Component packaging for high power asic thermal management

Abstract

This application relates to a component package for thermal management of high-power ASICs, and more specifically, to a cooling plate for cooling a microchip with redundant cooling fluid circulation. The cooling plate includes: a main fluid cooling circuit that directly removes heat from the microchip; and a secondary cooling circuit that acts as a condenser for the two-phase unit, indirectly removing heat from the microchip. The cooling plate can be manufactured as a lower plate and an upper plate, wherein the main cooling circuit is formed in the lower plate and the secondary cooling circuit is formed in the upper plate. Independent two-phase units can be immersed in the main cooling circuit and serve to transfer heat from the microchip to the secondary cooling circuit.

Description

Technical Field

[0001] Embodiments of the present invention generally relate to enhanced and reliable cooling of advanced microchips, such as ASICs and other microchips used in servers within data centers. Background Technology

[0002] Cooling is a critical factor in computer system and data center design. The number of high-performance electronic components, such as high-performance processors packaged inside servers, has been steadily increasing, thereby increasing the heat generated and dissipated during normal server operation. The proper operation of these processors is highly dependent on the reliable removal of the heat they generate. Therefore, proper cooling of the processors can provide higher overall system reliability.

[0003] Cooling electronic components is critical for computing hardware and other electronic devices, such as CPU servers, GPU servers, storage servers, network devices, edge and mobile systems, and automotive computing boxes. All these devices and computers are used for mission-critical purposes and are fundamental to the day-to-day operations of businesses. The design of hardware components and electronic packaged devices requires continuous improvement to support performance requirements. Cooling these electronic devices is becoming increasingly challenging, ensuring their proper functioning by consistently providing a properly designed and reliable thermal environment.

[0004] Many advanced chips, especially high-power-density chips, require liquid cooling. These chips are extremely expensive, necessitating efforts to ensure proper heat removal. Furthermore, liquid cooling equipment must be highly reliable, as any irregular heat removal can lead to chip loss, resulting in a loss of available computing power during replacement runs and potentially impacting service level agreements processed by the failed chip. Importantly, existing electronic cooling and thermal management solutions for processors do not provide module-level redundancy, meaning they are single points of failure in the system. Specifically, inadequate circulation of cooling fluid within the cooling plate can cause the corresponding processor to fail. Therefore, enhanced reliability can be achieved by developing complete end-to-end redundant cooling solutions for the chip, allowing for backup against individual failures through redundant design.

[0005] While liquid cooling solutions must deliver the required thermal performance and reliability, the cost of liquid cooling systems must remain acceptable, given that data centers may have thousands of chips requiring liquid cooling. The cost of a liquid cooling system can include the introduction of redundancy to enhance reliability. Furthermore, because different chips have different cooling requirements, an adaptable and scalable cooling design will be needed to accommodate different server architectures and be compatible with different chip packages. Attached Figure Description

[0006] Embodiments of the invention are illustrated by way of example and not limitation in the accompanying drawings, in which similar reference numerals denote similar elements.

[0007] Figure 1 This is a block diagram illustrating an example of a cold plate configuration according to one embodiment.

[0008] Figure 2 This is a schematic diagram showing a cross-section of the cooling plate according to an embodiment.

[0009] Figure 3 An implementation of manufacturing a cold plate prior to final assembly into the package design is shown.

[0010] Figure 4 Another implementation of manufacturing a cold plate prior to final assembly into the package design is shown.

[0011] Figure 5 This illustrates another implementation of manufacturing the cold plate prior to final assembly into the package design.

[0012] Figure 6 A top view of the cooling plate according to an embodiment is shown.

[0013] Figure 7 This is a schematic diagram illustrating the flow of cooling fluid in a cooling plate according to an embodiment.

[0014] Figure 8 This is a conceptual schematic diagram illustrating the cooling function of a cold plate according to an embodiment.

