Cold plate

Abstract

A cold plate 200 for cooling a circuit board comprises a plurality of cold sections and a channel 202. Each cold section is for cooling a separate electronic component on the circuit board. The channe

Description

Field of the disclosure The present disclosure relates to a cold plate, in particular a cold plate for use with processing devices. Background to the Disclosure Computer processes, in particular artificial intelligence (Al) computations, such as those involving deep learning and large-scale data processing, are typically performed using high-performance processors such as graphical processing units (GPUs) and tensor processing units (TPUs). These components generate a significant amount of heat during operation. Problematically, without proper cooling, the component temperatures inside the racks can rise to levels that can damage the hardware. High temperatures in data centre racks can lead to numerous negative consequences, for example: high temperatures can lead to thermal throttling, where processing units are forced to reduce their performance to prevent overheating; prolonged exposure to high temperatures can shorten the lifespan of electronic components; overheating can lead to system instability, crashes, and data corruption. Therefore, to prevent components from being damaged by this heat, data centres generally contain mechanisms for cooling the components. For example, data centres may comprise air conditioning units that provide chilled air to fans cooling the equipment in the racks. However, existing structures and methods for cooling components are often ineffective or inefficient. Therefore, improved cooling structures and methods are desired. Summary of the Disclosure According to an aspect of the present disclosure, there is described a cold plate for cooling a circuit board comprising a plurality of electronics components, the cold plate comprising: a plurality of cold sections, wherein each cold section is arranged to cool a separate component on the circuit board; and a channel comprising: an inlet; an outlet; a plurality of passages, wherein each passage is associated with a respective one or more cold sections, and wherein the channel comprises a branching channel section located between the inlet and the cold sections and a recombining channel section located between the cold sections and the outlet, the plurality of passages being located between the branching channel section and the recombining channel section. Preferably, the cold plate comprises: a metal base plate; and a metal top plate; wherein the channel is formed between the metal base plate and the metal top plate. Preferably, the cold plate comprises a roll-bonded cold plate. Preferably, the channel comprises a continuous (e.g. unbroken) channel between the inlet and the outlet, preferably wherein the channel is an integral part of the cold plate. Preferably, the inlet and the outlet are provided on a same side and / or a same edge of the cold plate. Preferably, the inlet and the outlet are adjacent, preferably wherein the inlet and the outlet are separated by no more than 250mm, no more than 100mm, and / or no more than 50mm. Preferably, each of the passages is associated with a plurality of cold sections. Preferably, each of the passages is associated with two cold sections. Preferably, the channel comprises at least two passages, at least three passages, and / or four passages. Preferably, each of the passages is shaped so as to provide an equal flow rate through the passages. Preferably, the passages are shaped such that a first passage has a different flow rate to a second passage. Preferably, the passages being shaped comprises the passages being differently sized and / or one or more of the passages comprising an obstruction. Preferably, a first passage with a first flow rate is located adjacent a first cold section and a second passage with a second, different, flow rate is located adjacent a second cold section, the first cold section and the second cold section being provided to cool electronic components with different maximum operating temperatures. Preferably, the inlet and the outlet of the channel are spaced from the cold sections. Preferably, the inlet and the outlet are each spaced from the cold sections by at least 50mm, at least 100mm, and / or at least 250mm. Preferably, the cold plate comprises one or more impingement structures at the one or more cold sections, the impingement structures being arranged to direct a flow of fluid onto a cooling surface of the cold plate, preferably wherein the impingement structures comprise holes. Preferably, the impingement structures comprise one or more hotspot cooling structures, wherein the passage is shaped so as to provide a different flow profile or cooling rate at the hotspot structures (as compared to the other impingement structures). Preferably, the cold plate comprises a pressed part arranged to cap the impingement structures, preferably wherein the pressed part is arranged to contact one or more electronics components of the circuit board. Preferably, the pressed part comprises one or more features for increasing heat transfer from the impingement structures. Preferably, at the location of each cold section the passages branch into a plurality of channel sections. Preferably, the plurality of channel sections comprise one or more inlet channel sections and one or more outlet channel sections. Preferably, the cold plate comprises an arrangement of alternating inlet channel sections and outlet channel sections, the inlet channel sections and the outlet channel sections being connected by holes, preferably being connected by holes of impingement structures. Preferably, the cold plate comprises cut-outs and / or thinned sections at the cold sections of the cold plate. Preferably, the cold plate comprises a thermal spreader. Preferably, the thermal spreader comprises a pressed part that is located adjacent impingement structures of the cold plate. Preferably, the thermal spreader comprises one or more sealed chambers. Preferably, the sealed chambers form heat pipes and / or thermosyphons. Preferably, each sealed chamber comprises an evaporator section and a condenser section. Preferably, the evaporator end and the condenser section are connected by a wick. Preferably, a condenser section of each sealed chamber is located adjacent impingement structures of the cold plate. Preferably, the thermal spreader comprises a clad metal, wherein a first part of the thermal spreader is formed of a first metal and a second part of the thermal spreader is former of a second metal. Preferably, the first metal and the second metal are dissimilar metals. Preferably, the first part is formed of aluminium and the second part is former of copper. Preferably, the first part of the thermal spreader is attached to the cold plate. Preferably, the second part of the thermal spreader is arranged to be located adjacent a component of the circuit board. Preferably, the cold plate comprises a capping sheet arranged to be located adjacent an electronics component being cooled by the cold plate. Preferably, the cold plate comprises a first coolant located between a first part of the thermal spreader and the remainder of the cold plate and a second, different coolant, located between a second part of the thermal spreader and the capping sheet. Preferably, the thermal spreader provides one or more sealed chambers for each of the cold sections of the cold plate. Preferably, the cold plate comprises a pillar, the pillar extending from the thermal spreader. Preferably, the pillar provides a heat pipe and / or a thermosyphon. Optionally, the pillar provides a vapour chamber and / or a solid conductor. Preferably, the pillar comprises a support insert, preferably wherein the support insert comprises an extruded component. Preferably, the cold plate is suitable for and / or arranged for and / or configured for use in an immersion chamber. Preferably, the cold plate is arranged to thermally isolate a first electronics component on the electronics board from a second electronics component on the electronics board. According to another aspect of the present disclosure, there is described a cold plate for cooling a circuit board located in an immersion chamber, the circuit board comprising a plurality of electronics components, wherein the cold plate is arranged to thermally isolate a first electronics component on the electronics board from a second electronics component on the electronics board. Preferably, the immersion chamber comprises a dielectric fluid, preferably a two-phase dielectric fluid. Preferably, wherein the two-phase dielectric fluid is arranged to change phase at a predetermined temperature. Preferably, the cold plate comprises a condenser coil, the condenser coil being arranged to remove heat from the dielectric fluid. Preferably, the cold plate comprises a cavity arranged to be located adjacent the first electronics component. Preferably, the cold plate comprises an inlet and an outlet for circulating coolant through the cavity, preferably wherein a temperature of the coolant at the inlet is lower than a temperature of a dielectric fluid in the immersion chamber. Preferably, the cold plate comprises an insulation chamber for insulating the coolant in the cold plate from the dielectric fluid in the immersion chamber. Preferably, the insulation chamber comprises an insulating material and / or a vacuum. Preferably, the insulation chamber is located on an external side of the cavity. Preferably, the outlet of the cold plate is connected to an inlet of a condenser coil, the condenser coil being arranged to receive heat from the dielectric fluid. Preferably, the outlet is thermally connected to the inlet of the condenser coil and or fluidly connected to the inlet of the condenser coil. Preferably, the cold plate comprises a vapour chamber, wherein an evaporator section of the vapour chamber is arranged to be located adjacent the first electronics section. Preferably, a condenser section of the vapour chamber is arranged to be located adjacent a condenser coil that extracts heat from the immersion chamber. Preferably, the cold plate comprises a wick for transferring a condensed liquid from a condenser section of the vapour chamber to an evaporator section of the vapour chamber. Preferably, the cold plate comprises a central wick, the central wick being encapsulated between two encapsulating sheets of the vapour chamber, the central wick being arranged for transferring a condensed liquid from a condenser section of the vapour chamber to an evaporator section of the vapour chamber. Preferably, the cold plate comprises a cold plate for cooling a data center board in a data center. According to another aspect of the present disclosure, there is described an immersion chamber comprising the aforesaid cold plate. According to another aspect of the present disclosure, there is described a structure comprising: an immersion chamber for containing a dielectric fluid, the dielectric fluid being useable to cool a plurality of components on an electronics board to a first temperature; and a cold plate arranged to thermally isolate one or more electronic components on the electronics board so as to cool these electronic components to a second temperature that is lower than the first temperature. According to another aspect of the present disclosure, there is described a housing comprising one or more cold plates as aforesaid. Preferably, the housing comprises mounting structures for one or more electronics boards. Preferably, the mounting structures are arranged to locate the electronics boards adjacent cold plates. According to another aspect of the present disclosure, there is described a method of manufacturing the aforesaid cold plate. Preferably, the method comprises roll bonding and inflating a sheet to form the channel. Optionally, the method comprises welding and / or brazing a plurality of sheets to form the channel. Preferably, the method comprises cutting material out of, or thinning material of, the sheet at the locations of the cold sections. Preferably, the method comprises forming impingement structures at the cold sections. Preferably, forming the impingement structures comprises punching holes through the sheet. Preferably, the method comprises attaching a pressed plate over the impingement structures. Preferably, the method comprises inflating the channel by providing an inflating fluid through the inlet of the channel. Preferably, the method comprises attaching a thermal spreader onto the cold plate. Preferably, the method comprises welding the thermal spreader onto the cold plate. Preferably, the method comprises forming the thermal spreader by attaching a first sheet of