By Prof. Leo Lorenz, President of the European Centre for Power Electronics (ECPE)
The performance of power electronic systems has always been, and continues to be, directly linked to the characteristics of the available semiconductor technologies. In this sense, power semiconductors have been the driving force behind electrification for decades.
The bipolar era: The Dawn of Modern Power Electronics
The history of power semiconductors dates back to the mid-1950s. The first qualified power diodes came onto the market at this time, enabling more compact and efficient rectifiers than the electromechanical or valve-based solutions that had been available until then. However, the real breakthrough followed shortly afterwards with the introduction of the thyristor by General Electric in 1956 and by Siemens in 1957. This made it possible to control the switching of high electrical power using a semiconductor device for the first time.
The thyristor was a revolutionary development for its time. It could block high voltages and switch large currents, opening up entirely new possibilities in industrial power conversion. In the decades that followed, this technology developed rapidly. Starting with the first devices in the early 1960s, which had voltage ratings of 1.5 kV and 100 A, increasingly powerful switches were developed for large-scale industrial applications, traction systems, and power transmission using light-triggered thyristors with ratings of up to 8 kV and 6 kA. These components remain crucial to high-power applications in power transmission and industrial processes in the high megawatt range to this day.
Physically, the thyristor is based on a bipolar charge carrier distribution, also known as an electron/hole plasma. Its excellent conduction properties are due to conductivity modulation: the injection of electrons and holes significantly increases the conductivity of the semiconductor structure, enabling low conduction losses. This property has made the thyristor the component of choice for high-power applications for decades.
However, the classic thyristor had one crucial limitation: while it could be switched on actively, it could not be switched off actively. Once triggered, it remained conductive until the current fell below a certain holding current, either at the natural zero-crossing point in AC operation or with the aid of external commutation circuits. This made system architecture complex.
A significant development was therefore the Gate Turn-Off Thyristor (GTO), which became important in industry in the late 1960s and 1970s. For the first time, it enabled this thyristor structure to be switched off actively via the gate. However, this progress came at a high cost. In the case of large devices, switching off requires very high gate currents, placing high demands on the design of the driver circuits and requiring considerable expertise. Furthermore, there was a fundamental physical conflict: a device whose conductivity is strongly modulated offers excellent conduction properties, but it also stores high charges that must be discharged during switching off. With the GTO, therefore, a compromise had to be found between good conduction losses and controllable turn-off.
Meanwhile, bipolar power transistors (BJTs) became increasingly important in the 1960s and 1970s. Unlike thyristors, BJTs could be switched on and off actively at any time, just like GTOs. This significantly simplified the implementation of many power electronics circuits, as the complex commutation networks mentioned earlier were no longer necessary. The key advantage of BJTs over GTOs was their significantly simpler and faster controllability at low to medium power levels. Furthermore, switching frequencies in the kilohertz range were possible, which was a considerable advance on earlier designs.
However, bipolar power transistors also had their limitations. While they were easier to drive than GTOs, the process was still relatively complex due to the requirement for a continuous base current. Even more problematic was their limited safe operating area (SOA). The so-called ‘second breakdown’ in particular posed a serious problem, as local thermal instabilities could lead to the device failing suddenly, even if the nominal limit values appeared not to have been exceeded at first glance. In practice, this meant that bipolar power transistors could only be operated within a restricted SOA range.
A crucial and deep going technology cut came with the MOS revolution
The transition to MOS-controlled power semiconductors in the late 1970s was a significant technological development. A key milestone was reached in 1979 when International Rectifier introduced one of the first commercially successful power MOSFETs. This ushered in the era of modern MOS-controlled power electronics.
Initially, this development took place within the low-voltage range. The first power MOSFETs were designed for 50 V and 100 V voltage ranges – unremarkable by today’s standards, but highly significant from a technological perspective at the time. This is because these components combined the electrical, thermal, and drive characteristics that system developers had long sought.
