JP2010080455A - Cooling device and cooling method for electronic equipment - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a cooling device and a cooling method for an electronic device capable of shortening a wiring length distance so that components can be mounted closer to an LSI and improving a signal transmission speed.
A single liquid cooling unit is thermally bonded to a single heating element, a surface defined as a region where a plurality of fins are present, and the single heating element with respect to the single liquid cooling unit. Since the thermal bonding surfaces have substantially the same shape and the same area, the cooling device can be reduced as much as possible. For example, peripheral components of LSI can be mounted in the immediate vicinity of LSI as an electronic device. As a result, the wiring length can be shortened and the signal transmission speed can be improved.
[Selection] Figure 1

Description

  The present invention relates to a cooling device and a cooling method for an electronic device, and more particularly to a cooling device and a cooling method for an electronic device having a liquid cooling module structure that uses a liquid such as water as a coolant for cooling LSIs such as processors of the electronic device. .

The degree of integration of LSIs used in electronic devices is increasing at every generation.
With the increase in the degree of integration of LSIs, the amount of heat generation tends to increase, and in some cases, LSIs that generate heat of 200 W or more have come out.
However, in order for the LSI to operate at high speed, it is necessary to control the operating temperature of the LSI below a certain temperature, and this requires the need to attach a cooling module that matches the amount of heat generated by the LSI.

On the other hand, in recent years, not only has the speed of computation inside the LSI improved, but also the speed of signal transmission between the LSI and its peripheral parts, such as between the LSI and the memory, has increased the speed of the entire system that constitutes the electronic device and The need to further increase the speed of electronic devices such as these has become important.
Such increase in signal transmission speed is specifically achieved by reducing the wiring distance. However, such a reduction in the wiring long distance is accompanied by a result that it is difficult to secure a heat radiation area on which the cooling module can be mounted on the LSI, and the heat radiation area on which the cooling module can be mounted on the LSI is reduced. There arises a problem that it cannot be increased to the extent that the increase in amount can be commensurate.

  In view of this, a cooling module having a chamber called a cold plate in which fins are provided on an LSI and a liquid refrigerant can flow between the fins has been proposed. In particular, in order to improve the cooling performance, fine fins with a width of 0.2 mm or less of the flow path between fins are formed, and the heat transfer rate is increased by increasing the contact area between the liquid refrigerant and the fins. Many attempts have been made to improve this.

  For example, in Patent Document 1, in order to improve the cooling performance, the fin is finely processed by wire cutting or micromachining to form a fine fin having a width of 0.2 mm or less of the flow path between the fins. And its cooling capacity.

  Patent Document 2 proposes a liquid cooling type cooling module that has a bent channel and cools a plurality of LSIs simultaneously.

Patent Document 3 discloses a semiconductor element having a first header that distributes a refrigerant to each of parallel flow paths including a plurality of fine flow paths, and a second header in which the refrigerant that has flowed out of the parallel flow paths merges. Coolers have been proposed.
JP 05-129485 A WO00-16397 JP 2001-35981 A

  In micromachining in which fins are formed at a level where the channel width is 0.2 mm or less disclosed in Patent Document 1, the aspect ratio, which is the ratio of the fin height to the channel width, is limited to about 5. The fin height cannot exceed about 1 mm at the maximum. Therefore, in such a fin, since the cross-sectional area of the flow path is small and the pressure loss becomes too large, the flow rate of the liquid refrigerant is limited, and the cooling performance when combined with an actual pump is about 150 W at the maximum. It is insufficient. In addition, the formation of fine fins causes an increase in manufacturing cost, and also causes a risk of clogging with foreign substances in the liquid refrigerant.

  In the liquid cooling module disclosed in Patent Document 2, the path through which the liquid refrigerant passes between the fins becomes long, and a sharp bend in the bent flow path increases the pressure loss when the liquid refrigerant flows. Further, in the liquid cooling module disclosed in Patent Document 2, it is possible to simultaneously cool a plurality of LSIs, and accordingly, the volume of the cold plate can be increased and the flow path can be bent. it can. However, when paying attention to a single LSI, which is a single heating element itself, it is difficult to apply a liquid cooling unit having a bent flow path and fine fins in terms of production technology. This problem greatly increases costs.