[0015] Figure 9 This is a conceptual schematic diagram illustrating the manufacture of a cold plate according to an embodiment. Detailed Implementation

[0016] Various embodiments and aspects of the invention will be described with reference to the details discussed below, and the accompanying drawings will illustrate the various embodiments. The following description and drawings are illustrative of the invention and should not be construed as limiting the invention. Numerous specific details are described to provide a full understanding of various embodiments of the invention. However, in some cases, well-known or conventional details have not been described in order to provide a concise discussion of embodiments of the invention.

[0017] The references to "one embodiment" or "implementation" in this specification mean that a specific feature, structure, or characteristic described in connection with this embodiment may be included in at least one embodiment of the invention. The phrase "in one embodiment" appearing throughout the specification does not necessarily refer to the same embodiment.

[0018] Incidentally, in the following embodiments, similar elements are identified by similar reference numerals in the form of x##, wherein in each embodiment, the first digit x is replaced with the same digit as the figure number, while the next two digits ## remain consistent. Therefore, in subsequent embodiments, previously disclosed elements need not be interpreted repeatedly.

[0019] The disclosed embodiments provide a cooling plate for electronic devices that utilizes multiple circulation loops to enhance the reliability of the cooling system. The cooling plate can be implemented to cool various electronic devices, such as single-chip modules (SCMs), system-on-a-chip (SoCs), multi-chip modules (MCMs), system-in-package (SIPs), etc. For the sake of brevity, these electronic devices are referred to herein as microchips or simply chips, and any such reference should be understood to include these electronic devices as well as any similar variations of dies and packages.

[0020] In the various disclosed embodiments, the cooling plate includes multiple fluid ports coupled to individual cooling circuits. Alternatively, embodiments may include a multiphase cooling unit connected in parallel with the cooling circuits. This embodiment enhances the cooling capacity of the cooling plate and improves the reliability of the cooling system. In embodiments using a two-phase cooling unit, one fluid cooling circuit can be used to directly extract heat from the microchip, while another fluid cooling circuit can be used to indirectly extract heat from the microchip as vapor condenses into liquid within the unit. The structure and function of these and other features will be described in more detail below.

[0021] Figure 1 This is a schematic diagram illustrating a chip cold plate configuration that can be implemented or modified according to the embodiments disclosed herein. The chip / cold plate assembly 100 can represent any processor / cold plate structure including fluid cooling for a server or other computing platform. Reference Figure 1 Chip 101 (e.g., ASIC) is inserted into a socket mounted on a printed circuit board (PCB) or motherboard 102, which is connected to other electronic components or circuitry of a data processing system or server. For fluid cooling, chip 101 also includes a cold plate 103 attached thereto, which is connected to a rack manifold, for example, via a blind-mating connector, to a liquid supply line 132 and / or a liquid return line 131. A portion of the heat generated by chip 101 is removed by the cold plate 103. The remaining heat enters the air space below or above, where it can be removed by airflow generated by a cooling fan 104.

[0022] Figure 2This is a side section of the cooling assembly 200, which includes a multi-loop cooling plate 203. The cooling plate 203 includes a lower plate or core plate 205 and an upper plate or secondary plate 215. The lower plate 205 and the upper plate 215 can be made of thermally conductive materials such as copper or aluminum. The bottom surface of the lower plate 205 is in physical contact with the chip package 201, and the top surface of the lower plate 205 is in physical contact with the upper plate 215, for example, through bonding, brazing, or welding. It should be noted that the interfaces between the lower plate 205 and the chip 201, and between the lower plate 205 and the upper plate 215, should be made with high thermal conductivity.

