Future Electronics – How Far Has New Technology Allowed the MOSFET to Encroach on the IGBT’s Territory?

By: Daniele Viganò, Power and Linear Specialist Engineer, Future Electronics (Italy)

The two dominant types of power transistor, the MOSFET and the IGBT, are so familiar and have been used in power-system designs for so many years that it is easy to assume that the differences between them remain unchanged from one year to the next.

Very broadly, this assumption is correct: MOSFETs, as they have always done, support faster switching speeds and higher efficiency, but are less rugged and have lower maximum current ratings. IGBTs switch more slowly and have higher switching and conduction losses, but are more rugged and handle higher peak- and continuous-current values.

The general rule governing the choice of MOSFET or IGBT is constant, then, and for most applications it will be obvious which is the more suitable device. But in fact both device types are constantly evolving, thanks to the continuous product and technology development programs of the main suppliers, such as STMicroelectronics, ON Semiconductor and Fairchild. And so the grey area in which an application might be served equally well by an IGBT or by a MOSFET gradually shifts its position as first one then the other device type gains a new feature or benefits from a performance enhancement.

By explaining the operating characteristics of the latest generation of MOSFETs and IGBTs, this article gives the user a better understanding of the application requirements which best suit each device type, clarifying the nature of today’s grey area for power transistor choices.

In Search of Speed and Efficiency
The developments in IGBTs and MOSFETs are in large part aimed at increasing switching speed for more precise and accurate control of power output and for quicker responses to transient power loads – and at increasing efficiency, by reducing switching and conduction losses.

In the bi-polar transistors, this process has concentrated on ameliorating the relatively slow turn-off characteristics (which give rise to the device’s large ‘current tail’). In addition, IGBT manufacturers worked to reduce the collector-emitter saturation voltage, referred to in datasheets as VCE(sat), which determines the device’s on-state voltage – in other words, its conduction loss.

While early versions of the IGBT suffered from disadvantages including a large current tail and a tendency to latch up, in the latest generation of IGBTs these problems have been largely eliminated.

Another problem with some early IGBT types was their negative temperature co-efficient, which could lead to thermal run-away: this makes it hard to run multiple devices in parallel to provide a high power output.

pg22_TV_image1

Figure 1: Comparison of the structure of a planar (left) and a trench gate field stop IGBT (right).
(Source: STMicroelectronics)

The solution to this problem came with the development of, first, Punch-Through (PT) planar technology, then Non-Punch-Through (NPT) planar, and then in today’s IGBTs, trench gate field stop technology (see Figure 1). These technologies for wafer fabrication have enabled manufacturers to continually reduce the mass of silicon inside the device. This has the advantages of:
• Reducing unit cost – because more devices can be cut from each wafer
• Enabling faster switching
• Reducing the length of the current tail for lower switching losses
• Reducing the collector-emitter saturation voltage

The reduction in power dissipation has the effect of enabling an increase in power density, so that today’s IGBTs can handle as much as 50% higher average currents than the first IGBTs. Parts benefiting from the latest IGBT technology are shown in Figure 2.

ManufacturerIGBT Family designationSwitching SpeedVoltage Rating
STMicroelectronicsH seriesUp to 30kHz600V
STMicroelectronicsHB seriesUp to 60kHz650V
STMicroelectronicsV seriesUp to 80kHz650V
FairchildSHD seriesUp to 60kHz650V
ON SemiconductorL2 familyUp to 20kHz600V
ON SemiconductorFL2 familyUp to 50kHz650V
ON SemiconductorIHR familySoft switchingUp to 1,350V
STMicroelectronicsM seriesUp to 20kHz1,200V
STMicroelectronicsH seriesUp to 50kHz1,200V
FairchildSMD seriesUp to 60kHzUp to 1,200V
ON SemiconductorL2 familyUp to 20kHz1,200V
ON SemiconductorFL2 familyUp to 50kHz1,200V

Figure 2: The latest families of IGBTs offering low losses and high switching speeds

MOSFET Technology: Ever Lower On-Resistance
Like IGBTs, MOSFETs have been through many evolutions over the past two decades.

In the early years, the structure of a MOSFET was planar: the gate pin was placed horizontally on the silicon body of the device. Newer MOSFETs have benefited enormously from the drastic improvements achieved by trench gate technology, and by the introduction of vertical super-junctions. In these new technologies, the gate pin is embedded more deeply inside the bulk of the silicon material, enabling it to make far better use of the available silicon.

As a result, trench technology has become the preferred structure for MOSFETs, even though planar devices are still available on the market.

pg21_Source_image8

Figure 3: Performance of a planar MOSFET (FDB44N25) compared with a similar device implemented in trench-gate technology (FDB2710) at turn on (left) and turn off (right) reverse recovery. (Source: Fairchild)

The planar MOSFET survives because comparisons with trench gate MOSFETs show that the planar devices have superior Forward Biased Safe Operating Area (FBSOA) and Unclamped Inductive Switching (UIS) avalanche capabilities. These studies also show, however, that the reverse-recovery performance of the body diode in trench MOSFETs, as characterized by the reverse-current densities, is superior to that of equivalent planar MOSFETs (see Figure 3). This is mainly because the structure of the planar MOSFET requires more silicon material – a greater thickness and a larger contact surface – than trench MOSFETs need. This gives the planar devices a higher thermal inertia and better thermal dissipation, since the larger contact surface area results in lower thermal resistance. On the other hand, it increases the intrinsic parasitic capacitance of the body diode, making it slower.

