Integrated energy recovery and pixel-level electrocaloric refrigeration function partitioned electrocaloric refrigeration device

Abstract

The application relates to an integrated energy recovery and pixel-level electrocaloric refrigeration device, comprising: a substrate layer; a pixel-level electrocaloric refrigeration function partition, arranged on the substrate layer and containing a plurality of independently controllable electrocaloric pixel units, used for heat coupling refrigeration with a hot spot area; an energy recovery and energy storage function partition, arranged in the same plane with the pixel-level electrocaloric refrigeration function partition, used for capturing and temporarily storing the depolarization released electric energy; a power supply excitation function partition, arranged in the same plane, used for providing an excitation signal; an interconnection layer electrically connecting the three partitions; a bidirectional switch matrix, comprising an excitation branch, a recovery branch and a return branch; and a control unit, configured to sequentially turn on the excitation branch to drive the target pixel to polarize, turn on the recovery branch to guide the depolarization electric energy into the energy recovery and energy storage function partition, and turn on the return branch to return the stored electric energy to the same or another pixel for subsequent polarization. The application realizes pixel-level on-demand refrigeration and depolarization electric energy recovery reuse.

Description

Technical Field

[0001] This application relates to the field of card cooling technology, and in particular to a card cooling device, system, computer equipment, and storage medium that integrates energy recovery and pixel-level card cooling functional partitions. Background Technology

[0002] As integrated circuits develop towards high density, 3D integration, and high computing power, chips exhibit significant spatial non-uniform heating characteristics during operation, resulting in multiple dynamically migrating local hotspots within a limited area. Electrocard cooling, as a solid-state active cooling technology, possesses advantages such as compressor-free operation, ease of miniaturization, and compatibility with electronic system electric drives, and is considered an important candidate technology for chip-level localized cooling. However, existing electrocard cooling solutions mainly focus on macroscopic coolers or single-unit electrocard thin-film devices, lacking pixel-level cooling functional partitions that correspond to the distribution of localized hotspots on the chip, leading to a mismatch between cooling capacity distribution and thermal load distribution. Simultaneously, the electrocard unit experiences significant charging and discharging energy exchange during high-field driving and depolarization processes. Traditional driving methods typically dissipate this energy through resistance or uncontrolled loss, resulting in low cooling efficiency and low system energy utilization. Summary of the Invention

[0003] Therefore, it is necessary to provide an integrated energy recovery and pixel-level card cooling function partition that can achieve pixel-level on-demand cooling and recover and reuse depolarized electrical energy, addressing the aforementioned technical problems.

[0004] In a first aspect, this application provides an electric card cooling device that integrates energy recovery and pixel-level electric card cooling functional partitions, comprising: basal layer; A pixel-level card cooling function partition is set on the base layer, including multiple independently controllable card pixel units, which are used to thermally couple with the hot spot area of ​​the target to be cooled and generate a cooling effect; An energy recovery and storage functional zone is set on the base layer and arranged coplanarly with the pixel-level card cooling functional zone. It is used to capture and temporarily store the electrical energy released during the depolarization stage of the pixel-level card cooling functional zone. A power supply excitation function partition is disposed on the base layer and arranged coplanarly with the pixel-level card cooling function partition, and is used to provide excitation signals to the pixel-level card cooling function partition. The interconnection layer electrically connects the pixel-level card cooling function partition, the energy recovery and storage function partition, and the power supply and excitation function partition. A bidirectional switch matrix includes at least an excitation branch, a recovery branch, and a feedback branch. The excitation branch connects the power supply excitation functional zone to the pixel-level card cooling functional zone. The recovery branch connects the pixel-level card cooling functional zone to the energy recovery and storage functional zone. The feedback branch connects the energy recovery and storage functional zone to the pixel-level card cooling functional zone. A control unit, connected to the bidirectional switch matrix, is configured to: activate the excitation branch during the excitation phase to drive the target card pixel unit to polarize; activate the recovery branch during the recovery phase to depolarize the target card pixel unit and import the electrical energy released by the target card pixel unit into the energy recovery and storage functional zone; and activate the feedback branch during the feedback phase to feed back the electrical energy stored in the energy recovery and storage functional zone to the same or another card pixel unit for subsequent polarization.

[0005] In one embodiment, the power supply excitation function partition and the energy recovery and storage function partition operate in a cooperative working mode. The cooperative working mode includes: the excitation energy provided by the power supply excitation function partition includes electrical energy fed back by the energy recovery and storage function partition, so that the external power supply input only provides the reference energy difference per drive cycle, so as to increase the instantaneous effective drive voltage on the card pixel unit when the external power supply voltage is limited.

[0006] In one embodiment, the equivalent capacitance of a single card pixel unit in the pixel-level card cooling function partition and the equivalent energy storage capacitor configured in the energy recovery and storage function partition satisfy the following: ,in This refers to the equivalent capacitance of the individual card pixel unit. The equivalent energy storage capacitor configured therefor.

[0007] In one embodiment, the equivalent capacitance of a single card pixel unit in the pixel-level card cooling function partition and the equivalent energy storage capacitor configured in the energy recovery and storage function partition satisfy the following: .

[0008] In one embodiment, a resonant transfer inductor is connected in series between the pixel-level card cooling function zone and the energy recovery and storage function zone. The resonant transfer inductor is used to control the peak discharge current and improve the charge transfer efficiency.

[0009] In one embodiment, the control unit is further configured to: perform at least one control strategy on the card pixel unit according to the hotspot mapping information, including dynamic gating, drive frequency adjustment, drive duty cycle adjustment, excitation voltage amplitude adjustment, or recycling path switching; for areas with high hotspot intensity, increase the drive frequency, pulse amplitude, or drive duty cycle of the corresponding card pixel unit; and for non-hotspot areas, reduce or turn off the corresponding card pixel unit.

[0010] In one embodiment, the energy recovery and storage functional zoning is any one of a centralized public energy storage architecture, a grouped local energy storage architecture, or a distributed architecture in which each card pixel unit is independently configured with a micro energy storage unit.

[0011] In one embodiment, the switching devices of the bidirectional switching matrix are any one or a combination of MOSFETs, GaN devices, IGBTs, MEMS switches, or relays.

[0012] In one embodiment, the power supply excitation function partition is any one or a combination of a charge pump circuit, a resonant boost circuit, a flyback boost circuit, a half-bridge pulse drive circuit, a full-bridge pulse drive circuit, or a multi-stage series-parallel pulse network circuit.

[0013] In one embodiment, a thermal management auxiliary structure is provided above or below the pixel-level card cooling function partition to improve the directional heat flow transmission and suppress thermal crosstalk between adjacent card pixel units. The thermal management auxiliary structure is any one or a combination of a heat diffusion layer, a heat sink layer, a micro thermal switch layer, a phase change interface layer, a thermal insulation trench, or a microstructure support layer.