  In the semiconductor device cooler disclosed in Patent Document 3, a study focusing solely on the cooling performance is performed regardless of the size of the cooling device itself, and cooling is performed by reducing the wiring long distance, which is advanced from the request for higher signal transmission speed. It does not provide a means for directly solving the problem that it is difficult to secure a heat radiation area where a module can be mounted on an LSI. Furthermore, the semiconductor element cooler disclosed in Patent Document 3 does not actively contribute to the technical requirement of shortening the wiring length and improving the signal transmission speed.

  The present invention has been made in view of such problems, and the object of the present invention is to shorten the wiring distance so that components can be mounted closer to the LSI, and the signal transmission speed can be reduced. It is in the point which provides the cooling device and the cooling method of the electronic device which can improve this.

In order to solve the above problems, the present invention has the following configuration.
That is, the electronic device cooling device of the present invention is formed by thermally bonding a single liquid cooling unit having a plurality of fins inside one chamber to a single heating element, and The thermal bonding surface for a single liquid cooling unit and the surface defined as the existence region of the plurality of fins have substantially the same shape and the same area.

  The electronic device cooling apparatus according to the present invention further includes a single liquid cooling system having a plurality of fins having a flow path width between adjacent fins of 0.2 mm to 0.8 mm and a height of 2 mm to 5 mm inside one chamber. It is characterized in that the unit is thermally bonded to a single heating element.

  The thermal bonding surface of the single heating element with respect to the single liquid cooling unit and the surface defined as the existence area of the plurality of fins have substantially the same shape and the same area. Also good.

  The electronic device cooling method of the present invention is a method for cooling an electronic device using a liquid cooling unit, and the liquid cooling unit is a single liquid cooling unit having a plurality of fins inside one chamber. The single liquid cooling unit is thermally bonded to a single heating element, the surface defined as the presence area of the single plurality of fins, and the single liquid cooling of the single heating element. The thermal bonding surface for the unit has substantially the same shape and the same area.

  In addition, the electronic device cooling method of the present invention is a method of cooling an electronic device using a liquid cooling unit, and the flow path width between adjacent fins of the liquid cooling unit is 0.2 mm to 0.8 mm. In addition, a single liquid cooling unit having a plurality of fins each having a height of 2 mm to 5 mm within one chamber is thermally bonded to a single heating element. .

[Action]
According to the cooling apparatus and cooling method for an electronic device of the present invention, a single liquid cooling unit is thermally bonded to a single heating element, and a surface defined as a region where a plurality of fins are present and the single cooling unit. Since the thermal bonding surface of the heating element to the single liquid cooling unit has substantially the same shape and the same area, the cooling device can be reduced as much as possible. LSI peripheral components can be mounted nearby, and as a result, the wiring length can be shortened and the signal transmission speed can be improved.

  Moreover, the fin structure in the cooling apparatus and cooling method for an electronic device according to the present invention has a plurality of fins having a flow path width between adjacent fins of 0.2 mm to 0.8 mm and a height of 2 mm to 5 mm inside one chamber. Since the fin pitch and the fin height are not finely processed regions, they can be processed by an easy manufacturing method, that is, an inexpensive cooling module is provided.

  Furthermore, there is provided a cooling structure capable of cooling 200 W or more in an LSI size that does not clog the flow path due to foreign matters contained in the liquid refrigerant, and that significantly exceeds the conventional cooling performance.