a first material to a second sheet of a second material. Preferably, the method comprises: providing a first sheet and a second sheet; applying a paint to a surface of the first sheet; locating the first sheet adjacent the second sheet with the painted surface of the first sheet being in contact with the second sheet; feeding the first sheet and the second sheet into a rolling machine to bond the first sheet and the second sheet so as to form a thermal spreader; and inflating the thermal spreader so as to form a channel between the first sheet and the second sheet at the locations of the paint. According to another aspect of the present disclosure, there is described a method of operating the aforesaid cold plate, the method comprising providing a flow of coolant through the cold plate. Preferably, the method comprises chilling the coolant prior to providing the coolant to the cold plate. Any feature in one aspect of the disclosure may be applied to other aspects of the invention, in any appropriate combination. In particular, method aspects may be applied to apparatus aspects, and vice versa. Furthermore, features implemented in hardware may be implemented in software, and vice versa. Any reference to software and hardware features herein should be construed accordingly. Any apparatus feature as described herein may also be provided as a method feature, and vice versa. As used herein, means plus function features may be expressed alternatively in terms of their corresponding structure, such as a suitably programmed processor and associated memory. It should also be appreciated that particular combinations of the various features described and defined in any aspects of the disclosure can be implemented and / or supplied and / or used independently. The disclosure also provides a computer program and a computer program product comprising software code adapted, when executed on a data processing apparatus, to perform any of the methods described herein, including any or all of their component steps. The disclosure also provides a computer program and a computer program product comprising software code which, when executed on a data processing apparatus, comprises any of the apparatus features described herein. The disclosure also provides a computer program and a computer program product having an operating system which supports a computer program for carrying out any of the methods described herein and / or for embodying any of the apparatus features described herein. The disclosure also provides a computer readable medium having stored thereon the computer program as aforesaid. The disclosure also provides a signal carrying the computer program as aforesaid, and a method of transmitting such a signal. The disclosure extends to methods and / or apparatus substantially as herein described with reference to the accompanying drawings. The disclosure will now be described, by way of example, with reference to the accompanying drawings. Description of the Drawings Figure 1 shows a data center board comprising a plurality of electronic components. Figures 2a and 2b show a cold plate for cooling a data center board. Figures 3a - 3c show a cold plate that comprises impingement structures. Figures 4a, 4b, and 5a - 5d show methods of manufacturing a cold plate. Figure 6 shows a thermal spreader for use with a cold plate. Figures 7a and 7b show a cold plate comprising the thermal spreader. Figures 8a - 8e show a thermal spreader that comprises a pillar. Figures 9a and 9b show an immersion chamber for cooling a data center board. Figures 10a and 10b show a cold plate for use with an immersion chamber. Figures 11 and 12 show a cold plate for use with an immersion chamber, which cold plate comprises a vapour chamber. Description of the preferred embodiments Referring to Figure 1, there is shown a data center board 100 that may be used within a data center. This board comprises a printed circuit board (PCB) 102 on which are mounted a plurality of electronic components 104. These electronic components, which may for example be processors, GPUs, TPUs, etc. typically generate heat so that it is desirable to cool these components. Certain methods of cooling such electronic components are known in the art. For example, it is known to provide fans in a data centre that provide airflow over electronic components (e.g. so that heat from the components is transferred to air flowing over the components and this heated air is then moved out of the data centre to remove heat from the data centre). Furthermore, direct-to-chip liquid cooling, also known as direct liquid cooling (DLC) or cold plate cooling may be used to cool electronic components. With this method of cooling: a cold plate -which comprises an internal channel forthe flow of liquid - is attached to an electronic component; coolant fluid is supplied to the channel via a piping; and a secondary heat exchanger (e.g. a radiator or a cooling tower) is used to extract heat from the coolant so that the cooled coolant can be recirculated. With such a cold plate, heat is transferred from the electronic components 104 to the coolant, and then from the coolant to the secondary heat exchanger. The secondary heat exchanger can then vent the heat to the environment (e.g. out of the data centre). Typically, the coolant comprises water, a dielectric fluid, and / or a fluid with a high heat capacity. Another method of cooling components is submersion liquid cooling. With this method, the electronic components 104 are submerged into a, e.g. thermally conductive, fluid. This fluid absorbs the heat generated by the electronic components. This heat can then be removed from the fluid via a secondary heat exchanger. The present disclosure provides improvements to conventional methods of cooling electronic components, in particular the present disclosure provides improved structures for providing direct-to-chip liquid cooling and submersion cooling, Direct-to-chip liquid cooling Referring to Figure 2a, there is shown a top view of a cold plate 200 according to an embodiment of the present disclosure. The cold plate, which is typically a metal plate, comprises a sheet that is formed, e.g. deformed, to generate a (typically internal) channel 202. A first end of the channel forms an inlet 204 and a second end of the channel forms an outlet 206. A fluid such as a coolant is thus able to flow through the cold plate via the channel. More specifically, in use, the cold plate 200 is located adjacent to the data center board 100 and then a working fluid (e.g. a coolant) is passed through the channel 202 from the inlet 204 to the outlet 206. This may involve use of a pump in order to drive the fluid through the channel. The working fluid is generally introduced to the cold plate at a temperature significantly below the temperature of the electronic devices 104 that are being cooled so that the coolant absorbs heat generated by these devices as it passes through the data center board. Maintaining a (e.g. constant) flow of coolant therefore allows heat to be removed from the circuit board in order to ensure that the electronic devices remain below a maximum operating temperature. Typically, the channel 202 is formed as a single continuous channel (instead of being formed as a plurality of separate channels that are then joined). In this regard, conventional cold plates typically use systems of pipes that are combined together after a manufacturing stage. In contrast, the channel of the present cold plate 200 is formed as one continuous length of material during the manufacture of the cold plate so that there are no joins between sections of channels or between the cold sections that are formed on the cold plate. Reducing the number of joins reduces the risk of leaks. Typically, the only joins required for the use of the disclosed cold plate 200 are at the inlet 202 and the outlet 204. The present disclosure considers a cold plate that provides an inlet and outlet that can be positioned away from the circuit board so that if any leaks occur from these points, they will not cause damage to the data center board 100. This provides both a decrease in the risk of leaks and also provides a cold plate that is relatively easy to install (as compared to conventional cold plates). Typically, the inlet 204 and the outlet 206 of the cold plate 200 are provided on a same side and / or a same edge of the cold plate. Typically, the inlet and the outlet are adjacent, e.g. they may be separated by no more than 250mm, no more than 100mm, and / or no more than 50mm (depending, e.g., on the size of the equipment that is to be connected to the inlet and outlet). Such an arrangement simplifies any checking for leaks and enables a user to monitor each of the joins in a simple and time-effective manner. Equally, the inlet 204 and the outlet 206 could be located at opposing sides of the cold plate, e.g. to provide an inlet on a first side of a server rack and an outlet on a second (e.g. opposite) side of the server rack. It will be appreciated that various arrangements of the inlet and the outlet are possible and that a particular arrangement is typically selected in dependence on the intended use of the cold plate. Regarding the use of the single inlet and outlet, conventional cold plates are typically arranged to cool a single component. That is, with conventional arrangements, a cold plate would be provided for each electronic component 104 located on the PCB 102. Separate cold plates and cooling channels are then required for each of the components, which increases the number of joins required to provide coolant to the components of the PCB. The present disclosure envisages a cold plate that comprises a plurality of cold sections, e.g. a cold section for each of the electronic components on the PCB 102. As described above, each of these cold sections (and the channel and passages that pass over the various cold sections) are part of a single component so that there are no joins between the cold sections. Therefore, the cold plate is leak-proof as compared to conventional arrangements in which separate cold plates are provided for each component. As will be described in more detail below, the cold plate comprises a single inlet that branches so as to provide a plurality of sub-channels (or ‘passages’) that pass through the plurality of cold sections. These internal channels are then recombined before the single outlet of the cold plate. Such an arrangement provides a leak-proof and cost-efficient cold plate at the cost of requiring a specific arrangement of electronic devices so that the cold sections are located adjacent these electronic devices. In this regard, the cold plate is typically designed for a specific arrangement of electronic devices (so might not be suitable for any other arrangement of devices). Referring to Figure 2b, there is shown a more detailed view of a cold plate according to the present disclosure, which cold plate is located adjacent the data center board 100. As described with reference to Figure 2a, this cold plate comprises a channel 202 with an inlet 204 and an outlet 206. As described above, the channel comprises a plurality of sub-channels (that are referred to as ‘passages’ for the sake of clarity) that pass through separate cold sections of the cold plate. More specifically, as shown in Figure 2b, the inlet leads to a branching channel section that directs the coolant into four separate passages. Each of these passages passes over two separate cold sections (so that the cold plate provides eight cold sections in total) before the passages flow into a recombining channel section that recombines the fluid from each passage prior to the passage of the fluid out of the outlet 206. It will be appreciated that the cold plate shown in Figure 2b that has eight cold sections and two cold sections per passage is exemplary; the cold plate may be provided with any number of cold sections and / or passages. To control the flow rates in the passages, the passages may comprise depressions and / or blockages that increase or decrease the fluid resistance in a given channel. Equally, the fluid resistance may be managed by narrowing or expanding the passages. In general, the skilled person being aware of known flow rate equations would be able to size and shape the passages in order to obtain a desired flow rate in each of the passages. Typically, the passages are formed so as to provide the same flow rate (e.g. same mass flow rate or the same velocity of flow) in each of the channels. In some embodiments, the cold plate comprises a plurality of different passages, which different passages may provide different flow rates (e.g. different mass flow rates or different velocities of flow). For example, a first passage of the cold plate