Cooling device for a system-on-chip

Abstract

A cooling device for a system-on-chip includes a plurality of cooling units including a first side to be coupled to a system-on-chip, each cooling unit spaced apart from the other cooling units to define a gap between adjacent cooling units, and an interposer coupled to a second side of the plurality of cooling units, the interposer including a flexible joint to enable relative movement of the adjacent cooling units.

Description

COOLING DEVICE FOR A SYSTEM-ON-CHIPBACKGROUND

[0001] Electronic printed circuit boards (PCBs), integrated circuits, and systems-on-chip (SOCs), such as those used in computers or other electronic systems, commonly have a board with electronic components mounted thereon. The electronic components can generate a considerable amount of heat due to electrical power consumption. The heat may be dissipated from the components and the board to ensure proper functioning of the components and to prevent damage to any part of the SOC and the overall electronic system.BRIEF DESCRIPTION OF THE DRAWINGS

[0002] FIG. 1A and IB are side views of examples of a system-on-chip assembly including a cooling device;10003] FIG. 2 is a side view of an example of a system-on-chip assembly including a cooling device;

[0004] FIG. 3 is a side view of an example of a system-on-chip assembly including a cooling device;

[0005] FIGS. 4A and 4B areside views of examples of a system-on-chip assembly including a cooling device;

[0006] FIGS. 5-8 are top views of examples of interposers with flexible joints;

[0007] FIG. 9 shows cross-sectional side views of examples of flexible joints in interposers;

[0008] FIG. 10 is an example of a method of manufacturing, assembling, and installing a cooling device.

[0009] FIG. 11 illustrates the cooling device in the process of being manufactured according to the method of FIG. 10.|0010] It will be recognized that the figures are schematic representations for purposes of illustration. The figures are provided for the purpose of illustrating the implementations withthe explicit understanding that the figures will not be used to limit the scope of the meaning of the claims.DETAILED DESCRIPTION

[0011] Heat can be removed from a system-on-chip (SOC) using a variety of techniques. For example, the heat can simply be transferred by convection to ambient air. However, the amount of heat transferred using this method may be relatively low. To enhance the effectiveness of dissipation by convection, a fan can be added to force the air to flow over the component. A heat sink or heat spreader may be attached to the component to further enhance heat dissipation by conducting the heat away from the heat source to a large surface area from which the heat may be dissipated to the air. A heat sink may be a solid piece of material with high thermal conductivity, such as metal, with multiple fins extending from a base coupled to the SOC. The fins increase the surface area of the heat sink, increasing the amount of heat that can be carried away by convection as air or another fluid flows over the heat sink through the fins. For even greater heat dissipation, liquid-cooled heat dissipation systems can be used.

[0012] Liquid-cooled systems may include a pump or compressor, a cooling manifold including fluid channels, and a heat sink. The cooling manifold can be coupled to the SOC to allow the heat to transfer from the cooling plate to the cooling fluid. The compressor pumps the cooling fluid through the manifold channels and through pipes or conduits that carry the fluid to the heat sink at a location remote from the SOC. In some cases, the heat sink may include a compressor for active refrigeration of the cooling fluid.

[0013] A thermal interface material (TIM) may be positioned between the SOC and the heat sink or cooling manifold to improve the thermal conductivity and heat transfer therebetween. Numerous materials can be used as the TIM, including thermal grease, polymer-based TIMs, carbon-based TIMs, and metal -based TIMs. Metal-based TIMs may have the highest thermal conductivity, as well as low thermal contact resistance, and may thus provide the best heat transfer from the SOC to the heat sink or cooling manifold. These metal-based TIMs may include silver sintering paste, gallium-based liquid metal, solder material, or TIMS including gold, silver, or copper bond interfaces.

[0014] Some TIMs, including metal-based TIMs such as silver sintering paste, may rigidly bond the heat sink or cooling manifold directly to the SOC. As the SOC heats up during use, differences in temperature, as well as differences in thermal expansion coefficients betweenthe SOC, the TIM, and the heat sink or cooling manifold, may cause stress on the TIM and the SOC. In larger SOCs, this stress can result in cracking of the SOC or the TIM and / or delamination of the TIM from the SOC or the heat sink or cooling manifold. Thus, less rigid TIMs, such as thermal grease, two-phase material, or polymer-based TIMs, have typically been used when the SOC size exceeds about 20mm x 20mm. However, as discussed above, these TIMs may have lower thermal conductivity than metal-based TIMs like silver sintering paste. Accordingly, a solution that allows rigid TIMs to be used on larger SOCs is desirable.

[0015] Referring now to FIGS. lA and IB, a SOC assembly 100 including a SOC 102 bonded to a heat sink 104 by a TIM 106 is shown in a typical arrangement before and after the SOC assembly heats up during use, respectively. The SOC 102 may include processing components 108 mounted to a substrate 110 (e.g., an organic or inorganic substrate). As shown in FIG. 1A, due to manufacturing imperfections, the substrate 110 and the SOC 102 as a whole may be slightly convex, while the lower surface of the heat sink 104 may be substantially flat. It should be understood that the figures are not to scale, and the bends shown in the SOC are exaggerated for clarity. The TIM 106 may be applied when the SOC 102 and heat sink 104 are “cold” (e.g., at room temperature). For example, as the SOC 102 heats up during use and heat is transferred through the TIM 106 to the heat sink 104 and the SOC 102, the TIM 106, and the heat sink 104 undergo thermal expansion. Temperature rise in the SOC during operation may cause a change in SOC shape. For example, as shown in FIG. IB, the amount of bow in the convex SOC 102 may reduce as the SOC 102 heats up, and the SOC 102 may become less convex or even concave. This may increase the thickness of the TIM 106 more at the center than at the edges when thermal grease or two phase TIM materials are used. This may cause the SOC 102 to straighten or become concave. The heat sink 104 may be stiffer than the SOC 102 and may resist the bending along with the SOC 102. This may impart significant stresses into the SOC 102 with a rigid TIM, such as a metal TIM, which can result in cracks in the substrate 110 and / or cracks in or delamination of the TIM 106. Differences in thermal expansion coefficients between the heat sink 104, the SOC 102, and the TIM 106 may further contribute to these stresses.

