Is innovation still possible with silicon transistors? STMicroelectronics Superjunction MDmeshTM case study

Is innovation still possible with silicon transistors? STMicroelectronics Superjunction MDmeshTM case study

Strategic Marketing, Innovation and Major Project Manager, Power Transistor Division, STMicroelectronics

Preface

Since solid-state transistors replaced vacuum tubes, the semiconductor industry has made amazing breakthroughs that have changed the way we live and work. Without these technological advances, it would be impossible for us to work remotely and stay connected to the outside world during the lockdown. In short, without the technological advancement of semiconductors, mankind would not be able to enjoy the miracle of science and technology.

举个例子，处理器芯片运算能力的显著提高归功于工程师的不断努力，在芯片单位面积上挤进更多的晶体管。根据摩尔定律，晶体管密度每18个月左右就提高一倍，这个定律控制半导体微处理器迭代50多年。现在，我们即将到达原子学和物理学的理论极限，需要新的技术，例如，分层垂直堆叠技术。

At the same time, we are also in the midst of another revolutionary wave. Wide-bandgap semiconductors such as silicon carbide (SiC) and gallium nitride (GaN) are developing rapidly. These new materials have unique physical properties that can improve the energy efficiency and power density of devices and enable them to work safely in harsher thermal environments.

STMicroelectronics has started mass production of STPOWER SiC MOSFETs, which will help advance electric vehicle (EV) applications, usher in an era of large-scale electrification of vehicles, and ultimately achieve autonomous driving and green travel.

The superjunction MOSFET, which appeared at the beginning of this century, was a technological revolution for high-voltage (i.e., above 200V) silicon-based power transistors. Until the late 1990s, chip designers had to accept the "axiom" that the quality factor (the product of on-resistance and chip area) of planar transistors was proportional to the breakdown voltage BV, with a ratio of up to 2.5. This axiom meant that the only solution to achieve a lower on-resistance value at a given voltage was to increase the chip area, and the result was that small package applications became increasingly difficult. By making the above relationship close to linear, superjunction technology saved the high-voltage MOSFET. STMicroelectronics named the technology MDmesh and listed it as a sub-brand of STPOWER.

Principle of superjunction transistor

The working principle of the superjunction transistor is to use a simplified Maxwell equation, for example, a one-dimensional coordinate system with only the vertical axis y, which stipulates that the slope of the electric field on the y-axis is equal to the charge density  divided by the dielectric constant , expressed in symbols: dE/dy=/. Another equation is to represent the relationship between the voltage V and the electric field component E on the y-axis, that is, E=-dV/dy. In other words, the voltage V is the integral of E, or in geometric terms, the area under the E curve is a function of y. We can understand how they work by comparing the vertical structure of a standard planar MOSFET and a superjunction MOSFET of the same size. In essence, the superjunction is to extend the p-body of the basic transistor inside the vertical drain to achieve a p-type column.

In the planar structure (Figure 1 left and diagram), starting from the chip surface and going down the y-axis, we see the p-body, and the slope of the field is positive until we reach point A. From point A to point B, the drain has a negative charge polarity, so the slope of the field changes from positive to negative. From point B to the substrate, the negative charge density becomes larger (n-), so the slope of the field increases. The green area in the figure shows the voltage that can be maintained in the off state. In the superjunction diagram on the right, the addition of the p-type pillar changes the electric field distribution. In fact, from point C to point A, the electric field distribution remains constant (same polarity for the body and pillar), and then, because of the drain and substrate, the field changes the same as the planar structure, and the slope becomes negative. Therefore, the area under the electric field is larger, and the voltage V2 can be maintained, which is the magic of the p-type pillar. Now, for a given voltage, we can reduce the drain resistivity and the on-resistance.

Figure 1 Planar MOSFET (left) and superjunction MDmesh MOSFET (right)

Technological evolution

Since its introduction, MDmesh transistors have been continuously improved and perfected, and today there are still a large number of substations using MDmesh transistors. The manufacturing process of the vertical p-type column has been greatly optimized to ensure that the transistor has better manufacturing yield and working robustness. Depending on the topology and application requirements of the target circuit, engineers can now choose different dedicated product series. This versatility and flexibility at the technical level brings more choices to system designers. In the voltage range of 400V to 650V, the general M2 series has a very high cost-effectiveness. In addition, there are dedicated products with a withstand voltage of up to 1700V covering PFC, soft-switching LLC and bridge topologies.

