Heat Dissipation Technology for Computer CPU Chips

With the continuous improvement of electronic computer manufacturing technology and process levels, the microprocessor CPU has developed towards high integration, miniaturization, and high frequency, leading to an increase in CPU heat dissipation. In 1971, Intel launched the world’s first microprocessor 4004, with an integration of 2300 transistors. By 2004, Intel launched the next generation 64-bit processors Madison 9M and Deerfield, with an integration exceeding 500 million. The characteristic size of CPUs decreased from 0.35m to 0.18m between 1990 and 2000, and the heat dissipation increased from 10W to 50W. By 2005, Intel’s desktop CPUs used 0.09m packaging technology, with a maximum heat output of 115W. In 2000, the American Semiconductor Industry Association estimated that by 2011, using 0.04m technology, high-performance microprocessors would integrate 7.053109 transistors within an area of 8.17cm2, with an internal clock of 10GHZ, working voltage of only 0.6V, and chip power consumption reaching 177W. Therefore, efficient heat dissipation technology is needed to solve the heat dissipation problem of CPUs, and the quality of heat dissipation design directly relates to the performance and reliability of CPU chips. Intel pointed out in its technology forum that due to leakage current and heat dissipation issues when line widths enter the nanoscale, a proper solution has yet to be found, thus temporarily abandoning the development of CPUs with higher frequencies in favor of dual-core or even multi-core CPUs. This indicates that CPU heat dissipation faces serious challenges.

To solve the CPU heat dissipation problem, two approaches can be taken:

(1）Utilizing power-saving technologies to reduce heat generation in computers, especially laptops, such as Intel’s Speedstep frequency conversion power-saving technology, AMD’s PowerNow technology, and Crusoe’s LongRun energy management technology, which can all reduce heat and noise output from computers. These energy-saving processors not only have higher performance but also provide a new avenue for effectively addressing CPU heat dissipation issues;

(2）Improving the heat dissipation system to achieve better cooling effects.

This article first discusses the mechanism of temperature’s impact on CPU performance, then reviews several commonly used CPU heat dissipation technologies’ principles and research progress, discussing their advantages and disadvantages.

According to electronic theory, heat does not directly damage the CPU (except for overheating explosions), but rather the phenomenon of electronic migration caused by overheating damages the internal chips of the CPU. Electronic migration refers to the phenomenon where electrons, when flowing in a directed manner, collide with metal atoms, causing the movement of these atoms. In the high current intensity metal conductors inside the CPU, the flow of electrons provides momentum to the metal atoms; when they collide with metal atoms, those atoms will detach from their surface and move around, forming pits or hills on the metal surface, causing irreversible permanent damage. The phenomenon of electronic migration does not immediately damage the chip; the harm is a slow process. If the CPU continues to operate at high temperatures, it will eventually cause short circuits or disconnections in the core’s internal circuits, ultimately leading to complete CPU failure. Moreover, the higher the temperature, the faster the CPU is destroyed, and the shorter its lifespan. Experimental analysis shows that if the surface temperature of the CPU exceeds 50℃ during normal operation, and the internal temperature exceeds 80℃, it will cause permanent damage due to electronic migration.

In terms of heat dissipation methods, CPU cooling can be divided into passive and active cooling systems. The active cooling method is characterized by the CPU’s temperature always being above the ambient temperature, without refrigeration mechanisms, relying on thermal radiation and natural convection to dissipate heat. Early CPU cooling adopted this method, using heat sinks to transfer CPU heat through heat dissipation plates, leaving heat dissipation holes and keyboards to naturally transfer heat to the external environment. The passive cooling method always includes refrigeration mechanisms to achieve lower temperatures, thereby lowering the chip temperature below the ambient temperature. Clearly, the passive cooling method is more beneficial for improving CPU performance, and with the continuous increase in CPU heat dissipation power, almost all CPU cooling now adopts passive cooling methods.