[0014] The aforementioned card cooling device, system, computer equipment, and storage medium, which integrate energy recovery and pixel-level card cooling functional zones, achieve point-to-point on-demand cooling of local hotspots by coplanarly integrating pixel-level card cooling functional zones, energy recovery and storage functional zones, and power supply and excitation functional zones on the substrate layer. A bidirectional switch matrix containing excitation, recovery, and feedback branches is set up. The control unit sequentially activates the corresponding branches according to the excitation, recovery, and feedback stages, realizing point-to-point on-demand cooling of local hotspots. Simultaneously, the electrical energy released during the depolarization stage is captured, temporarily stored, and fed back to the same or another card pixel unit for subsequent polarization, improving the utilization efficiency of electrical energy in single and multi-cycle periods and reducing the external power supply burden. The coplanar arrangement of the three zones and the shortened interconnection layer reduce parasitic inductance and resistance, facilitating high-frequency drive and fast response. The control unit's sequential control of the polarization, depolarization recovery, and feedback of the target card pixel unit allows energy to be dynamically reused within the device, enhancing the instantaneous effective driving electric field without significantly increasing the external power supply voltage, thereby improving cooling capacity. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a card cooling device that integrates energy recovery and pixel-level card cooling function partitions in one embodiment; Figure 2 This is an exemplary circuit diagram of a card cooling device that integrates energy recovery and pixel-level card cooling function partitions in one embodiment; Figure 3 This is a schematic diagram of a complete cooling cycle in one embodiment. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. First, to facilitate understanding of the technical solutions provided by the embodiments of this application, the background technology involved in the embodiments of this application will be described below.

[0017] As integrated circuits develop towards high density, three-dimensional integration, and high computing power, chips exhibit significant spatial non-uniform heating characteristics during operation. This means that multiple local hotspots with dynamically migrating locations and heat flux densities far exceeding the average level can appear within the limited area of ​​the same chip. For such hotspots, traditional monolithic, uniform heat dissipation solutions (such as macroscopic heat sinks and thermoelectric coolers) often fail to accurately distribute cooling energy to the hotspot areas. This often results in insufficient cooling and continuous temperature rise in the hotspot areas, while non-hotspot areas experience redundant cooling, leading to increased system power consumption, delayed temperature control, and in severe cases, device performance degradation or decreased reliability.

[0018] Electrocaloric cooling is a solid-state technology that achieves active cooling based on the polarization entropy change effect of electrocaloric materials. When an electric field is applied to the electrocaloric material, the dipoles tend to become ordered, and the material temperature rises (polarization exothermic). After the electric field is removed, the dipoles return to disorder, and the material temperature drops (depolarization endothermic). This technology has inherent advantages such as no compressor working fluid, ease of miniaturization, compatibility with integrated circuit electric drive systems, and fast theoretical response speed. Therefore, it is widely recognized as one of the ideal candidate technologies for chip-level local thermal management.

[0019] However, research has revealed that existing card cooling solutions suffer from two interrelated technical bottlenecks that limit their practical application in chip-level local hotspot management.

[0020] First, there is a lack of pixel-level cooling functional partitions tailored to the spatial distribution of hotspots. Most existing research and patent disclosures focus on macroscopic rotary card devices, fluid heat exchange card devices, or simply thermally coupling a single card film to the integrated circuit. None of these solutions divide the card cooling unit into independently controllable pixel-level arrays that dynamically match the location and intensity of chip hotspots. This results in the entire card element operating in the same mode regardless of changes in hotspot distribution, failing to achieve on-demand distribution of cooling capacity and precise localized cooling.

[0021] Second, the electrical energy released by the card unit during the depolarization phase is not effectively recovered and utilized. The card material can essentially be considered a capacitive load. During the polarization phase, an external power source is needed to establish an electric field, while during the depolarization phase (when the electric field is removed), the stored charge and electric field energy are released. Existing driving methods typically release this energy through resistive dissipation or simple short-circuiting, resulting in significant energy loss. This not only reduces the overall energy efficiency of the card-based cooling system but also makes it difficult to obtain a sufficiently high driving electric field to fully activate the card effect without increasing the supply voltage, especially under conditions of limited external power supply voltage, thus limiting the improvement of cooling capacity.

[0022] Furthermore, at the device architecture level, the power supply excitation unit, the card cooling unit, and the possible energy processing unit are often distributed in different modules. The modules are connected by long leads, which leads to an increase in parasitic inductance and parasitic resistance, which is not conducive to high-frequency driving and fast dynamic response, and also increases the complexity of system integration.

[0023] In summary, the core problem with existing technologies is that they cannot simultaneously achieve pixel-level on-demand cooling for dynamic hotspots, recovery and reuse of depolarized electrical energy, and a compact device architecture with high integration and low parasitic losses. Therefore, embodiments of this application provide an electronic card cooling device that integrates energy recovery and pixel-level electronic card cooling functional partitioning.

[0024] As mentioned above, existing card cooling devices lack pixel-level cooling functional partitions that correspond to the local hotspot distribution of the chip. Furthermore, the electrical energy released during the depolarization phase is wasted, and the dispersed arrangement of functional units results in large parasitic parameters. Therefore, in one embodiment, such as... Figure 1 As shown, Figure 1A schematic diagram of an integrated energy recovery and pixel-level card cooling functional partition is shown. As shown, the device includes, from bottom to top: a base layer, a pixel-level card cooling functional partition (Area A) coplanarly arranged on the base layer, an energy recovery and storage functional partition (Area B), and a power supply and excitation functional partition (Area C), wherein Area A is located in the center, and Areas B and C are located on the left and right sides of Area A, respectively; an interconnect layer is disposed above the three partitions A, B, and C, integrating a bidirectional switch matrix and a control unit; a thermal management auxiliary structure can also be disposed above the interconnect layer. Area A is thermally coupled to the chip to be cooled or the packaged heat source layer through a thermal interface. Specifically, the integrated energy recovery and pixel-level card cooling functional partition of this embodiment includes: basal layer; The pixel-level card cooling function partition (Area A) is set on the base layer and includes multiple independently controllable card pixel units, which are used to thermally couple with the hot spot area of ​​the target to be cooled and generate a cooling effect. The energy recovery and storage functional zone (Zone B) is located on the base layer and is arranged on the same plane as the pixel-level electrical card cooling functional zone. It is used to capture and temporarily store the electrical energy released during the depolarization stage of the pixel-level electrical card cooling functional zone. The power supply excitation function zone (C zone) is located on the base layer and is arranged on the same plane as the pixel-level card cooling function zone. It is used to provide excitation signals to the pixel-level card cooling function zone. The interconnection layer electrically connects pixel-level card cooling function zones, energy recovery and energy storage function zones, and power supply and excitation function zones. A bidirectional switch matrix includes at least an excitation branch, a recovery branch, and a feedback branch. The excitation branch connects the power supply excitation function zone and the pixel-level card cooling function zone. The recovery branch connects the pixel-level card cooling function zone and the energy recovery and storage function zone. The feedback branch connects the energy recovery and storage function zone and the pixel-level card cooling function zone. The control unit, connected to a bidirectional switch matrix, is configured to: conduct the excitation branch during the excitation phase to drive the target card pixel unit to polarize; conduct the recovery branch during the recovery phase to depolarize the target card pixel unit and guide the electrical energy released by the target card pixel unit into the energy recovery and storage functional zone; and conduct the feedback branch during the feedback phase to feed back the electrical energy stored in the energy recovery and storage functional zone to the same or another card pixel unit for subsequent polarization.

[0025] The components of the above technical solution will be described in detail below.