[0023] The lower plate 205 includes one or more fluid channels 212 through which cooling fluid from the cooling system circulates. A plurality of two-phase cooling units 214 are immersed between the fluid channels 212. In one embodiment, the channels 212 can be simply defined or formed by the arrangement and orientation of the two-phase cooling units 214, as will be further illustrated below. The two-phase cooling units 214 are independent and sealed, such that no fluid flows into or out of the two-phase cooling units 214. Instead, the fluid at the bottom of each unit is heated by the chip 201 and evaporates. The vapor rises to the top surface of the lower plate 205, condenses, and then flows back down. In this example, this process is enhanced by providing a wicking material or structure 216 inside the two-phase cooling units 214. Thus, the lower plate removes heat from the chip 201 partly by the fluid circulating within the fluid channels 212 and partly by evaporation within the two-phase units 214.

[0024] In this embodiment, the lower plate 205 transfers some of the heat removed from the chip 201 to the upper plate 215. The upper plate in this embodiment includes a cooling channel 213 that accommodates circulating cooling fluid, separate from the fluid circulation of the lower plate 205. Here, the fluid channel 213 of the upper plate 215 includes fins 217 to enhance heat removal. As the cooling fluid circulates within the upper plate 215, heat is extracted from the bottom of the upper plate to keep it cool, thereby enhancing the condensation effect in the two-phase cooling unit 214 of the lower plate 205. Through this effect, at least a portion of the heat removed from the chip 201 by the two-phase cooling unit 214 is transferred to the upper plate, allowing the fluid in the cooling channel 213 to indirectly remove heat from the chip.

[0025] like Figure 2As shown, two separate inlet and outlet ports are provided to transfer redundant cooling fluid to cooling plate 203. Specifically, inlet port 208 transfers cooling fluid to lower plate 205, while inlet port 218 transfers cooling fluid to upper plate 215. Similarly, outlet port 209 returns fluid from lower plate 205, while outlet port 219 returns cooling fluid from upper plate 215. This redundancy ensures that even if one fluid circulation system fails, the remaining circulation systems can still remove heat from the chip. For example, if fluid circulation in cooling channel 212 stops, the fluid remaining stagnant in cooling channel 212 will be heated by heat from chip 201. As the fluid in upper plate continues to circulate and keeps the upper plate at a low temperature, the heat from the fluid in channel 212 will be transferred to the fluid circulating in cooling channel 213.

[0026] Redundancy in two cooling fluid loops can be implemented in various ways. For example, different types of cooling fluid can be supplied to the lower and upper plates. Different fluids can circulate in two independent cooling systems using separate pumps and conduits. Conversely, the type of fluid can be the same, but can be handled by two independent cooling circulation systems, i.e., flowing in different independent loops. In this way, if one cooling fluid circulation system fails (e.g., pump failure or need to be shut down due to leakage), that circulation will still function in the other system / loop to provide cooling for the microchip.

[0027] Therefore, an arrangement of a microchip and a cooling plate is provided, wherein heat generated by the microchip is partially removed directly from the microchip via a first or main cooling fluid loop, and heat generated by the microchip is also partially removed indirectly from the microchip via a secondary cooling fluid loop that circulates cooling fluid independently of the main cooling loop. The secondary cooling fluid loop indirectly removes heat from the microchip by enhancing condensation in the two-phase units immersed in the main cooling loop.

[0028] Figure 3 An embodiment of multiple fluid port cooling plates before assembly is shown. Figure 3 Components that can be designed, constructed, and then assembled as a cooling enclosure are also shown. In one example, the lower plate 305 is manufactured by forming (e.g., milling) a tub or cavity 306 in a metal plate such as aluminum or copper. The tub is manufactured with multiple sets of fins 307. The upper plate 315 is manufactured by forming a fluid channel 313 in a plate 310 and optionally including fins 317 inside the channel 313. A first set of fluid inlet / outlet ports 318 and 319 are configured to fluidly communicate with the fluid channel 313. A second set of ports 308 and 309 are formed with openings to fluidly communicate with the tub 306 when the lower and upper plates are assembled. The second set of ports 308 and 309 are not fluidly communicated with the fluid channel 313.

