Heat Dissipating Structure and Heat Dissipating Method Thereof

Abstract

A heat dissipation structure and a heat dissipation method are disclosed. The heat dissipation structure includes a duct having a channel formed therein and an airflow generating device. The heat dissipation structure is disposed on, adjacent to, or above a semiconductor device and is configured to dissipate heat generated from the semiconductor device. The airflow generating device is configured to dissipate heat generated from the semiconductor device by generating an airflow that flows through the channel within the duct.

Description

Technology Field

[0001] This application claims the benefit of U.S. Provisional Application No. 63 / 690,314 filed September 4, 2024. Additionally, this application claims the benefit of U.S. Provisional Application No. 63 / 693,207 filed September 11, 2024. The contents of these applications are incorporated herein by reference.

[0002] The present invention relates to a heat dissipation structure and a heat dissipation method thereof, and more specifically, to a heat dissipation structure and a heat dissipation method thereof for cooling electronic products, etc. Background Technology

[0003] System-on-Chip (SoC) technology increases integration density but also concentrates heat in smaller areas. As electronic devices become smaller and more powerful due to the demands of AI, 5G, and advanced applications, cooling has become a critical challenge. Constrained space and water resistance requirements prevent the use of traditional active cooling (e.g., fans). Meanwhile, conventional passive cooling solutions, such as heat spreaders, vapor chambers, heat sinks, and advanced thermal materials, are limited to merely suppressing temperature. When heat cannot be effectively dissipated, reliance is placed on thermal management methods like throttling, which can cause a 50% performance degradation or even lead to device shutdown. As power density continues to rise, future compact devices (e.g., Dual Inline Memory Module (DIMM) cards) may require innovative active cooling solutions to ensure high performance and user safety.

[0004] Therefore, the main objective of the present application is to provide a heat dissipation structure and a heat dissipation method thereof to improve upon the disadvantages of the prior art.

[0005] One embodiment of the present application discloses a heat dissipation structure, wherein the heat dissipation structure comprises a duct—with a channel formed within the duct—and an airflow generating device; wherein the heat dissipation structure is disposed on a semiconductor device, or on a side of a semiconductor device, or on top of a semiconductor device, and is configured to dissipate heat generated from the semiconductor device; and the airflow generating device is configured to generate an airflow flowing through a channel within the duct to dissipate heat generated from the semiconductor device.

[0006] Another embodiment of the present application discloses a heat dissipation method, wherein the heat dissipation method comprises the steps of: forming a channel on top of a semiconductor device; and an airflow generating device generating an airflow through the channel, and the airflow dissipates heat generated from the semiconductor device.

[0007] These and other objects of the present invention will undoubtedly become apparent to those skilled in the art after reading the following detailed description of preferred embodiments illustrated in the various drawings and figures. Brief explanation of the drawing

[0008] FIGS. 1 to 3 are schematic diagrams of a heat dissipation structure according to one embodiment of the present invention. FIG. 4 is a side schematic diagram of a heat dissipation structure according to one embodiment of the present invention. FIGS. 5 and 6 are schematic cross-sectional views of a heat dissipation structure according to one embodiment of the present invention. FIG. 7 is a schematic diagram of a heat dissipation structure according to one embodiment of the present invention. Figure 8 is a schematic cross-sectional view of the heat dissipation structure shown in Figure 7. FIG. 9 is a schematic diagram of a heat dissipation structure according to one embodiment of the present invention. FIG. 10 is a schematic diagram of the wiring scheme for the AFG device shown in FIG. 5. FIG. 11 is a schematic diagram of the waveforms of the modulation signal and demodulation signal for the AFG device shown in FIG. 5. FIG. 12 is a schematic diagram of the differential mode movement and common mode movement of the AFG device shown in FIG. 5 in slow motion. FIG. 13 is a schematic diagram of an air pulse according to one embodiment of the present invention. FIGS. 14 to 16 are schematic diagrams of a heat dissipation structure according to one embodiment of the present invention. FIG. 17 is a schematic top view of a semiconductor device according to one embodiment of the present invention. Specific details for implementing the invention

[0009] The contents of U.S. Patent No. 12,356,141, U.S. Application No. 19 / 007,580, and U.S. Application No. 19 / 303,389 are incorporated herein by reference.