[0016] Referring now to FIG. 2, a SOC assembly 200 (e.g., a SOC module) is shown, according to some examples. Instead of a single, unitary heat sink (e.g., heat sink 104), the SOC assembly 200 includes a cooling device 202 that includes multiple cooling units (e.g., heat dissipation modules), shown in FIG. 2 as microchannel manifolds 204 (e.g., siliconmicrochannel manifolds) each bonded to the SOC 102 by separate TIMs 206 on a first side of each microchannel manifold 204. In some examples, the cooling units may be physical heat sinks, each with a plurality of fins to increase the surface area of the heat sink to improve convection or, as shown in FIG. 2, liquid cooling manifolds (e.g., microchannel manifolds, silicon microchannel manifolds, etc.). Each microchannel manifold 204 (or other type of cooling unit) may be spaced apart from the other microchannel manifolds 204, allowing each microchannel manifold 204 to move relative to the other microchannel manifolds 204. Because the microchannel manifolds 204 are separated from each other, as the SOC 102, microchannel manifolds 204, and TIMs 206 heat up, the microchannel manifolds 204 can move toward or away from each other and / or move vertically (as shown) relative to each other as the bend in SOC 102 changes (e.g., becoming less convex) without imparting excess stress into the SOC 102 or the TIMs 206. Unlike the SOC 100 assembly of FIGS. 1A and IB, in which the stiffness of the large unitary heat sink 104 resists the bending of the SOC 102 causing stress at the bond interface, the spaces between the microchannel manifolds 204 allow the SOC 102 to deform with less restriction, thus reducing the bond interface stress. This allows rigid materials, including metal-based TIMs with higher thermal conductivity such as silver sintering paste, to be used for the TIMs 206. In some examples, rigid thermal interface materials may refer to thermal interface materials with a modulus of elasticity higher than 15 GPa at room temperature. Thus, in contrast to the SOC assembly 100, the SOC assembly 200 allows for TIMs with very high thermal conductivity and very low contact resistance to be used despite their rigidity. And in contrast to what is illustrated in FIG. 2, in some examples, the TIM 206 may be continuous while the microchannel manifolds 204 are separated from each other. For example, the SOC assembly 200 may include a single section of TIM 206 bonded to two or more microchannel manifolds 204, rather than including a separate section of TIM 206 for each microchannel manifold 204. The separation of the microchannel manifolds 204 may be sufficient to reduce stress without also separating the TIM 206 as shown in FIG. 2.

[0017] In some examples, the SOC assembly 200 may include a bridge (e.g., a connector, a distributor, a common member, etc.) that connects two or more of the cooling units. For example, as shown in FIG. 2, the bridge is a coolant distribution manifold 208 (e.g., a coolant delivery manifold, a fluid distribution manifold, a liquid distribution manifold, etc.) that supplies liquid to the fluid channels of each microchannel manifold 204 to which it is connected. In some examples, neither the bridge nor any other incompressible, non-viscousmaterials may extend into the gaps between the microchannel manifolds 204. The coolant distribution manifold 208 and the microchannel manifolds 204 may thus enable independent movement of individual microchannel manifolds 204 relative to the other microchannel manifolds 204, for example, in response to thermal expansion. In some examples, thermal grease or other viscous or compressible materials may be applied to fill the gaps between the microchannel manifolds 204, for example, to increase heat conduction between the microchannel manifolds 204. In some examples, no material may be positioned between the microchannel manifolds 204. In a SOC assembly 200 in which the cooling units are solid, aircooled heat sinks, the bridge may be a heat-conducting structure that further conducts heat away from the cooling units and the SOC 102. For example, the bridge may itself be a heat sink with a plurality of fins to increase the surface area of the bridge to improve convection. Heat may conduct from the SOC 102 into the cooling units, and from the cooling units into the bridge. Heat may be removed from both the cooling units and the bridge either by natural convection or by forced convection using a fan.

[0018] The coolant distribution manifold 208 may be coupled to the microchannel manifolds 204 by flexible seals 210 on a second side of each microchannel manifold 204, opposite the first side of each microchannel manifold 204, which may be bonded to the SOC 102 by a rigid TIM 206, such as a metal -based TIM. The flexible seals 210 may be made of elastomer or other flexible materials. The flexible seals 210 may allow the microchannel manifolds 204 to move relative to the coolant distribution manifold 208 without imparting significant stresses into the microchannel manifolds 204, the flexible seals 210, or the coolant distribution manifold 208. The flexible seals 210, in combination with the gaps between the cooling units, may enable independent movement of individual microchannel manifolds 204 relative to the other microchannel manifolds 204. In some examples, a single flexible seal 210 (e.g., a gasket) may be positioned between the coolant distribution manifold 208 and all of the microchannel manifolds 204 or a subset of the microchannel manifolds 204. Because the interfaces between the microchannel manifolds 204 are farther from the SOC 102 than the TIM 206 is, a flexible material with less thermal conductivity than the TIM 206, such as the flexible seals 210, may be used. A coolant distribution manifold 208 coupled to two or more of the microchannel manifolds 204 may improve the thermal integration of the system, allowing heat to transfer between microchannel manifolds 204, and may also improve the efficiency of coolant distribution. Liquid-cooled cooling devices 202 are discussed in further detail below with regard to FIGS. 3 A-3C. In some examples, a coolant distribution manifold208 may be coupled to one microchannel manifold 204. Thus, each microchannel manifold 204 may include a separate coolant distribution manifold 208.

