowered and a very large number of minority carriers are injected to
both sides of the junction. The injected minority a carrier eventually recombines with
the majority carries as they defuse further into the electrically neutral drift region. The
excess free carrier density in both p and n side follows exponential decay
characteristics. The characteristic decay length is called the "minority carrier diffusion
length" Carrier density gradients on either side of the junction are supported by a
forward current IF (flowing from p side to n side) which can be expressed as
IF = IS(exp (qv/kT )-1
(1.1)
Where, Is = Reverse saturation current ( Amps)
v = Applied forward voltage across the device (volts)
q = Change of an electron
k = Boltzmans constant
T = Temperature in Kelvin
To arrive at the structure shown in Fig 1.3 (c) a lightly doped n- epitaxial layer of
specified width (depending on the required break down voltage) and donor atom
density (NdD) is grown on a heavily doped n+ substrate (NdK donor atoms.Cm -3) which
acts as the cathode. Finally the p-n junction is formed by defusing a heavily doped (N aA
acceptor atoms.Cm-3) p+ region into the epitaxial layer. This p type region acts as the
anode. Impurity atom densities in the heavily doped cathode (Ndk .Cm -3) and anode
(NaA.Cm -3) are approximately of the same order of magnitude (10 19 Cm -3) while that
of the epitaxial layer (also called the drift region) is lower by several orders of
magnitude (NdD ~ 10 14 Cm 3) . In a low power diode this drift region is absent.
4. Switching Characteristics of Power Diodes
Power Diodes take finite time to make transition from reverse bias to forward bias
condition (switch ON) and vice versa (switch OFF). Behavior of the diode current and
voltage during these switching periods are important due to the following reasons.
Severe over voltage / over current may be caused by a diode switching at
different points in the circuit using the diode.
Voltage and current exist simultaneously during switching operation of a diode.
Therefore, every switching of the diode is associated with some energy loss. At high
switching frequency this may contribute significantly to the overall power loss in the
diode. Observed Turn ON behavior of a power Diode: Diodes are often used in circuits
with di/dt limiting inductors. The rate of rise of the forward current through the diode
during Turn ON has significant effect on the forward voltage drop characteristics. A
typical turn on transient is shown in Fig. 1.4.
Fig. 1.4: Forward current and voltage waveforms of a power diode during
Turn On
Operation.
It is observed that the forward diode voltage during turn ON may transiently
reach a significantly higher value Vfr compared to the steady slate voltage drop at the
steady current IF. In some power converter circuits (e.g voltage source inverter) where
a freewheeling diode is used across an asymmetrical blocking power switch (i.e GTO)
this transient over voltage may be high enough to destroy the main power switch.VFr
(called forward recovery voltage) is given as a function of the forward di/dt in the
manufacturers data sheet. Typical values lie within the range of 10-30V. Forward
Thyristors
Although the large semiconductor diode was a predecessor to thyristors, the
modern power electronics area truly began with advent of thyristors. One of the first
developments was the publication of the P-N-P-N transistor switch concept in 1956 by
J.L. Moll and others at Bell Laboratories, probably for use in Bells Signal application.
However, engineers at General Electric quickly recognized its significance to power
conversion and control and within nine months announced the first commercial Silicon
Controlled Rectifier in 1957. This had a continuous current carrying capacity of 25A and
a blocking voltage of 300V. Thyristors (also known as the Silicon Controlled Rectifiers or
SCRs) have come a long way from this modest beginning and now high power light
triggered thyristors with blocking voltage in excess of 6kv and continuous current
rating in excess of 4kA are available. They have reigned supreme for two entire
decades in the history of power electronics. Along the way a large number of other
devices with broad similarity with the basic thyristor (invented originally as a phase
control type device) have been developed. They include, inverter grade fast thyristor,
Silicon Controlled Switch (SCS), light activated SCR (LASCR), Asymmetrical Thyristor
(ASCR) Reverse Conducting Thyristor (RCT), Diac, Triac and the Gate turn off thyristor
(GTO).
From the construction and operational point of view a thyristor is a four layer,
three terminal, minority carrier semi-controlled device. It can be turned on by a current
signal but can not be turned off without interrupting the main current. It can block
voltage in both directions but can conduct current only in one direction. During
conduction it offers very low forward voltage drop due to an internal latch-up
mechanism. Thyristors have longer switching times (measured in tens of s) compared
to a BJT. This, coupled with the fact that a thyristor can not be turned off using a control
input, has all but eliminated thyristors in high frequency switching applications
involving a DC input (i.e., choppers, inverters). However in power frequency ac
applications where the current naturally goes through zero, thyristor remain popular
due to its low conduction loss its reverse voltage blocking capability and very low
control power requirement. In fact, in very high power (in excess of 50 MW) AC DC
(phase controlled converters) or AC AC (cyclo-converters) converters, thyristors still
remain the device of choice.
Constructional Features of a Thyristor
Fig 1.6 shows the circuit symbol, schematic construction and the photograph of a
typical thyristor.