This means that, when selecting MOSFETs for their specific application, system designers should pay particular attention to:
• The parasitics in their circuit
• The thermal characteristics of the operating environment
• The relative ruggedness or vulnerability of their chosen MOSFET technology

The most important parasitic parameters to take account of are two stray inductances. They can cause over-voltage transients, slow the switching speed, and cause unexpected imbalances of the current between devices connected in parallel, as well as unwanted oscillations.

The first is an inductance in series to the Source pin. This inductance is present in the gate-driving control loop and acts as a sort of feedback, slowing the gate-driver signal. The designer will already be taking care not to exceed the gate-to-source maximum voltage rating. But even if the applied gate voltage is kept below the maximum, this stray inductance coupled with the gate capacitance might generate ringing voltages which could lead to the destruction of the oxide layer.

pg22_TV_image3

Figure 4: The effects of Source stray inductance at turn-on

The second inductance is in series to the Drain pin. If not clamped, it causes an over-voltage spike when the device switches off. This effect can be minimized with the use of snubbers or clamping devices. Moreover, when switching on, another effect of this inductance is that the drain voltage falls, resulting in a discharge of the Miller capacitance, causing the gate driver to draw more current, and making for a slower overall commutation-transition edge (see Figure 4).

In order to minimize these effects, stray circuit inductance must be kept as low as possible. This is done by keeping conduction paths as short as possible, by minimizing the area of current loops, by using twisted pairs of leads, and by using a ground plane construction.

These measures have the effect of controlling the stray inductances in a MOSFET system, but in fact they constitute good layout practice that is equally valid for users of IGBTs.

pg22_TV_image2

Figure 6: A typical efficiency v current plot for IGBTs and MOSFETs

Finding the Boundary Between MOSFET and IGBT Applications
As shown above, the characteristics of IGBTs and MOSFETs make the choice of device simple in most applications. But at the cross-over, both device types have trade-offs that muddy the waters. And the development of MOSFETs based on silicon carbide (SiC) technology confuses matters further, since these offer higher performance (faster switching, lower losses) than silicon MOSFETs, but at a markedly higher unit cost. Today, after the many evolutions in the technology of IGBTs and MOSFETs described above, this cross- over affects applications operating at more than 250V, switching at a frequency between 10kHz and 200kHz, and operating at power levels over 500W (see Figures 5, 6).

In the MOSFET’s favour, its structure incorporates a diode which is very useful for handling free- wheeling currents. In low-voltage MOSFETs up to 200V, such as the STM F7 series, Fairchild’s Power Trench series, NXP Semiconductors’ PowerMOS Trench 9 and Trench 8 series, and the Vishay Generation IV series, this integrated diode is very fast. To achieve the same functionality with an IGBT, the designer must specify a so called co-packaged IGBT – a discrete fast diode and IGBT housed in a single package, a bigger and more expensive solution than the standard MOSFET.

In applications operating at voltages higher than 500V, the picture becomes even more clouded: this is because of the development of Super-Junction (SJ) MOSFETs addressing high-voltage systems operating at more than 500V, such as the MDmesh II, MDmesh V, FDmesh II and SuperMESH 5 from STMicroelectronics, the SuperFet II, Easy Drive, Fast, and FRFET (Fast Recovery) series from Fairchild, and the E and EF series from Vishay. The SJ MOSFET appears to offer an alternative to the co-packaged IGBT at a higher voltage range than the ordinary MOSFET can support. The problem is that the SJ MOSFET’s internal body diode is inherently slower than that of the normal FRED co-packaged hyper-fast diode of an IGBT.

Where necessary, for instance in an H-bridge phase-shift topology, it is possible to select one of the special SJ MOSFETs supplied with a relatively fast body diode. Every large MOSFET manufacturer includes in its product portfolio these special high-speed SJ MOSFETS, such as the FDmesh II from ST or the SuperFet II from Fairchild, but even these devices are never as fast as the hyper-fast diodes found in a standard IGBT.

At high voltages, the SJ MOSFET may be suitable for relatively low power outputs. When operating at more than 600V and producing a high power output, the IGBT remains the only choice. This is because the IGBT’s saturation voltage remains almost constant over the whole current range, while the voltage drop caused by a MOSFET’s on-resistance increases as current increases. So at high power levels, the IGBT suffers markedly lower conduction losses than the MOSFET.

At voltages lower than 600V and with relatively low power outputs, the MOSFET has become a more viable choice in performance terms than ever before, supporting higher switching speeds than the IGBT and providing greater efficiency. But of course, the most important engineering parameter of all might end up determining the choice after all the performance considerations have been made.

That parameter is, of course, unit cost.

Leave a Reply

Your email address will not be published. Required fields are marked *

Protected by WP Anti Spam