may be formed so as to have an obstruction that slows the flow of fluid through this passage. Therefore, the flow rate of the fluid through this first passage may be lower than the flow rate of the fluid through a second, unobstructed, passage. Such a cold plate may be used to provide different flow rates in each of the cold sections and thus to provide different amounts of cooling to different electronic components. In particular, relatively low (e.g. mass) flow rates of coolant may be provided to components requiring lower power dissipation so as to provide efficient cooling (that ensures device performance without overcooling components with high maximum operating temperatures). The branching channel section, the passages, and the recombining channel section are typically formed as a single channel (e.g. with no joins). Exemplary methods of manufacturing such an arrangement of channels are described below. Typically, the cold plate is formed from sheet metal. Manufacturing the cold plate 200 and providing the channel may comprise use of a roll bonding technique. Equally, manufacturing the cold plate may comprise use of a stamping and pressing technique (where two sheets are stamped to form depressions in these sheets and then the sheets are pressed together so that the depressions form channels). The pressing step may involve sealing the sheets using laser welding, brazing or adhesives. Other manufacturing methods are also possible. Typically, the inlet 204 and the outlet 206 are each arranged to be substantially spaced from the cold sections (e.g. by at least 50mm, at least 100mm, and / or at least 250mm). This may involve the sheet being sized so as to space the inlet and outlet from these sections. Therefore, any leaks - which leaks are most likely to occur at the inlet and the outlet - will be spaced from the electronics components on the data center board 100. Referring to Figures 3a and 3b, there is shown an arrangement of a passage at the location of a cold section 210 of the cold plate. As shown in Figure 3b, each passage (at each cold section) typically comprises a passage inlet 214, a passage outlet 216, and a plurality of (e.g. parallel) channel sections between the passage inlet and the passage outlet so that fluid flows through these channel sections and extracts heat from an electronic component adjacent the cold section. Typically, each channel section comprises a meandering section to maximise the amount of time for which the fluid is adjacent the electronic component. The channel sections may be designed to provide equal flow rates in each channel section. Equally, the passage may comprise differently sized or shaped channel sections that provide different flow rates through the cold section 210. For example, the channel sections may be shaped so as to provide a higher rate of flow across a central portion of the cold section, where this enables the cold plate to target component hot spots of the devices being cooled by the cold plate. Similarly to the cold plate as a whole, each of the cold sections typically comprises a splitting passage section 211 and a recombining passage section 212, where fluid enters the splitting passage section from the passage inlet 214, flows through the channel sections of the passage, and is then recombined in the recombining passage section before exiting through the passage outlet 216. Referring to Figure 3c, in some embodiments the cold plate (e.g. the passages of the cold plate) comprises one or more impingement structures 217 for providing a flow of fluid onto a surface of the cold plate. In particular, the cold plate 200 may be arranged to provide an impinging jet of fluid onto a surface adjacent an electronic component so as to remove an increased amount of heat from that surface. The location of the impingement structures 217 may depend on target hotspots associated with the electronic component. In this regard, the cold plate 200 is typically designed for use with a specific data center board 100, where this board may have known hotspots. By providing suitably located impingement structures, these hotspots can be effectively cooled without (unnecessarily) overcooling other spots on the data center board. Jet impingement improves the rate of heat flux that may be achieved through the cold plate 200, at least because the impinging jet thins (or removes) a thermal boundary layer of fluid at the point of impingement, which a thick thermal boundary layer otherwise acts to insulate the cold plate. As shown in Figures 3a-3c, at each cold section the passage typically comprises a plurality of inlet channel sections (that are connected to the branching passage section 211) and a plurality of outlet channel sections (that feed into the recombining passage section 212). These inlet and outlet channel sections are typically connected via holes (or, e.g. slots), which holes form the impingement structures 217. Specifically, fluid is typically arranged to flow from the branching passage section 211 into the inlet channel sections, to flow through holes (e.g. impingement structures 217) in the inlet channel sections so as to impinge upon a surface 218 adjacent an electronic component, to flow through holes in the outlet channel sections, and then to flow through the outlet channel sections and into the recombining passage section 212. The impingement structures may comprise narrowed holes that result in a high pressure jet of fluid being expelled onto the surface. These holes that form the impingement structures may be formed to be a variety of sizes or shapes (e.g. rectangular, oval, round, slot or freeform). The outlet channel sections help to prevent any jet side wash, where the impingement jets in an array stop other jets from impinging on the surface 218 due to interference between the jets. In this regard, if the passages comprised only a single set of channel sections connecting the branching passage section 211 to the recombining passage section 212, then the impingement jets of the various channel sections would interfere with each other leading to inefficient and / or inconsistent cooling across the cold section. The disclosed arrangement that comprises each of the inlet channel sections and the outlet channel sections avoids this problem. Typically, the inlet channel sections and the outlet channel sections are interlaced / alternating in order to minimise any jet side wash and to ensure a consistent and regular flow through the channel sections of the passages (and thus to provide consistent cooling). In other words, a potential problem with jet impingement is that jets provided in a first area may impede jets in a nearby second area, because the fluid flowing through the jets in the first area may create a high pressure region that prevents fluid from passing through the thermal boundary layers so as to impinge on a surface to be cooled. With the described arrangement of inlet channel sections and outlet channel sections, jet impingement only occurs in alternating sections of the passage, and so impinging jets from an inlet channel section are able to flow through neighbouring outlet channel sections without preventing jet impingement from any other inlet channel sections. Referring to Figures 4a, 4b, and 5a - 5c, there is described a method of forming a cold plate as described in Figures 2a - 3c. This method may be performed by a manufacturing device that is controlled by a computer device. This cold plate is typically customised so as to be suitable for a specific data center board 100 a specific arrangement of components used on that board. The cold plate is typically formed of sheet metal (e.g. aluminium). As shown in Figures 5a and 5b, in a first step 11, a user and / or a computer device (e.g. a device controlling a large press or a roller) roll bonds and inflates a sheet (e.g. a metal sheet) in order to form the channel. Any material not used for cooling can be cut from this inflated sheet using a laser cutter in order to reduce the weight of the cold plate. Roll bonding is a cost-effective manufacturing process, especially for larger format devices. Therefore, the cold plate 200 is typically formed using a roll-bonding process. Equally, as described above, the cold plate could also be formed by soldering, brazing, gluing, or welding two plates together where a suitable manufacturing method may depend on the available equipment and also the number of cold plates being manufactured. The first step 11 may comprise inflating the sheet to form the channels and then cutting the cold plate out from the larger sheet. In some embodiments, the cold plate comprises cut-outs 222 that may be formed using a cutting process that occurs after inflation, to allow each of the cold sections to be flexed so as to conform to the surface of the data center board 100, thereby reducing the need for tight tolerances during manufacture. More generally, the cold plate 200 may comprise cut-outs or thinned sections at the locations of the cold sections in order to enable a user to deform the cold plate at the cold sections so as to ensure a close fit between the cold plate and a board onto which the cold plate is mounted. The use of these cut-outs enables higher tolerance during a manufacture process and increases the efficiency of the cold plate in use by reducing the amount of thermal interface material required to obtain a thermal connection between the cold sections and the components being cooled by the cold sections. Furthermore, the cut-outs 222 allow air to pass through the cold plate 200 so that components that do not require direct liquid cooling but still requiring some degree of cooling may be cooled by air flowing through these cut-outs. Yet further, these cut-outs reduce the overall weight and resource use of the cold plate. As shown in Figure 5c, in a second step 12, the user forms the impingement structures 217, e.g. by cutting holes or jets (or slots etc.) into the inflated sheet using a high-speed CNC, a turret punch, a punch ora laser cutter. The second step may comprise identifying the locations of target hotspots and forming the impingement sections in dependence on said locations. As shown in Figure 5d, in a third step 13, the user may cap the holes that form the impingement structures. For example, the user may cap the holes with a pressed part 224 (that may simultaneously cap each impingement structure 217 of a cold section). This pressed part is typically attached to the cold plate in a watertight manner, e.g. by laser welding the pressed part to the inflated sheet. It will be appreciated that other attachment techniques such as sonic welding, brazing, or adhesives can also be used to attach the pressed part to the inflated sheet. In some embodiments, each cold section is capped using a separate pressed part. In some embodiments, a single pressed part is used to cap each of the plurality of cold sections, where this can further simplify the manufacture of the cold plate. Regarding the roll bonding technique, a more detailed exemplary process for roll bonding and inflating the cold plate is described with reference to Figure 4b. In a first step 21, a paint is applied to a first (typically metal) sheet; for example, a graphite paint is silk screened onto an aluminium sheet. The paint defines areas of the first metal sheet that will not be bonded together in the later steps of the roll bonding method. In a second step 22, a second (typically metal) sheet is placed on top of the first sheet with the painted side of the first sheet being in contact with the second sheet. In a third step 23, the two sheets are fed into a (hot or cold) rolling machine to bond the two sheets into a combined plate (e.g. to form the cold plate 200). Typically, the cold plate is annealed following the rolling process to reduce any stresses that have built up during the rolling process. Equally, the two sheets may be bonded using a static press or another bonding technique such as welding, brazing, gluing, sintering, or soldering. In a fourth step 24, the cold plate 200 is inflated using a high pressure fluid (e.g. a liquid or gas) so that each of the areas that have been painted expand. This provides a plate with each of flat areas and inflated areas / channels / passages (e.g. where the inflated chambers provide the channel through which the coolant flows). Typically, the inlet 204 and / or the outlet 206 is used to provide the inflation point (where this point may then be left unsealed to form the inlet or the outlet following the inflation of the cold plate 200. The paint is applied so as to form a single, unbroken, channel that comprises each of the passages, which channel extends to one or more edges of the cold plate (to provide the inlet 204 and the outlet 206). In some embodiments, a