[0026] Substrate Layer: The substrate layer serves as the mechanical support and electrical insulation base for the entire device. In one implementation, the substrate layer can be a ceramic substrate, such as alumina ceramic or aluminum nitride ceramic. Alumina ceramic has high mechanical strength and good insulation properties, while aluminum nitride ceramic has higher thermal conductivity (approximately 170-200 W / (m·K)), which is beneficial for conducting heat from the device to the external heat dissipation structure. The substrate layer can also be a glass-ceramic substrate or a low-temperature co-fired ceramic substrate. In practical applications, other insulating materials such as silicon and silicon carbide can also be used for the substrate layer; this embodiment does not limit this choice. One or more of the following can be provided on the substrate layer as needed: a heat diffusion layer, an insulating layer, a buffer layer, or a stress relief layer. The heat diffusion layer is used to uniformly diffuse local heat flow; the insulating layer is used to prevent unnecessary electrical connections; the buffer layer is used to alleviate stress caused by thermal expansion coefficient mismatch; and the stress relief layer is used to absorb mechanical stress generated during packaging or operation, thereby improving the thermal stability and packaging compatibility of the device.

[0027] Pixel-level card cooling functional zone: This zone is located on the substrate layer and consists of multiple card pixel units distributed along the plane. Each card pixel unit includes at least a card material layer, an upper electrode, a lower electrode, a local wiring port, and a thermally conductive interface thermally coupled to the target to be cooled. The upper and lower electrodes can be made of conductive metal materials such as gold, platinum, aluminum, copper, or conductive oxides such as indium tin oxide (ITO). The thickness of the electrodes is typically between 10 nanometers and 1 micrometer, with the specific value determined based on the thickness of the card material layer and the required electric field strength. Each card pixel unit is physically independently controllable; that is, the upper or lower electrode of each pixel unit is individually connected to the corresponding branch of the bidirectional switch matrix through the local wiring port, allowing it to be individually driven or turned off, achieving spatially selective cooling. The planar arrangement of the card pixel units can adopt a regular matrix array (e.g., a square grid, with the number of rows and columns set according to the chip area as 16×16, 32×32, etc.), a cellular array (hexagonal tiling, used to maximize the fill factor), a strip array (long strip pixels, suitable for strip-shaped heat sources), or a non-uniform patterned layout customized according to the chip hotspot distribution (e.g., arranging smaller, higher-density pixels in areas with concentrated hotspots). The lateral dimension (i.e., the width or diameter of the pixel) of a single card pixel unit can be designed according to the scale of the target hotspot, with a design range of 10 micrometers to 10 millimeters. In practical applications for chip-level hotspot management, the preferred lateral dimension is 50 micrometers to 2 millimeters. The electrostatic material layer can be made of ferroelectric ceramics (such as lead zirconate titanate Pb(Zr,Ti)O3, abbreviated as PZT), antiferroelectric ceramics (such as lead zirconate PbZrO3), relaxor ferroelectric ceramics (such as lead magnesium niobate-lead titanate (1-x)Pb(Mg1 / 3Nb2 / 3)O3-xPbTiO3, abbreviated as PMN-PT), high-entropy ferroelectric ceramics, or any of the following: ferroelectric thick film (thickness from 1 micrometer to 100 micrometers), ferroelectric thin film (thickness less than 1 micrometer), or ferroelectric multilayer stacked structure. In practical applications, the selection of the electrostatic material needs to comprehensively consider factors such as dielectric constant, breakdown field strength, electrostatic effect intensity (i.e., adiabatic temperature change ΔT and isothermal entropy change ΔS), fatigue life, and compatibility with the electrode. This embodiment does not limit these factors.

[0028] Energy recovery and storage functional zone: This zone is coplanar with the pixel-level card cooling functional zone on the substrate layer, i.e., located in the same plane, and the two are electrically connected through an interconnect layer. The energy recovery and storage functional zone is used to capture and temporarily store the electrical energy released by the pixel-level card cooling functional zone during the depolarization phase. In one implementation, this zone may include one or more of the following: a recovery capacitor, a buffer capacitor, an energy storage capacitor, a resonant transfer inductor, a rectifier device, or a clamping device. Specifically, the recovery capacitor is used to directly receive the charge released from the card pixel unit; the buffer capacitor is used to smooth voltage fluctuations; the energy storage capacitor is used to store the recovered energy for a long period; the resonant transfer inductor is used to form a resonant circuit with the capacitor to improve transfer efficiency; the rectifier device (such as a diode) is used to prevent backflow of electrical energy; and the clamping device (such as a transient voltage suppressor diode (TVS)) is used to limit voltage spikes and protect the switching devices. The energy storage architecture for the energy recovery and storage functional zones can be a centralized public energy storage architecture (i.e., all card pixel units share a common energy storage unit, such as a single large-capacity capacitor or a capacitor bank formed by multiple capacitors connected in parallel), a grouped local energy storage architecture (dividing the card pixel units into several groups, with each group configured with one energy storage unit), or a distributed architecture where each card pixel unit is independently configured with a micro-energy storage unit. In practical applications, an appropriate architecture can be selected based on the requirements of recovery efficiency, area cost, and response speed; this embodiment does not impose any limitations on this.

[0029] Power Supply Excitation Function Zone: This zone is also located on the base layer and is coplanar with the pixel-level card cooling function zone. The power supply excitation function zone is used to convert external input power (e.g., low-voltage DC from a battery or power management unit, typically 3.3V, 5V, or 12V) into high-voltage pulses or high-electric-field excitation signals (typically ranging from 20V to 500V, with pulse widths from microseconds to milliseconds) required to drive the card pixel units. In one implementation, the power supply excitation function zone may include one or more of the following: an external power input interface, a boost circuit, a pulse drive circuit, a waveform shaping circuit, and an overvoltage protection circuit. The external power input interface receives power from an external power source and typically includes a filter capacitor and a reverse connection protection diode. The boost circuit boosts the input voltage to the required high-voltage level. The pulse drive circuit generates a high-voltage pulse with a specific pulse width, duty cycle, and repetition frequency. The waveform shaping circuit adjusts the leading and trailing edge slopes of the pulse to reduce electromagnetic interference. Overvoltage protection circuits (such as Zener diode clamping or comparator shutdown circuits) are used to prevent abnormal increases in output voltage from damaging the card pixel units. The circuit topology of the power supply excitation functional area can be any combination of one or more of the following: charge pump circuit, resonant boost circuit, flyback boost circuit, half-bridge pulse drive circuit, full-bridge pulse drive circuit, or multi-stage series-parallel pulse network circuit. For example, the charge pump circuit uses a switched capacitor network to achieve voltage multiplication. Its advantages are simple structure, no inductor required, and low electromagnetic interference, making it suitable for low-current, low-power applications (e.g., 5V input, 20V-40V output). The flyback boost circuit achieves a higher boost ratio (e.g., 5V input, 200V output) based on a transformer, while providing electrical isolation. In practical applications, the power supply excitation functional area can be implemented using discrete components (such as independent MOSFETs, diodes, capacitors, and inductors implemented on a printed circuit board), or using hybrid integration (integrating multiple bare chips on the same substrate) or single-chip integration (integrating all circuits on the same silicon wafer using semiconductor technology). This embodiment does not limit this approach.

[0030] Interconnect Layer: The interconnect layer is the top metal wiring layer that enables low parasitic electrical connections between pixel-level card cooling functional zones, energy recovery and storage functional zones, and power supply and excitation functional zones. In one implementation, the interconnect layer is made of metals such as copper, aluminum, or gold, formed by sputtering, evaporation, or electroplating processes, and patterned using photolithography to form the required wiring pattern. The thickness of the interconnect layer is typically 0.5 micrometers to 5 micrometers, and the linewidth is designed according to the current carrying capacity, with typical linewidths ranging from 1 micrometer to 100 micrometers. Ohmic contacts are formed between the interconnect layer and the electrodes or pads of each functional zone, with a contact resistance of less than 1 ohm to ensure low signal and energy transmission loss. The interconnect layer can also include multiple layers of metal and interlayer dielectrics to form a complex wiring network, enabling independent addressing connections between each card pixel unit and the bidirectional switch matrix.

