Design Article
Application-Specific Power Semiconductors Part 1 - Attack of the IGBT
Sampat Shekhawat and Bob Brockway, Fairchild Semiconductor
9/23/2009 3:12 PM EDT
Application-specific IGBTs are taking a major change in direction like any other power devices and adopting distinct bifurcated approaches. The first pertains to innovations improving the silicon and process technologies to overcome the limitations of the current technology and the other approach is towards packaging and the driving of these devices. These IGBTs are targeted for induction heating (IH), motion control, uninterruptible power supplies (UPS), welding, steel cutting, switched mode power supplies (SMPS) and renewable energy (wind power & solar inverter) market,etc. The demand for electricity is increasing and at the same time, the cost of power generation is also going up. There is increasing pressure from governmental agencies to reduce the emission of harmful gases. This is forcing equipment designers to increase efficiency and performance. Governmental agencies will set new minimum efficiency limits. Just one IGBT technology is not suitable for all of the above applications. Each application needs to use its own application-specific IGBT. This is forcing device designers to design application-specific IGBTs. Each application needs its own unique topology variation. In all these topologies, device parameters play a vital role to improve circuit efficiency and performance. Fairchild Semiconductor provides application-specific IGBTs for many types applications.
The insulated gate bipolar transistor (IGBT), also known as the conductivity-modulated field-effect transistor, is one of the most commonly available advanced switching power devices in the market today. This IGBT structure is very similar to that of MOSFET. This is because a MOSFET IGBT is also a voltage-controlled device. One difference between these two devices is the starting material. The starting material for a MOSFET is N+, where as for an IGBT it is P+. Generally the MOSFET has high resistive n-epi region so conduction losses are high. For an IGBT, on the other hand, the n-epi region is placed on P+ substrate forming a p-n junction where conductivity modulation takes place and conduction losses are reduced. This is shown as variable resistance RDRIFT shown in the first figure.
The RDRIFT is also known as modulation resistance. An IGBT is available right from a 300V voltage rating to several kilo-volts. The IGBT has a high forward conduction current density and very low drive since it is a voltage-controlled device like a MOSFET. It has significantly superior characteristics for low and medium switching frequency and some of the Fairchild Semiconductor 300V & 600V IGBTs can be used in applications up to 100 kHz. Generally, a 600V IGBT has lower conduction losses compared to a 600V MOSFET at high current operating condition. The IGBT is a minority carrier device compared to a MOSFET, which is a majority carrier device. The recombination of minority carriers at turn-off accounts for current tail and, in turn, increases turn-off losses. Due to this, there is a tradeoff between IGBT and MOSFET applications. Generally at low frequency the turn-off losses become less of an issue so it is better to choose a device that has lower conduction losses for applications such as motor drives, UPS, welding and low frequency PFC applications, etc. IGBT is ideal for these applications.
An IGBT is made up of thousands and thousands of small cells connected in parallel like a MOSFET. The equivalent circuit model of this cell is shown in the figure below. This circuit consists of P-N-P and N-P-N bipolar transistor connected like four layer parasitic thyristor. The N-P-N transistor is shunted by MOSFET structure. The resistance between MOSFET and the base of the P-N-P transistor represents the N- drift region of IGBT. The shorting resistance Rsc shunts the base and emitter of the parasitic N-P-N transistor. This resistance is so small that the base-emitter junction of this transistor does not turn-on even when very high current flows through it. One can assume that for all practical purposes this N-P-N parasitic transistor remains inactive. When an IGBT is fully on the VCE(SAT) voltage across it is given by Equation 1.
The drift resistance of IGBT is much less than the MOSFET drift resistance because an IGBT is a conductivity-modulated device. This drift voltage is less for PT IGBT.
Equivalent circuit of IGBT -- Click on image to enlarge. |
Equation 1 -- Click on image to enlarge. |
Fairchild's IGBT technologies have been optimized to reduce VCE(SAT). The trench gate eliminates the parasitic JFET resistance (RJFET) of the MOSFET part of the structure. This results in reduced VCE(SAT).. In case of field stop IGBT RDRIFT is reduced reducing VCE(SAT). whereas for trench field stop IGBT the JFET resistance is eliminated and also RDRIFT is reduced reducing VCE(SAT). The transconductance of the IGBT is very high compared to the MOSFET. This increases the short circuit current through the IGBT and one has to take action to protect IGBT under short circuit or fault conditions.
Basically, two types of planar IGBT exists, which are punch-through (PT) and non-punch-through (NPT). Both of them normally have similar top cell structures. The main difference between the two is in the vertical device structure as shown in figure below.