Unlike a bipolar power transistor, a power MOSFET does not require continuous control current. The gate is electrically isolated, so only its input capacitance, which is determined by the cell structure, needs to be recharged to drive it. This significantly simplifies the driver circuit while enabling precise control of the switching edges and management of the protection functions. In addition, the power MOSFET operates as a unipolar device without minority charge storage, enabling high switching frequencies.
Another decisive advantage lies in its operational behavior. While the safe operating area (SOA) range of bipolar power transistors was severely limited by second-breakdown effects, that of a power MOSFET approaches a rectangular characteristic curve much more closely. This does not mean that MOSFETs can withstand unlimited stress, as thermal, electrical and robustness limits still apply, but they can be operated much closer to their specified operating limits than bipolar power transistors.
The fact that this technology was quickly recognized as groundbreaking is evident from the subsequent entry of other major manufacturers, including Siemens, into the market with their own power MOSFETs.
However, the transition to MOS technologies meant more than just a new component; it also represented a fundamental shift in manufacturing. While many bipolar power semiconductors could still be manufactured using traditional discrete processes, power MOSFETs required IC-compatible technologies, necessitating high investment. This resulted in a significant shift in the market in favor of financially robust manufacturers, as smaller companies could not afford the necessary production facilities.
The development of the power MOSFET represented a revolutionary step not only in manufacturing technology, but also in the development of power electronic energy converters. With the power MOSFET, switch-mode power supplies (SMPS) could now be manufactured with high power density, i.e. small size and weight, while maintaining high efficiency. This opened up a high-volume field of application for power supplies in the consumer sector, ICT (information and communications technology), and data processing. The next step, starting around the 1990s, was the integration of the drive and protection circuitry – the so-called SMART FET – which led to the electrification of automobiles. Many previously mechanical switches, such as relays and fuses, were replaced by SMART FETs. With increasing system integration, this led to significant energy savings, higher reliability and a smaller footprint.
However, this technology has also reached its physical limits. The classic silicon MOSFET is a unipolar device. Unlike bipolar switches, it cannot modulate its conductance when switched on. As the reverse voltage increases, the drift region must be made thicker and less heavily doped, significantly increasing the on state-resistance. Consequently, the economically viable operating range for traditional silicon power MOSFETs has historically been limited to a few hundred volts.
The path to higher voltages: BiMOS, IGBT and super-junction
Following the initial success of low-voltage power MOSFETs, the appeal of MOS-controlled operation became apparent. The obvious question, therefore, was how these advantages could be extended to higher voltage ranges.
One early interim solution was the BiMOS concept, which combined MOS and bipolar transistor technologies. The aim was to combine the simple, voltage-controlled operation of MOS technology with the high-voltage capability of bipolar power semiconductors. While these concepts were important intermediate technological steps, they were superseded by the IGBT (Insulated Gate Bipolar Transistor), which possessed precisely these characteristics: a MOS control gate, a long drift region for voltage build-up and a back-doped p-region for hole injection.
Its breakthrough came in the mid-1980s. The IGBT combines the simple gate control of a MOSFET with the excellent conductivity of bipolar power semiconductors. Physically, it utilizes conductivity modulation in the drift region through electron and hole injection, achieving significantly lower conduction losses at high voltages than a purely unipolar MOSFET.
Consequently, the IGBT has become a vital component in many modern power electronic converters across various industrial, mobility and energy supply applications. The IGBT has revolutionized the entire field of electric drive technology with variable speed control, including industrial electric drives and high-power converters extending into the multi-megawatt range for use in process technology, renewable energy generation and power transmission. Without this component, traction converters in e-mobility on road, rail or water would be almost inconceivable in their current form. Electric motors account for more than 50 per cent of global electricity consumption, and it is precisely here that the IGBT has made energy-efficient, precisely controllable drive technology economically viable on a large scale. Reduced manufacturing costs, significantly increased power density for chips and power modules, and excellent robustness and reliability have enabled the successful implementation of electric drives ranging from low power – for example, domestic appliances such as fridges and washing machines – to several megawatts in the high-power range.