[0010] For heat dissipation, the present invention provides a heat dissipation structure comprising a duct and an airflow generating (AFG) device. The AFG device is operated to generate air pulses at an ultrasonic pulse rate in a direction toward or away from the opening of the duct. These air pulses produce a net airflow in a single direction, inducing airflow(s) within a channel formed within the duct. The duct is in contact with an electronic device (e.g., a memory module) and forms a heat transfer path from the electronic device to the surrounding environment.

[0011] To ensure a substantial airflow, an AFG device is introduced that may include a modulation means and a demodulation means. The modulation means is an ultrasonic carrier frequency. f UC It generates an ultrasonic air pressure wave / variation (UAW) having . The amplitude of the UAW is the input signal (S INIt is modulated according to ). This amplitude-modulated ultrasonic air pressure wave / variation (AMUAW) is subsequently synchronously demodulated by a demodulation means, so that the spectral components embedded in the AMUAW are integer multiples of the ultrasonic carrier frequency, ± n · f UC It is shifted by (where n is a positive integer). As a result of this synchronous demodulation, the spectral components of AMUAW are partially transferred to the baseband. In this way, the AFG device can be made compact while generating enough airflow or pressure to function as a (small) air pump or bladeless fan.

[0012] For example, FIG. 1 (a) and (b) are schematic top and side views, respectively, of a heat dissipation structure (10). The heat dissipation structure (10) includes AFG devices (100a, 100b) and a duct (or conduit) (150). The heat dissipation structure (10) is positioned adjacent to a memory module (190) (e.g., Dual Inline Memory Module (DIMM)) to transfer heat away. The duct (150) may or may not be a thermally conductive duct.

[0013] The AFG devices (100a and 100b) can initiate airflow to promote heat transfer. Specifically, the AFG device (100a) is operated to generate air pulses at an ultrasonic pulse rate toward the (first) opening of the duct (150). These air pulses generate a continuous net (first) airflow in a single direction (e.g., -Z). Similarly, the AFG device (100b) is operated to generate air pulses at an ultrasonic pulse rate in a direction away from the (second) opening of the duct (150). These air pulses generate a continuous net (second) airflow in a single direction (e.g., +Z). The (first and second) airflows (e.g., 500aF and 500bF in FIG. 5) can induce a (third) airflow within a channel (e.g., 555 in FIG. 5) of the duct (150).

[0014] Since the duct (150) is hollow to form a channel, the (third) airflow directed through the channel can cool the memory module (190). Specifically, the channel is connected to the AFG devices (100a and 100b) respectively through the (first and second) openings. The (first) airflow from the AFG device (100a) enters the channel through the (first) opening, introducing cold air (e.g., ambient temperature) into the channel. The cold air then flows through the channel and absorbs heat from the memory chip (191) of the memory module (190). The heated air is then drawn out of the channel through the (second) opening by the AFG device (100b), forming the (second) airflow.

[0015] FIG. 5 is a schematic cross-sectional view of a heat dissipation structure (50) according to one embodiment of the present invention. The cross-section of the heat dissipation structure (50) can be taken along the cross-sectional line (A-A') shown in FIG. 1.

[0016] The heat dissipation structure (10 and 50) can be implemented using a similar mechanism. For example, the AFG device (500a and 500b) may be placed directly on one surface of the duct (550), while the opposite surface of the duct (550) may be in contact with or close to a heat source(s) (e.g., a printed circuit board (PCB) (192) or a memory chip (191) of a memory module (190). In this way, cold air flowing through the duct (550) may be heated by the heat source(s), and the duct (550) and the AFG device (500a and 500b) send the heated air away from the heat source(s).

[0017] To facilitate heat dissipation, the duct (550) may be made of a thermally conductive material (e.g., copper, aluminum, pyrolytic graphite, or graphene). Thermal interface material(s) may be used to eliminate air gaps between the duct (550) and the heat source(s).