The PT has a thick P+ substrate injector and an N-buffer layer that controls the injection in to the N- region (base of the PNP transistor). The switching speed and VCE(SAT) of this device is controlled by controlling the buffer charge (doping) and the minority carrier lifetime (approx. 0.25us) in the N- region.
The NPT has no buffer and has a thicker N- region. The switching speed and VCE(SAT) of this device is controlled by controlling the injection efficiency of the P-emitter. This is done by making the P- emitter very shallow (<0.5um) and also by limited activation of the P- type atoms in the emitter. The NPT typically does not use lifetime control and hence the minority carrier life times in the N- region are quite long. NPT is fabricated on float zone silicon material. No epitaxial process is needed where as PT IGBT need epitaxial process.
Basic differences in characteristics with brief explanations:
1. VCE(SAT) vs EOFF trade off: PT has better trade off due to thinner N- region.
2. VCE(SAT) temperature coefficient: NPT will always have a positive temperature coefficient. This is due to the lack of any lifetime killers resulting in long life times. So NPT is good for paralleling. PT may or may not have positive temperature coefficient depending on the type of lifetime killer used and also the absolute value of the lifetime. Typically, fast devices will have negative temperature coefficient and slow parts will have positive temperature coefficient.
3. EOFF vs. temperature: With increase in temperature there is a small increase in EOFF or NPT, since life time remains constant with temperature. For PT, Eoff increases rapidly (~2x from 25°C to 150°C) due to increase in lifetime with temperature. This effect is more pronounced on the fast PT parts.
4. TFALL: NPT have high fall times due to the long tail current. This is because of the high lifetimes in the NPT. PT has extremely short fall times and therefore low EOFF losses. But in some cases, this may cause voltage and current ringing problems.
5. Eon vs. temperature: Remains almost constant for both.
Planar IGBT cell configuration -- Click on image to enlarge. |
6. UIS: NPT are inherently more rugged than PT for the same cell structure. This is because they have reasonable N- resistivities and have very thick N- layers. They normally have UIS capability close to that of a MOSFET. PT has poor UIS unless specifically designed for UIS. The UIS capability of PT devices decreases as you make them faster.
7. SCWT: NPT generally have better SCWT capability due to thicker N- layer. The extra thickness at the bottom adds resistance which helps in limiting the SC current. PT could be easily designed to have high SCWT by changing the top cell structure.
8. Both have JFET resistance
Fairchild 300V, 450V and 600V SMPS IGBTs are amongst the fastest IGBT available in the market for high frequency applications. Fairchild has recently released a new planar field stop IGBT using almost similar cell structure as a SMPS IGBT. Again there are two different types of these IGBTs that have been released. The trade-off curve for these IGBTs is the same. UF/D series planar field stop is for low frequency application where conduction loss has been reduced for moderate switching speed and soft switching IH applications. The other one is a SF/D series planar field stop IGBT for high frequency applications where conduction loss has been compromised for a low switching loss. This SF series IGBT can replace a Fairchild SMPS IGBT and is good for high frequency hard-switching applications such as solar and UPS inverters, PFC and other SMPS topologies. Efficiency is one of the most important parameters for solar inverters. An increase in efficiency helps the reduce the cost of PV solar panels with the same power output. The SFD series, in particular, being the fastest IGBT that can improve the efficiency of solar inverter.
Turn-off switching energy comparison of SFD and next best Trench Field Stop IGBT at same gate resistance -- Click on image to enlarge. |
The figure below shows another comparison of the FGH40N60SFD against its competition, IKW30N60T, at the same gate resistance and Ice of 20A. The turn-off loss of FGH40N60SFD was measured against IKW30N60T, an another low VCE(SAT) IGBT. The turn-off energy of FGH40N60SFD is 267 micro Joules where the turn-off energy for the competition is 530 micro Joules as shown in the figure below. This means a turn-off loss of FGH40N60SFD is 50% less compared to the IKW30N60T. Even though VCE(SAT) of FGH40N60SFD is still higher than IKW30N60T, the overall loss will be less for FGH40N60SFD and it will perform better at high frequency for solar and UPS inverters and mid-frequency DC/DC converters such as welding customers. In all of these applications, the selection of a free-wheeling diode (co-pack) is very important to reduce turn-on loss as well as EMI. The free wheeling-diodes for these IGBTs have been optimized for these applications. These diodes, Trr and Qrr, have been reduced at the same time the diodes are soft.
Turn-off switching energy comparison of SFD and another competition trench FS IGBT at same Rg -- Click on image to enlarge. |
Stay Tuned for Part 2!