[0019] As discussed above, in a liquid-cooled system, the cooling units may be microchannel manifolds 204 (e.g., silicon microchannel manifolds) and the bridge may be a coolant distribution manifold 208. Due to manufacturing constraints, it may be difficult to manufacture the channels of the coolant distribution manifold 208 (which may be made of a molded thermoplastic material) to accurately provide fluid directly to the smallest microchannels of the microchannel manifolds 204. Instead, the silicon microchannel manifolds 204 may have several (e.g., three, four, etc.) layers of fluid channels, with the smallest, most tightly packed channels in the layer closest to the SOC 102 and larger channels in the layers approaching the coolant distribution manifold 208, which may include even larger fluid channels. Most of the heat exchange occurs in the layer closest to the SOC 102, with the other layers used to maintain high coolant flow through the channels, uniform coolant temperature distribution, and low coolant pressure drop at the coolant input and output ports. The additional layers are also used to account for a mismatch in coefficients in thermal expansion (CTE) between the silicon microchannel manifolds 204 and the coolant distribution manifold 208, which may be made of a thermoplastic material. For example, silicon has a CTE of about 3ppm / °C, and thermoplastic materials may have a CTE in the range of 25ppm / °C to 50ppm / °C. An excess increase in temperature of the coolant distribution manifold 208 may cause stress in the bond interface between the microchannel manifold 204 and the coolant distribution manifold 208, which may include flexible seals 210 and may include an adhesive bond layer. The layer of a microchannel manifold 204 closest to the SOC 102 may experience the highest temperature change due to its proximity to the SOC 102, while each subsequent layer approaching the coolant distribution manifold 208 may experience progressively lower temperature change, essentially insulating the coolant distribution manifold 208 from the SOC 102. This reduces the temperature change of the coolant distribution manifold 208, thus minimizing the effect of the CTE mismatch between the silicon microchannel manifold 204 and the coolant distribution manifold 208. Additional layers of silicon may thus reduce the stress at the bond interface. However, silicon is a relatively expensive material, and additional layers increase the overall size of the cooling device 202, which may be undesirable in electronics where space is a concern.[00201 Referring now to FIG. 3, a SOC assembly 200 is shown, according to some examples. To reduce cost and the overall height of the cooling device 202, layers (or portions) of the cooling units (e.g., the microchannel manifolds 204) may be replaced by an interposer 209. The interposer 209 is positioned between the coolant distribution manifold 208 and microchannel manifolds 204. In some examples, all of the microchannel manifolds 204 may be coupled to one interposer 209. In other examples, less than all of the microchannel manifolds 204 may be coupled to one interposer 209. For example, the cooling device 202 may include multiple interposers 209 each coupled to a subset of the microchannel manifolds 204. In some examples, there may be a separate interposer 209 coupled to each cooling unit. However, it may be desired to use one interposer 209 to couple together multiple microchannel manifolds 204, so that the spaces between the microchannel manifolds 204 can be tightly controlled. For example, it may be desired to have the microchannel manifolds 204 very close together such that substantially an entire surface of the SOC 102 (e.g., the upper surface of the SOC 102 as shown in FIG. 3) is covered by the microchannel manifolds 204, with the gaps positioned over an insubstantial portion of the surface. For example, the gap between each microchannel manifold 204 may be about 200 microns wide or less. Such a small gap may be difficult to maintain if each microchannel manifold 204 is individually bonded to the SOC 102. Accordingly, each microchannel manifold 204 may first be coupled to the interposer 209, maintaining a very small gap between the microchannel manifolds 204, and the complete cooling device 202 can be to the SOC 102. This may also allow the cooling devices 202 to be pre-assembled for installation on existing SOCs 102, for example, to replace an existing heat sink. The cooling devices 202 may be manufactured and / or assembled using precision machinery to maintain the very small gaps between the microchannel manifolds 204.[00211 In a liquid-cooled cooling device 202, the interposer may include fluid channels that fluidly connect the fluid channels of the coolant distribution manifold 208 to the fluid channels of the microchannel manifolds 204. The interposer 209 may be made of an epoxy mold compound, which may have a CTE of about 7 ppm / °C. Because the interposer 209 may have a CTE between that of the microchannel manifold 204 (e.g., 3 ppm / °C) and the coolant distribution manifold 208 (e.g., 25-50ppm / °C), the interposer 309 may provide a transition layer that reduces stress at the interfaces compared to a cooling device 202 in which the microchannel manifold 204 and the coolant distribution manifold 208 are directly coupled together. The interposer 209 with fluid channels may replace layers of silicon microchannelmanifolds 204 with the lower-cost EMC while providing the same or more effective cooling. For example, in a cooling device 202 in which each silicon microchannel manifold 204 includes four layers, the upper two layers of each silicon microchannel manifold 204 may be removed and replaced by the interposer. The complete cooling device 202 with two silicon layers and an interposer may provide equivalent or improved cooling compared to the cooling device with four silicon layers. The interposer 209 may also be thinner than the combined thickness of the two silicon layers, resulting in a lower overall height of the cooling device 202. The interposer 209 may include several thin layers (e.g., thinner than the layers of the silicon microchannel manifolds 204), allowing for the fluid channels to fan out from the lower layer adjacent the microchannel manifolds 204 to the layer adjacent the coolant distribution manifold 208.