[0029] Additionally, the two-phase unit 314 is attached (e.g., welded) to the bottom surface of the upper plate 315, thereby ensuring a high thermal conductivity interface between the two-phase unit 314 and the upper plate 315. Furthermore, a fluid channel 312 is formed in the upper plate 315, aligned with the fins 307 formed in the lower plate. Therefore, when assembling the lower and upper plates, the fins 307 are positioned within the fluid channel 312. During assembly of the lower and upper plates, the two-phase unit 314 is aligned to be immersed in the main cooling fluid and positioned between the fins 307. The shape, positioning, and orientation of the two-phase unit 314 can be used to define the fluid channel 312. That is, the fluid flowing in the lower plate will follow the path defined by the fins 307 and the two-phase unit 314, thus the fins 307 and the two-phase unit 314 together define the channel 312. Each of the two-phase units 314 may include a wicking structure 316. Once the lower and upper plates are assembled together, it can be attached to the chip 301.

[0030] Therefore, a method for manufacturing a cooling plate for a microchip is provided, comprising: constructing a lower plate by constructing a first metal plate and forming a main cooling channel in the first metal plate; constructing an upper plate by constructing a second metal plate and forming a secondary cooling channel; constructing a first set of inlet / outlet ports in the upper plate, the first set of inlet / outlet ports having fluid channels leading to the secondary cooling channels; constructing a second set of inlet / outlet ports in the upper plate, the second set of inlet / outlet ports having openings at the bottom surface of the upper plate to form fluid channels leading to the main cooling channels after the upper plate is attached to the lower plate; and attaching the upper plate to the lower plate.

[0031] Figure 4 Another example of manufacturing the cooling plate is shown, along with the overall package design before final assembly. Figure 4 Design and Figure 3 The designs are somewhat similar, the difference lies in Figure 4 The upper plate includes an outer frame 431. The outer frame may include various sealing mechanisms, such as an O-ring 433 and / or a leak detector, such as a liquid sensor 437.

[0032] Therefore, according to the disclosed embodiments, a cooling plate for cooling a microchip is provided, comprising: a lower plate including a main fluid cooling arrangement; and an upper plate attached to the lower plate and having a secondary fluid cooling arrangement fluidly separated from the main fluid cooling arrangement, wherein the secondary fluid cooling arrangement includes a fluid cooling channel formed in the upper plate, a fluid inlet port fluidly connected to the fluid channel, and a fluid outlet port fluidly connected to the fluid channel.

[0033] Figure 5 Another embodiment for manufacturing and assembling the cooling plate is shown. Figure 5In this embodiment, the fluid channel 512, fins 507, and two-phase unit 514 including wicking structure 516 are all fabricated on the lower plate. Incidentally, the dashed curved arrows indicate independent fluid circulation within the two-phase unit 514. In this embodiment, two sets of inlet / outlet ports 508 and 509, and 518 and 519, are provided on the upper plate 515; however, in other embodiments, inlet / outlet ports 508 and 509 lead to fluid channel 512, while inlet / outlet ports 509 and 519 lead to fluid channel 516, forming two separate and independent cooling circuits.

[0034] Figure 6 This is a top perspective view of the cooling plate according to an embodiment. In this view, some structural elements are omitted to highlight the heat removal function of the cooling plate. For example, the cooling fluid delivered via inlet port 608 is confined to the flow region indicated by 612 and then flows out via outlet port 609. Meanwhile, fins 607 are provided within the flow region 612 to increase heat transfer of the fluid. The number, size, and orientation of the fins can be designed to provide the appropriate flow resistance required for adequate heat transfer without overloading the pump in the cooling circuit.

[0035] Two-phase units 614 can be distributed within flow region 612, allowing some heat to be transferred from the two-phase units 614 to the fluid in region 612. However, most of the heat from the two-phase units 614 is intended to be transferred to the fluid flowing in the channel indicated by 613. The fluid is transferred to channel 613 via inlet port 618 and returns to the loop via outlet port 619, thus removing heat from the two-phase units 614 through outlet port 619. This enhances fluid condensation within the individual two-phase units 614. (See reference...) Figure 7 This allows for a deeper understanding of heat transfer and transport.