thermal interface material (TIM) may be provided on the cold plate, e.g. on the exterior of the pressed part 224, in order to further improve the transfer of heat from electronic components to the coolant in the passages. The heights of the inflated chambers of the cold plate 200 are typically selected so as to minimise the thickness of this thermal interface material and to contour to features on the adjacent data center board that is to be cooled by the cold plate. In particular, the method of Figure 4b may be performed so as to obtain a plurality of inflated chambers of variable (or different) height (or thickness) so that there is no necessity to compensate for differences in height of electronic components 104 using thermal interface material (in other words, this enables the inflated chambers to be sized so that a surface to be cooled (or any other hardware, forthat matter) can be positioned accurately and in close proximity to a corresponding chamber regardless of the height of that surface). Furthermore, the cold plate is typically inflated so as to obtain inflated chambers with flattened surfaces, to improve a thermal connection between said chambers and components located adjacent said chambers. This may involve the cold plate (and a mould used to form the cold plate) being formed so as to provide inflated chambers of differing heights. In order to accurately control the size and shape of the inflated chambers, two different grades, thicknesses, or types of metal may be used for the first and second metal sheets. This can be used to ensure that the inflation occurs only in one location (e.g. if the first sheet is much thicker than the second sheet, then the first sheet will stay roughly unchanged as the second sheet deforms to form the inflated chambers). Therefore, the first metal sheet may be: thicker, stiffer, and / or harder than the second metal sheet (or vice versa). In some embodiments, chambers of different heights are achieved by modifying (e.g. machining) the first sheet, the second sheet, and / or the cold plate 200 before inflation so as to change the thickness of a section of the first sheet, the second sheet, and / or the cold plate. Benefits of the cold plate 200 and the method of manufacture described above include: 1. Mass Producibility: The described approach is highly mass producible, making it inexpensive to manufacture the cold plate 200 (or, more specifically, to manufacture a large number of cold plates) and thus cost-effective. 2. Customisation: The described approach is highly customisable for a given data center board and its specific electronic components. For example, the impingement structures can be configured to target specific hotspots on components for efficient cooling. 3. Leak-Free Design: As the cold plate is a single part, it is leak-free by its very nature. The inlet 204 and exit 206 ports can be positioned away from the cold sections to ensure no drips on critical components during installation or servicing. 4. Simplified Assembly: Combining the described cold plate (with a plurality of cold sections) with the data center board 100 is simplified as compared to an arrangement in which each electronic component on the data center board 100 is provided with a separate cold plate. 5. Flexibility and Optimal Contact: The use of sheet metal allows the cold plate to flex and bend into place, ensuring optimal contact with the electronic components, eliminating the need for tight manufacturing tolerances, and allowing very thin thermal interface materials to be used. 6. Low Profile Design: The height profile of the cold plate 200 is typically between 3-5mm. This means that if only the main components are liquid-cooled while the rest of the board is air-cooled, the cold sections and any related pipes do not obstruct the airflow to the other elements. In this regard, in some embodiments the data center board 100 comprises different electronic components, where some of these components may be cooled using the cold plate 200 and other components may be cooled by other means (e.g. air cooling). The cold plate may then be provided only the necessary number of cold sections where, for example, if a central component of the data center board is arranged to be air cooled, then this cold section of the cold plate may not be inflated (e.g. no paint may be applied to this (potential) cold section during the first step 21 of the roll bonding process). In other words, the cold plate may comprise a number of (e.g. eight) potential cold sections, but the cold plate may be manufactures so as not to inflate each of these potential cold sections but rather to only inflate the required (e.g. seven) cold sections. Equally, these ‘potential’ cold sections may be cut out (e.g. using a laser cutter) to better enable air to flow around the air-cooled sections. 7. Integrated Design: There is a reduced need for pipes, seals, and connectors as the functions of these components are largely provided by the cold plate, reducing the bill of materials (BOM) required to provide cooling structures and reducing subsequent hardware and assembly costs. Referring to Figure 6, there is shown an embodiment of the cold plate 200, which embodiment comprises a thermal spreader 300, e.g. a two-phase thermal spreader. Figure 6 also shows a cross-section of this cold plate. The thermal spreader 300 is typically a part of (or the whole of) the pressed part 224. In this regard, the pressed part or the thermal spreader may be arranged to efficiently transfer heat between the electronic components and the remainder of the cold plate (e.g. the coolant in the passages) by using the thermal spreader to receive heat from the electronic component and to distribute this heat across a surface of the thermal spreader, which surface is adjacent a cold section of the cold plate. Therefore, the surface of the thermal spreader that is adjacent the cold section, and adjacent the impingement structures 217 provides a surface of consistent heat. In other words, the thermal spreader 300 is arranged to reduce the thermal flux of the component to a level that can be more effectively cooled by jet impingement cooling. This enables the cold plate 200 to effectively cool higher flux electronic components such as high power microprocessors or high power Al GPU chips. The thermal spreader 300 is typically formed from a plurality of different materials where a first part 300a of the thermal spreader may be formed of aluminium and a second part 300b of the thermal spreader may be formed of copper. Typically, the thermal spreader comprises a clad metal, which clad metal is a combination of two or more dissimilar metals that are metallically bonded together into one sheet. The use of a clad metal removes galvanic corrosion problems that are associated with using materials with different galvanic potentials with an electrolyte. With the clad metal thermal spreader 300, e.g. water can be used on the second, copper, part 300b of the thermal spreader while e.g. an aluminium compatible coolant (e.g. a coolant that does not react with aluminium to cause, e.g. corrosion or a degradation in performance) can be used on the first, aluminium, part 300a of the thermal spreader. The second material, e.g. the copper, may be removed around an edge 302 of the thermal spreader so that the thermal spreader can be welded to the cold plate 200 at this edge without generating potential galvanic corrosion issues. In this regard, the cold plate and the first part of the thermal spreader are typically formed of the same material, e.g. aluminium, and as shown in Figure 7a the thermal spreader is typically attached to, e.g. welded to, the cold plate at an edge of the thermal spreader that is formed of this material. In some embodiments, a first working fluid (e.g. an aluminium compatible coolant) is located between the (e.g. first part 300a of the) thermal spreader 300 and the cold plate 200 and a second working fluid (e.g. water) is located between the (e.g. second part 300b of the) thermal spreader and a capping sheet 308. Typically, the thermal spreader comprises a plurality of sealed chambers 306, where the thermal spreader 300 may comprise one or more separate chambers for each of the cold sections of the cold plate 200. These separate chambers may each be capped with a, e.g. metal, sheet 308, which capping sheet 308 may be made of the same material as the second part 300b of the thermal spreader, e.g. copper, and which capping sheet may be attached to, e.g. welded to, the second part 300b of the thermal spreader. The capping sheet is arranged to contact an electronic component that is being cooled by the sealed chamber associated with the capping sheet (and by the cold section that is associated with the sealed chamber). Typically, the sealed chamber 306 forms an evaporator section (e.g. an evaporator section of a two-phase chamber) near this surface of the capping sheet 308, with a condenser section of the sealed chamber being formed at an opposite side of the sealed chamber, this opposite side being near a surface of the second part 300b of the thermal spreader 300. Heat from the electronic component is then received at the capping sheet 308 and this heat transfers, via the capping sheet, into the evaporator section of the sealed chamber 306 so as to heat a fluid that is present in the evaporator section. This fluid is heated and evaporated so that the vapour moves into the condenser section of the sealed chamber that is adjacent the second part 300b of the thermal spreader 300. The vapour then transfers heat to the second part of the thermal spreader and, in the process of transferring this heat, condenses and moves back to the evaporator section of the sealed chamber. The heat then transfers to the first part 300a of the thermal spreader via the second part and transfers into the coolant in the passages of the cold plate due to the impingement of the coolant onto the cold sections of this cold plate. Typically, the sealed chambers 306 comprise a wicking structure that encourages the transfer of a working fluid within the sealed chambers from the condenser section back to the evaporator section. The wicking structure may, for example, comprise grooves in the surface of the sealed chambers, a sintered metal powder, ora wire mesh. The wicking structure enables a condensed fluid to move, e.g. under capillary action, from the condenser section to the evaporator section. In some embodiments, the sealed chambers 306 comprise thermosyphons, where the working fluid is arranged to move from the condenser section to the evaporator section under the force of gravity. A hybrid configuration is also possible, where a combination of gravity and wicking structures are used to return the condensed fluid to the evaporator section. Referring to Figures 7a and 7b, there is shown a thermal spreader 300 that has been attached to (e.g. welded to) the cold plate 200 (so as to become a part of the cold plate). These figures show, for the purposes of illustration, a single sealed chamber 306 that is located adjacent a cold section 210 of the cold plate. Typically, the cold plate comprises a plurality of cold sections and the thermal spreader comprises a plurality of sealed chambers with each sealed chamber being located adjacent a respective cold section of the cold plate. Equally, the thermal spreader may comprise a plurality of sets of sealed chambers, with each cold section being located adjacent a respective set of sealed chambers. It will be appreciated that any number and combination of sealed chambers, impingement structures, cold sections, etc. may be provided. The attaching of the thermal spreader onto the cold plate may comprise laser welding the cold plate onto the thermal spreader so as to form a seal between the cold plate and the thermal spreader. It will be appreciated that other attachment techniques are possible, such as adhesives, soldering, orbrazing. Figure 7a shows the impingement structures 217 that are present in the cold section 210 of the cold plate 200, these impingement structures being located adjacent the condenser section of the sealed chamber 306 of the thermal spreader 300 (and the first part 300a of the thermal spreader). Thermal energy is removed from the thermal spreader by the impingement of fluid onto this first part of the thermal spreader and is then absorbed into the coolant that is flowing through the cold plate. Figure 7b shows the sealed chambers 306 of the thermal spreader 300, which sealed chambers are arranged to aid the transfer of heat from an electronic component to the associated cold section 210. Referring to Figures 8a and 8b, there is shown an embodiment of the thermal spreader 300 that comprises a pillar 310 that is integrated into the thermal spreader, the