[0031] Bidirectional Switching Matrix: The bidirectional switching matrix is ​​the core switching network connecting three functional zones, and it includes at least an excitation branch, a recovery branch, and a feedback branch. The excitation branch connects the power supply excitation functional zone and the pixel-level card cooling functional zone, used to apply drive signals to selected card pixel units during the excitation phase. The recovery branch connects the pixel-level card cooling functional zone and the energy recovery and storage functional zone, used to guide the depolarized released energy to the storage unit during the recovery phase. The feedback branch connects the energy recovery and storage functional zone and the pixel-level card cooling functional zone, used to feed back the stored energy to the card pixel units requiring polarization during the feedback phase. Each branch can consist of one or more switching devices, which should have bidirectional or unidirectional conduction capabilities combined with anti-parallel diodes to adapt to the current flow direction at different stages. The switching devices in the bidirectional switching matrix can be any combination of one or more of the following: MOSFETs, GaN devices, IGBTs, MEMS switches, or relays. In low-voltage, high-frequency scenarios (e.g., drive voltage <100V, switching frequency >100kHz), MOSFETs or GaN devices are preferred due to their fast switching speed (nanosecond level) and low on-resistance (milliohm level). In high-voltage, low-frequency scenarios (e.g., drive voltage >200V, switching frequency <10kHz), IGBTs or hybrid solutions (e.g., IGBTs in parallel with fast recovery diodes) are more suitable. MEMS switches have extremely low on-resistance (subohm level) and extremely high turn-off insulation (gigaohm level), but their switching speed is relatively slow (microsecond to millisecond level), making them suitable for low-frequency, high-isolation scenarios. Relays are suitable for scenarios where space is not critical and high reliability is required. In practical applications, the specific model of the switching device needs to be selected based on the voltage level, switching frequency, and current capacity; this embodiment does not impose any limitations on this.

[0032] Control Unit: The control unit is connected to the bidirectional switch matrix and integrates sequential logic circuits or a microprocessor. It is configured to control the on / off state of each branch in the bidirectional switch matrix according to a preset phase sequence. Specifically, during the excitation phase, the control unit outputs a high-level control signal to the gate or base of the switching device in the excitation branch, turning on the switching device in that branch. This applies the excitation signal generated by the power supply excitation functional zone to the target card pixel unit, driving the pixel unit into a polarization state. After the excitation phase ends (usually triggered by a timer or a polarization completion signal), the control unit turns off the excitation branch and outputs a control signal during the recovery phase to turn on the switching device in the recovery branch. At this time, the depolarization current of the target card pixel unit flows into the energy recovery and storage functional zone through the recovery branch, realizing the capture and temporary storage of electrical energy. The duration of the recovery phase is determined by the time constant of the card material, typically ranging from microseconds to milliseconds. After the recovery phase ends, the control unit shuts off the recovery branch and outputs a control signal during the feedback phase to turn on the switching devices in the feedback branch, allowing the electrical energy stored in the energy recovery and storage functional zone to be fed back to the same card pixel unit or another card pixel unit that needs polarization for subsequent polarization excitation. The feedback phase may involve voltage reshaping (e.g., adjusting the voltage level via a charge pump or DC-DC converter), and the control unit needs to synchronously control these auxiliary circuits. The control unit can be implemented using any of the following: an on-chip driver chip (e.g., an application-specific integrated circuit integrating a high-voltage gate driver), a field-programmable gate array (FPGA), a microcontroller (MCU), or a dedicated timing controller. In practical applications, the control unit also needs to be connected to a temperature sensor or hotspot detection module (e.g., an on-chip diode temperature array or an infrared thermal sensor) to receive hotspot mapping information (i.e., the real-time temperature value of the area corresponding to each pixel) and dynamically adjust the driving parameters based on this information; this embodiment does not limit this.

[0033] Based on the above, this device coplanarly integrates pixel-level card cooling functional zones, energy recovery and storage functional zones, and power supply and excitation functional zones on the substrate layer, and achieves low parasitic electrical connections through the interconnect layer, thereby shortening the interconnect length, reducing parasitic inductance and parasitic resistance, and facilitating high-frequency drive and fast dynamic response.

[0034] One feasible implementation method is, for example Figure 2 As shown, the diagram illustrates an exemplary circuit schematic of a card cooling device integrating energy recovery and pixel-level card cooling functional zones. Zones A, B, and C share a common bottom electrode (labeled "Common Bottom Electrode / GND") and are electrically connected through a switching network and an interconnect layer. Zone A contains an equivalent nonlinear capacitor. Area B contains energy storage capacitors. The three zones are connected via a switch network, wherein: SCA To excite the branch switch, with optional inductor L CA Holding diode D hold Connected to node_A after concatenation; S AB1 and S AB2 To recover the branch switch, and resonant transfer inductor L AB 1. Rectifier device D AB After being connected in series, it is connected between node_A and node_B; node_A is the connection point of the electrode on the card unit in area A, and node_B is the energy storage capacitor in area B. The positive extreme connection point; L AB Used to control peak discharge current and improve charge transfer efficiency; V 外部 It is an external DC power supply.

[0035] During operation, the control unit controls the bidirectional switching matrix in three phases: In the excitation phase, the excitation branch is activated, and the power supply excitation functional zone applies an excitation voltage to the selected target card pixel unit, polarizing the card material and releasing heat, which is then dissipated through an external heat dissipation path. In the recovery phase, the recovery branch is activated, and the target card pixel unit depolarizes to absorb heat from the hot spot to be cooled. Simultaneously, the released electrical energy is temporarily stored in the energy recovery and storage functional zone via the recovery branch. In the feedback phase, the feedback branch is activated, feeding the temporarily stored electrical energy back to the same card pixel unit or another card pixel unit for subsequent polarization. This mechanism achieves the capture and reuse of depolarized electrical energy, avoiding the drawbacks of energy dissipation by resistors or short-circuit losses in traditional methods, thus improving the utilization efficiency of electrical energy in single and multi-cycle periods. Furthermore, since electrical energy can be dynamically shared between pixels, the external power supply only needs to provide the baseline energy difference lost per cycle, reducing the burden on the external power supply.

[0036] Under conditions where external power supply voltage is limited (e.g., only a single lithium battery at 3.7V can power the system), existing card drive solutions struggle to simultaneously achieve sufficiently high drive field strength (typically requiring tens or even hundreds of volts) and system energy efficiency. Therefore, in one embodiment, the power supply excitation functional zone and the energy recovery and storage functional zone operate in a cooperative mode. This cooperative mode includes: the excitation energy provided by the power supply excitation functional zone includes electrical energy fed back from the energy recovery and storage functional zone, ensuring that the external power input only provides the baseline energy difference per drive cycle, thereby increasing the instantaneous effective drive voltage on the card pixel unit when the external power supply voltage is limited.