[0022] As shown in FIG 3, the interposer 209 is bonded to each of the microchannel manifolds 204, for example, with an adhesive. If the interposer 209 is rigid, this may cause a result similar to that discussed above with respect to FIG. 1. As the SOC 102 deforms due to thermal stress, the rigid interposer 209 may resist the ability of the microchannel manifolds 204 to move relative to one another. If the TIM 206 is also rigid, this may cause stress at the interface between the microchannel manifolds 204 and the SOC 102. For example, as shown in FIG. 3, the TIM 206 may be nonuniform or missing in certain areas when the SOC 102 deforms and may experience stress that can lead to cracking. It should be understood that FIG. 3 is not to scale, and the bending is exaggerated to show its effects. Further, although a single section of TIM 206 is shown between the SOC 102 and all of the microchannel manifolds 204, similar effects may result from the deformation of the SOC 102 even if each microchannel manifold 204 is coupled to the SOC 102 by a separate section of TIM 206.10023] Referring now to FIG. 4A, a SOC assembly 200 with a cooling device 302 is shown, according to some examples. The cooling device 302 may be substantially similar to the cooling device 202 shown in FIG. 3, except as shown and described. Unlike the relatively rigid interposer 209 of the cooling device 202, the cooling device 302 includes an interposer 309 with flexible joints 311 in the areas between the microchannel manifolds 204. The flexible joints 311 may allow the microchannel manifolds 204 to move relative to and substantially independent of one another when the SOC 102 deforms, as described above with reference to FIG. 2. Because the microchannel manifolds 204 are separated from each other and the interposer 309 includes flexible joints 311, as the SOC 102, microchannel manifolds204, and TIMs 206 heat up, the microchannel manifolds 204 can move relative to each other as the SOC 102 deforms (e.g., becomes less convex) without imparting excess stress into the SOC 102 or the TIMs 206. The SOC assembly 200 thus allows for TIMs with very high thermal conductivity and very low contact resistance to be used, despite their rigidity. In some examples, the TIM 206 may be continuous while the microchannel manifolds 204 are separated from each other. For example, the SOC assembly 200 may include a single section of TIM 206 bonded to two or more microchannel manifolds 204, rather than including a separate section of TIM 206 for each microchannel manifold 204. The separation of the microchannel manifold 204 may be sufficient to reduce stress without also separating the TIM 206. The interposer 309 may thus create insubstantial resistance to the movement of the microchannel manifolds 204 relative to one another compared to the interposer 209 of FIG. 3. However, because the interposer 309 still couples the microchannel manifolds 204 together, the microchannel manifolds 204 may be precisely positioned and coupled to the interposer 309 to maintain tightly controlled gaps between the microchannel manifolds 204 (e.g., gaps of about 200 micrometers or less). The flexible joints 311 may be, for example, in the form of partial or complete cuts in the interposer 309 or a combination of partial and complete cuts.10024] The interposer 309 may allow the cooling device 302 to be fully assembled before being installed on the SOC 102. It may be helpful to have the microchannel manifolds 204 very close together such that substantially an entire surface of the SOC 102 (e.g., the upper surface of the SOC 102 as shown in FIG. 2) is covered by the cooling units, with the gaps positioned over an insubstantial portion of the surface. For example, the gap between each microchannel manifold 204 may be about 200 microns wide or less. Such a small gap may be difficult to maintain if each microchannel manifold 204 is individually bonded to the SOC 102. Accordingly, each microchannel manifold 204 may first be coupled to the interposer 309, maintaining a very small gap between the microchannel manifolds 204 aligned with the flexible joints 311, which may then be coupled to a coolant distribution manifold 208 to form a complete cooling device 302, which may then be coupled to the SOC 102. This may also allow the cooling devices 302 to be pre-assembled for installation on existing SOCs 102, for example, to replace an existing heat sink. The cooling devices 202 may be manufactured and / or assembled using precision machinery to maintain the very small gaps between the microchannel manifolds 204.[00251 FIG. 4B shows an example of the cooling device 202 in a liquid-cooled SOC assembly 200. The cooling units are shown as microchannel manifolds 204 with microchannels 249 extending therethrough to carry fluid near the surface of the SOC 102. The microchannels are fluidly coupled to fluid channels 316 in the bond portions 313 of the interposer 309. The fluid channels 316 are also fluidly coupled to fluid channels 247 in the coolant distribution manifold 208. Liquid coolant (e.g., water, oil) may be supplied to the inlet 212 of the coolant distribution manifold 208, may travel through the fluid channels 247 and fluid channels 316 to the microchannels 249, where the coolant may absorb heat from the SOC 102. The coolant may then return to the coolant distribution manifold 208 and be expelled from the outlet 213. The coolant may then be cooled by a heat sink and / or a refrigeration unit and recirculated to the inlet 212. As shown in FIG. 4B, the fluid channels 316 do not extend through the flexible joint 311 but instead return coolant from each microchannel manifold 204 to the coolant distribution manifold 208. Although the fluid channels 316 in the bond portions are shown in series, with the coolant from the outlet side of the first bond portion 313 being delivered to the inlet side of the second bond portion 313, in other examples, the fluid channels 316 may be arranged in series, with the coolant supplied to the inlet 212 being split and separately supplied to each microchannel manifold 204 before being delivered to the outlet 213.|0O26] FIG. 5-8 show examples of interposers 309 with ten bond portions 313 to be coupled to ten microchannel manifolds 204. The example interposers 309 shown in FIGS. 5-8 include different flexible joint designs that may produce varying levels of flexibility and structural support. In the example interposer 309 of FIG. 5, a flexible joint 311 is positioned between each of the ten bond portions 313 (which may also be referred to as interposer tiles, tile portions, etc.). The flexible joints 311 include both through-cut portions 315 and partial-cut portions 317. The through-cut portions 315 extend all the way through the interposer 309, while the partial-cut portions 317 are slots or channels that do not extend all the way through the interposer 309. The through-cut portions 315 provide no resistance to movement of the microchannel manifolds 204 relative to each other. The partial-cut portions 317 provide relatively insubstantial resistance to movement of the microchannel manifolds 204 relative to each other (e.g., compared to an interposer with no cut portions) while coupling the individual bond portions 313 together. As shown in FIG. 5, each bond portion 313 includes fluid channels 316 for fluidly coupling a coolant distribution manifold to a microchannel manifold cooling unit. The interposer 309 includes interruptions 319 in the through-cut portions 315 between channels 316, such that the channels 316 are fluidly sealed, and at the intersectionsof the vertical through-cut portions 315 and the horizontal through-cut portion 315. The flexible joints 311 may broadly be described as having a lower cross-sectional area on a plane extending longitudinally through the flexible joint 311 than a cross-sectional area of an adjacent bond portion 313 along a parallel plane. Stated another way, the thickness (i.e., into the page as shown in FIGS. 5-8) of the interposer 309 at the flexible joint 311 or a portion of the flexible joint 311 may be less than the thickness of the interposer 309 at an adjacent bond portion 313. The interposer 309 of FIG. 6 similarly includes ten bond portions 313 but does not include a horizontal flexible joint 311. Instead, the interposer 309 includes four vertical flexible joints 311 in the form of uninterrupted through cuts. Thus, the interposer 309 may have compliance (e.g., flexibility) about a vertical axis (as shown in FIG. 6) but may not have substantial compliance about a horizontal axis (as shown in FIG. 6). This interposer 309 may be used in a cooling device 302, for example, when the SOC 102 to which the cooling device 302 is coupled is expected to bow primarily about a single, known axis. For example, as shown in FIGS 1A and IB, the SOC 102 bows about an axis extending into the page. If the SOC 102 does not also bow about a horizontal (as shown) axis, the interposer 309 of FIG. 6 may be arranged with the flexible joints 311 extending into the page, parallel to the bend axis. If the SOC 102 does not also bow about the horizontal axis, flexible joints extending parallel to the horizontal axis may not be included. The interposer 309 of FIG. 7 is substantially similar to the interposer 309 of FIG. 5, except that the interposer 309 of FIG. 7 does not include partial-cut portions 317. Instead, the flexible joints 311 are in the form of through cuts that do not extend to the edges of the interposer 309, leaving solid edge portions 312 at the edges. Further, the horizontal flexible joint 311 includes interruptions between the fluid channels 316 but does not include interruptions at the intersections of the vertical flexible j oints 311 and the horizontal flexible j oint 311. The interposer of FIG. 7 may be less flexible than the interposer of FIG. 5 due to the lack of partial-cut portions 317, but the solid edge portions 312 may couple the bond portions 313 together more strongly than the partial-cut portions 317 of the interposer 309 of FIG. 5. The interposer 309 of FIG. 8 includes flexible joints 311 in the form of partially cut slots 321 that do not extend completely through the interposer 309. Because the slots 321 do not extend completely through the interposer 309, the slots 321 may extend to the edges of the interposer 309 while the bond portions 313 remain coupled together. The slots 321 may provide flexibility while coupling the bond portions 313 together more strongly than the partial-cut portions 317 of the interposer 309 of FIG. 5 than the flexible joints 311 that include through-cut portions 319 as shown in FIGS. 6-8.[O027| Referring now to FIG. 9, several cross-sectional views of flexible joints 31 la-31 Ih are shown. As discussed above, a flexible joint 311 may also include through-cut portion 315 along the length of the j oint 311. In some examples, the flexible j oints 311 may be cut into a flat interposer 309 using a laser, saw, mill, or router. In other examples, the flexible joints 311 may be molded into a flat interposer 309. For example, a layered molding process, such as the molded interconnect substrate manufacturing process, may be used to form the flexible joints 311. The flexible joints 311 may alternatively be molded into the interposer 309 Flexible joint 311a includes two straight partial cuts 323 (which may also be referred to as slots) in a first side of the interposer 309 extending past the centerline of the interposer 309 and one straight partial cut 323 in the other side of the interposer 309 extending past the centerline of the interposer 309 and positioned in between the two straight partial cuts 323 on the first side. The partial cuts 323 extend along the interposer 309 parallel to each other to form the flexible joint 311a. Flexible joint 311b includes three straight partial cuts 323 in a first side of the interposer 309 extending past the centerline of the interposer 309 and two straight partial cuts 323 in the other side of the interposer 309 extending past the centerline of the interposer 309 and positioned in between the three straight partial cuts 323 on the first side. Flexible joint 311c includes one straight partial cut 323 in each side of the interposer 309 and aligned with the partial cut 323 on the opposite side. Flexible joint 3 l id includes one straight partial cut 323 in a first side of the interposer 309 extending past the centerline of the interposer 309 and one straight partial cut 323 in the other side of the interposer 309 extending past the centerline of the interposer 309 and offset from the straight partial cut 323 on the first side.|0028] Flexible joint 3 l ie includes a channel 322 on each side of the interposer 309. The channels 322 each include a vertical portion 324 extending past the centerline of the interposer 309 and a horizontal portion 325 extending from the end of the respective vertical portion 324 toward the vertical portion 324 of the other channel 322. The horizontal portions 325 overlap in the horizontal direction. As it may be challenging to form the shapes of the channels 322 using machining processes, a molded interconnect substrate manufacturing process may be used to form the channels 322. Layers of epoxy mold compound and metallic (e.g., copper) traces may be formed, with the metallic traces forming the shape of the channels. The metallic traces may then be chemically etched to remove the metallic material, leaving the channels 322 in the epoxy mold compound.[00291 Flexible joint 31 If includes two straight partial cuts 323 in a first side of the interposer 309 but not extending past the centerline of the interposer 309 and one straight partial cut 323 in the other side of the interposer 309 but not extending past the centerline of the interposer 309 and positioned in between the two straight partial cuts 323 on the first side. Flexible joint 311g includes one pyramid-shaped partial cut 327 in a first side of the interposer 309 extending past the centerline of the interposer 309, becoming progressively narrower as its depth increases, and one pyramid-shaped partial cut 327 in the other side of the interposer 309 extending past the centerline of the interposer 309, becoming progressively narrower as its depth increases, and offset from the straight partial cut 323 on the first side. Flexible joint 31 Ih includes a single through cut 331 extending completely through the interposer 309. A layer of high-temperature flexible tape 329 couples the two bond portions 313. Each of these flexible joints 311 reduces the resistance to the movement of the bond portions 313 relative to each other, thereby reducing the resistance to the movement of the microchannel manifolds 204 relative to each other when coupled to the bond portion 313.