[0036] Figure 7 The diagram schematically illustrates heat transfer and delivery using a cooling plate according to the disclosed embodiment. In this example, processor 701 generates heat, quantified as Q1. Some heat, quantified as Q2, is transferred to a cooling fluid, denoted as coolant #1, flowing within channel 712. The heat Q2 is then discharged from the cooling plate via outlet port 709. The remaining heat, quantified as Q3, generated by the processor, is transferred to a fluid, denoted as coolant #3, within two-phase unit 714. Coolant #3 carries away heat Q3 from the processor by evaporating and rising to the top of each unit 714. At the top of unit 714, the vapor condenses by transferring heat Q3 to a cooling fluid, denoted as coolant #2, flowing within channel 713. Coolant #2 then delivers heat Q3 via outlet port 719.

[0037] Figure 8This diagram further illustrates the heat removal using the cooling plate disclosed herein. The dashed lines represent two independent heat removal processes that operate independently and simultaneously to provide redundant heat removal loops. To the right of the dashed lines is the circulation of coolant #1, which circulates within the fluid channels of the lower plate and removes heat directly from the heat source (i.e., the processor). To the left of the dashed lines is a two-step process for further heat removal from the heat source. The first step removes heat through the evaporation of coolant #3 within the two-phase unit. Heat is then transferred from coolant #3 to coolant #2 through the condensation of coolant #3 vapor. The heat is then transported away through coolant #2, while the condensed coolant #3 optionally flows back to the bottom of each two-phase unit via a wicking structure. In this sense, fluid #2 indirectly removes heat from the microchip.

[0038] Figure 9 This is a general flow diagram illustrating the manufacturing process of a cooling plate according to an embodiment. In step 990, an upper plate (also referred to as the upper core) is formed with two sets of inlet and outlet ports. Additionally, a cooling channel for coolant #2 is formed within the upper core. At 992, a two-phase unit is formed and attached to the upper or lower core (also referred to as the base core). Additionally, a fluid channel for coolant #1 is formed within the base core. Optionally, at 994, fins are also formed within the fluid channels of the base core. As shown in 996, the fins for the base core are staggered, alternating, or interwoven with the two-phase unit, such that the two-phase unit is mounted between the fins. In 998, the two cores are aligned and attached together, such as by welding, brazing, or bonding. For example, indium paste can be placed on the mating surfaces, and the two cores can be pressed together for cold welding or annealing in a hot welding furnace.

[0039] Therefore, the disclosed embodiment provides a cooling device comprising two independent cooling channels. The cooling device includes a lower plate having a main fluid channel for directly removing heat from the microchip. An upper plate is attached to the lower plate and includes a secondary cooling channel for indirectly removing heat from the microchip and for providing redundant cooling loops. Thus, two sets of fluid inlet / outlet ports are provided, one for circulating cooling fluid in the main channel and the other for circulating fluid in the secondary channel. Multiple two-phase units are disposed in the lower plate, transferring heat from the microchip to the fluid circulating in the secondary fluid channel through the evaporation and condensation of the fluid contained therein. The cooling device is attached to the microchip or included in the microchip package.

[0040] According to a further disclosed aspect, a method for manufacturing a cooling plate for a microchip includes: fabricating a lower plate by setting a first metal plate and forming a main cooling channel in the first metal plate; fabricating an upper plate by setting a second metal plate and forming a secondary cooling channel; fabricating a first set of inlet / outlet ports in the upper plate, the first set of inlet / outlet ports having fluid channels leading to the secondary cooling channels; fabricating a second set of inlet / outlet ports in the upper plate, the second set of inlet / outlet ports having openings at the bottom surface of the upper plate to form fluid channels leading to the main cooling channels after the upper plate is attached to the lower plate; and attaching the upper plate to the lower plate.