[0037] In this embodiment, the power supply excitation functional zone does not independently obtain all excitation energy from the external power supply in each cycle, but rather works in conjunction with the energy recovery and storage functional zone. Specifically, during the feedback phase, the energy recovery and storage functional zone sends the previously recovered and stored electrical energy back to the output of the power supply excitation functional zone or directly to the card pixel unit through the feedback branch. When the power supply excitation functional zone generates an excitation signal, the input energy of its internal boost circuit or pulse drive circuit includes the electrical energy fed back from the energy recovery and storage functional zone. Therefore, the external power supply input only needs to provide the energy difference caused by switching losses, leakage losses, etc., in each drive cycle, while most of the polarization energy comes from internal recovery. When the external power supply voltage is low, through this cooperative working mode, the power supply excitation functional zone can utilize the superposition of recovered electrical energy to generate a higher instantaneous output voltage or electric field strength, thereby obtaining a higher effective drive voltage on the card pixel unit than the external power supply voltage.

[0038] In one implementation, a voltage reshaping circuit can be set between the power supply excitation functional zone and the energy recovery and storage functional zone to boost or adjust the recovered electrical energy to a suitable voltage level before feeding it into the excitation path. This voltage reshaping circuit can be implemented using a switched-capacitor converter or a boost converter.

[0039] Based on the above, through the coordinated operation of the power supply excitation functional zone and the energy recovery and storage functional zone, the external power input only needs to provide the reference energy difference per drive cycle, rather than the entire excitation energy. This allows for an increase in the instantaneous effective drive voltage applied to the card pixel unit under conditions of limited external power supply voltage, thereby more fully stimulating the card effect and enhancing cooling capacity. Simultaneously, since most of the energy is recycled, the overall system energy utilization efficiency is improved.

[0040] If the energy storage capacitor capacity of the energy recovery and storage functional zone is not properly selected relative to the equivalent capacitance of the card pixel unit, it will affect the recovery efficiency and return speed. Therefore, in one embodiment, the equivalent capacitance of a single card pixel unit in the pixel-level card cooling functional zone satisfies the following relationship with the equivalent energy storage capacitor configured in the energy recovery and storage functional zone: ,in This refers to the equivalent capacitance of a single pixel unit in a digital card. The equivalent energy storage capacitor is configured.

[0041] The equivalent capacitance of a card pixel unit depends on the dielectric constant of the card material, the electrode area, and the thickness of the card material layer. The equivalent energy storage capacitance configured in the energy recovery and storage functional zone refers to the total equivalent capacitance of the capacitor elements used to temporarily store recovered electrical energy. When the recovery branch is active, a charge transfer loop is formed between the card pixel unit and the energy storage capacitor. If... If the value is too small (less than 0.5), the energy storage capacitor capacity will be insufficient to effectively absorb most of the charge released by the pixel unit of the card, resulting in low recycling efficiency; if... If the value is too large (greater than 50), the energy storage capacitor capacity will be too large, increasing the charging time constant, resulting in insufficient feedback voltage and affecting the feedback speed and subsequent polarization effect. Therefore, [the following is implied:] By limiting the value to the range of 0.5 to 50, both recycling efficiency and return speed can be balanced.

[0042] In one implementation, the equivalent energy storage capacitor configured in the energy recovery and storage functional zone can be achieved by connecting multiple capacitors in parallel or series to meet the required equivalent capacitance value.

[0043] By comparing the equivalent capacitance of a single card pixel unit with the equivalent energy storage capacitor configured in the energy recovery and storage functional zone... Setting the value between 0.5 and 50 ensures that the charge released by the pixel unit of the card during the recycling phase can be transferred to the energy storage capacitor, while the energy storage capacitor can provide sufficient voltage and energy to drive subsequent polarization during the return phase, thereby achieving a balance between recycling efficiency and return speed.

[0044] In practical design, further optimizing the capacitance ratio range can achieve better energy recovery performance. Specifically, in one embodiment, the equivalent capacitance of a single pixel unit in the pixel-level card cooling function zone and the equivalent energy storage capacitor configured in the energy recovery and storage function zone satisfy the following relationship: .

[0045] This embodiment further reduces the capacitance ratio to 1 to 10. When the value is between 1 and 10, the capacity of the energy storage capacitor is on the same order of magnitude as or slightly larger than the capacitance of the card pixel unit, resulting in smaller voltage fluctuations during charge transfer and higher energy transfer efficiency. In one implementation, for a typical card pixel unit, the equivalent energy storage capacitor configured in the energy recovery and storage functional zone can be used... Designed for It is 1 to 10 times larger, which ensures both charge absorption capacity and avoids the area overhead caused by excessively large capacitors.

[0046] Based on the above, within a ratio range of 1 to 10, the energy recovery and storage functional zone can capture and return electrical energy with higher efficiency, while also achieving better area utilization and cost control of the device, thus achieving a better trade-off between cooling performance and implementation cost.

[0047] During the conduction of the recycling branch, the charge transfer between the card pixel unit and the energy storage capacitor may generate a large peak current, which may not only damage the switching devices but also reduce the energy transfer efficiency. To address this, in one embodiment, a resonant transfer inductor is connected in series between the pixel-level card cooling function zone and the energy recovery and storage function zone. The resonant transfer inductor is used to control the discharge peak current and improve the charge transfer efficiency.

[0048] Specifically, a resonant transfer inductor is connected in series between the pixel-level card cooling function zone and the energy recovery and storage function zone (usually on the recovery branch). When the recovery branch is on, the equivalent capacitance of the card pixel unit, the resonant transfer inductor, and the equivalent energy storage capacitance of the energy recovery and storage function zone together form a second-order LC resonant circuit. By appropriately selecting the inductance of the resonant transfer inductor, the charge transfer process can be made to exhibit a critically damped or underdamped state, thereby limiting the peak current to within the range that the switching devices and card materials can withstand. At the same time, during the resonant transfer process, energy is transferred in the form of electromagnetic energy oscillating between the capacitor and the inductor, resulting in lower energy loss and higher charge transfer efficiency compared to purely resistive discharge.

[0049] In one implementation, the resonant transfer inductor can be an air-core coil or a magnetic core inductor. The inductance value is roughly estimated based on the target recovery time constant and the pixel equivalent capacitance. The specific value can be determined by those skilled in the art through actual circuit simulation.

[0050] Based on the above, by setting a resonant transfer inductor in series, the peak current during the discharge process of the card pixel unit is effectively controlled, avoiding the impact of excessive current on the switching devices and card materials, and extending the device life. At the same time, since the energy transfer process is dominated by resonance rather than resistive dissipation, the transfer efficiency of charge to the energy recovery and storage functional area is improved, thereby improving the overall energy recovery efficiency.

[0051] Chip hotspots exhibit dynamic migration and intensity variations. If fixed driving parameters are applied, on-demand cooling allocation cannot be achieved, leading to energy waste. Therefore, in one embodiment, the control unit is further configured to: execute at least one control strategy among dynamic gating, driving frequency adjustment, driving duty cycle adjustment, excitation voltage amplitude adjustment, or recycling path switching on the card pixel unit based on hotspot mapping information; for areas with high hotspot intensity, increase the driving frequency, pulse amplitude, or driving duty cycle of the corresponding card pixel unit; for non-hotspot areas, reduce or shut down the corresponding card pixel unit.

[0052] The above technical solution will be described in detail below.

[0053] Hotspot mapping information refers to chip surface temperature distribution data obtained through on-chip temperature sensors, infrared thermal imaging, or thermal simulation models. This data identifies the location and intensity of each hotspot area. The control unit executes one or more of the following control strategies based on this information.