According to the electronic apparatus cooling device and cooling method of the present invention, the surface defined as the existence region of the plurality of fins and the thermal bonding surface of the single heating element to the single liquid cooling unit are substantially provided. Since they have the same shape and the same area, the components can be mounted closer to the LSI. As a result, the wiring distance can be shortened and the signal transmission speed can be improved.
Moreover, by making the flow path width between adjacent fins 0.2 mm to 0.8 mm and the height 2 mm to 5 mm, without increasing the pressure loss, without reducing the flow rate of the liquid refrigerant, Since the contact area between the liquid refrigerant and the heat radiating fins can be increased, the cooling performance can be significantly improved as compared with the conventional case.
In particular, by setting the flow path width between fins to 0.2 mm or more, the manufacturing method can be facilitated by pressing, and an inexpensive liquid cooling module can be provided.

Next, the best mode for carrying out the present invention will be described in detail with reference to the drawings.
FIG. 1 is a perspective view of a cold plate 100 according to an embodiment of the present invention. The cold plate 100 includes a chamber 1 having an internal space 2 and a fin structure 3 provided in the internal space 2.

  The chamber 1 includes a chamber bottom plate 11, a flat plate-shaped chamber top lid 12, and a chamber frame 13. The chamber bottom plate 11 closes one opening of the chamber frame 13, and the chamber top lid 12 is a chamber frame. Other openings of the body 13 are closed.

The upstream space 21 in the internal space 2 of the chamber 1 is a space located upstream of the fin structure 3, and the downstream space 22 is a space located downstream of the fin structure 3, and the upstream space 21 and Each of the downstream spaces 22 forms part of the internal space 2.
The chamber top cover 12 is provided with an inlet 18 for allowing the liquid refrigerant to flow in so that the liquid refrigerant can flow from the upper part of the upstream space 21, and the chamber upper cover 12 so that the liquid refrigerant can flow from the upper part of the downstream space 22. Is provided with an outlet 19 for allowing the liquid refrigerant to flow out.

  The inflow port 18 can be attached to any location of the chamber frame 13 that is in contact with the upstream space 21 as long as the liquid refrigerant can flow into the upstream space 21. The outflow port 19 can also be attached to an arbitrary location on the chamber frame 13 in contact with the downstream space 22 as long as the liquid refrigerant can flow out from the downstream space 22.

  FIG. 2 shows an AA cross-sectional view of the cold plate 100 in FIG. For example, an object to be cooled 99 such as an LSI attached to a wiring board (not shown) is thermally bonded to the chamber bottom plate 11 with a heat conductive bonding material 98 such as grease or a heat conductive sheet. The fin structure 3 includes, for example, fins 31 to 33, and a plurality of fins including the fins 31 to 33 are fixed to the chamber bottom plate 11 so as to be parallel to each other.

  FIG. 3 is a plan view showing the internal structure of the cold plate 100 in FIG. The fin 32 and the fin 33 are adjacent to each other with the inter-fin passage 23 interposed therebetween. The fin structure 3 is provided with a plurality of inter-fin channels such as the inter-fin channel 23. These inter-fin flow paths are in parallel with each other.

  The liquid refrigerant that has flowed into the upstream space 21 from the outside of the cold plate 100 through the inlet 18 passes through one of a plurality of inter-fin channels provided in the fin structure 3, for example, the inter-fin channel 23. It flows into the downstream space 22, and then flows out of the cold plate 100 through the outlet 19. Therefore, the heat generated in the cooled object 99 and moved to the fins such as the fins 31 to 33 via the heat conductive bonding material 98 and the chamber bottom 11 is transferred from there to the liquid refrigerant flowing through the inter-fin flow path. The heat is dissipated and is carried out of the cold plate 100 by the liquid refrigerant.

  FIG. 4 shows an example of the structure of the cold plate 100 applied to the cooling apparatus and cooling method for electronic equipment according to the present invention. The chamber bottom plate 11, the chamber top cover 12, and the chamber frame 13 are omitted, and only the fin structure 3 is shown. It is shown. Here, if the direction in which the liquid refrigerant flows is the vertical direction and the direction perpendicular to the direction in which the liquid refrigerant flows is the horizontal length, W is the horizontal length of the fin structure 3 and L is the vertical length of the fin structure. The length, Z, represents the height of the fin structure. Wc represents the width of the channel between fins, and Ww represents the length of the thickness of one fin.