[0030] Referring now to FIG. 10 an example of a method 400 of manufacturing, assembling, and installing a cooling device is shown. FIG. 11 further illustrates the operations of the method 400 performed to manufacture, assemble, and install an example cooling device 302. At operation 402 of the method 400 an interposer with bond portions separated by flexible joints (e.g., interposer 309) is manufactured. The interposer 309 may be molded using molded interconnect substrate technology, may be injection molded, or may be machined. For example, the flexible joints 311 may include slots or channels cut using a laser, saw, mill, or router. The flexible joints 311 may alternatively be molded into the interposer 309. Fluid channels (e.g., fluid channels 316 as shown in FIGS. 5-8) may also be molded or cut into the interposer 309. At operation 404 of the method 400, cooling units (e.g., microchannel manifolds 204 as shown in FIG. 4) are coupled to the bond portions 313 of the interposer 309. A layer of adhesive 333 may be applied to each bond portion 313 of the interposer, to each microchannel manifold 204, or to both the microchannel manifolds 204 and the bond portions 313. Each microchannel manifold 204 may be separated from the other cooling units by a gap 211. The width of the gap 211 may be approximately 200 micrometers or less. When the flexible joint 311 between adjacent bond portions 313 is a slot or channel, the gap 211 may substantially align with the gap 211. For example, a plane extending through the gap 211 may also extend longitudinally through the flexible joint 311. The microchannel manifolds 204 may be microchannel manifolds (e.g., silicon microchannel manifolds) withmicrochannels that are fluidly coupled to the channels of the interposer 309 when the microchannel manifolds 204 are coupled to the interposer 309. At operation 406 of the method 400, the interposer 309 is coupled to a bridge (e.g., coolant distribution manifold 208 as shown in FIG. 4) to form a complete cooling device (e.g., cooling device 302 as shown in FIG. 4). In some examples, the interposer 309 may be coupled to the coolant distribution manifold 208 by flexible seals 210. In some examples, the interposer 309 may be coupled to the coolant distribution manifold 208 by an adhesive bond layer. The coolant distribution manifold 208 includes fluid channels that are fluidly coupled to the fluid channels of the interposer 309 when the interposer 309 is coupled to the coolant distribution manifold 208.