(1.6)
Where Ico Ico1 +Ico2 is the total reverse leakage current of J2
Now as long as VAK is small Ico is very low and both 1 & 2 are much lower than
unity. Therefore, total anode current IA is only slightly greater than Ico. However, as VAK
is increased up to the avalanche break down voltage of J2, Ico starts increasing rapidly
due to avalanche multiplication process. As Ico increases both 1 & 2 increase and 1
+ 2 approaches unity. Under this condition large anode current starts flowing,
restricted only by the external load resistance. However, voltage drop in the external
resistance causes a collapse of voltage across the thyristor. The CB junctions of both Q1
& Q2 become forward biased and the total voltage drop across the device settles down
to approximately equivalent to a diode drop. The thyristor is said to be in ON state.
Just after turn ON if Ia is larger than a specified current called the Latching
Current IL, 1 and 2 remain high enough to keep the thyristor in ON state. The only
way the thyristor can be turned OFF is by bringing IA below a specified current called
the holding current (IH) where upon 1 & 2 starts reducing. The thyristor can regain
forward blocking capacity once excess stored charge at J2 is removed by application of
a reverse voltage across A & K (ie, K positive with respect A).
It is possible to turn ON a thyristor by application of a positive gate current
(flowing from gate to cathode) without increasing the forward voltage across the device
up to the forward break-over level. With a positive gate current equation 1.5 can be
written as
IK=IA+IG
(1.7)
Combining with Eqns. 1.2 to 1.4
(1.8)
Obviously with sufficiently large IG the thyristor can be turned on for any value of
Ico (and hence VAK). This is called gate assisted turn on of a Thyristor. This is the usual
method by which a thyristor is turned ON.
When a reverse voltage is applied across thyristor (i.e, cathode positive with
respect to anode.) junctions J1 and J3 are reverse biased while J2 is forward biased. Of
these, the junction J3 has a very low reverse break down voltage since both the n+ and
p regions on either side of this junction are heavily doped. Therefore, the applied
reverse voltage is almost entirely supported by junction J1. The maximum value of the
reverse voltage is restricted by
Turn ON is sufficient to effect control. The minimum gate pulse width is decided by
the external circuit and should be long enough to allow the anode current to rise above
the latching current (IL) level. The left hand side of Fig 1.7 shows the reverse i-v
characteristics of the thyristor. Once the thyristor is ON the only way to turn it OFF is by
bringing the thyristor current below holding current (IH). The gate terminal has no
control over the turn OFF process. In ac circuits with resistive load this happens
automatically during negative zero crossing of the supply voltage. This is called
natural commutation or line commutation. However, in dc circuits some
arrangement has to be made to ensure this condition. This process is called forced
commutation.
During reverse blocking if ig = 0 then only reverse saturation current (Is) flows
until the reverse voltage reaches reverse break down voltage (VBRR). At this point
current starts rising sharply. Large reverse voltage and current generates excessive
heat and destroys the device. If ig > 0 during reverses bias condition the reverse
saturation current rises as explained in the previous section. This can be avoided by
removing the gate current while the thyristor is reverse biased.
Switching Characteristics of a Thyristor
During Turn on and Turn off process a thyristor is subjected to different voltages
across it and different currents through it. The time variations of the voltage across a
thyristor and the current through it during Turn on and Turn off constitute the switching
characteristics of a thyristor.
Turn on Switching Characteristics
A forward biased thyristor is turned on by applying a positive gate voltage
between the gate and cathode as shown in Fig 1.8.
divided into three separate intervals namely, (i) delay time (td) (ii) rise time (tr) and (iii)
spread time (tp). These times are shown in Fig 1.8 for a resistive load.
Delay time (td): After switching on the gate current the thyristor will start to conduct
over the portion of the cathode which is closest to the gate. This conducting area starts
spreading at a finite speed until the entire cathode region becomes conductive. Time
taken by this process constitute the turn on delay time of a thyristor. It is measured
from the instant of application of the gate current to the instant when the anode
current rises to 10% of its final value (or VAK falls to 90% of its initial value). Typical
value of td is a few micro seconds.
Rise time (tr): For a resistive load, rise time is the time taken by the anode current
to rise from 10% of its final value to 90% of its final value. At the same time the voltage
VAK falls from 90% of its initial value to 10% of its initial value. However, current rise
and voltage fall characteristics are strongly influenced by the type of the load. For
inductive load the voltage falls faster than the current. While for a capacitive load VAK
falls rapidly in the beginning. However, as the current increases, rate of change of
anode voltage substantially decreases.
If the anode current rises too fast it tends to remain confined in a small area. This can
give rise to local hot spots and damage the device. Therefore, it is necessary to limit
the rate of rise of the ON state current (dia/dt) by using an inductor in series with the
device. Usual values of maximum allowable dia/dt is in the range of 20-200 A/s.
Spread time (tp): It is the time taken by the anode current to rise from 90% of its
final value to 100%. During this time conduction spreads over the entire cross section
of the cathode of the thyristor. The spreading interval depends on the area of the
cathode and on the gate structure of the thyristor.