  In the cold plate 100 applied to the cooling apparatus and cooling method for an electronic device according to the present invention, the lateral length W of the fin structure 3 and the vertical length L of the fin structure are defined as a plurality of fin existence regions. The shape W · L and the area W × L obtained by the above are substantially the same shape and the same area as the thermal bonding surface 99a of the single heating element to be cooled, for example, the cold plate 100 of a single LSI 99. It is said. As a result, heat can be transferred from the cold plate 100 to the liquid refrigerant more quickly, and at the same time, space efficiency can be improved.

  When the above cold plate 100 is used, in order to efficiently cool the cooled object 99, if a large number of fine fins are formed, the contact area with the liquid refrigerant flowing in the inter-fin flow path can be increased. In addition, more heat can be efficiently removed, that is, the cooling performance of the cold plate 100 can be improved. However, if the fins are too fine, the flow path between the fins is also narrowed, so that it is difficult for the liquid refrigerant to flow, and as a result, the cooling performance decreases. That is, when optimizing the cooling performance of the cooling module, it is necessary to consider the number and size of fins according to the capacity of the pump.

  From such a viewpoint, in the electronic apparatus cooling apparatus and cooling method of the present invention, the inter-fin channel width Wc is set to 0.2 mm to 0.8 mm. When the fin-to-fin channel width is less than 0.2 mm, the pressure loss increases, and in order to have a cooling capacity of 200 W or more, the current pump is difficult to produce a cooling performance and the manufacturing method is difficult. When the inter-fin channel width is 0.8 mm or more, the surface area of the fins in contact with the liquid refrigerant is small, so that satisfactory cooling performance is not achieved.

  Furthermore, by setting the width of the channel between fins to 0.2 to 0.8 mm, options for the manufacturing method are expanded as compared with the manufacturing of fine fins. For example, press processing, cutting processing, extrusion processing, etc., which are low-cost manufacturing methods that could not be applied to the fin-to-fin flow path of 0.2 mm or less, can be used.

  The fin height is preferably about 5 times to about 25 times the channel width (fin height is 1 mm to 5 mm), that is, the aspect ratio is 5 to 25. When the aspect ratio is 5 or less, a reduction in heat transfer rate due to a reduction in fin surface area and an increase in pressure loss due to a small cross-sectional area of the fin-to-fin flow path lower cooling performance. Also, if the fin height is 25 times or more of the flow path width, even if copper with high thermal conductivity is used as the fin material, the fin efficiency will deteriorate and the increase in surface area will not affect the cooling performance. Since the cross-sectional area between the fins and the flow path between the fins becomes too large, the flow velocity between the fins decreases, and the cooling performance decreases.

  Further, the fin does not necessarily have a plate shape, and the shape of the fin such as a curved shape or a pin-like structure is not limited. In that case, the flow path width can be specified as the width between adjacent fins or pins in the direction perpendicular to the liquid flow direction. The fin height can be specified as the shortest distance between the longest part of the fin or pin and the bottom of the chamber.

  Further, the fin structure 3, the chamber bottom 11, the chamber top lid 12, the chamber frame 13, the inlet 18, the outlet 19 and the like do not necessarily have to be one part, and any combination may be integrally formed. It doesn't matter. For example, it is conceivable to manufacture a part in which the chamber top lid 12, the inlet 18, and the outlet 19 are integrated. Each material is not particularly limited, but considering heat dissipation performance, copper or aluminum that is easy to process with a metal member having high thermal conductivity is preferable.

  By applying a liquid cooling module connected to the cold plate and pump, tank, and heat exchanger of the present invention with a hose so that liquid refrigerant can circulate in electronic equipment, the temperature of LSIs can be lowered and stabilized. It can be operated.