[0031] At operation 408 of the method 400, the cooling device 302 is coupled to a SOC (e.g., SOC 102 as shown in FIG. 4). Operation 408 may include applying a TIM 206 to one or both of the SOC 102 or the microchannel manifolds 204, positioning the cooling device 302 on the SOC 102, and curing the TIM 206. As discussed above, a rigid TIM 206, such as silver sintering paste, may be used because the flexible interposer 309 allows the cooling units to move relative to each other. A mask may be applied to the SOC 102 such that when the TIM 206 is applied, each portion of TIM 206 is separated. During testing, tape was used to simulate an interposer 309 to determine the effects of flexible joints 311 in interposer 309. In a first test setup, the tape was bonded to each of the microchannel manifolds 204 and at each end to a stiffener frame coupled to the SOC 102 to simulate a rigid interposer 309 that does not allow the microchannel manifolds 204 to conform to the surface of the SOC 102. In a second test setup, the tape was bonded to each of the microchannel manifolds 204 but not bonded to the stiffener frame in order to simulate an interposer 309 with flexible joints allowing the microchannel manifolds 204 to conform to the surface of the SOC 102. The bond line thicknesses of the TIM 206 with each test setup were measured at various locations across the surface of the SOC 102. It was determined that in the first test setup, representing a cooling device 202 with a rigid interposer 309, the standard deviation of the bond line thickness measurements was 175 micrometers. In the second test setup, representing the cooling device 302 with an interposer 309 with flexible joints 311, the same measurements had a standard deviation of 21 micrometers. Thus, the test setup representing the cooling device 302 with an interposer 309 with flexible joints 311 had a much more consistent and controllable bond line thickness, indicating that the cooling of the SOC 102 can be more tightly controlled and consistent across the entire surface of the SOC 102.[00321 In an aspect, a cooling device for a system-on-chip is provided. The cooling device includes a plurality of cooling units including a first side to be coupled to a system-on-chip, each cooling unit spaced apart from the other cooling units to define a gap between adjacent cooling units. The cooling device further includes an interposer coupled to a second side of the plurality of cooling units, the interposer including a flexible joint to enable relative movement of the adjacent cooling units.[00331 In some examples, the flexible joint separates two tile portions of the interposer, the adjacent cooling units each coupled to one of the two tile portions with the gap therebetween.|0034] In some examples, a thickness of the interposer at the flexible joint is lower than a thickness of the interposer at an adjacent tile portion.