Turn off Switching Characteristics
Once the thyristor is on, and its anode current is above the latching current level
the gate loses control. It can be turned off only by reducing the anode current below
holding current. The turn off time tq of a thyristor is defined as the time between the
instant anode current becomes zero and the instant the thyristor regains forward
blocking capability. If forward voltage is applied across the device during this period the
thyristor turns on again. During turn off time, excess minority carriers from all the four
layers of the thyristor must be removed. Accordingly tq is divided in to two intervals,
the reverse recovery time (trr) and the gate recovery time (tqr). Fig 1.9 shows the
variation of anode current and anode cathode voltage with time during turn off
operation on an expanded scale.
the thyristor (and hence allow them to operate at higher switching frequency).
However, large voltage spike due to this snappy recovery will appear across the
device after the device turns off. Typical turn off times of converter and inverter grade
thyristors is in the range of 50-100 s and 5-50 s respectively. As has been
mentioned in the introduction thyristor is the device of choice at the very highest
power levels. At these power levels (several hundreds of megawatts) reliability of the
thyristor power converter is of prime importance. Therefore, suitable protection
arrangement must be made against possible overvoltage, over current and unintended
turn on for each thyristor. At the highest power level (HVDC transmission system)
thyristor converters operate from network voltage levels in excess of several hundreds
of kilo volts and conduct several tens of kilo amps of current. They usually employ a
large number of thyristors connected in series parallel combination. For maximum
utilization of the device capacity it is important that each device in this series parallel
combination share the blocking voltage and on state current equally. Special equalizing
circuits are used for this purpose.
Triac
The Triac is a member of the thyristor family. But unlike a thyristor which conducts only
in one direction (from anode to cathode) a triac can conduct in both directions. Thus a
triac is similar to two back to back (anti parallel) connected thyristosr but with only
three terminals. As in the case of a thyristor, the conduction of a triac is initiated by
injecting a current pulse into the gate terminal. The gate looses control over conduction
once the triac is turned on. The triac turns off only when the current through the main
terminals become zero. Therefore, a triac can be categorized as a minority carrier, a
bidirectional semi-controlled device. They are extensively used in residential lamp
dimmers, heater control and for speed control of small single phase series and
induction motors.
Construction and operating principle
Fig. 1.10 (a) and (b) show the circuit symbol and schematic cross section of a triac
respective. As the Triac can conduct in both the directions the terms anode and
cathode are not used for Triacs. The three terminals are marked as MT1 (Main Terminal
1), MT2 (Main Terminal 2) and the gate by G. As shown in Fig 4.12 (b) the gate terminal
is near MT1 and is connected to both N3 and P2 regions by metallic contact. Similarly
MT1 is connected to N2 and P2 regions while MT2 is connected to N4 and P1 regions.
Fig. 1.10: Circuit symbol and schematic construction of a Triac (a) Circuit
symbol (b) Schematic construction.
Since a Triac is a bidirectional device and can have its terminals at various
combinations of positive and negative voltages, there are four possible electrode
potential combinations as given below
1. MT positive with respect to MT , G positive with respect to MT
2
1
1
2. MT positive with respect to MT , G negative with respect to MT
2
1
1
3. MT negative with respect to MT , G negative with respect to MT
2
1
1
4. MT negative with respect to MT , G positive with respect to MT
2
1
1
The triggering sensitivity is highest with the combinations 1 and 3 and are generally
used. However, for bidirectional control and uniforms gate trigger mode sometimes
trigger modes 2 and 3 are used. Trigger mode 4 is usually averded. Fig 1.11 (a) and (b)
explain the conduction mechanism of a triac in trigger modes 1 & 3 respectively.
From a functional point of view a triac is similar to two thyristors connected in anti
parallel. Therefore, it is expected that the V-I characteristics of Triac in the 1st and 3rd
quadrant of the V-I plane will be similar to the forward characteristics of a thyristors. As
shown in Fig. 1.12, with no signal to the gate the triac will block both half cycle of the
applied ac voltage provided its peak value is lower than the break over voltage (VBO) of
the device. However, the turning on of the triac can be controlled by applying the gate
trigger pulse at the desired instance. Mode-1 triggering is used in the first quadrant
where as Mode-3 triggering is used in the third quadrant. As such, most of the thyristor
characteristics apply to the triac (ie, latching and holding current). However, in a triac
the two conducting paths (from MT1 to MT2 or from MT1 to MT1) interact with each other
in the structure of the triac. Therefore, the voltage, current and frequency ratings of
triacs are considerably lower than thyristors. At present triacs with voltage and current
ratings of 1200V and 300A (rms) are available. Triacs also have a larger on state
voltage drop compared to a thyristor. Manufacturers usually specify characteristics
curves relating rms device current and maximum allowable case temperature as shown
in Fig 1.13. Curves relating the device dissipation and RMS on state current are also
provided for different conduction angles.
Fig. 1.14: Circuit symbol and schematic cross section of a GTO (a) Circuit
Symbol, (b) Anode shorted GTO structure, (c) Buffer layer GTO structure.
