L
Power Electronics
Power electronics is defined as electronics where the processing of energy is of concern, as opposed to signal-level electronics where the purpose is to process information. Power electronics is ubiquitous in energy systems as the interface between energy sources and the systems that they supply, providing the necessary conversion of the source characteristics (e.g., voltage, frequency, stability) to those required by the powered apparatus (e.g., constant voltage, constant power, specific or variable frequency). Because the power semiconductor devices used in power electronic circuits operate as switches, they ideally carry zero current when they are off, and support zero voltage when they are on. Consequently, they produce zero loss in operation. However, this ideal case is never realized and there is some loss associated with the “on” state. The “off” state is, for practical purposes, lossless. For this reason, the conversion of energy using power electronics can achieve efficiencies that are typically in the high 90 percent range.
The switches used in power electronic circuits can be of various types. The most ubiquitous in today’s systems, and those applicable to the Army’s needs, are two types of transistors: the metal-oxide-field-effect-transistor (MOSFET) and the integrated gate bipolar transistor (IGBT). The Si power MOSFET can be applied at voltages up to about 1 kV, while the SiC MOSFET can support voltages approaching 3 kV. The IGBT can be used at voltages as high as 10 kV. A significant difference between the two devices is that the MOSFET can switch at much higher frequencies than the IGBT which gives it a distinct advantage where light weight and small size are important.
CHARACTERISTICS OF SEMICONDUCTOR DEVICES
As noted earlier, when a power transistor functioning as a switch is “on,” the voltage across its terminals is not zero. Therefore, there is energy being dissipated in the switch, which is known as on-state loss. The transition from on-to-off or vice versa of a transistor is not instantaneous, resulting in there being simultaneously a voltage across its terminals and a current through them, creating an additional energy loss known as switching loss. While the former is relatively constant with switching frequency, the latter increases linearly. Therefore, there is always a trade-off between going to higher frequencies to reduce filtering requirements or minimize component sizes (particularly inductors and transformers), and a countervailing concern that such benefits not be compromised by increased switching losses in the circuit.
The bipolar transistor has been superseded in practice by the power MOSFET and the IGBT. The IGBT can be viewed as the combination of a bipolar transistor whose base is driven by a MOSFET. The structure of the power MOSFET is distinct from MOS transistors used to process information, typically in an integrated circuit, and permits the blocking of high voltages and the carrying of high currents. The IGBT can switch at maximum frequencies in the 50-100 kHz range, while the power MOSFET can switch at frequencies in the 10’s of MHz range for silicon based devices, and in the 100’s of MHz for devices fabricated in gallium nitride (GaN).
GENERAL PROPERTIES OF SEMICONDUCTOR DEVICE MATERIALS
Early transistors were fabricated in germanium (Ge) but because of its small bandgap the transistor properties were a strong function of temperature. Ge is very seldom used for power semiconductor devices. Silicon (Si) is the dominant device material and provides for an upper temperature limit of the junction of approximately 125°C. More recently what are known as wide band-gap materials have become available which have permitted both the switching frequencies and temperature limits to be increased. These materials are silicon carbide (SiC) and GaN. Table L.1 shows the electrical properties of semiconductor materials practical for fabricating transistors. The bandgap determines the concentration of charge carriers due to thermal excitation. The smaller the bandgap the higher the concentration of carriers at a specific temperature and the lower the temperature limit of a device fabricated with the material. The wide bandgaps of SiC and GaN account for their high temperature applicability. The critical field is the electric field at which the material breaks down. It is closely correlated with the upper voltage limit of a semiconductor device fabricated with that material. The electron mobility
TABLE L.1 Parameters of various semiconductor materials at 25°
PARAMETER | Si | Ge | GaN | SiC | UNITS |
---|---|---|---|---|---|
Bandgap (Eg) | 1.12 | 0.66 | 3.4 | 3.26 | eV |
Critical field (Ec) | 3 × 105 | 105 | 3 × 106 | 3 × 106 | V/cm |
Intrinsic concentration (ni) | 1.4 × 1010 | 3 × 1013 | 1.6 × 10−10 | 8.2 × 10−9 | /cm3 |
Electron mobility (me) | 1360 | 3900 | 1250 | 900 | cm2/V-s |
Hole mobility (mh) | 490 | 1900 | 200 | 100 | cm2/V-s |
Saturation drift velocity (nsat) | 107 | 6 × 106 | 2.5 × 107 | 2.7 × 107 | cm/s |
Electron diffusion constant (De) | 34 | 100 | 25 | 22 | cm2/s |
Hole diffusion constant (Dh) | 12 | 50 | 5 | 3 | cm2/s |
Permittivity (∈) | 11.8 | 16 | 8.9 | 9.7 | ∈o (F/m) |
Thermal conductivity (k) | 1.5 | 0.6 | 1.6 | 3.6 | W/cm-K |
determines how much current flows under the influence of an electric field. The electron saturation velocity, which is related to mobility, is a more accurate metric of a material’s suitability for application to power devices. The higher the saturation velocity, the better suited is the material. Thermal conductivity determines how easily heat can be extracted from a device, and SiC is clearly superior in this regard to Si or GaN.
The thermal constraints of passive components also are currently an obstacle to decreasing the size and weight of power electronics, suggesting that development of high temperature materials for passive components could enable hotter power electronics, thereby improving fuel efficiency by reducing cooling system losses. Commercial work in this area may not be adequate for the Army’s needs due to commercial application cost constraints.