Meanwhile, the MOSFET continued to evolve. A major technological breakthrough was the introduction of the super-junction MOSFET. Unlike conventional silicon MOSFETs, the drift region of super-junction MOSFETs is based on a structure of alternating p- and n-doped columns. The space charges of these columns compensate each other out during reverse-bias operation. This enables the drift region to be heavily doped without affecting the breakdown voltage. This greatly reduces the turn-on resistance and, to some extent, overcomes the classic physical limitations of conventional silicon MOSFETs. Consequently, silicon MOSFETs have been able to penetrate 600 V, 650 V and 800 V voltage ranges cost-effectively. Although the area-specific on-resistance increases with rising reverse voltage, it does so much more slowly than in conventional structures. Through consistent technological advancements, the area-specific on-resistance has decreased by over tenfold in the last two decades while maintaining the same reverse voltage.
It’s not just the cell structure that matters
Unlike in digital semiconductors, where technological development is primarily driven by ‘feature size’, power semiconductor devices are significantly influenced by the cell structure and drift region, which determine the quality of the material.
The drift region, which is responsible for the breakdown voltage, is a vital part of the active semiconductor material. Its thickness, doping and material quality are key determinants of the switch’s performance. Therefore, expertise in power semiconductor development is not only required for the layout of the semiconductor cell or process miniaturization, but also for materials science, defect control, and mastery of vertical structures.
This also explains why power semiconductors have historically been outsourced to traditional foundries far less frequently than digital components. Mastering the material and vertical structure enables significant control over performance.
The wide-bandgap era: SiC and GaN
The development of wide-bandgap materials, particularly silicon carbide (SiC) and gallium nitride (GaN), has marked a significant milestone in power electronics converter and system technology. These extremely fast-switching components, which have low dynamic and static losses, open up new possibilities in the development of power electronic converters in terms of power density, efficiency, and dynamic system behavior whilst maintaining high robustness. Compared to silicon-based semiconductors, both materials offer physical advantages in terms of a higher breakdown field strength, greater temperature resistance and better switching characteristics as unipolar switches.
SiC is particularly well-suited to high voltages and power levels. It allows for one magnitude thinner drift zones together with two magnitude higher doping while maintaining the same breakdown strength, resulting in significantly lower dynamic and on state losses. SiC is therefore ideal for applications requiring high efficiency, high voltages or high switching frequencies. These include traction converters in electric vehicles, which following the mission profile typically experience partial-load and dynamic operating conditions, as well as high-power DC/DC converters, fast-charging infrastructure, industrial power converters, and highly efficient power supplies.
GaN truly shines at very high switching frequencies extending into the MHz range, while simultaneously delivering high power densities at the chip and system levels. Its excellent dynamic properties make it ideal for compact power supplies, data centres, chargers, and increasingly, more demanding power electronics applications. GaN HEMT devices, which are commonly used today, are lateral transistors offering the major advantage of system integration. This enables control and protection functions, as well as complex system functions, to be implemented in a highly targeted, application-specific manner. A significant development is the MBGaN (monolithic bidirectional GaN) transistor. These transistors significantly simplify established power electronics topologies by reducing the required number of components and the associated control overhead, consequently reducing overall system complexity.
However, technological progress does not necessarily mean the rapid and complete replacement of silicon, despite what some experts may claim. In many segments, the transition is likely to be much slower than the shift from bipolar power semiconductor devices to MOS-controlled technologies. This is due to the excellent quality of Si wafer material, the enormous maturity of silicon manufacturing, low costs and ongoing challenges regarding material defects, yield, packaging and processing wide-bandgap materials.
In the field of GaN, there is also an intense debate about whether to use horizontal or vertical structures. Horizontal HEMT-based designs currently dominate due to their excellent switching characteristics and good integration potential. However, vertical structures could become relevant in the future for higher voltages. Nevertheless, they are significantly more technically challenging.
Is it better to have one large switching stage or many small ones?