[0054] Dynamic gating: The control unit selects the pixel unit corresponding to the hot spot position from all the pixel units of the pixel-level card cooling function zone as the target card pixel unit, and puts it into working state, while other pixel units remain silent.

[0055] Drive frequency adjustment: For areas with high hot spot intensity, the control unit increases the drive frequency of the corresponding card pixel unit, that is, executes more polarization-depolarization cycles per unit time, thereby increasing the cooling power; for areas with low hot spot intensity, the drive frequency is reduced to save energy.

[0056] Drive duty cycle adjustment: The control unit adjusts the duty cycle of the excitation voltage pulse (i.e., the proportion of the excitation time to the entire cycle). When the hot spot intensity is high, the duty cycle is increased to prolong the depolarization heat absorption time after each polarization heat release, thereby enhancing the cooling effect.

[0057] Excitation voltage amplitude adjustment: The control unit controls the power supply excitation function zones to generate excitation voltages of different amplitudes. When the hot spot intensity is high, the excitation voltage amplitude is increased, allowing the electrical card material to withstand a higher electric field, thereby obtaining a greater adiabatic temperature change (i.e., stronger cooling capacity); when the hot spot intensity is low, the excitation voltage amplitude is reduced to reduce energy consumption.

[0058] Recovery path switching: The control unit can select to send the recovered electrical energy back to the pixel unit that is currently being cooled (the same pixel) or to another pixel unit that is about to be cooled (another pixel) according to the real-time energy status, so as to realize the dynamic scheduling of energy on demand.

[0059] For areas with high hotspot intensity, the control unit increases one or more of the following: driving frequency, pulse amplitude, or driving duty cycle of the corresponding card pixel unit; for non-hotspot areas, the control unit decreases or completely shuts down the driving of the corresponding card pixel unit.

[0060] Through the aforementioned adaptive control strategy, this device can dynamically match the spatial distribution and intensity changes of hot spots on the chip, achieving on-demand allocation of cooling capacity. For areas with strong hot spots, more cooling resources are concentrated (higher frequency, amplitude, or duty cycle), while cooling is reduced or stopped in non-hot spot areas, thereby minimizing overall energy consumption while ensuring controllable hot spot temperatures. Simultaneously, the dynamic switching of the recovery path ensures that the recovered electrical energy can be used by the pixels that need it most, further improving energy utilization efficiency.

[0061] The architectural design of the energy recovery and storage functional zones affects recovery efficiency, area cost, and response speed; different application scenarios require different architectural solutions. Therefore, in one embodiment, the energy recovery and storage functional zones can be any of the following: a centralized public energy storage architecture, a grouped local energy storage architecture, or a distributed architecture where each energy card pixel unit is independently configured with a micro-energy storage unit.

[0062] A centralized public energy storage architecture refers to an entire energy recovery and storage functional zone containing only one public energy storage unit (such as a large-capacity capacitor or capacitor bank). All energy card pixel units release electrical energy into this public energy storage unit during the recovery phase and also obtain electrical energy from this public energy storage unit during the feedback phase. This architecture is simple in structure and has a high area utilization rate.

[0063] The grouped local energy storage architecture divides the pixel units of a card into multiple groups, with each group configured with a local energy storage unit. Pixels within each group share the group's energy storage unit, but the energy storage units between different groups are independent of each other. This architecture strikes a trade-off between area overhead and performance, and can reduce coupling between different groups.

[0064] A distributed architecture refers to a system where each pixel unit is independently configured with a micro-energy storage unit (such as a miniature thin-film capacitor). Energy recovery and feedback for each pixel are completed within its own corresponding micro-energy storage unit, and there is no energy sharing between pixels. This architecture can completely eliminate electrical interference between pixels, but it has a larger area overhead.

[0065] In practical applications, the most suitable architecture can be selected based on the chip size, hotspot distribution characteristics, and packaging density requirements. For example, for scenarios where hotspots are dispersed and change frequently, a grouped local energy storage architecture can be used; for scenarios with extremely limited area, a centralized public energy storage architecture can be used; and for scenarios requiring strict independent control, a distributed architecture can be used.

[0066] Each of the three architectures has its advantages, and designers can choose according to actual application requirements. Centralized public energy storage architecture minimizes area overhead, grouped local energy storage architecture balances area and performance, and distributed architecture maximizes independent control capabilities. This application does not limit the specific implementations, thus covering a wider range of possibilities.

[0067] The switching devices in a bidirectional switching matrix need to be selected appropriately based on voltage level, switching frequency, and current capacity. Different device types are suitable for different application scenarios. Therefore, in one embodiment, the switching devices in the bidirectional switching matrix are any one or a combination of MOSFETs, GaN devices, IGBTs, MEMS switches, or relays.

[0068] Each switch branch in a bidirectional switch matrix consists of one or more switching devices. The function of the switching devices is to quickly and reliably connect or disconnect the circuit path under the control signal issued by the control unit.

[0069] Metal-oxide-semiconductor field-effect transistors (MOSFETs) are voltage-controlled devices with advantages such as fast switching speed, low on-resistance, and ease of parallel connection, making them particularly suitable for low-voltage, high-frequency applications.

[0070] Gallium nitride high electron mobility transistors (GaN devices) have higher electron mobility and lower gate charge, and their switching speed is faster than silicon-based MOSFETs. They also have lower on-resistance, making them suitable for applications requiring higher frequencies and higher efficiency.

[0071] Insulated Gate Bipolar Transistor (IGBT) integrates the high input impedance of MOSFET and the low on-state voltage drop of bipolar transistor, making it suitable for high voltage and low frequency applications.

[0072] Microelectromechanical system (MEMS) switches are miniature mechanical switches with extremely low on-resistance and extremely high off-resistance, but their switching speed is relatively slow, making them suitable for low-frequency, high-isolation applications.

[0073] Relays are traditional electromagnetic switches with high voltage resistance and high current carrying capacity, but they are large in size and slow in switching speed. They are suitable for scenarios where space is not critical and reliability is a high priority.

[0074] In practical applications, one or more devices can be selected and used in combination based on specific design specifications (operating voltage, switching frequency, current capacity, and board area).

[0075] By providing a variety of switching device types and their combinations, this implementation ensures that the bidirectional switching matrix can adapt to different voltage levels, switching frequencies, and current capacity requirements, thereby expanding the applicability of the device.

[0076] The power supply excitation function needs to convert external input electrical energy into high-voltage pulses or high-electric-field signals suitable for driving the pixel units of the card. Different circuit topologies are suitable for different input and output voltage requirements. Therefore, in one embodiment, the power supply excitation function is any one or a combination of a charge pump circuit, a resonant boost circuit, a flyback boost circuit, a half-bridge pulse drive circuit, a full-bridge pulse drive circuit, or a multi-stage series-parallel pulse network circuit.

[0077] The core function of the power supply excitation functional zone is to convert the relatively low external input voltage into the high-voltage pulse or high-electric field excitation signal required by the card pixel unit. This implementation describes several feasible circuit topologies.

[0078] Charge pump circuits utilize switched capacitor networks to achieve voltage multiplication. Their advantages include simple structure, no need for inductors, low electromagnetic interference, and suitability for low-current, low-power applications.

[0079] The resonant boost circuit utilizes the resonant effect of inductors and capacitors to achieve efficient voltage boosting through zero-voltage or zero-current switching of the switching transistor. It is suitable for medium-power applications and has low switching losses.

[0080] The flyback boost circuit is a transformer-based isolated boost topology that can achieve a high boost ratio while providing electrical isolation, making it suitable for scenarios where input and output isolation is required.