Air cooling is currently the main method for cooling desktop CPUs. Its principle is to conduct heat out through heat sinks and then enhance air movement through fan rotation, transferring heat from the heat sink to the surrounding environment through forced convection. Air cooling relies on air as the heat transfer medium, which has low thermal efficiency. To enhance heat exchange capability, using powerful fans generates noise. To improve cooling capacity, it mainly uses heat sinks to increase the heat dissipation area; by increasing wind speed, improving airflow organization, and adopting convection heat exchange or boiling heat exchange methods to increase the convective heat transfer coefficient; utilizing low-temperature air supply or directly employing cooling systems to achieve a larger temperature difference. However, relying solely on conduction and convection, the conventional air cooling method’s metal heat sinks are gradually approaching their heat transfer limits as the CPU’s surface heat flux density increases. To address this, Zhou Jianhui, Yang Chunxin, and others proposed a new concept of an integrated design for forced air convection cooling systems, conducting aerodynamic design on CPU fans and designing a series of radial heat sinks based on numerical simulation results of the fan’s outlet flow field characteristics. The simulation results indicate that the curved heat sink has a thermal resistance value reduced by 15.9% compared to traditional vertical heat sinks, achieving high-performance cooling effects.

Water cooling uses a water circulation system to carry heat away from the CPU through water flow. In water cooling systems, thermal spikes rarely occur, effectively addressing the issue of system crashes due to poor CPU cooling. Water cooling systems can maintain the CPU temperature at room temperature, significantly improving the internal working environment of the computer. However, water cooling systems are inconvenient, large, and complicated to install, and they pose leakage risks that could cause short circuits. These issues need further resolution.

Zeng Ping and others developed a new type of computer chip water cooling system powered by a dual-chamber parallel piezoelectric pump to achieve quiet and efficient cooling. Experimental results show that this integrated water cooling system has significant cooling effects and high development and application value.

Thermoelectric cooling, also known as semiconductor cooling, utilizes the Peltier effect in physics, where electrons (holes) directly transfer heat during movement. Figure 1 illustrates its working principle, consisting of thermoelectric pairs made of P-type and N-type semiconductor materials. When powered, it generates a thermal effect, with one side being the cold end (heat absorption) and the other side being the hot end (heat release). The cold end is tightly attached to the CPU surface to absorb heat and transfer it to the hot end for dissipation. The cold end of the semiconductor cooler cannot directly contact the CPU surface; it requires auxiliary heat sinks to transfer heat from the CPU core to the cold end, which is beneficial for fully utilizing the semiconductor cooler. This cooling method’s advantages include compact structure, quiet operation, no mechanical components or vibration, and long lifespan; the cooling capacity and speed can be adjusted by changing the current. Its disadvantages include low efficiency, high cost, and immature technology, which may lead to condensation on the CPU due to excessively low temperatures, causing short circuits.

Figure 1 Thermoelectric Cooling Working Principle

Heat pipes, also known as thermal superconductors, were first proposed by R. S. Gaugler of General Motors in Ohio, USA, in a patent (No. 2350348) in 1944. In 1965, Cotter proposed a more complete thermal management theory, laying the theoretical foundation for heat pipe research and becoming the basis for heat pipe performance analysis and design. A typical heat pipe consists of a pipe shell, a wicking core, and end caps, with its basic working principle illustrated in Figure 2. The working fluid evaporates in the evaporation section due to heat flow, with its vapor traveling through the adiabatic section to the condensation section, where the vapor is cooled by external cold fluid, releasing latent heat and condensing into liquid, which is then drawn back to the evaporation section by the capillary action of the wicking core to absorb heat and evaporate again. The heat pipe exhibits the following characteristics during operation:

(1）High axial heat transfer;

(2）Small axial and radial temperature gradients;

(3）Negligible axial conduction compared to convection.