[Example 1]
Embodiments of the electronic device cooling method of the present invention will be described below.
FIG. 4 shows an example of a specific structure of the cold plate of the present invention used in this embodiment. In FIG. 4, the chamber bottom plate 11, the chamber top lid 12, and the chamber frame 13 are omitted, and only the fin structure 3 is shown. Here, if the direction in which the liquid refrigerant flows is the vertical direction and the direction perpendicular to the direction in which the liquid refrigerant flows is the horizontal length, W is the horizontal length of the fin structure 3 and L is the vertical length of the fin structure. The length, Z, represents the height of the fin structure. Wc represents the width of the channel between fins, and Ww represents the length of the thickness of one fin.

  In this embodiment, W: 20 mm, L: 20 mm, Z: 2 mm, Wc: 0.25 mm, and Ww: 0.25 mm. The chamber bottom plate 11, the chamber top lid 12, the chamber frame 13, the fin structure 3, the inlet 18 and the outlet 19 were all made of copper, and the respective members were joined by brazing. In addition, the experiment was performed assuming that water was used as the liquid refrigerant and the heat generating element was a 20 mm square LSI having a size equivalent to the bottom area of the fin structure.

  FIG. 5 shows a graph in which the thermal resistance R [° C./W], which is an index of the cooling performance, is plotted against the flow rate Q [ml / min] of the liquid refrigerant in the cold plate having the fin structure shown in FIG. Even when the flow rate was 200 ml / min and the flow rate was relatively small, the thermal resistance was 0.08 [° C./W]. It was found that as the flow rate was increased, the thermal resistance was saturated from around 1000 ml / min and decreased to about 0.065 [° C./W]. Now, considering that the maximum allowable LSI temperature is 85 ° C. and the refrigerant temperature is 35 ° C., the allowable temperature rise with respect to the LSI refrigerant temperature is 50 ° C. The cooling capacity is 50 [° C.] / (0.08 + 0) even when the flow rate is 200 ml / min, assuming that the thermal resistance of the thermally conductive bonding material between the LSI and the cold plate is 0.1 [° C./W]. .1) [° C./W]=277 [W], which is about 280 W, and can be cooled by 200 W or more.

  Next, when the commercially available pump was used using FIG. 6, the flow volume which flows into the cold plate which has the fin structure shown in FIG. 4 was derived | led-out. As shown in FIG. 6, the intersection of the pressure loss curve (hereinafter referred to as the system impedance curve) generated with respect to the flow rate of the liquid refrigerant in the cold plate of the present invention and the PQ curve indicating the performance of the pump is the actual pump operation. It becomes a point and the flow rate can be derived. Since the pumps have different PQ characteristics depending on the manufacturer, two pumps A and B having different PQ characteristics were selected from among commercially available liquid cooling systems for electronic devices. Pump A is a low discharge pressure but high flow pump, and pump B is a high discharge pressure but low flow pump. On the other hand, for the system impedance of the cold plate of the present invention, the pressure loss with respect to the flow rate was derived by experiments.

  As a result, it was found that when the cold plate having the fin structure shown in FIG. 4 is used, the pump A flows about 800 [ml / min] and the pump B flows about 480 [ml / min]. When the cooling performance is derived from this derived flow rate using FIG. 5, it can be predicted that the pump A can be cooled by 263 W and the pump B can be cooled by 284 W, and a good agreement was confirmed in actual experiments.

[Example 2]
Next, a second embodiment will be described. In the second embodiment, a case in which the flow path width is wider than that of the first embodiment and the fin height is increased is used. W: 20 mm, L: 20 mm, Z: 5 mm, Wc: 0.4 mm, Ww : 0.4 mm. As in the first embodiment, the chamber bottom plate 11, the chamber top lid 12, the chamber frame 13, the fin structure 3, the inflow port 18 and the outflow port 19 are all copper, and the respective members are joined by brazing. In addition, the experiment was performed assuming that water was used as the liquid refrigerant and the heat generating element was a 20 mm square LSI having a size equivalent to the bottom area of the fin structure.