Like a thyristor, a GTO is also a four layer three junction p-n-p-n device. In order
to obtain high emitter efficiency at the cathode end, the n+ cathode layer is highly
doped. Consequently, the break down voltage of the function J3 is low (typically 2040V). The p type gate region has conflicting doping requirement. To maintain good
emitter efficiency the doping level of this layer should be low, on the other hand, from
the point of view of good turn off properties, resistively of this layer should be as low as
possible requiring the doping level of this region to be high. Therefore, the doping level
of this layer is highly graded. Additionally, in order to optimize current turn off
capability, the gate cathode junction must be highly interdigitated. A 3000 Amp GTO
may be composed of up to 3000 individual cathode segments which are a accessed via
a common contact. The most popular design features multiple segments arranged in
concentric rings around the device center.
The maximum forward blocking voltage of the device is determined by the
doping level and the thickness of the n type base region next. In order to block several
kv of forward voltage the doping level of this layer is kept relatively low while its
thickness is made considerably higher (a few hundred microns). Beyond the maximum
allowable forward voltage either the electric field at the main junction (J2) exceeds a
critical value (avalanche break down) or the n base fully depletes, allowing its electric
field to touch the anode emitter (punch through).
The junction between the n base and p+ anode (J1) is called the anode junction.
For good turn on properties the efficiency of this anode junction should be as high as
possible requiring a heavily doped p+ anode region. However, turn off capability of
such a GTO will be poor with very low maximum turn off current and high losses. There
are two basic approaches to solve this problem.
In the first method, heavily doped n+ layers are introduced into the p+ anode
layer. They make contact with the same anode metallic contact. Therefore, electrons
traveling through the base can directly reach the anode metal contact without causing
hole injection from the p+ anode. This is the classic anode shorted GTO structure as
shown in Fig 1.14 (b). Due to presence of these anode shorts the reverse voltage
blocking capacity of GTO reduces to the reverse break down voltage of junction J3 (2040 volts maximum). In addition a large number of anode shorts reduces the efficiency
of the anode junction and degrades the turn on performance of the device. Therefore,
the densities of the anode shorts are to be chosen by a careful compromise between
the turn on and turn off performance.
In the other method, a moderately doped n type buffer layer is juxtaposed
between the n- type base and the anode. As in the case of a power diode and BJT this
relatively high density buffer layer changes the shape of the electric field pattern in the
n- base region from triangular to trapezoidal and in the process, helps to reduce its
width drastically. However, this buffer layer in a conventional anode shorted GTO
structure would have increased the efficiency of the anode shorts. Therefore, in the
new structure the anode shorts are altogether dispensed with and a thin p+ type layer
is introduce as the anode. The design of this layer is such that electrons have a high
probability of crossing this layer without stimulating hole injection. This is called the
Transparent emitter structure and is shown in Fig 1.14 (c).
continues to flow in the form of a current tail for a further duration of ttail. A GTO is
normally used with a R-C turn off snubber. Therefore, VAK does not start to rise
appreciably till tf. At this point VAK starts rising rapidly and exceeds the dc voltage Vd
(VdM) (due to resonance of snubber capacitor with didt limiting inductor) before setting
down at its steady value Vd . A GTO should not be retriggered within a minimum off
period off (min) to avoid the risk of failure due to localized turn ON. GTOs have typically
low turn off gain in the range of 4-5.
transistor. Power transistors have poor reverse voltage withstanding capability due to
low break down voltage of the base-emitter junction. Therefore, reverse voltage (C
negative with respect to E) should not appear across a power transistor.
In the active region the ratio of collector current to base current (DC current Gain
()) remains fairly constant up to certain value of the collector current after which it
falls off rapidly. Manufacturers usually provide a graph showing the variation of as a
function of the collector current for different junction temperatures and collector
emitter voltages. This graph is useful for designing the base drive of a Power transistor.
Typically, the value of the dc current gain of a Power transistor is much smaller
compared to their signal level counterpart.
The maximum collector-emitter voltage that a power transistor can withstand in
active region is determined by the Base collector avalanche break down voltage. This
voltage, denoted by VSUS in Fig, 1.18 is usually smaller than VCEO. The voltages VSUS can
be attained only for relatively lower values of collector current. At higher collector
current the limit on the total power dissipation defines the boundary of the allowable
active region as shown in
Fig 1.18.
At still higher levels of collector currents the allowable active region is further
restricted by a potential failure mode called the Second break down. It appears on
the output characteristics of the BJT as a precipitous drop in the collector-emitter
voltage at large collector currents. The collector voltage drop is often accompanied by
significant rise in the collector current and a substantial increase in the power
dissipation. Most importantly this dissipation is not uniformly spread over the entire
volume of the device but is concentrated in highly localized regions. This localized
heating is a combined effect of the intrinsic non uniformity of the collector current
density distribution across the cross section of the device and the negative
temperature coefficient of resistively of minority carrier devices which leads to the
formation of current filamements (localized areas of very high current density) by a
positive feed-back mechanism. Once current filaments are formed localized thermal
runaway quickly takes the junction temperature beyond the safe limit and the device
is destroyed.