pillar comprising an elongated sealed chamber (or, optionally, a solid conductor). The pillar is arranged to transfer heat from a spaced component into the cold plate 200 so as to enable, for example, the cold plate 200 to be moved towards the top of a rack (e.g. a server rack) to avoid clashes with other components on the data center board 100 and / or clashes with other components located on the same rack (e.g. on lower tiers of the rack). The pillar may, for example, extend at least 5mm, at least 10mm, at least 20mm, and / or at least 50mm from the remainder of the thermal spreader 300 and / or from the cold plate 200. Referring to Figure 8b, the pillar 310 comprises a pillar cavity 312 that allows vapor to move up the pillar and to spread across the thermal spreader 300 and the base of the cold plate 200. In this regard, the pillar is typically arranged to extend from a first portion of the thermal spreader and the pillar cavity is typically arranged to extend across a second portion of the thermal spreader, the second portion being substantially larger than the first portion. Therefore, the pillar may receive heat from a small area of an electronic component and may spread this heat across a larger area of the cold section via the pillar cavity. The pillar 310 comprises an evaporator section 316 that is arranged to receive heat from an electronic component located adjacent (e.g. in contact with) the pillar and / or the evaporator section. A fluid in the evaporator section (e.g. a fluid that is absorbed into a first section 311 of one or more wicks of the pillar) is heated so that this fluid evaporates and the vapour then flows towards a condenser section of the pillar via a cavity 312 of the pillar. This cavity of the pillar is connected to (and / or a part of) a larger vapour cavity of a sealed chamber of the thermal spreader and so the vapour spreads from this cavity across the condenser section of the sealed chamber. This heat present in the vapour is then transferred to the cold section 210 of the cold plate 200 as described above, which transfer of heat causes the vapour in the cavity to condense. This condensed vapour (fluid) is then absorbed by a second section 313 of the one or more wicks of the cavity 312. Through capillary action, the liquid in the wick travels back to the evaporator section of the pillar and into the first sections 311 of the wicks, thus completing the heat transfer loop. To prevent the pillar 310 from collapsing, a support insert 315 may be placed inside the pillar. This support insert may be extruded or may be machined, and the support insert may be formed from various materials, including copper, plastic, and aluminium. The support insert could be a part of the walls of the pillar so that the support insert provides - in a single component - the internal structure and the walls of the pillar. The void volume within the cavity 312 of the pillar may be reduced due to the presence of the support insert. In some embodiments, the support insert 315 is arranged to guide the wick 311, 313 down the walls 314 of the pillar 310. In some embodiments, these walls 314 are separate to the support insert; in some embodiments, the walls are an integrated part of the support insert 315. Referring to Figure 8c, there is shown another embodiment of a pillar that may be provided on the cold plate 200 and / or the thermal spreader 300. As with the embodiment of Figure 8b, the pillar of Figure 8c comprises an evaporator section 316, wherein heat from a component to be cooled conducts into the evaporator section so as to evaporate a working fluid saturated in an evaporator wick 319 of the evaporator section of the pillar. As shown in Section B-B of Figure 8c, the pillar of Figure 8c comprises urging structures 323 that help to push the evaporator wick 319 against the evaporator section 316 so as to improve the transfer of heat into the evaporator wick. Similarly to the pillar of Figure 8b, the evaporated fluid flows into a cavity 312 of the pillar, travels up the cavity, and then spreads across the thermal spreader condenser section 300. In Figure 8c, the cavity in the pillar section 317 is shown as a single slot shape, but it will be appreciated that the pillar may comprise multiple chambers with any cross-sectional geometry (e.g., circular, square, triangular, etc.). Similar to the embodiment of Figure 8b, the vapour then condenses in a condenser end of the pillar and the condensed liquid is absorbed into a condenser wick 321 at this condenser end. To prevent vapor that is traveling up through the pillar from impinging on this condenser wick, the pillar comprises a vapour deflector 322, where this vapor deflector may be formed from a cut and bent section of a capping sheet 308 of the pillar. This vapor deflector prevents the travelling vapor from impinging and entraining condensed liquid that is moving from the condenser end to the evaporator end of the pillar. To join the evaporatorwick319 to the condenser wick so as to enable this movement of condensed liquid, a chamber 318 of the pillar is typically filled with a wicking material, such as wire wool, small particles, or another porous structure that enables capillary wicking of the working fluid from the condenser wick to the evaporator wick. In some embodiments, a wire mesh or a cloth 320 is placed at the top of the chamber 318 to prevent any material filling the chamber from escaping from this chamber. In particular, the chamber may be filled with a resilient material that has a spring-like quality, where this resilient material acts to push the cloth 320 into the condenser wick 321 so that this resilient material acts like a cushion. This arrangement provides a constant and consistent contact between the evaporator wick 319 and the condenser wick so that the working fluid can move seamlessly between these two wick sections. The returned working fluid can then be re-evaporated at the evaporator end 316 of the pillar so as to provide a fluid loop. Referring to Figure 8d, there is shown another embodiment of a pillar. The embodiment of Figure 8d comprises only a single wick 324, which wick comprises cuts at a first locations 324d and which wick comprises a bend at a second section 324c. This wick is attached to a condenser surface 300 at the condenser end of the pillar, e.g. using welded tacks. In a separate subassembly, a pillar section 317 of the pillar is attached to the capping sheet 308, which capping sheet is then attached to the thermal spreader 300 so as to form the cavity 312 of the pillar. In this embodiment, the wick 324 is threaded or posted through cavity 318 in the pillar 317. At this stage, the evaporator end 316 of the pillar is secured to the (single) wick. The wick is then simultaneously folded at third and fourth locations 324b, 324a so that the evaporator end of can be attached to the remainder of the pillar 317. For example, the evaporator end may be attached by welding, brazing, adhesives, etc. to form a seal e.g. an airtight, watertight, or hermetic seal. In some embodiments, the cavity 318 of the pillar is filled with a wicking material such as wire wool, small particles, or a porous structure that promotes capillary wicking, in order to aid the wicking of a fluid in the pillar and to help to bias a flow of vapour into the cavity 312 of the pillar. Referring to Figure 8e, there is shown a further embodiment of the pillar. In this embodiment, the pillar 310 comprises integrated bolt guides 325 that are arranged to receive bolts 327. More specifically, the pillar is arranged such that the bolts 327 can pass through the thermal spreader 300, then the capping sheet 308, and then into the bolt guides 325, ultimately connecting to receiving nuts or threaded sections positioned on the far side of the PCB 102 at a receiving location 329. The cold section 210 (which is not shown in Figure 8e) can then be attached above the thermal spreader 300 as has been described above. With the embodiment of Figure 8e, the screwing in of the bolts 827 causes the cold section 210 to flex until the bolt guide end stops 330 make contact with the PCB 102. This contact sets the relative positioning for the connection between the component 104 and the pillar 310, this positioning being determined by the distance between the end stops 330 and the top of pillar evaporator cap at the evaporator end 316 of the pillar. This arrangement allows for very tight tolerances between the component 104 and the top of the pillar, enabling the use of very thin thermal interface materials and thus reducing the thermal resistance associated with the cold plate. Flexing the cold plate 200 to conform to tight tolerances around the components ensures optimal contact with the component to be cooled, which is important because thick thermal interface materials can be detrimental to heat transfer, especially for the high-power devices that are often found in data centres. This approach offers performance comparable to individual cold plates that can move independently of each other. Furthermore, the use of the pillar enables accurate positioning of the cold plate relative to other hardware (e.g. to enable the cold plate to be used within various server racks with different structures). In general, the use of the pillar enables the cold plate to be used in a variety of contexts and with a variety of different structures while accurately positioning one or more cold sections of the cold plate (and / or one or more pillars) adjacent one or more components to be cooled thereby ensuring high performance and high cooling. It will be appreciated that any features described with reference to Figures 8a - 8e may be provided in any combination. For example, the arrangement of bolts 827 described with reference to Figure 8e may be provided with the pillar of any of Figures 8b, 8c, and 8d.The thermal spreader is typically formed using a roll bonding process, as has been described with reference to Figure 4b. In contrast to the cold plate 200, following the fourth step 24 of the method of Figure 4b, the individual sealed chambers of the thermal spreader are typically sealed off from an inflation line used to inflate the thermal spreader so as to form individual sealed chambers on the same thermal spreader. In this regard, the paint is typically applied so as to form connected areas and to provide an inflation channel to an edge of the thermal spreader. This enables an inflating fluid to be provided via the inflation channel so as to inflate each of the painted areas. Following this process, separate inflated chambers may then be formed by sealing the channel and by sealing connections between various inflated areas from each other (e.g. by deforming the thermal spreader to close any channels connecting adjacent inflated chambers). Such a method of forming and separating inflated chambers enables each sealed chamber to be inflated from a single inflation point before these chambers are separated. Where the process of Figure 4b is used to form the thermal spreader 300, in a fifth step a (e.g. small amount of) working fluid is typically provided to the sealed chambers of the thermal spreader. The method may also comprise, prior to the provision of this working fluid, creating a vacuum in the chambers to remove undesirable gasses and contaminants from these chambers. Thereafter, the working fluid is transferred into the chambers, and then the chambers are sealed. In some embodiments, wicks are added to the chambers; for example, by inserting scrolled up fine metal meshes (that act as wicks) through holes that have been drilled into the thermal spreader. Equally, metal powders may be inserted into these holes. Where a metal powder is used, a vibration plate may then be used to level the powder, with the thermal spreader then being placed in a high temperature oven to sinter the metal powder so as to form a sintered wick. Equally, the powdered metal could be pressed externally to fix it in place. More generally, in some embodiments, the paint and inflation process is implemented so as to provide heat pipes within the thermal spreader, which heat pipes comprise wicking sections. In some embodiments, in the fifth step, a thermally conductive material is added to the chambers to promote higher heat transfer rates compared to the base materials; for example, materials such as graphite may be injected into the chambers. In some embodiments, phase change materials are inserted into the chambers to absorb heat and to manage thermal power peaks. Typically, the above-described method is used to provide a thermal spreader 300 comprising a plurality of thermosyphons, vapour chambers, or heat pipes - and this method enables a quick, effective, efficient, and cheap method of providing a thermal spreader that can