  FIG. 7 shows a graph in which the thermal resistance R [° C./W], which is an index of the cooling performance, is plotted against the flow rate Q [ml / min] of the liquid refrigerant in the cold plate having the fin structure of the second embodiment. When the flow rate was 200 [ml / min] and the flow rate was relatively small, the thermal resistance was 0.15 [° C./W]. When the flow rate was increased to 1200 [ml / min], the thermal resistance decreased to about 0.096 [° C./W]. When the value of this thermal resistance is assumed to be 0.1 [° C./W], the cooling performance is converted to 200 W at 200 [ml / min] and 255 W at 1200 [ml / min]. .

FIG. 8 shows a graph when the flow rate is obtained from the system impedance curve of the cold plate and the PQ curves of the pump A and the pump B in the second embodiment.
From FIG. 8, the flow rates flowing when using pump A and pump B were 930 [ml / min] and 480 [ml / min], respectively. When the cooling performance is derived from the derived flow rate using FIG. 7, it can be predicted that the pump A can be cooled by 250 W and the pump B can be cooled by 227 W, and good agreement was confirmed in an actual experiment.

  In addition to the sizes shown in the first and second embodiments, experiments were conducted using various fin-to-fin channel widths and channel heights as parameters. It was confirmed that the channel width is preferably 0.2 mm to 0.8 mm.

  In addition, it was confirmed that the fin height is preferably about 5 times to about 25 times the fin width (from 1 mm to 5 mm in fin height), that is, the aspect ratio is 5 to 25.

The perspective view which shows 1st Embodiment of the cooling device of this invention. Sectional drawing which shows 1st Embodiment of the cooling device of this invention. The top view which shows 1st Embodiment of the cooling device of this invention. The fin structure of the 1st Embodiment of this invention. Graph of thermal resistance against flow rate in the first embodiment of the present invention The graph which shows the relationship between the pressure loss and flow volume in the 1st Embodiment of this invention Graph of thermal resistance against flow rate in the second embodiment of the present invention The graph which shows the relationship between the pressure loss and flow volume in the 2nd Embodiment of this invention

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Chamber, 2 ... Internal space, 3 ... Fin structure, 11 ... Chamber bottom part, 12 ... Chamber upper cover, 13 ... Chamber frame, 18 ... Inlet, DESCRIPTION OF SYMBOLS 19 ... Outlet, 21 ... Upstream space, 22 ... Downstream space, 23 ... Inter-fin flow path, 31-33 ... Fin, 41 ... Base of heat sink, 98. .. Thermally conductive bonding material, 99... Object to be cooled, 99 a... Surface defined as fin existing area, 100.

Claims (5)

A single liquid cooling unit having a plurality of fins in one chamber is thermally bonded to a single heating element, and the single heating element is thermally bonded to the single liquid cooling unit. And a surface defined as a region where the plurality of fins are provided have substantially the same shape and the same area. A single liquid cooling unit having a plurality of fins having a flow path width between adjacent fins of 0.2 mm to 0.8 mm and a height of 2 mm to 5 mm inside one chamber is thermally formed into a single heating element. A cooling apparatus for electronic equipment, characterized by being joined. 3. The electronic device cooling apparatus according to claim 2, wherein a surface of the single heating element to be bonded to the single liquid cooling unit is substantially the same as a surface defined as the existence area of the plurality of fins. A cooling device for electronic equipment having a shape and the same area. A method of cooling an electronic device using a liquid cooling unit, wherein the liquid cooling unit is a single liquid cooling unit having a plurality of fins inside one chamber, and the single liquid cooling unit is a single unit. The surface defined as the existence area of the single plurality of fins and the thermal bonding surface of the single heating element to the single liquid cooling unit are substantially the same. A method for cooling an electronic device, characterized by having a shape and the same area. A method of cooling an electronic device using a liquid cooling unit, wherein the flow path width between adjacent fins of the liquid cooling unit is 0.2 mm to 0.8 mm, and the height of each fin is 2 mm to 5 mm. A method for cooling an electronic device, comprising: thermally bonding a single liquid cooling unit having a plurality of fins as described above to a single heating element.

Images