It is in the saturation region that the output characteristics of a Power transistor
differ significantly from its signal level counterpart. In fact the saturation region of a
Power transistor can be further subdivided into a quasi saturation region and a hard
saturation region. Appearance of the quasi saturation region in the output
characteristics of a power transistor is a direct consequence of introducing the drift
region into the structure of a power transistor. In the quasi saturation region the basecollector junction is forward biased but the lightly doped drift region is not completely
shorted out by excess minority carrier injection from the base. The resistivity of this
region depends to some extent on the base current. Therefore, in the quasi saturation
region, the base current still retains some control over the collector current although
the value of decreases significantly. Also, since the resistivity of the drift region is still
significant the total voltage drop across the device in this mode of operation is higher
for a given collector current compared to what it will be in the hard saturation region.
In the hard saturation region base current looses control over the collector current
which is determined entirely by the collector load and the biasing voltage VCC. This
behavior is similar to what happens in a signal transistor except that the drift region of
a power transistor continues to offer a small resistance even when it is completely
shorted out (by excess carrier injection from the base). Therefore, for larger collector
currents the collector-emitter voltage drop is almost proportional to the collector
current. Manufacturers usually provide the plots of the variation of VCE (sat) vs. iC for
different values of base current and junction temperature. Curves showing the variation
of VCE (sat) with iB for different values of iC and junction temperature are also provided
by certain manufacturers.
Applicable operating limits on a power transistors are compactly represented in two
diagrams called the Forward Bias Safe Operating Area (FBSOA) and the Reverse Bias
Safe Operating Area. (RBSOA) applicable to iB > 0 and iB 0 conditions respectively.
Typical safe operating areas of power transistors are shown in Fig 1.19.
The third limit of the FBSOA (red line) arises due to the second break down failure
mode of a Power transistor. It shows the limiting combinations of collector voltage and
current so that second break down does not occur. On the log log scale of the FBSOA
this limit also appears as a straight limit. Like the maximum power dissipation limit, the
second break down limit is also different for DC and Pulsed operation of different
pulse durations. The interpretation of the pulse duration (marked on the right side of
Fig 1.19 (a)) corresponding to a particular limit is also same.
The final limit of the FBSOA corresponds to the forward biased avalanche break
down voltage (VSUS) of the transistor and appear as a vertical line in the FBSOA at VCE =
VSuS
The FBSOA of a Power transistor is given at a specified case temperature. Both the
maximum power dissipation limit and the second break down limits are to be rated as
per the rating characteristics provided by the manufacturers when the case
temperature exceeds the specified value.
In contrast to the FBSOA, the RBSOA (Fig 1.19 (b)) is plotted on a linear scale and
has a more rectangular shape. RBSOA is a switching SOA since a transistor cannot
conduct current for any appreciable duration under reverse biased condition. It
essentially shows the limiting permissible combinations of VCE & iC with base emitter
junction reverse biased. The upper horizontal limit corresponds to the maximum
allowable collector current (ICM) and is same as that in the FBSOA. The right hand side
vertical limit corresponds the avalanche break down voltage of the transistor with
reverse bias. If the base terminal is open (i,e, iB = 0) then this voltage is VCEO. If a
negative voltage is applied across the BE junction the right hand side limit of the
RBSOA increases somewhat to the value VCBO at low value of the collector current.
In addition to the applicable limits on the output characteristics as represented in
the FBSOA and the RBSOA, limiting specification with respect to the base emitter
junction is also provided by the manufacturer. Typical specifications that are provided
are
VEBO : This is maximum allowable reverse bias voltage across the B-E junction
IB
: Maximum allowable average base current at a given case temperature.
IBM : Maximum allowable peak base current at a given case temperature and of
specified
pulse duration.
The input characteristics (iB Vs VBE) at a given case temperature is also provided.
Switching characteristics of a Power Transistor
Turn On characteristics of a Power Transistor
From the description of the basic operating principle of a power transistor presented
in the previous sections it is clear that minority carriers must be moved across different
regions of a power transistor in order to make it switch between cut off and saturation
regions of operation. The time delay in the switching operation of a power transistor is
due to the time taken by the minority carriers to reach appropriate density levels in
different regions. The exact level of minority carrier densities (and depletion region
widths) required for proper switching is determined by the collector current and biasing
collector voltage during switching, both of which are determined by external circuits.
The rate at which these densities are attained is determined by the base current
waveform. Therefore, the switching characteristics of a power transistor is always
specified in relation to the external load circuit and the base current waveform as
shown in Fig 3.6 which shows a clamped inductive switching circuit with a flat base
drive.
state values. During this period, called the Turn ON delay time no appreciable
collector current flows. The values of iO and VCE remains essentially at their OFF state
levels.
At the end of the delay time (td ON) the minority carrier density at the base region
quickly approaches its steady state distribution and the collector current starts rising
while the diode current (id) starts falling. At t = tdON + tri the collector current becomes
equal to the load current (and id becomes zero) IL. At this point D starts blocking reverse
voltage and VCE becomes unclamped. tri is called the current rise time of the transistor.
At the end of the current rise time the diode D regains reverse blocking capacity.