Figure 2 Heat Pipe Working Principle

In 1998, the Sandia National Laboratories in the USA first used heat pipe technology to cool computer chips, and IBM was the first manufacturer to introduce heat pipe cooling into laptop cooling, using small heat pipes with an outer diameter of 3-5mm, an inner diameter of 2.6-4mm, and a length of less than 300mm, which can be bent into various shapes. The installation and cooling method of small heat pipes in laptops are determined by the heat dissipation power. Generally, for heat dissipation below 6W, they are installed below the keyboard, using the keyboard as a means of external heat dissipation (as shown in Figure 3(a)), or installed on the bottom plate of the chassis (as shown in Figure 3(b)). When the CPU’s heat dissipation reaches 710W, hinge-type or forced convection cooling should be employed (as shown in Figures 4a and 4b). The hinge-type cooling method first uses a heat pipe to transfer CPU heat to the hinge block connecting the display and the chassis, and then another heat pipe transfers this heat to the aluminum plate behind the display; the forced convection method transfers CPU heat to an aluminum plate containing flat micro heat pipes, which then transfer the heat to aluminum heat sinks with many thin fins, with a mini fan in front of the heat sink to expel the heat.

Figure 3 Schematic of Laptop CPU Cooling (1)

Figure 4 Schematic of Laptop CPU Cooling (2)

In the application of small heat pipes for laptop cooling, foreign scholars have conducted research on heat transfer performance using experimental data. Meiling Li and others studied the heat transfer performance of micro heat pipes in portable electronic products such as laptops and PDAs, using copper for the heat pipe shell and pure water as the working fluid. They obtained heat transfer coefficients and various heat transfer limits for heat pipes and concluded through comparative experiments that the triangular cross-section heat pipe’s performance is superior to that of rectangular cross-section heat pipes. SeokHwan Moon and others determined the heat transfer characteristics and limits of triangular and rectangular cross-sections through experiments and proposed methods and suggestions for improving the thermal performance of small heat pipes for laptop cooling based on studies of the extrusion thickness, cross-sectional height, total wall length, heat flow, and tilt angle.

With the development of microelectronics technology, the heat dissipation of high-performance desktop CPUs has reached 50-100W or even higher, and conventional forced fan cooling methods can hardly meet these requirements, making heat pipes the preferred cooling method. Kwang-Soo Kim and others conducted analytical research on desktop CPU cooling technologies through experiments. When the fan operates at high speed, the CPU cooling effect is good, but there is a significant noise problem. When combined with heat pipe cooling, even if the fan operates at low speed, the cooling effect remains prominent, alleviating noise issues. The experiments also compared the effects of different installation methods of heat pipes in desktop computers on cooling performance. Fujikura developed a so-called cactus heat pipe, whose cooling effect is related to the flow speed of the cold air.

Microchannel heat sinks are processed into grooves with cross-sectional dimensions of only tens to hundreds of microns on very thin silicon or other suitable substrates using photolithography, etching, and precision cutting methods. The heat transfer medium flows through these channels and exchanges heat with other heat transfer media through the substrate. This microchannel heat sink structure was first proposed by Tuckerman and Pease in 1981, theoretically demonstrating that the cooling capacity of water-cooled microchannels can reach 1000W/cm2.

The Panasonic laptop with a working frequency of 1.2GHz and a 12W microprocessor uses highly thermally conductive graphite sheets for cooling. This graphite sheet is combined with aluminum to improve its bending process, and then sealed around the graphite sheet using aluminum pressing technology to prevent dust from affecting its appearance. The heat generated by the microprocessor can be diffused to the graphite sheet and conducted to the back of the keyboard and the shell for external dissipation. This cooling method’s overall weight, including a 14-inch SXGA LCD screen and DVD-ROM, is less than 1.5kg. The SONY laptop uses a 7W microprocessor, utilizing graphite sheets located above and below the motherboard for cooling. This cooling mechanism’s weight is half that of traditional cooling fan designs, with the same cost.

Wang Peng and Gu Bo proposed an evaporation cooling technology that effectively combines air cooling and water cooling, using evaporation to carry away latent heat, effectively improving CPU cooling performance, pointing out related ideas such as the calculation and control of permeable water, permeability testing, selection of permeable materials, and heat sink structure design.

As CPU computing speeds increase, functionalities become more powerful, and packaging technology develops towards high density, effective heat dissipation has become a highly focused research hotspot. This article reviews the main heat dissipation technologies currently employed for CPU chips, their research progress, and existing issues. With the expansion and deepening of research, new cooling solutions and heat dissipation theories will continue to be proposed and applied to newer fields, enriching and extending the connotation of CPU chip heat dissipation technology.