efficiently transfer heat between regions of the thermal spreader. A thermal spreader that is suitable for use with the cold plate 200, and methods and systems for forming such a thermal spreader are further described in the application GB2311882.1 that has previously been filed by the Applicant. This document is hereby incorporated by reference in its totality. Immersion cooling Referring to Figure 9a, there is shown an alternative structure for cooling a component, such as the data center board 100. This structure 400 comprises an immersion chamber that is arranged to receive a dielectric fluid (that is thermally conductive, but electrically non-conductive) and a structure for holding the printed circuit board in the immersion chamber. Typically, the dielectric fluid is a fluid with a high heat capacity; for example, the dielectric fluid may comprise mineral oil hydrocarbons, synthetic fluorocarbons, and / or silicone fluids. In use, the data center board 100 is inserted into the immersion chamber, along with the dielectric fluid. The dielectric fluid fully or partially submerges the electronic components of the data center board so that these components are able to transfer heat to the dielectric fluid. A secondary heat exchanger, e.g. located adjacent the walls of the chamber, is then able to remove the heat from the dielectric fluid. Referring to Figure 9b, there is shown an exemplary immersion chamber 400 for providing two-phase immersion cooling. This chamber is arranged to cool an exemplary electronics board 500 (e.g. the data center board 100) that comprises a first electronics component 501 and a second electronics component 502. The chamber is arranged to hold a dielectric fluid 404 into which the electronics board 500 can be placed. With this two-phase chamber, the dielectric fluid is selected so as to change phase at the temperature of a component with a lowest maximum operating temperature. When this phase change occurs, vapour bubbles rise to the surface of the dielectric fluid. The immersion chamber 400 further comprises a condenser coil 401 that has a condenser inlet 402 and a condenser outlet 403. The condenser coil is placed above (or, optionally, in contact with) the dielectric fluid 404 so that the vapour bubbles formed due to the heating of the dielectric fluid condense on the condenser coil and transfer heat to the condenser coil. Typically, a condenser fluid, e.g. a chilled coolant, is provided at the condenser inlet so that this condenser fluid flows from the condenser inlet to the condenser outlet and receives heat from the dielectric fluid during the movement from inlet to outlet. Typically, the exit temperature ofthe condenser coil fluid is close to the bulk temperature of the immersion fluid, while the inlet temperature is significantly lower than the bulk temperature ofthe immersion fluid. It will be appreciated that the condenser coil is simply an example of a secondary heat exchanger that removes heat from the immersion chamber and that different secondary heat exchangers may be used (and references below to the condenser coil should be more broadly interpreted as references to a secondary heat exchanger). While some embodiments of the immersion chamber 400 use a two-phase dielectric fluid that vaporises at a maximum operating temperature, it will be appreciated that other methods of maintaining a maximum temperature within the immersion chamber are possible (e.g. using a single phase dielectric fluid). As has been described above, the dielectric fluid is selected so as to change phase at the temperature of a component with a lowest maximum operating temperature - or, more generally, so as to maintain a temperature beneath a maximum operating temperature. More generally, the dielectric fluid is selected so as to maintain a temperature at or below a maximum operating temperature (which may be held beneath the lowest maximum operating temperature of a component on the electronics board 500 by a safety factor). However, this can lead to overcooling of certain electronics components on the electronics board. Considering an example where the first electronics component 501 is a device with a maximum temperature of 80°C (e.g. where the first electronics component is an optical component) and the second electronics component 502 is a device with a maximum temperature of 95°C (e.g. an FPGA type chip), the immersion chamber 400 may need to be kept at 80°C even though this results in a substantial overcooling of the second electronics component. Such an immersion chamber will require a larger cooling energy input than is strictly necessary to maintain each chip at the maximum operating temperature of that chip, since a large mass of bulk fluid needs to be cooled to the lower max operating temperature. Typically, the immersion chamber 400 is arranged to hold a plurality of electronics boards (which may have different electronics components). The aforementioned overcooling issue may then be exacerbated if a single component on a single board is determining the max bulk temperature for the immersive fluid 404 and for the inlet condenser fluid. The present disclosure provides structures and methods for reducing the energy loss due to overcooling. In particular, the present disclosure provides an immersion chamber in which an electronics component with a low maximum operating temperature may be thermally isolated from other electronics components in the immersion chamber enabling a bulk temperature of an immersion fluid to be higher than this low maximum operating temperature. This approach not only provides a more efficient method of cooling, directly consuming less energy, but it also leads to an increase in the temperature of the condenser fluid at the condenser outlet 403, which enables an increase in a usefulness of energy that can be recovered from this condenser fluid (e.g. the recovered fluid may be useable for heating fluid in a separate heating system of a building). Figure 10a shows an embodiment of a cold plate 600 that is used to cool the second electronics component. The cold plate of Figure 10a may comprise any of the features of the cold plates described with reference to the previous figures and vice versa (e.g. the cold plate 600 of figure 10 may be formed to comprise sealed chambers that comprise wicking elements). The cold plate 600 comprises a cavity 602 that is arranged to receive a cooling fluid from a cold plate inlet 605 and to output the cooling fluid from a cold plate outlet 606. This flow of fluid may, for example, be driven by a pump. The cooling fluid is provided and / or maintained at a temperature that is lower than a temperature of the dielectric fluid 404 in the immersion chamber 400. The cold plate is arranged to be attached to (e.g. mounted on) the first electronics component 501 such that the cavity 602 of the cold plate is located adjacent the first electronics component. Therefore, the cooling fluid that is flowing through the cavity cools the first electronics component. Since this cooling fluid is provided at a lower temperature than the bulk temperature of the dielectric fluid, the cold plate is arranged to provide additional cooling to the first electronics component. Therefore, the first electronics component can be kept at a lower temperature than the second electronics component 502 to ensure that each component is kept below a maximum operating temperature without overcooling the second electronics component. Typically, the cavity 602 is surrounded by an insulation chamber 601 that is arranged to insulate the cooling fluid in the cold plate 600 from the immersion fluid in the immersion chamber 400 (and to prevent the heating of this cooling fluid. In various embodiments, the insulation chamber comprise a vacuum and / or an insulating material. In some embodiments, the cold plate outlet 605 is connected to the condenser inlet 402 and the coolant fluid is the same as the condenser fluid. Therefore, this cooling fluid is able to flow through the cavity 602 so as to specifically cool the first electronics chip 501 before flowing through the condenser coil 401 to cool the dielectric fluid 404 in the immersion chamber 400. Equally the cold plate cooling fluid and the condenser fluid may be separated so that separate cooling loops are used to cool the cold plate and to cool the dielectric fluid. The cold plate outlet 605 may be fluidly connected to the condenser inlet 402 so that coolant flowing out of the cold plate outlet flows into the condenser inlet. Equally, the cold plate outlet may be thermally connected to the condenser inlet so that heat flows from the cold plate outlet to the condenser inlet (even through the cold plate and the outlet may be separated so as to use different and separated flows of coolant. The cold plate 600 typically comprises sheet metal and the cold plate may be formed using a roll bonding process (as described with reference to Figures 4a-5d). An exemplary mould that may be used within such a roll bonding process is shown in Figure 10b, which figure shows a first, male, part 608 of a mould and a second, female, part of the mould. A sheet of metal is located within this mould before being inflated to form the cold plate 600. Typically, this comprises forming two symmetrical parts of the cold plate using the mould and then connecting, e.g. laser welded the parts of the cold plate together at a seam. It will be appreciated that other methods of forming, and / or joining sheets of, the cold plate are possible for example, the two parts of the cold plate could be bonded using adhesives, soldering, or brazing. By using a roll-bonding process and providing a roll-bonded cavity 602, a cavity can be provided that has an insulation chamber 601 which covers only a first side of the cavity (e.g. where this insulation chamber does not cover a heat transfer surface 604 of the cavity). This enables a heat transfer surface to be located directly adjacent the first electronics component 501 so as to be cooled by the cold plate (e.g. without an insulation chamber in between the heat transfer surface and the first electronics cavity). This ensures efficient heat transfer and thus efficient cooling of a target component. Providing such a cavity 602, and such a cold plate 600, may require the use of a plurality of moulds, where a first mould is used to form the exterior of the cold plate, this exterior having an insulation chamber that extends around the entirety of the exterior and a second mould is used to form the interior of the cold plate and the heat transfer surface 604, where the second mould may provide a flat surface (e.g. without an insulation chamber) at the location of the heat transfer surface. The described cold plate is useable to cool the first electronics component 501 to a lower temperature than the second electronics component 502. More generally, the cold plate may be arranged to thermally isolate, and to cool, one or more of, or a plurality of, electronics components on a data board so that these isolated components are at a lower temperature than one or more non-isolated components. The insulating chamber 601 that prevents the transfer of heat from the immersion fluid may be provided for other components. For example, the insulating chamber may be integrated into the cold plate 200 of Figures 2b - 8e so as to provide thermal segregation between different fluids (e.g. so as to provide thermally segregated sections on the cold plate). The disclosure further extends to a structure that comprises an immersion chamber used to cool a plurality of components on an electronics board to a first temperature, wherein the structure further comprises a cold plate that is arranged to thermally isolate one or more components on the electronics board and to cool these components to a second temperature that is lower than the first temperature. Figure 11 illustrates an embodiment of a cold plate that comprises an insulated vapour chamber 700 (which vapour chamber may be structurally similar to the sealed chamber 306 of the thermal spreader 300). The vapour chamber is arranged to be thermally isolated from the dielectric fluid 404 so that the vapour chamber can be located in the dielectric fluid while operating at a lower temperature than the dielectric fluid so as to maintain the first electronics component 501 at a temperature that is lower than a bulk temperature of the dielectric fluid. The vapour chamber may, for example, be a part of the cavity 602 and / or may replace the cavity. Beneficially, the vapour chamber can provide a sealed structure for removing heat from the first electronics component at a first end of the vapour chamber (where heat is then removed from the vapour chamber at a second end of the vapour chamber). Typically, the thermal