As voltages rise, the focus shifts to mastering the overall system as well as the individual component. Single switches capable of withstanding extremely high voltages — for example, in the range of 10, 15 or 20 kV — seem attractive at first glance. Fewer components, less complexity and a higher voltage withstand capacity would be achieved. In practice, however, the expertise lies in mastering the overall system. Today, many experts rely on multi-level topologies with several IGBTs (insulated-gate bipolar transistors) or other power semiconductors connected in series or cascaded when developing high-power applications. The advantage is that, instead of a single large voltage step, there are many smaller ones. This improves the current-voltage waveform, reduces dv/dt stress, minimizes electromagnetic compatibility (EMC) issues, and reduces the need for filtering.
While a single high-voltage switch may be technically feasible, it results in significantly larger voltage spikes, causing additional problems at the system level. Furthermore, many new requirements arise for packaging technologies. Therefore, the best component is not necessarily the best system solution.
In modern power electronics system engineering, packaging plays a key role
It is no longer just the semiconductor that determines the performance of a power converter. A significant proportion of the electrical and thermal performance, as well as the reliability and service life, is determined by the packaging — that is, the housing design, assembly and interconnection technologies, materials used for heat dissipation and matching the coefficients of thermal expansion, and the arrangement of the chips to minimize distributed capacitances and parasitic stray inductances. While the chip performs the actual switching function in conjunction with the module’s internal parasitic components, the packaging must simultaneously meet several requirements, which are sometimes in conflict with each other:
- High electrical insulation
- Efficient heat dissipation
- Minimal parasitic inductance
- Low distributed parasitic capacitance
- High mechanical and electrical reliability and robustness
- Long service life
Packaging becomes a critical factor, particularly given the high switching speeds of modern silicon carbide (SiC) and gallium nitride (GaN) devices. Even the smallest parasitic inductances can cause significant overvoltages, high-frequency oscillations, and electromagnetic compatibility (EMC) problems at high current-change rates.
At the same time, as power density increases, the challenge of thermal management also increases. Higher power dissipation within a smaller chip volume results in greater thermal stress and more stringent requirements for matching the thermal expansion coefficients of the materials used, as well as the interface technologies between the chip and heat sink and the cooling system itself. The development trend for all power semiconductors, including silicon and wide-bandgap technologies, is towards chip scaling, i.e. higher power densities. However, this also results in higher power dissipation densities. As power densities increase, the complexity of packaging technology and thermal management also rises.
New packaging concepts therefore rely on chip embedding, direct copper connections and sintering technologies instead of conventional solder joints, as well as high-performance ceramics that offer good thermal conductivity and high electrical insulation.
Higher voltages further intensify these requirements. In such cases, controlling the electric field distribution and ensuring creepage distances, as well as considering the partial discharge and insulation strength of the entire module, are paramount.
Another development involves highly integrated modules that bring power semiconductors, drivers, protection functions and, in some cases, sensor technology closer together.
Monolithically integrated bidirectional GaN switches also represent a promising development for the future. These will significantly simplify certain power electronics topologies, such as direct AC/AC converters and the cost-effective implementation of constant-current converters for motor control and matrix converters.
Modern electrification is impossible without power semiconductors
It is clear that power semiconductors have made many applications possible. A large proportion of the world’s electrical energy is used to power electric motors, ranging from industrial drives and pumps to compressors and rail systems. It is only thanks to high-performance, fast-switching and precisely controllable semiconductors that the variable speed control we now take for granted is possible, enabling enormous energy savings.Electric mobility in its current form would also be almost inconceivable without modern power semiconductors. Traction inverters, on-board chargers, DC/DC converters and fast-charging infrastructure all rely on these technologies.
In addition, inverters are essential for renewable energies, making energy from photovoltaic or wind power plants suitable for the grid. They are also vital for data centres, whose highly efficient power supplies would be impossible to realize without modern power electronics.
The new GaN switches and excellent FP (field-plate) low-voltage MOSTFETs provide a solid foundation for the vast field of robotics, improving the performance and capability of everything from humanoid to industrial robots.
The complete electrification of vehicles featuring up to 150 ECUs (Electronic Control Units) has also been realized through the development of specialized power semiconductors, ranging from smart power components for lighting, pumps, and comfort and safety functions, to complex high-voltage drive modules.
Power semiconductors are thus much more than just a component class. They are among the key enabling technologies of modern electrification. st