The collector voltage VCE which has so far been clamped to VCC because of the
conducting diode D starts falling towards its saturation voltage VCE (sat). The initial
fall of VCE is rapid. During this period the switching trajectory traverses through the
active region of the output characteristics of the transistor. At the end of this rapid fall
(tfv1) the transistor enters quasi saturation region. The fall of VCE in the quasi
saturation region is considerably slower. At the end of this slow fall (tfv2) the transistor
enters hard saturation region and the collector voltage settles down to the saturation
voltage level VCE (sat) corresponding to the load current IL. Turn ON process ends here.
The total turn on time is thus, TSW (ON) = td (ON) + tri + tfv1 +tfv2.
Power loss occurs at all time during the operation of a power transistor. However,
the collector leakage current is usually negligibly small and power loss due it can be
safely neglected in comparison to the power loss during ON condition. Power loss
occurs during Turning ON a Power transistor due to simultaneous existence of non-zero
VCE and ic during tri, tfv1, and tfv2. The energy lost during these periods is called the Turn
ON loss and given by the area under the Pl curve in Fig 1.20 (b). The average Turn ON
loss is obtained by dividing this area by (tri + tfv1 + tfv2). For safe Turn ON this average
power loss must be less than the limit set on the maximum power dissipation in the
FBSOA corresponding to a pulse width greater than tri + tfv1 + tfv2. Similar restriction with
respect to second break down should also be observed.
Turn ON time can be reduced by increasing the base current. However large base
current increases the quantity of excess carrier in the base and collector drift region
which has to be removed during Turn Off. As will be seen later this increases the Turn
OFF time. The Turn ON delay time can however be reduced by boosting the base
current at the beginning of the Turn ON process. This can be achieved by connecting a
small capacitance across RB. This increases the rate of rise of VBE & iB. Therefore, Turn
ON delay time decreases. However, in steady state iB settle downs to a value
determined by RB & VBB and no adverse effect on the Turn OFF time is observed.
In figure 1.20 (b) the reverse recovery current of D has been neglected. If this
current is not negligible then for safe Turn ON operation the sum of the load current and
the diode reverse recovery current must be less than the ICM rating of the transistor.
Thermal and second break down limits must also be observed.
It should be noted that there is some power loss at the BE junction as well. This
power loss depends on the current gain of the transistor during hard saturation. Since
current gain reduces during saturation (typically between 5 to 10) this power loss may
become significant. Manufacturers usually provide the values of td (ON), tri, tfv as
functions of ic for a given base current and case temperature.
Turn Off Characteristics of a Power Transistor
During Turn OFF a power transistor makes transition from saturation to cut off
region of operation. Just as in the case of Turn ON, substantial redistribution of minority
charge carriers is involved in the Turn OFF process. Idealized waveforms of several
important variables in the clamped inductive switching circuit of Fig. 1.20 (a) during the
Turn OFF process of Q are shown in Fig 1.21 (a)
through the collector and the collector current starts falling. At the end of the current
fall time tfi the collector current becomes zero and the load current freewheels through
the diode D. Turn OFF process of the transistor ends at this point. The total Turn OFF
time is given by Ts (OFF) = ts + trv1 + trv2 + tfi
As in the case of Turn ON considerable power loss takes place during Turn OFF due to
simultaneous existence of ic and VCE in the intervals trv1, trv2 and tfi. The last trace of Fig
1.21 (a) shows the instantaneous power loss profile during these intervals. The total
energy last per turn off operation is given by the area under this curve. For safe turn off
the average power dissipation during trv1 + trv2 + tfi should be less than the power
dissipation limit set by the FBSOA corresponding to a pulse width greater than trv1 + trv2
+ tfi. Turn OFF time intervals of a power transistor are strongly influenced by the
operating conditions and the base drive design. Manufacturers usually specify these
values as functions of collector current for given positive and negative base current and
case temperatures. Variations of these time intervals as function of the ratio of positive
to negative base currents for different collector currents are also specified.
In this section and the precious one inductive load switching has been considered.
However, if the load is resistive. The freewheeling diode D will not be used. In that case
the collector voltage (VCE) and collector current (ic) will fall and rise respectively
together during Turn ON and rise and fall respectively together during Turn OFF. Other
characteristics of the switching process will remain same. The switching Power loss in
this case will also be substantially lower.
At first glance it would appear that there is no path for any current to flow
between the source and the drain terminals since at least one of the p n junctions
(source body and body-Drain) will be reverse biased for either polarity of the applied
voltage between the source and the drain. There is no possibility of current injection
from the gate terminal either since the gate oxide is a very good insulator. However,
application of a positive voltage at the gate terminal with respect to the source will
convert the silicon surface beneath the gate oxide into an n type layer or channel,
thus connecting the Source to the Drain as explained next. The gate region of a
MOSFET which is composed of the gate metallization, the gate (silicon) oxide layer and
the p-body silicon forms a high quality capacitor. When a small voltage is application to
this capacitor structure with gate terminal positive with respect to the source (note that
body and source are shorted) a depletion region forms at the interface between the Si O2
and the silicon as shown in Fig 1.24 (a).
1.24: Gate control of MOSFET conduction. (a) Depletion layer formation; (b)
Free electron accumulation; (c) Formation of inversion layer.