[0081] The half-bridge pulse drive circuit consists of two switching transistors and a capacitor, and can generate unipolar or bipolar pulse waveforms, making it suitable for scenarios that require alternating positive and negative driving of electrical card materials.

[0082] The full-bridge pulse drive circuit consists of four switching transistors and can generate bipolar pulses. The peak output voltage can be close to twice the input voltage, making it suitable for scenarios that require large-amplitude pulses.

[0083] A multi-stage series-parallel pulse network circuit consists of multiple capacitors and switches. By controlling the timing of the switches, the capacitors are connected in series and parallel to generate a high-voltage pulse at the output. Its advantage is that it does not require a transformer or inductor and is easy to integrate.

[0084] In practical applications, one or more circuits can be selected and used in combination based on the input voltage range, output voltage amplitude, output power, efficiency requirements, and cost constraints.

[0085] This implementation provides examples of various power supply and excitation circuit topologies, enabling designers to flexibly select appropriate circuit schemes according to specific application requirements, thereby optimizing efficiency, cost, and integration while meeting the driving requirements of the card pixel unit.

[0086] In the pixel-level card cooling function partition, thermal crosstalk may exist between adjacent card pixel units, meaning that heat generated by one pixel can be conducted to adjacent pixels, affecting the cooling effect. Simultaneously, the directional transmission of heat flow also needs to be guided. Therefore, in one embodiment, a thermal management auxiliary structure is provided above or below the pixel-level card cooling function partition to improve the directional transmission of heat flow and suppress thermal crosstalk between adjacent card pixel units. The thermal management auxiliary structure is any one or a combination of a heat diffusion layer, a heat sink layer, a micro-thermal switch layer, a phase change interface layer, a thermal insulation trench, or a microstructure support layer.

[0087] The thermal management auxiliary structure can be set above (i.e., the side closer to the target to be cooled) or below (i.e., the side closer to the substrate) the pixel-level card cooling function zone to guide heat to flow in the desired direction and reduce thermal interference between pixels.

[0088] Thermal diffusion layer: A thin layer made of highly thermally conductive materials (such as copper, aluminum, graphene, aluminum nitride ceramics, etc.). Its function is to evenly distribute the concentrated heat in the hot spot area in the horizontal direction and prevent local overheating. In one implementation, the thermal diffusion layer can cover the entire pixel-level card cooling function partition and contact the upper electrode or thermal interface of each card pixel unit.

[0089] Heat sink layer: A structure made of materials with high thermal conductivity and high heat capacity (such as copper blocks, aluminum blocks, diamond, etc.) used to absorb and store heat and conduct heat to external heat sinks or the environment. Heat sink layers are usually placed on the outermost layer of a device.

[0090] Micro-thermal switch layer: Composed of materials with adjustable thermal conductivity (e.g., using phase change materials or thermodeformable structures), it can change its thermal conductivity at a specific temperature, thereby realizing the switching control of the heat flow path.

[0091] Phase change interface layer: A thin layer filled with phase change materials (such as paraffin, hydrated salts, etc.). When the phase change material absorbs heat, it undergoes a solid-liquid phase change, thereby absorbing a large amount of latent heat and playing a role in temperature buffering.

[0092] Thermal insulation trenches: In the pixel-level card cooling function partition, trenches are etched between adjacent card pixel units. The trenches are filled with materials with low thermal conductivity (such as silicon dioxide, air, etc.) to block heat conduction between pixels in the planar direction and suppress thermal crosstalk.

[0093] Microstructure support layer: a micro-column, mesh, or porous structure prepared by microfabrication technology, which provides mechanical support and reduces lateral heat conduction by reducing the contact area.

[0094] In practical applications, one or more of the above structures can be combined according to specific thermal management requirements. For example, thermal insulation trenches can be set between pixels to suppress thermal crosstalk, while a thermal diffusion layer can be set on top of the whole to dissipate heat evenly.

[0095] By incorporating a thermal management auxiliary structure, this device improves the directional transport of heat flow: it rapidly dissipates heat generated during the polarization phase to the external environment, while concentrating the cooling energy generated during the depolarization phase in the hot spot area to be cooled. Structures such as thermally insulating trenches effectively suppress thermal crosstalk between adjacent pixel units, preventing the heat dissipation of one pixel from affecting the cooling effect of adjacent pixels, thereby improving the accuracy of temperature control and cooling efficiency.

[0096] In addition to the technical solutions covered in the above embodiments, there are also a variety of alternative implementation methods in the embodiments of this application. These alternative methods can also achieve the technical objectives of this application. The following is a supplementary explanation.

[0097] Regarding the thermal coupling method between the pixel-level card cooling functional area and the target to be cooled: Thermal coupling between the pixel-level card cooling functional area and the target to be cooled (such as a hot spot area on the chip) can be achieved through any of the following methods: direct mounting (directly bonding the device to the back of the chip using thermally conductive adhesive), thermal interface material bonding (filling interface gaps using thermally conductive grease, thermal pads, etc.), flip-chip interconnect (connecting the pixel unit electrode to the corresponding pad on the chip using solder balls or copper pillars, achieving both electrical and thermal connection), co-firing integration (co-firing the card device and the chip on the same ceramic substrate), thin-film deposition (depositing the card material directly onto the back of the chip), and clamp-on packaging (pressing the device onto the chip using spring clips or clamps). The specific choice depends on the integration process, thermal resistance requirements, and mechanical reliability.

[0098] Regarding the preferred range for pixel size: In chip-level thermal management, for typical hotspot sizes (such as processing units within a central processing unit core, ranging from approximately 100 μm × 100 μm to 1 mm × 1 mm), the lateral dimension of the card pixel unit is preferably 50 μm to 2 mm. Smaller pixels (less than 50 μm) allow for finer hotspot matching, but increase manufacturing difficulty and cost; larger pixels (greater than 2 mm) cannot accurately match small hotspots. For package-level or module-level thermal management, the pixel size can be increased to 10 mm.

[0099] Regarding other components in the energy recovery and storage functional zone: The energy recovery and storage functional zone may also include auxiliary circuits such as voltage detection circuits (for monitoring the voltage of the energy storage capacitor), overvoltage protection circuits (e.g., parallel Zener diodes to limit the maximum voltage), and discharge resistors (for safely releasing residual charge when shutting down) to ensure the safety and reliability of the recovery process.

[0100] Regarding other circuits in the power supply excitation function zone: The power supply excitation function zone may also include auxiliary circuits such as temperature compensation circuit (adjusting the output voltage according to the ambient temperature to maintain the optimal operating point of the electrical card material), output enable control (for quickly shutting off the output), and soft-start circuit (to prevent inrush current during power-on).

[0101] Regarding the specific implementation details of the control unit: The hotspot mapping information stored in the control unit can be a two-dimensional temperature array, where each element of the array corresponds to the temperature value of a pixel unit. The control unit can run a proportional-integral-derivative control algorithm or a more advanced predictive control algorithm to calculate the driving parameters (including frequency, duty cycle, and voltage amplitude) for the next cycle based on the difference between the current temperature and the target temperature. In practical applications, the control algorithm can be designed according to specific requirements, and this application does not limit this.

[0102] Regarding the operating steps of the device: as follows Figure 3 As shown, a complete refrigeration cycle includes the following six steps.