[0035] In some examples, the flexible joint comprises flexible tape extending across a channel between the two tile portions.

[0036] In some examples, the flexible joint comprises a slot or channel in the interposer.

[0037] In some examples, a first cooling unit of the plurality of cooling units is a microchannel manifold including a first plurality of fluid channels, and the interposer comprises a second plurality of fluid channels fluidly coupled to the first plurality of fluid channels.10038] In some examples, the cooling device includes a coolant distribution manifold comprising a third plurality of fluid channels fluidly coupled to the second plurality of fluid channels.

[0039] In some examples, the plurality of cooling units are coupled to the interposer by an adhesive.[0040[ In some examples, the flexible joint comprises a first slot in a first face of the interposer and a second slot in a second face of the interposer extending parallel to and proximate the first slot.|0041] In another aspect, a cooling device for a system-on-chip is provided. The cooling device includes a plurality of cooling units including a first side to be coupled to a system-on-chip, each cooling unit spaced apart from the other cooling units to define a gap between adjacent cooling units, each cooling unit including fluid microchannels. The cooling device further includes an interposer coupled to a second side of the plurality of cooling units, the interposer comprising a flexible joint to enable relative movement of the adjacent cooling units and a first plurality of fluid channels fluidly coupled to the microchannels of each cooling unit.[00421 In some examples, none of the first plurality of fluid channels in the interposer extends through the flexible joint.

[0043] In some examples, the cooling device includes a coolant distribution manifold comprising a second plurality of fluid channels fluidly coupled to the first plurality of fluid channels.

[0044] In another aspect, a cooling device for a system-on-chip is provided. The cooling device includes a coolant delivery manifold including a first plurality of fluid channels and an interposer coupled to the coolant delivery manifold. The interposer includes a first section including a second plurality of fluid channels fluidly coupled to the first plurality of fluid channels, a second section including a third plurality of fluid channels fluidly coupled to the first plurality of fluid channels, and a flexible joint connecting the first section and the second section. The cooling device further includes a first cooling unit coupled to the first section of the interposer and including a fourth plurality of fluid channels fluidly coupled to the second plurality of fluid channels and a second cooling unit spaced apart from the first cooling unit to define a gap therebetween. The second cooling unit is coupled to the second section of the interposer and inlcudes a fifth plurality of fluid channels fluidly coupled to the third plurality of fluid channels.[0045| In some examples, the flexible joint comprises a slot or channel in a surface of the interposer, wherein a thickness of the interposer at the flexible joint is lower than a thickness of the interposer at the first section or second section.10046] In some examples, the first cooling unit and the second cooling unit are silicon microchannel manifolds, and the interposer comprises epoxy mold compound.

[0047] It should be noted that certain passages of this disclosure may reference terms such as “first” and “second” in connection with devices, surfaces or sides of devices, modes ofoperation, transmit chains, antennas, etc., for purposes of identifying or differentiating one from another or from others. These terms are not intended to merely relate entities (e.g., a first device and a second device) temporally or according to a sequence, although in some cases, these entities may include such a relationship. Nor do these terms limit the number of possible entities (e.g., devices) that may operate within a system or environment.[0048[ While this specification contains specific implementation details, these should not be construed as limitations on the scope of what may be claimed but rather as descriptions of features specific to particular implementations. Certain features described in this specification in the context of separate implementations can also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation can also be implemented in multiple implementations separately or in any suitable sub-combination. Moreover, although features may be described as acting in certain combinations and even initially claimed as such, features from a claimed combination can, in some cases, be excised from the combination, and the claimed combination may be directed to a sub-combination or variation of a sub-combination.

[0049] As utilized herein with respect to structural features (e.g., to describe shape, size, orientation, direction, relative position, etc.), the terms “approximately,” “about,” “substantially,” and similar terms are meant to cover minor variations in structure that may result from, for example, the manufacturing or assembly process and are intended to have a broad meaning in harmony with the common and accepted usage by those of ordinary skill in the art to which the subject matter of this disclosure pertains. Unless otherwise noted, the use of the words “approximate,” “about,” “around,” “substantially,” etc., when used to refer to a numerical value mean plus or minus ten percent.