conductivity of the vapour core of a two-phase vapour chamber device is low, making it an effective insulator. Therefore, the vapour chamber may be located within the dielectric fluid without transferring heat from the dielectric fluid to the first electronics component 501. The vapour chamber may also be surrounded by an insulation chamber as described with reference to Figure 10a. The insulated vapour chamber 700 is arranged to be attached to the first electronics component 501 at or near an evaporator section 701 of the vapour chamber. Typically, the attachment point comprises a thermal transfer surface with a high conductivity to enhance the transfer of heat from the first electronics component to the evaporator section of the vapour chamber. Therefore, the heat from the first electronics component conducts through the walls of the vapour chamber and into a wick 703 that is attached to the walls of the vapour chamber. This heat causes the evaporation of a working fluid that saturates the wick so that this working fluid changes phase and evaporates. This phase change increases the pressure in the vicinity of the wick (since as vapor is less dense than liquid) so that the vapour is pushed towards a condenser section 702 of the vapour chamber due to the pressure difference in an evacuated outer chamber 705 of the vapour chamber. Once the vapour reaches the condenser end 702 of the vapour chamber, the heat in the vapour is transferred into a further structure, e.g. the environment. In some embodiments, the condenser end is located adjacent, or is thermally connected to, the condenser coil 401 near the inlet 402 of the condenser coil, where the temperature of the condenser fluid is lowest. Therefore, the condenser end transfers heat to the condenser fluid. It will be appreciated that numerous structures may be used to receive heat from the condenser end of the vapour chamber. This transfer of heat out of the condenser end of the vapour chamber causes the vapour to condense to its liquid state and soak into the wick near a condenser end 702 of the wick. This change of phase results in a pressure drop so that the condensed liquid is urged back towards the evaporator end of the wick 703. This pressure drop drives the vapour from the evaporator section 701 of the vapour chamber to the condenser section 702. In the evaporator section the pressure increases and at the condenser section the pressure decreases. This pressure difference pushes the vapour from the evaporator section to the condenser section (e.g. at supersonic speeds). The condensed fluid is then returned to the evaporator section via the capillary action ofthe wick(s) of the vapour chamber. Equally, in some embodiments, the vapour chamber comprises a thermosyphon (where a thermosyphon does not contain a wick, but instead provides the return ofthe condensed fluid to the evaporator section using gravity). Typically, the vapour chamber comprises a plurality of wicks, e.g. a condenser wick, an evaporator wick, a centre core wick, and / or a plurality of outer wicks. In some embodiments, the vapour chamber comprises dimples 704, 711 in the outer skin 712 ofthe vapour chamber, which dimples connect one or more outer wicks 703, 706, to one or more centre core wicks 707 to enable the flow of fluid between these various wicks. Typically the centre core wick(s) is encapsulated between two encapsulating sheets 708 of metal such that this centre core wick transfers condensed liquid from the condenser end 702 back to the evaporator end 701, completing a heat transfer loop. The encapsulating sheets 708 ensure that the liquid that is absorbed into the centre core wick 707 is contained within the encapsulating sheets and so these sheets prevent the entrainment of the condensed liquid into the evacuated outer chamber 705 of the vapour chamber. This encapsulating ofthe fluid between the encapsulating sheets helps to reduce the transfer of heat from the dielectric fluid into the working fluid ofthe vapour chamber by ensuring that no working fluid (that might receive heat) is present in the cavity ofthe vapour chamber. It will be appreciated that while typically beneficial, the encapsulating sheets are an optional component ofthe vapour chamber. Methods for forming the insulated chamber have been described above with reference to the sealed chamber306 ofthe thermal spreader 300 (and, e.g. with reference to Figure 4b). In some embodiments, the vapour chamber 700 is formed by pressing and cutting outer surfaces 712, 713 from sheet metal. For example, a turret punch may be used to cut and shape the sheet metal, or other pressing and punching methods may be used to shape this metal. Once the sheet metal has been shaped, outer wicks 703, 706 that connect the evaporator end 701 and the condenser end 702 ofthe vapour chamber are thermally and mechanically attached (e.g. tacked into place); these wicks may, for example, comprise layers of wire mesh, sintered powder, or grooves. Then, a number of wick layers are sandwiched between two thin encapsulating sheets 708 so as to form the centre core wicks 707. Slots 709 may then be cut through the encapsulating sheets 708 and the centre core wicks to insulate the liquid core of the vapour chamber from any heat that is conducted from the dielectric fluid 404 in the immersion tank 400. Edges 714 of these slots can be welded or otherwise sealed to prevent the entrainment of any liquid into the vapor cavity 705 (from the centre core wicks), and this operation may also prevent the centre core wicks from unravelling or otherwise separating. Finally, the centre core wick is sandwiched between the outer surfaces and a perimeter weld 710 (or other sealing method) is used to fully seals the vapour chamber. At this stage, a vacuum may be pulled on the device, a small known volume of fluid injected via a filling hole, and the device then sealed. While this vapour chamber is described with reference to cooling a single component, it will be appreciated that a single vapour chamber may be used to cool multiple components on an electronics board. In general, any number of vapour chambers may be provided to cool any number of components. Referring to Figure 12, there is shown a further embodiment of the vapour chamber 700of the cold plate. In this embodiment the vapour chamber comprises embossed sections 717, 718 that are located in in the outer surfaces 712, 713 of the vapour chamber. These embossed sections allow the bending of the vapour chamber without collapsing this chamber (and it will be appreciated that these embossed sections may be provided in a variety of shapes and sizes to achieve this effect). With the vapour chamber of Figure 12, the wick section 707 is typically located in the centre axis of a bending section of the vapour chamber so that the wick section does not need to be stretched or compressed as the vapour chamber is bent like it would if it was attached to the outer walls 712 and 713. More specifically, the vapour chamber may comprise a central section 717, where a first and second outer section of the vapour chamber can be bent around this central section to provide a bent vapour chamber. The wick section may be located in this central section so as to be unaffected by the bending. While the embodiment of Figure 12 shows one bending point (and one central section 717), it will be appreciated that a plurality of bend points could be implemented to allow the vapour chamber to conform to a specific application. These bending points can enable the thermal coupling of fixed infrastructure pieces together (e.g. the coupling of a board component in a server rack and a condenser coil). Integration into the vapour chamber of fixing points similar to those shown in figure 8e may also be performed with the vapour chamber of Figure 12. This vapour chamber of Figure 12 (and indeed any other disclosures herein) may be particularly useful for the aerospace industry where the vapour chamber may be used as part of a thermal strap. As mentioned above, any methods of providing and filling the sealed chamber 306 that have been described above are equally applicable for providing and filling the vapour chamber 700 and vice versa. The structures and methods described above provide cold plates that are able to cool high heat density components so as to enable efficient and effective operation of electronics components, including components with high densities of heat-generating equipment, such as data centre boards running Al and machine learning workloads. These structures and methods provide improvements in efficiency over existing components as well as minimising overcooling, minimising operational costs and reducing the risk of leaks (or other incidents). Furthermore, these structures and methods enable the efficient use of space within data centres by providing cold plates that can operate efficiently with only a small profile. Alternatives and modifications It will be understood that the present invention has been described above purely by way of example, and modifications of detail can be made within the scope of the invention. For example, while the detailed description has primarily described embodiments with only a single means for selectively cooling a region of a circuit board, in other examples a plurality of such means could be provided. For example, a vapour chamber could be provided to cool an optical connector, while direct liquid cooling may be used to cool a processor. More generally, the structures used for the cold plates described with reference to Figures 2a-8e may be combined with the structures used for the cold plates described with reference to Figures 9b-11 so that a single cold plate 200, 600 may be provided that comprises any one or more of: a plurality of passages and / or cold sections 210, impingement structures 217, a thermal spreader 300, a sealed chamber 306, a cavity 602 for coolant flow, and a vapour chamber 700. While in the example cold plates of Figures 7a - 8e, both jet impingement and the thermal spreader 300 are used, in other examples jet impingement cooling may be provided to cool a circuit board without the thermal spreader. In still further examples, the thermal spreader may be provided along with direct liquid cooling, but the direct liquid cooling may not comprise the impingement structures 217 to cause jet impingement. The combination of both jet impingement and the thermal spreader cooling may however allow a greater heat flux from the data center board than would be possible with only jet impingement in direct liquid cooling, or with only the thermal spreader but without jet impingement. In some embodiments, the cold plate 200 may comprise parallel flow cooling channels instead of, or alongside, the impingement structures 217. In some embodiments, the cold plate may comprise cut-outs that enable air to flow over the thermal spreader 300 so as to air cool the thermal spreader. The cold plates described herein are particularly useful for cooling electronics components on boards, e.g. PCBs, in data centres. It will be appreciated that the described cold plates may equally be used in other situations, e.g. to cool telecoms equipment or electronics components in vehicles. In some embodiments, e.g. for edge compute nodes, the cold plate may be mounted to a pole or a mast. In some embodiments, the cooling loop may be used to provide weather and / or EMI shielding, e.g. to provide such shielding to edge nodes that are deployed outdoors. The cold plate may also be mounted in a housing that comprises separate weatherproofing and / or EMI protection structures. In some embodiments, the cold plate 200 may be integrated into a housing (e.g. a server rack), where the housing comprises one or more mounting structures for mounting data center boards. This enables a data center board to be places into the housing so as to be correctly located adjacent a cold section and / or a vapour chamber of the integrated cold plate. The present disclosure extends to a housing for one or more electronics boards (e.g. data center boards), the housing comprising one or more cold plates for cooling the electronics boards. 5 This housing may further comprise a housing inlet or a housing outlet, which housing inlet and housing outlet are arranged to receive and / or output coolant, which coolant can then be provided to the one or more cold plates of the housing via respective cold plate inlets and outlets. The disclosure further extends to a housing that comprises, or provides, an immersion chamber. For example, a housing that is arranged to be filled with a dielectric fluid so as to provide the 10 immersion chamber. Reference numerals appearing in the claims are by way of illustration only and shall have no limiting effect on the scope of the claims.

