THE IMPORTANCE OF ADVANCED COOLING SOLUTIONS FOR HIGH PERFORMANCE ELECTRONICS
Overheating is the main reason for degradation/failure in computing systems. Whether it is for gaming, science/engineering modeling, or graphics rendering, the microelectronics in the silicon chip of Central/Graphics Processing Units (CPUs/GPUs) require a certain voltage/current to operate at a specified frequency to perform the necessary operations. Simply put, the higher the operating frequency (or clock speed) of the processor, the faster the calculation proceeds. However, increasing operating frequency comes at the expense of higher input power to the cores, and therefore the challenge of removing higher rates of heat from the silicon chip. If the generated heat cannot be effectively removed, the core temperatures will easily surpass the 85 to 100 °C design limits. Long term operation at these temperatures may result in permanent damage, and to avoid this the chip designers incorporate thermal throttling, which results in undesirable performance reduction or shutdown. 
PCs based on flagship CPUs/GPUs require advanced liquid cooling solutions. Although significant advances have been made, current state-of-the-art coolers available in the market still hit chip temperature limits at high frequencies. The performance improvements in microprocessor and graphics chips that occurred over the past two decades have slowed recently due to the difficulties with further scaling down the dimensions of the transistors. The current trend is towards heterogeneous integration, which requires close placement of multiple chips, or chiplets, making careful chip level thermal management a crucial part of system design. EMCOOL is uniquely positioned to provide thermal solutions for these emerging applications.
HEAT TRANSFER FUNDAMENTALS
While we may get a little technical in terms of heat transfer terminology, we will also use simple analogies to better explain the fundamentals of microsystems cooling. A useful tool to describe heat transfer in such systems is called thermal resistance network; this tool accounts for each of the resistances (or obstacles) that the heat encounters along its path from the chip to the ambient. The ideal heat transfer system (impossible due to the laws of thermodynamics) is the one in which the thermal resistance is zero, meaning that any chip power could be dissipated, while keeping it at the ambient temperature. Therefore, the best (feasible) heat transfer solution is one in which the total resistance is the lowest. Thinking of heat as a vehicle that needs to go from point A (chip) to point B (ambient), the resistance as the path, and the goal as traveling that distance in the shortest time (fastest heat removal); then it makes sense that the overall resistance length to be as short as possible. Bearing those analogies in mind, it is easy to understand the comparison of conventional liquid cooling approaches and EMCOOL's embedded microfluidic cooling solution™.
The schematics below depict on the left side a cross-sectional cut of a CPU package with both technologies, in which the main components are indicated by the red lines and different temperature probes (red dots) are located at surfaces of interest. These dots help to characterize the different surface temperatures in the system, which account for the equivalent thermal resistance network shown at the center of the schematics. Finally, the representations on the left show a depiction of the heat path (red lines) and its spreading behavior. Recalling the objectives in the analogy, then we are looking for the shortest path, therefore smallest resistance network. 
CONVENTIONAL LIQUID COOLING SYSTEMS
Due to the small dimensions of the silicon chip (die) in many commercial CPUs, an Integrated Heat Spreader (IHS) is used to increase the area over which the heat is removed. While this may seem as a good way to remove heat, it creates what is called a "spreading resistance", as it makes the heat path longer. In other words, removing the heat from a small silicon chip comes at the expense of a larger resistance network. An extra spreading resistance is also generated at the base of the waterblock, which further spreads the heat to the fin area. In addition, two "contact resistances" are also part of the resulting network (which include contributions from the interfaces) as well as the Thermal Interface Materials (TIMs) used between the surfaces of the silicon, IHS and waterblock. Finally, the heat resistances due to conduction and convection mechanisms, as well as a resistance associated with the increase in coolant temperature within the waterblock are included to get the equivalent network shown at the center of the schematic below. Each resistance has a finite value, and its sum represents the total heat transfer resistance. When the value of overall heat transfer resistance (units of °C/Watt) is multiplied by the chip power dissipation (Watt), the resulting quantity is the temperature difference between the start and end points of the equivalent network, in this case the average temperature of the silicon CPU and the incoming cooling liquid. As it can be noted, there are multiple resistances and the higher their value, the higher the operating temperature of the silicon chip. The combination of all these factors is why overheating is a big challenge even for the most advanced commercial coolers, which limits the overclock potential and life of the microelectronics.  
Conventional liquid cooling technology: left) cross-section cut of a CPU package and waterblock assembly, center) thermal resistance network, right) heat spreading path.
EMCOOL EMBEDDED MICROFLUIDIC COOLING TECHNOLOGY™
EMCOOL Inc. has developed the most powerful and efficient cooling solution to date for commercial microprocessors. EMCOOL’s embedded microfluidic cooling™ system can unlock the high-frequency potential on flagship CPUs/GPUs while operating well below the thermal limits of its internal circuitry. This means that the computing system can be accelerated to its highest potential, while also protecting the chip from any form of thermal throttling/damage.
EMCOOL uses the exclusive Direct Silicon Footprint Microfluidics (DSFM) technology (U.S. Provisional Patent Application No. 62/805,119, and commercially referred as simply microfluidics), which features engineered cooling microstructures that are optimized based on the chip’s architecture and footprint; this approach allows unparalleled cooling performance when compared with any other cooling technology in the market.
​
As seen in the schematic below, with EMCOOL's technology the heat is removed directly at the source through a chip-specific microfluidic cooling layer, which is optimized for each of our supported models in order to mitigate the heat spreading effects. The ambitious goal of removing the heat in such a small area is accomplished through micro-machined structures of high-aspect ratio, which significantly enhance the fluid mixing and heat transfer. With the combination of these multiple features, the heat spreading resistances are eliminated from the overall network, while also reducing contact and convective resistances. With a significantly lower overall heat transfer resistance than conventional technologies, it is possible to dissipate higher power at lower silicon temperatures; the most important feature of microfluidics. To learn more about the history and development of this technology, please click here.
EMCOOL's embedded microfluidic cooling technology™: left) cross-section cut of a CPU package and EMCOOL assembly, center) thermal resistance network, right) heat spreading path.
WHAT ABOUT GPUs?
It is very important to mention that most commercial GPU dies have significantly larger sizes than CPU dies (between 2 and 6 times the area depending on the specific model), and therefore the heat can be removed without the use of an IHS (known as direct-die cooling). However, heat spreading is still present when in contact with either a waterblock or a heat pipe cooler, generating a resistance. Our GPU microfluidic solution does not allow such spreading behavior and instead directly addresses the heat at the silicon source, therefore providing a significantly higher performance when compared with any kind of cooling solution in the market (see benchmarks).
SUPPORT
FAQs
Shipping
Payment methods