Periodic Reporting for period 1 - UniSCool (UNISCOOL: SMART IN-CHIP LIQUID COOLING FOR ADVANCED MICROELECTRONIC SYSTEMS)
Reporting period: 2023-07-01 to 2024-06-30
The ever-increasing rate of data and communications, accentuated by the rapid growth of applications such as 5G, VR, or AI, coupled with advanced manufacturing and packaging technologies in micro and nanoelectronic systems resulted in a continuous increase of the power density of integrated circuits (ICs). However, the inability to follow Dennard's scaling law has incurred an exponential increase in heat dissipation, establishing thermal management as a major concern for the information and communication technologies (ICT) community. In addition, the growing demand for data processing and storage is not free of environmental impact: data centers account today for around 1% of global electricity demand, and up to 40% of this consumption is associated with cooling systems. To meet the increasing power density of microprocessors and reduce cooling power requirements, data centers are increasingly moving from air to liquid cooling alternatives due to their higher heat capacity, compactness, and higher cooling performance.
Today, liquid-cooled devices for advanced microelectronics are mainly based on microchannels cold plate, fixed to the backside of microprocessors and, although this technology offers low thermal resistances, also presents large pressure drops (that turns into high hydraulic pumping powers) and poor temperature uniformities (which implies electronic reliability and lifetime issues). Moreover, in real operating conditions of advanced multicore processors or 3D-IC, the heat flux distribution changes spatially and over time and current cooling systems cannot provide high-performance levels, leading to temperature non-uniformities and overcooled systems. Within the existing cooling solutions to improve the performance of liquid cooling systems, none focuses on developing a system capable of adapting to changing conditions in time and space, so heat transfer is improved even when not needed and additional pressure drops are induced in the fluid channel, causing oversized pump powers for changing conditions.
A promising approach for more efficient thermal management, proposed by several studies, is to directly embed liquid cooling inside the chip, which eliminates the thermal resistance between the semiconductor die and the packaging and dimensions of the IC and cooling power can be significantly reduced.
UniSCool aims to enter a new leading thermal management paradigm by developing an in-chip smart liquid cooling system, embedded at the chip stack, with a new patented and highly innovative liquid cooling system, based on an adaptive heat sink that includes a series of thermally activated fins capable of efficiently adapting the local heat extraction to variable heat fluxes. This solution can boost the heat extraction capacity of the system up to more than 1kW/cm2 with significantly reduced flow rate and pumping power consumption (x0.5) high performance, and significantly reduced space (x10) in a sustainable and environmentally friendly way.
Today, liquid-cooled devices for advanced microelectronics are mainly based on microchannels cold plate, fixed to the backside of microprocessors and, although this technology offers low thermal resistances, also presents large pressure drops (that turns into high hydraulic pumping powers) and poor temperature uniformities (which implies electronic reliability and lifetime issues). Moreover, in real operating conditions of advanced multicore processors or 3D-IC, the heat flux distribution changes spatially and over time and current cooling systems cannot provide high-performance levels, leading to temperature non-uniformities and overcooled systems. Within the existing cooling solutions to improve the performance of liquid cooling systems, none focuses on developing a system capable of adapting to changing conditions in time and space, so heat transfer is improved even when not needed and additional pressure drops are induced in the fluid channel, causing oversized pump powers for changing conditions.
A promising approach for more efficient thermal management, proposed by several studies, is to directly embed liquid cooling inside the chip, which eliminates the thermal resistance between the semiconductor die and the packaging and dimensions of the IC and cooling power can be significantly reduced.
UniSCool aims to enter a new leading thermal management paradigm by developing an in-chip smart liquid cooling system, embedded at the chip stack, with a new patented and highly innovative liquid cooling system, based on an adaptive heat sink that includes a series of thermally activated fins capable of efficiently adapting the local heat extraction to variable heat fluxes. This solution can boost the heat extraction capacity of the system up to more than 1kW/cm2 with significantly reduced flow rate and pumping power consumption (x0.5) high performance, and significantly reduced space (x10) in a sustainable and environmentally friendly way.
Within this project, UniSCool has shown its capability to design, fabricate and characterize a cooling system embedded at the chip stack, able to provide an adaptive behaviour in a passive way to optimize the heat extraction in function of the local and instantaneous needs.
The first objective of the project was to develop microscale self-adaptive actuators. Although the development and integration of self-adaptive fins is still a challenge in terms of performance and manufacturability, UniSCool team proposed and developed an alternative approach to reach the desired regulation, as identified in risk 1 and risk 2 of the project. An alternative heat sink, as detailed in the risk-mitigation measures, based on an array of microfluidic cells that incorporate a thermally driven self-adaptive valve able to passively tailor the flow rate to the local heat extraction needs is proposed (Milestone 1 achieved).