[0041] In the foregoing description, embodiments of the invention have been described with reference to specific exemplary embodiments. It will be apparent that various modifications can be made to these embodiments without departing from the broader spirit and scope of the invention as set forth in the appended claims. Therefore, this specification and the accompanying drawings should be considered illustrative rather than restrictive.

Claims

1. A cooling plate for cooling a microchip, comprising: The lower plate includes a main fluid cooling arrangement, which includes a plurality of two-phase cooling units and a main cooling channel formed in the lower plate, wherein the plurality of two-phase cooling units are immersed in the main cooling channel; An upper plate is attached to the lower plate and has a secondary fluid cooling arrangement fluidly separated from the main fluid cooling arrangement, wherein the secondary fluid cooling arrangement includes: a secondary cooling channel formed in the upper plate, a fluid inlet port fluidly connected to the secondary cooling channel, and a fluid outlet port fluidly connected to the secondary cooling channel. The main cooling channel includes cooling fins; the two-phase cooling unit includes a wicking structure; and the cooling fins and the two-phase cooling unit are arranged alternately.

2. The cooling plate as described in claim 1, wherein, The main fluid cooling arrangement also includes: Fluidly connected to the main fluid inlet port of the main cooling channel, and It is fluidly connected to the main fluid outlet port of the main cooling channel.

3. The cooling plate as described in claim 1, wherein, The secondary cooling channel includes fins.

4. The cooling plate as claimed in claim 1 further includes a sealing ring disposed between the upper plate and the lower plate.

5. The cooling plate as claimed in claim 1 further includes a leakage sensor.

6. A method for manufacturing a cooling plate for a microchip, comprising: A first metal plate is provided, and a main cooling channel and a plurality of two-phase cooling units are formed in the first metal plate. The plurality of two-phase cooling units are immersed in the main cooling channel to manufacture the lower plate. A second metal plate is installed to form a secondary cooling channel, thereby manufacturing the upper plate; A first set of inlet and outlet ports is manufactured in the upper plate, the first set of inlet and outlet ports having fluid passages leading to the secondary cooling channel; A second set of inlet and outlet ports is manufactured in the upper plate, the second set of inlet and outlet ports having openings at the bottom surface of the upper plate to form a fluid passage leading to the main cooling channel after the upper plate is attached to the lower plate; as well as Cooling fins are formed in the main cooling channel, a wicking structure is provided inside the two-phase cooling unit, and the cooling fins and the two-phase cooling unit are arranged alternately.

7. The method of claim 6, wherein, Attaching the upper plate to the lower plate includes welding, brazing, or bonding the upper plate to the lower plate.

8. The method of claim 6 further includes providing a sealing ring between the upper plate and the lower plate.

9. The method of claim 8, further comprising attaching a leak detector to one of the upper plate and the lower plate to detect fluid leakage between the upper plate and the lower plate.

10. The method of claim 6, further comprising forming cooling fins in the secondary cooling channel.

11. The method of claim 10, further comprising attaching the two-phase cooling unit to the bottom surface of the upper plate.

12. A method for cooling a microchip using a cooling plate as described in any one of claims 1-5, comprising: Attach the cooling plate to the microchip; The main cooling fluid is circulated through the main cooling channels in the cooling plate to partially remove the heat generated by the microchip; The secondary cooling fluid is circulated through the secondary cooling channels in the cooling plate to further remove the heat generated by the microchip.

13. The method of claim 12, further comprising removing a portion of the heat generated by the microchip by evaporating the fluid within the two-phase cooling unit, and transferring the removed heat to the secondary cooling fluid by condensing the evaporated fluid.

14. The method of claim 12, wherein, The primary cooling fluid is used to partially and directly remove heat from the microchip, and the secondary cooling fluid is used to indirectly remove heat from the microchip.

Images