In addition, STMicroelectronics has introduced end-of-life technologies such as platinum ion implantation to improve the performance of parasitic diodes, reduce the reverse recovery time trr and reverse recovery charge Qrr, and improve dV/dt (DM series) tolerance. These product features are very suitable for bridges and high-power phase-shift circuits. The MDmesh fast diode model can even compete with IGBTs in low-power motor drives without adding another diode to the package. The 150W inverter for refrigerator compressors is a typical example of improving energy efficiency, as shown in Figure 2.

Figure 2: Comparison of energy efficiency curves of MDmesh fast diode MOSFET and DPAK packaged IGBT in compressor inverter application. Test conditions: 0.23Nm (load), 220V/50Hz (input voltage)

Billions of MDmesh transistors have been shipped in the market for a wide range of applications. The M6 ​​series is an MDmesh product optimized for resonant converters. Compared with the earlier M2 series, STMicroelectronics designers have put a lot of effort into iterative improvements, as shown in Figure 3.

Figure 3: From M2 to M6: Gate charge, threshold voltage and output capacitance are comprehensively improved

In Figure 3, from left to right, we see lower gate charge, higher threshold voltage, and a more linear voltage to output capacitance ratio, which results in higher switching frequencies, lower switching losses, and higher efficiency at lighter loads.

Combining the basic superjunction transistor technology with advanced manufacturing processes and paying special attention to important switching parameters such as dI/dt and dV/d, ST has created a high-performance high-voltage MOSFET, as shown in the safe operating area diagram in Figure 4. Thanks to these improvements, the DM6 MDmesh series is well suited for applications such as solar inverters, charging stations, and electric vehicle on-board chargers (OBCs).

Figure 4 dI/dt and dV/dt safe operating area

Application Areas

STMicroelectronics’ MDmesh transistors have a wide range of applications, and we can only select a few representative use cases to demonstrate the product advantages.

Mobile phone charging adapters are a large-scale MDmesh application area. Figure 5 shows a 120W charging adapter.

Figure 5: MDmesh in a mobile phone charging adapter

Figure 6 depicts the efficiency comparison of the “customized” M5 series and the basic M2 series in a higher power 1.5kW power factor correction circuit. Both MOSFETs have similar on-resistance (37 and 39 mOhm for M5 and M2 respectively) and reverse withstand voltage (650V).

Figure 6: How the M5 series (blue line) improves PFC efficiency at high power conditions

Figure 7 shows an interesting example of a 3kW half-bridge LLC circuit in an on-board charger OBC, where STMicroelectronics’ latest DM6 series (STWA75N65DM6) is compared with competing products at Vin = 380V-420V, Vout = 48V, and switching frequency f = 250Hz-140kHz.

Figure 7: 3kW full-bridge LLC: turn-off EMF vs. output power ratio; energy efficiency vs. output power ratio

Figure 8 is a loss classification analysis diagram, which shows that the optimal combination of conduction loss and switching loss is the key factor to achieve the lowest loss and highest energy efficiency.

Figure 8: Analysis of various loss sources in a 3kW full-bridge LLC converter

In addition, the rapidly growing 5G technology will also benefit from MDmesh technology innovation. The density of 5G system cellular cells continues to increase, while base stations continue to be miniaturized, from micro cells to pico cells. MDmesh, which has advantages in energy efficiency, production capacity, competitiveness and performance, is an excellent choice for repeater power chips.

为了使5G系统的工作能效超过98%，PFC级和DC-DC变换器级的能效都必需达到99%。PFC的解决方案可以是MCU数控三角形电流模式（TCM）三通道交错无桥图腾柱电路。TCM系统使变换器能够执行零电压开关操作，从而显著降低开关损耗。总体上，最后得到一个平滑的能效曲线，能效在低负载时表现良好，此外，还可以使用尺寸更小的电感器、EMI扰滤波器和输出电容。

MDmesh transistors pave the way for the rollout of 5G wireless systems.

Diffusion bonding and packaging

The diffusion welding process is another interesting innovation in the next generation of MDmesh products.

In the standard welding process (soft soldering), the formation of intermetallic compounds (IMP) is the basis of bonding. The IMP includes a thin layer of intermetallic compounds at the interface and unreacted solder between the layers. Analysis of the failure mechanism of standard soft solder joints after thermal cycling found that fatigue crack growth occurred within the volume of unreacted solder.

Hardness and brittleness are two important properties of all intermetallic compounds, which reduce the ductility of the material. It is well known that brittleness can lead to device failure under thermomechanical stress, thus reducing the reliability of electronic devices.