Claims

1. A cold plate for cooling a circuit board comprising a plurality of electronics components, the cold plate comprising:a plurality of cold sections, wherein each cold section is arranged to cool a separate component on the circuit board; anda channel comprising:an inlet;an outlet;a plurality of passages, wherein each passage is associated with a respective one or more cold sections, and wherein the channel comprises a branching channel section located between the inlet and the cold sections and a recombining channel section located between the cold sections and the outlet, the plurality of passages being located between the branching channel section and the recombining channel section.

2. The cold plate of any preceding claim, comprising a thermal spreader, preferably wherein the thermal spreader comprises a pressed part that is located adjacent impingement structures of the cold plate.

3. The cold plate of claim 2, wherein the thermal spreader comprises one or more sealed chambers, preferably wherein the sealed chambers form heat pipes and / or thermosyphons, preferably wherein each sealed chamber comprises an evaporator section and a condenser section, preferably wherein the evaporator end and the condenser section are connected by a wick.

4. The cold plate of claim 3, wherein a condenser section of each sealed chamber is located adjacent impingement structures of the cold plate.

5. The cold plate of any of claims 2 to 4, comprising a pillar, the pillar extending from the thermal spreader, wherein the pillar provides a heat pipe and / or a thermosyphon, preferably wherein the pillar comprises a support insert, more preferably wherein the support insert comprises an extruded component.

6. The cold plate of any of claims 2 to 5, wherein the thermal spreader comprises a clad metal, wherein a first part of the thermal spreader is formed of a first metal and a second part of the thermal spreader is former of a second metal, preferably wherein:the first metal and the second metal are dissimilar metals, more preferably wherein the first part is formed of aluminium and the second part is former of copper; and / orthe first part of the thermal spreader is attached to the cold plate; and / orthe second part of the thermal spreader is arranged to be located adjacent a component of the circuit board.

7. The cold plate of any of claims 2 to 6, comprising a capping sheet arranged to be located adjacent an electronics component being cooled by the cold plate, preferably comprising a first coolant located between a first part of the thermal spreader and the remainder of the coldplate and a second, different coolant, located between a second part of the thermal spreader and the capping sheet.

8. The cold plate of any of claims 2 to 7, wherein the thermal spreader provides one or more sealed chambers for each of the cold sections of the cold plate.

9. The cold plate of any preceding claim, comprising:a metal base plate; anda metal top plate;wherein the channel is formed between the metal base plate and the metal top plate.

10. The cold plate of any preceding claim, wherein the cold plate comprises a roll-bonded cold plate.

11. The cold plate of any preceding claim, wherein the channel comprises a continuous (e.g. unbroken) channel between the inlet and the outlet, preferably wherein the channel is an integral part of the cold plate.

12. The method of any preceding claim, wherein:the inlet and the outlet are provided on a same side and / or a same edge of the cold plate; and / orthe inlet and the outlet are adjacent, preferably wherein the inlet and the outlet are separated by no more than 250mm, no more than 100mm, and / or no more than 50mm.

13. The cold plate of any preceding claim, wherein each of the passages is associated with a plurality of cold sections, preferably wherein:each of the passages is associated with two cold sections; and / orthe channel comprises at least two passages, at least three passages, and / or four passages.

14. The cold plate of any preceding claim, wherein each of the passages is shaped so as to provide an equal flow rate through the passages.

15. The cold plate of any of claims 1 to 13, wherein the passages are shaped such that a first passage has a different flow rate to a second passage, preferably wherein the passages being shaped comprises the passages being differently sized and / or one or more of the passages comprising an obstruction.

16. The cold plate of claim 15, wherein a first passage with a first flow rate is located adjacent a first cold section and a second passage with a second, different, flow rate is located adjacent a second cold section, the first cold section and the second cold section being provided to cool electronic components with different maximum operating temperatures.

17. The cold plate of any preceding claim, wherein the inlet and the outlet of the channel are spaced from the cold sections, preferably wherein the inlet and the outlet are each spaced from the cold sections by at least 50mm, at least 100mm, and / or at least 250mm.

18. The cold plate of any preceding claim, comprising one or more impingement structures at the one or more cold sections, the impingement structures being arranged to direct a flow of fluid onto a cooling surface of the cold plate, preferably wherein the impingement structures comprise holes.

19. The cold plate of claim 18, comprising a pressed part arranged to cap the impingement structures, preferably wherein the pressed part is arranged to contact one or more electronics components of the circuit board.

20. The cold plate of any preceding claim, wherein, at the location of each cold section, the passages branch into a plurality of channel sections, preferably wherein the plurality of channel sections comprise one or more inlet channel sections and one or more outlet channel sections, more preferably comprising an arrangement of alternating inlet channel sections and outlet channel sections, the inlet channel sections and the outlet channel sections being connected by holes, preferably being connected by holes of impingement structures.

21. The cold plate of any preceding claim, comprising cut-outs and / or thinned sections at the cold sections of the cold plate.

22. A cold plate for cooling a circuit board located in an immersion chamber, the circuit board comprising a plurality of electronics components, wherein the cold plate is arranged to thermally isolate a first electronics component on the electronics board from a second electronics component on the electronics board.

23. The cold plate of any preceding claim, being a cold plate for cooling a data center board in a data center.

24. A housing comprising one or more cold plates according to any preceding claim, preferably wherein the housing comprises mounting structures for one or more electronics boards, more preferably wherein the mounting structures are arranged to locate the electronics boards adjacent cold plates.

25. A method of manufacturing the cold plate of any of claims 1 to 23, the method comprising roll bonding and inflating a sheet to form the channel; preferably, comprising one or more of: cutting material out of the sheet at the locations of the cold sections.forming impingement structures at the cold sections, preferably comprising:forming the impingement structures comprises punching holes through the sheet; and / orattaching a pressed plate over the impingement structures.inflating the channel by providing an inflating fluid through the inlet of the channel.attaching a thermal spreader onto the cold plate, preferably comprising:welding the thermal spreader onto the cold plate; and / orforming the thermal spreader by attaching a first sheet of a first material to a second sheet of a second material.

Images