The positive charge induced on the gate metallization repels the majority hole
carriers from the interface region between the gate oxide and the p type body. This
exposes the negatively charged acceptors and a depletion region is created. Further
increase in VGS causes the depletion layer to grow in thickness. At the same time the
electric field at the oxide-silicon interface gets larger and begins to attract free
electrons as shown in Fig 1.24 (b). The immediate source of electron is electron-hole
generation by thermal ionization. The holes are repelled into the semiconductor bulk
ahead of the depletion region. The extra holes are neutralized by electrons from the
source.
As VGS increases further the density of free electrons at the interface becomes
equal to the free hole density in the bulk of the body region beyond the depletion layer.
The layer of free electrons at the interface is called the inversion layer and is shown in
Fig 1.24 (c). The inversion layer has all the properties of an n type semiconductor and is
a conductive path or channel between the drain and the source which permits flow of
current between the drain and the source. Since current conduction in this device takes
place through an n- type channel created by the electric field due to gate source
voltage it is called Enhancement type n-channel MOSFET.
The value of VGS at which the inversion layer is considered to have formed is
called the GateSource threshold voltage VGS (th). As VGS is increased beyond VGS(th)
the inversion layer gets somewhat thicker and more conductive, since the density of
free electrons increases further with increase in VGS. The inversion layer screens the
depletion layer adjacent to it from increasing VGS. The depletion layer thickness now
remains constant.
Steady state output i-v characteristics of a MOSFET
The MOSFET, like the BJT is a three terminal device where the voltage on the gate
terminal controls the flow of current between the output terminals, Source and Drain.
The source terminal is common between the input and the output of a MOSFET. The
output characteristics of a MOSFET are then a plot of drain current (iD) as a function of
the Drain Source voltage (VDS) with gate source voltage (VGS) as a parameter. Fig 1.25
(a) shows such a characteristics.
Fig. 1.25: Output i-v characteristics of a Power MOSFET (a) i-v characteristics;
(b) Components of ON-state resistance; (c) Electron drift velocity vs Electric
field; (d) Transfer
With gate-source voltage (VGS) below the threshold voltage (VGS (th)) the MOSFET
operates in the cut-off mode. No drain current flows in this mode and the applied drain
source voltage (VDS) is supported by the body-collector p-n junction. Therefore, the
maximum applied voltage should be below the avalanche break down voltage of this
junction (VDSS) to avoid destruction of the device.
When VGS is increased beyond VGS (th) drain current starts flowing. For small
values of VDS
(VDS < (VGS VGS (th)) iD is almost proportional to VDS. Consequently this mode of
operation is called ohmic mode of operation. In power electronic applications a
MOSFET is operated either in the cut off or in the ohmic mode. The slope of the VDS iD
characteristics in this mode is called the ON state resistance of the MOSFET (rDS (ON)).
Several physical resistances as shown in Fig 1.25 (b) contribute to rDS (ON). Note that rDS
(ON) reduces with increase in VGS. This is mainly due to reduction of the channel
resistance at higher value of VGS. Hence, it is desirable in power electronic applications,
to use as large a gate-source voltage as possible subject to the dielectric break down
limit of the gate-oxide layer.
At still higher value of VDS (VDS > (VGS VGS (th)) the iD VDS characteristics
deviates from the linear relationship of the ohmic region and for a given vGS, iD tends
to saturate with increase in VDS. The exact mechanism behind this is rather complex. It
will suffice to state that, at higher drain current the voltage drop across the channel
resistance tends to decrease the channel width at the drain drift layer end. In addition,
at large value of the electric field, produced by the large Drain Source voltage, the
drift velocity of free electrons in the channel tends to saturate as shown in Fig 1.25 (c).
As a result the drain current becomes independent of VDS and determined solely by the
gate source voltage VGS. This is the active mode of operation of a MOSFET. Simple,
first order theory predicts that in the active region the drain current is given
approximately by
(1.9)
Where K is a constant determined by the device geometry.
At the boundary between the ohmic and the active region
(1.10)
Therefore,
(1.11)
Equation (1.11) is shown by a dotted line in Fig 1.25 (a). The relationship of Equation
(1.9) applies reasonably well to logic level MOSFETs. However, for power MOSFETs the
transfer characteristics
(iD Vs VGS) is more linear as shown in Fig 1.25 (d).At this point the similarity of the output
characteristics of a MOSFET with that of a BJT should be apparent. Both of them have
three distinct modes of operation, namely, (i)cut off, (ii) active and (iii) ohmic
(saturation for BJT) modes. However, there are some important differences as well.
Unlike BJT a power MOSFET does not undergo second break down.
The primary break down voltage of a MOSFET remains same in the cut off and in the
active modes. This should be contrasted with three different breakdown voltages
(VSUS, VCEO & VCBO) of a BJT.
The ON state resistance of a MOSFET in the ohmic region has positive temperature
coefficient which allows paralleling of MOSFET without any special arrangement for
current sharing. On the other hand, VCE (sat) of a BJT has negative temperature
coefficient making parallel connection of BJTs more complicated.