[0050] The term “coupled” and variations thereof, as used herein, means the joining of two members directly or indirectly to one another. Such joining may be stationary (e.g., permanent or fixed) or moveable (e.g., removable or releasable). Such joining may be achieved with the two members coupled directly to each other, with the two members coupled to each other using a separate intervening member and any additional intermediate members coupled with one another, or with the two members coupled to each other using an intervening member that is integrally formed as a single unitary body with one of the two members. If “coupled” or variations thereof are modified by an additional term (e.g., directly coupled), the generic definition of “coupled” provided above is modified by the plain language meaningof the additional term (e.g., “directly coupled” means the joining of two members without any separate intervening member), resulting in a narrower definition than the generic definition of “coupled” provided above. Such coupling may be mechanical, electrical, or fluidic.

[0051] References herein to the positions of elements (e.g., “top,” “bottom,” “upper,” “middle,” “lower,” “above,” “below,” vertical,” “horizontal,” etc.) are merely used to describe the orientation of various elements in the FIGURES. It should be noted that the orientation of various elements may differ according to other examples, and that such variations are intended to be encompassed by the present disclosure.|0052] With respect to the use of substantially any plural and / or singular terms herein, those having skill in the art can translate from the plural to the singular and / or from the singular to the plural as is appropriate to the context and / or application. The various singular / plural permutations may be expressly set forth herein for the sake of clarity.[00531 The foregoing description of illustrative examples has been presented for purposes of illustration and of description. It is not intended to be exhaustive or limiting with respect to the precise form disclosed, and modifications and variations are possible considering the above teachings or may be acquired from practice of the disclosed examples. It is intended that the scope of the claimed subject matter be defined by the claims appended hereto and their equivalents.

Claims

WHAT IS CLAIMED IS1. A cooling device for a system-on-chip, the cooling device comprising: a plurality of cooling units comprising a first side to be coupled to a system-on-chip, each cooling unit spaced apart from the other cooling units to define a gap between adjacent cooling units; and an interposer coupled to a second side of the plurality of cooling units, the interposer comprising a flexible joint to enable relative movement of the adjacent cooling units.

2. The cooling device of claim 1, wherein the flexible joint separates two tile portions of the interposer, the adjacent cooling units each coupled to one of the two tile portions with the gap therebetween.

3. The cooling device of claim 2, wherein a thickness of the interposer at the flexible joint is lower than a thickness of the interposer at an adjacent tile portion.

4. The cooling device of claim 2, wherein the flexible joint comprises flexible tape extending across a channel between the two tile portions.

5. The cooling device of claim 1, wherein the flexible joint comprises a slot or channel in the interposer.

6. The cooling device of claim 1, wherein a first cooling unit of the plurality of cooling units is a microchannel manifold including a first plurality of fluid channels, and wherein the interposer comprises a second plurality of fluid channels fluidly coupled to the first plurality of fluid channels.

7. The cooling device of claim 6, further comprising a coolant distribution manifold comprising a third plurality of fluid channels fluidly coupled to the second plurality of fluid channels.

8. The cooling device of claim 1, wherein the plurality of cooling units are coupled to the interposer by an adhesive.

9. The cooling device of claim 1, wherein the flexible joint comprises a first slot in a first face of the interposer and a second slot in a second face of the interposer extending parallel to and proximate the first slot.

10. A cooling device for a system-on-chip, the cooling device comprising: a plurality of cooling units comprising a first side to be coupled to a system-on-chip, each cooling unit spaced apart from the other cooling units to define a gap between adjacent cooling units, each cooling unit comprising fluid microchannels; and an interposer coupled to a second side of the plurality of cooling units, the interposer comprising a flexible joint to enable relative movement of the adjacent cooling units and a first plurality of fluid channels fluidly coupled to the microchannels of each cooling unit.

11. The cooling device of claim 10, wherein none of the first plurality of fluid channels in the interposer extends through the flexible joint.

12. The cooling device of claim 11, further comprising a coolant distribution manifold comprising a second plurality of fluid channels fluidly coupled to the first plurality of fluid channels.

13. A cooling device for a system-on-chip, the cooling device comprising: a coolant delivery manifold comprising a first plurality of fluid channels; an interposer coupled to the coolant delivery manifold, the interposer comprising: a first section comprising a second plurality of fluid channels fluidly coupled to the first plurality of fluid channels; a second section comprising a third plurality of fluid channels fluidlycoupled to the first plurality of fluid channels; and a flexible joint connecting the first section and the second section; a first cooling unit coupled to the first section of the interposer and comprising a fourth plurality of fluid channels fluidly coupled to the second plurality of fluid channels; and a second cooling unit spaced apart from the first cooling unit to define a gap therebetween, the second cooling unit coupled to the second section of the interposer and comprising a fifth plurality of fluid channels fluidly coupled to the third plurality of fluid channels.

14. The cooling device of claim 13, wherein the flexible joint comprises a slot or channel in a surface of the interposer, wherein a thickness of the interposer at the flexible joint is lower than a thickness of the interposer at the first section or second section.

15. The cooling device of claim 13, wherein the first cooling unit and the second cooling unit are silicon microchannel manifolds, and the interposer comprises epoxy mold compound.

Images