[0103] The first step is to design the number of pixels, pixel size and planar layout of the pixel-level card cooling function partition based on the chip hotspot distribution, the size of the area to be cooled and the allowable temperature rise range, and to determine the capacity level, number of groups and interconnection relationship between the energy recovery and energy storage function partition and the power supply and excitation function partition and the pixel-level card cooling function partition.

[0104] The second step involves constructing pixel-level card cooling functional zones, energy recovery and storage functional zones, power supply and excitation functional zones, and their top-level metal wiring on the base layer. The pixel-level card cooling functional zones form a card pixel array, the energy recovery and storage functional zones form a recovery and storage network, and the power supply and excitation functional zones form a boost and pulse drive network. The electrical connection of the three functional zones is completed through a bidirectional switch matrix.

[0105] The third step is to apply a first excitation voltage to the target card pixel unit by the power supply excitation function zone at the beginning of the operation, so that the card pixel unit enters the polarization state. During the polarization process, the dipoles inside the card material tend to be ordered, the temperature of the card pixel unit rises, and the generated heat is discharged through the heat diffusion layer, heat sink layer or external heat dissipation path.

[0106] Fourth, after the heat release is completed or the preset timing conditions are met, the control unit causes the target card pixel unit to switch from the polarized state to the depolarized state; the temperature of the card pixel unit decreases and it absorbs heat from the hot spot to be cooled. At the same time, part of the charge or electric field energy in the branch of the card pixel unit enters the energy recovery and storage functional zone through the A-B zone recovery switch branch.

[0107] The fifth step involves the energy recovery and storage functional zone temporarily storing, equalizing, current limiting, and voltage reforming the recovered energy from the pixel-level card cooling functional zone. In subsequent drive cycles, the recovered energy is reused for polarization excitation of the same card pixel unit or other target card pixel units through the B-A zone feedback branch or the B-C zone collaborative branch.

[0108] The sixth step involves the control unit dynamically adjusting the drive pixel set and recycling path based on the hotspot mapping. For areas with high hotspot intensity, the drive frequency, pulse amplitude, or drive duty cycle of the corresponding electrical card pixel unit is increased. For non-hotspot areas, the corresponding electrical card pixel unit is reduced or turned off to achieve on-demand cooling distribution and overall energy consumption optimization.

[0109] The above six steps provide a complete cooling and energy recovery operation method according to the embodiments of this application, which can enable the device to operate normally.

[0110] Furthermore, the card cooling device integrating energy recovery and pixel-level card cooling functional partitioning provided in this application embodiment can be mass-produced on the substrate layer using semiconductor processes such as thin film deposition, photolithography, etching, and electroplating, and is compatible with existing integrated circuit manufacturing processes. This device can be used for hotspot management at the chip, package, and module levels, and can also be extended to other micro-thermal management devices requiring synergistic optimization of local solid-state cooling and energy recovery (such as microprocessors, graphics processors, RF power amplifiers, laser diodes, and microelectromechanical systems devices), demonstrating clear industrial applicability.

[0111] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0112] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A card cooling device integrating energy recovery and pixel-level card cooling functional partitioning, characterized in that, include: basal layer; A pixel-level card cooling function partition is set on the base layer, including multiple independently controllable card pixel units, which are used to thermally couple with the hot spot area of ​​the target to be cooled and generate a cooling effect; An energy recovery and storage functional zone is set on the base layer and arranged coplanarly with the pixel-level card cooling functional zone. It is used to capture and temporarily store the electrical energy released during the depolarization stage of the pixel-level card cooling functional zone. A power supply excitation function partition is disposed on the base layer and arranged coplanarly with the pixel-level card cooling function partition, and is used to provide excitation signals to the pixel-level card cooling function partition. The interconnection layer electrically connects the pixel-level card cooling function partition, the energy recovery and storage function partition, and the power supply and excitation function partition. A bidirectional switch matrix includes at least an excitation branch, a recovery branch, and a feedback branch. The excitation branch connects the power supply excitation functional zone to the pixel-level card cooling functional zone. The recovery branch connects the pixel-level card cooling functional zone to the energy recovery and storage functional zone. The feedback branch connects the energy recovery and storage functional zone to the pixel-level card cooling functional zone. A control unit, connected to the bidirectional switch matrix, is configured to: turn on the excitation branch during the excitation phase to drive the polarization of the target card pixel unit; During the recycling phase, the recycling branch is activated to depolarize the target card pixel unit and guide the electrical energy released by the target card pixel unit into the energy recovery and storage functional zone; during the feedback phase, the feedback branch is activated to feed back the electrical energy stored in the energy recovery and storage functional zone to the same or another card pixel unit for subsequent polarization.

2. The card cooling device integrating energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, The power supply excitation function partition and the energy recovery and storage function partition operate in a cooperative working mode. The cooperative working mode includes: the excitation energy provided by the power supply excitation function partition includes electrical energy fed back by the energy recovery and storage function partition, so that the external power supply input only provides the reference energy difference per drive cycle, so as to increase the instantaneous effective drive voltage on the card pixel unit when the external power supply voltage is limited.

3. The card cooling device integrating energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, The equivalent capacitance of a single pixel unit in the pixel-level card cooling function zone and the equivalent energy storage capacitor configured in the energy recovery and storage function zone satisfy the following: ,in This refers to the equivalent capacitance of the individual card pixel unit. The equivalent energy storage capacitor configured therefor.

4. The card cooling device with integrated energy recovery and pixel-level card cooling function partitioning according to claim 3, characterized in that, The equivalent capacitance of a single card pixel unit in the pixel-level card cooling function zone and the equivalent energy storage capacitor configured in the energy recovery and storage function zone satisfy the following: .

5. The card cooling device integrating energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, A resonant transfer inductor is connected in series between the pixel-level card cooling function zone and the energy recovery and storage function zone. The resonant transfer inductor is used to control the peak discharge current and improve the charge transfer efficiency.

6. The card cooling device integrating energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, The control unit is further configured to: perform at least one of the following control strategies on the card pixel unit according to the hotspot mapping information: dynamic gating, drive frequency adjustment, drive duty cycle adjustment, excitation voltage amplitude adjustment, or recycling path switching; for areas with high hotspot intensity, increase the drive frequency, pulse amplitude, or drive duty cycle of the corresponding card pixel unit; for non-hotspot areas, reduce or turn off the corresponding card pixel unit.

7. The card cooling device integrating energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, The energy recovery and storage functional zoning can be any one of the following: a centralized public energy storage architecture, a grouped local energy storage architecture, or a distributed architecture in which each card pixel unit is independently configured with a micro energy storage unit.

8. The card cooling device integrating energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, The switching devices of the bidirectional switching matrix are any one or a combination of MOSFET, GaN device, IGBT, MEMS switch or relay.

9. The card cooling device integrating energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, The power supply excitation function partition is any one or a combination of charge pump circuit, resonant boost circuit, flyback boost circuit, half-bridge pulse drive circuit, full-bridge pulse drive circuit, or multi-stage series-parallel pulse network circuit.

10. The card cooling device with integrated energy recovery and pixel-level card cooling functional partitioning according to claim 1, characterized in that, A thermal management auxiliary structure is also provided above or below the pixel-level card cooling function partition to improve the directional heat flow transmission and suppress thermal crosstalk between adjacent card pixel units. The thermal management auxiliary structure is any one or a combination of a thermal diffusion layer, a heat sink layer, a micro thermal switch layer, a phase change interface layer, a thermal insulation trench, or a microstructure support layer.

Images