The second objective was the design of a microfluidic system embedded at the chip stack, which has been achieved as detailed in section 2.2. Milestones 2 and 3 have been reached within this section.
The third objective considered the fabrication of microchannels with self-adaptive fins embedded at the chip stack. As explained in section 2.3 self-adaptive fins have been replaced by thermally activated valves, and a prototype has been successfully fabricated at the University of Sherbrooke (Milestone 4 reached, risk 3 identified).
Finally, by the development of this in-chip cooling prototype, UniSCool demonstrated the feasibility of a high heat extraction capacity with reduced pressure drop cooling system, embedded at the chip stack that brings innovative and high-value thermal management solutions in the European chip environment (objective 4).
The first objective of the project was to develop microscale self-adaptive actuators. Although the development and integration of self-adaptive fins is still a challenge in terms of performance and manufacturability, UniSCool team proposed and developed an alternative approach to reach the desired regulation, as identified in risk 1 and risk 2 of the project. An alternative heat sink, as detailed in the risk-mitigation measures, based on an array of microfluidic cells that incorporate a thermally driven self-adaptive valve able to passively tailor the flow rate to the local heat extraction needs is proposed (Milestone 1 achieved).
The second objective was the design of a microfluidic system embedded at the chip stack, which has been achieved as detailed in section 2.2. Milestones 2 and 3 have been reached within this section.
The third objective considered the fabrication of microchannels with self-adaptive fins embedded at the chip stack. As explained in section 2.3 self-adaptive fins have been replaced by thermally activated valves, and a prototype has been successfully fabricated at the University of Sherbrooke (Milestone 4 reached, risk 3 identified).
Finally, by the development of this in-chip cooling prototype, UniSCool demonstrated the feasibility of a high heat extraction capacity with reduced pressure drop cooling system, embedded at the chip stack that brings innovative and high-value thermal management solutions in the European chip environment (objective 4).
The main results of the UniSCool project include data center energy savings of up to 70% compared with current air-cooled systems. If this technology was implemented in all existing data centers, we could save 40 TWh/year of electrical energy, which is equivalent to the whole year’s electrical consumption of countries like Portugal, and we could have avoided the emission of 175 Mt of CO2 into the atmosphere, equivalent to the one that 1% of the world’s forest mass is capable of fixing. Hence, UniSCool technology can have a significant impact on the environment and society in terms of energy savings and CO2 emissions avoided. Furthermore, as energy savings occur, this unconsumed energy will be available in the market, making it more readily available to consumers and lowering the price.
This innovation project is seen as a natural evolution for the company: starting from our existing smart on-chip liquid cooling system, our aim is to have a further impact on the IT thermal management market by developing an integrated cooling system at the chip stack that can boost current heat extraction capacities. Also, hotspot mitigation and space, cost, and energy savings open a new paradigm for the thermal management of advanced microelectronics. With the development of the UniSCool project, we seek to create a disruptive prototype in the market for embedded chip cooling that serves as a reference for semiconductor manufacturers and demonstrates its advantages in terms of thermal performance, energy, and space savings.
These advantages are fully aligned with USC's mission, based on a policy of energy savings and CO2 emissions reduction associated with microelectronics cooling. The possibility that each microchip sold in the market incorporates UniSCool technology directly from manufacture would mean a reduction in CO2 emissions and electricity consumption due to the avoidance of two important steps: fabrication and implementation of on-chip cooling in the server after the fabrication and placement of the chip.
This innovation project is seen as a natural evolution for the company: starting from our existing smart on-chip liquid cooling system, our aim is to have a further impact on the IT thermal management market by developing an integrated cooling system at the chip stack that can boost current heat extraction capacities. Also, hotspot mitigation and space, cost, and energy savings open a new paradigm for the thermal management of advanced microelectronics. With the development of the UniSCool project, we seek to create a disruptive prototype in the market for embedded chip cooling that serves as a reference for semiconductor manufacturers and demonstrates its advantages in terms of thermal performance, energy, and space savings.
These advantages are fully aligned with USC's mission, based on a policy of energy savings and CO2 emissions reduction associated with microelectronics cooling. The possibility that each microchip sold in the market incorporates UniSCool technology directly from manufacture would mean a reduction in CO2 emissions and electricity consumption due to the avoidance of two important steps: fabrication and implementation of on-chip cooling in the server after the fabrication and placement of the chip.