As in the case of a BJT the operating limits of a MOSFET are compactly
represented in a Safe Operating Area (SOA) diagram as shown in Fig 1.26. As in the
case of the FBSOA of a BJT the SOA of a MOSFET is plotted on a log-log graph. On the
top, the SOA is restricted by the absolute maximum permissible value of the drain
current (IDM) which should not be exceeded even under pulsed operating condition. To
the left, operating restriction arise due to the non zero value of rDS(ON) corresponding
to VGS = VGS(Max). To the right, the first operating restriction is due to the limit on the
maximum permissible junction temperature rise which depends on the power
dissipation inside the MOSFET. This limit is different for DC (continuous) and pulsed
operation of different pulse widths. As in the case of a BJT the pulsed safe operating
areas are useful for shaping the switching trajectory of a MOSFET. A MOSFET does not
undergo second break down and no corresponding operating limit appears on the
SOA. The final operation limit to the extreme right of the SOA arises due to the
maximum permissible drain source voltage (VDSS) which is decided by the avalanche
break down voltage of the drain -body p-n junction. This is an instantaneous limit.
There is no distinction between the forward biased and the reverse biased SOAs for the
MOSFET. They are identical.
(1.12)
The fall of VDS occurs in two distinct intervals. When the MOSFET is in the active region
(VDS > (VGS VGS (th)) CGD = CGD1.Since CGD1 << CGD2, VDS falls rapidly. This fast fall time of
VDS is marked tfv1 in Fig 1.27(b). However, once in the ohmic region, CGD = CGD2 >>CGD1.
Therefore, rate of fall of VDS slows down considerably (tfv2). Once VDS reaches its on state
value (rDS(ON) Io) VDS becomes unclamped and increases towards Vgg with a time
constant 2 = Rg (CGS + CGD2). Note that all switching periods can be reduced by
increasing Vgg or / and decreasing Rg. The total turn on time is tON = td(ON) + tri + tfv1 +
tfv2.
To turn the MOSFET OFF, Vgg is reduced to zero triggering the exact reverse process of
turn on to take place. The corresponding waveforms and switching intervals are show in
Fig 1.27(b). The total turn off time toff = td(off) + trv1 + trv2 + tfi.
MOSFET of the Darlington configuration remains off and hence the output p-n-p
transistor also remains off.
When the gate emitter voltage exceeds the threshold, an inversion layer forms in
the p type body region under the gate. This inversion layer (channel) shorts the emitter
and the drain drift layer and electron current flows from the emitter through this
channel to the drain drift region. This in turn causes substantial hole injection from the
p+ type collector to the drain drift region. A portion of these holes recombine with the
electrons arriving at the drain drift region through the channel. The rest of the holes
cross the drift region to reach the p type body where they are collected by the source
metallization.
Most IGBTs are designed to with stand this fault current for a few microseconds within
which the device must be turned off to prevent destruction of the device.
It is interesting to note that an IGBT does not exhibit a BJT-like second break
down failure. Since, in an IGBT most of the collector current flows through the drive
MOSFET with positive temperature coefficient the effective temperature coefficient of
VCE in an IGBT are slightly positive. This helps to prevent second break down failure of
the device and also facilitates paralleling of IGBTs.
Switching characteristics of IGBT
Switching characteristics of the IGBT will be analyzed with respect to the clamped
inductive switching circuit shown in Fig 1.31(a). The equivalent circuit of the IGBT
shown in Fig 1.31 (b) will be used to explain the switching waveforms.
increasing till VgE reaches the threshold voltage VgE (th). Thereafter, ic increases
following the transfer characteristics of the device till VgE reaches a value VgE IL
corresponding to ic = iL. This period is called the current rise time tri. The freewheeling
diode current falls from IL to zero during this period. After ic reaches IL, VgE becomes
clamped at VgE IL similar to a MOSFET. VCE also starts falling during this period. First VCE
falls rapidly (tfv1) and afterwards the fall of VCE slows down considerably. Two factors
contribute to the slowing down of voltage fall. First the gate-drain capacitance Cgd will
increase in the MOSFET portion of the IGBT at low drain-source voltages. Second, the
pnp transistor portion of the IGBT traverses the active region to its on state more slowly
than the MOSFET portion of the IGBT. Once the pnp transistor is fully on after tfv2, the on
state voltage of the device settles down to VCE (sat). The turn ON process ends here.
The turn off process of an IGBT follows the inverse sequence of turn ON with one
major difference. Once VgE goes below VgE (th) the drive MOSFET of the IGBT equivalent
circuit turns off. During this period (tfi1) the device current falls rapidly. However, when
the drive MOSFET turns off, some amount of current continues of flow through the
output p-n-p transistor due to stored charge in its base. Since there is no reverse
voltage applied to the IGBT terminals that could generate a negative drain current,
there is no possibility for removing the stored charge by carrier sweep-out. The only
way these excess carriers can be removed is by recombination within the IGBT. During
this recombination period (tfi2) the remaining current in the IGBT decays relatively
slowly forming a current fail. A long tfi2 is undesirable, because the power dissipation in
this interval will be large due to full collector-emitter voltage. tfi2 can be reduced by
decreasing the excess carrier life time in the p-n-p transistor base. However, in the
process, on state losses will increase. Therefore, judicious design tradeoffs are made in
a practical IGBT to give minimum total loss.