[0018] The duct (550) may be a long tube compared to a relatively small AFG device (500a or 500b) (e.g., length 10 to 15 millimeters, width 10 to 15 millimeters, thickness 2 to 3 millimeters). The size of the AFG device (500a or 500b) may be smaller than or equal to the size of a single heat source. The duct (550) may completely overlap the heat source(s) to increase the heat transfer area where heat can be exchanged.

[0019] The geometric features of the duct (550) may be associated with AFG devices (500a and 500b). For example, a channel (555) formed within the duct (550) includes openings (556a and 556b) for AFG devices (500a and 500b). Except for the openings (556a and 556b), the duct (550) has a closed, hollow structure. The ends (551 and 552) of the duct (550) may be designed to be closed because the AFG devices (500a and 500b) are located near the corresponding ends (551 and 552). Since the ends (551 and 552) are closed, air pushed or pulled by the AFG device (500a or 500b) mainly flows into or out of the channel (555) through the openings (556a and 556b). This push-pull configuration can facilitate heat dissipation. However, the invention is not limited thereto, and the duct (550) may otherwise be open at the ends (551 or 552).

[0020] An AFG device (500a) placed directly on a duct (550) at a position corresponding to an opening (556a) comprises a film structure (504a) (e.g., a membrane or diaphragm). The film structure (504a) comprises flaps (501a and 503a) located opposite each other. The operating principle of the AFG device (500a) is similar to that disclosed in U.S. Patent No. 11,943,585 B2, No. 12,317,034 B2 and U.S. Application No. 18 / 624,105, which are incorporated herein by reference. The flaps (501a and 503a) constituting the flap pair (502a) are ultrasonic carrier frequencies f UCIt is operated to perform a common mode movement to form an AMUAW with an ultrasonic opening rate (e.g., 192 or 96 kHz), which can be considered a modulation operation. Meanwhile, the flaps (501a and 503a) are also operated to perform a differential mode movement to form an opening or virtual valve (VV) with an ultrasonic opening rate (e.g., 192 or 96 kHz), which can be considered a demodulation operation.

[0021] A slit (512a) is formed between flap (501a) and flap (503a), and an opening or VV is created as a result of the slit (512a). In the present invention, the terms “slit,” “opening,” and “VV” share the same notation (e.g., 512a) because they share the same physical location and express similar concepts in other aspects. VV (512a) is intended to emphasize the ability to be controlled to open or close, and opening (512a) is intended to emphasize the state, particularly when it is open. By operating flaps (501a and 503a), the distance between the free ends of flaps (501a and 503a) increases, and an opening (512a) or VV (512a) is formed.

[0022] In the present invention, “flaps (501a and 503a) performing common mode movement” means that flaps (501a and 503a) are actuated to move in a common direction or actuated by a common driving signal (e.g., the modulation signal (SM) shown in FIG. 10). Furthermore, “flaps (501a and 503a) performing differential mode movement” means that flaps (501a and 503a) are actuated to move or bend in different / opposite directions relative to a common reference position or actuated by a differential pair of driving signals (e.g., the demodulation signals (+SV and -SV) shown in FIG. 10, but not limited thereto).

[0023] As differential mode movement (demodulation) and common mode movement (modulation) are performed simultaneously by the flap pair (502a), in-situ and simultaneous modulation-demodulation can be achieved through a specific wiring scheme. For example, as illustrated in FIG. 10, the AFG device (500a) may include an actuator (501aA) placed on flap (501a) and an actuator (503aA) placed on flap (503a). Each actuator (e.g., 501aA or 503aA) includes an upper electrode and a lower electrode. For example, FIG. 10 (a), (b), and (c) illustrate details of the circular area indicated by the dashed line in FIG. 5. As illustrated in FIG. 10 (a) and (b), the lower electrode of the actuator (501aA or 503aA) receives a modulation signal (SM), and the upper electrode of the actuator (501aA or 503aA) receives demodulation signals (+SV and -SV) having opposite polarity. An appropriate bias voltage (V BIAS ) can be applied to the bottom electrode shown in FIG. 10 (a) or the top electrode shown in FIG. 10 (b). As shown in FIG. 10 (c), one electrode of the actuator (501aA or 503aA) receives both the modulation signal (SM) and the demodulation signal (+SV or -SV) (but is not limited thereto), and the other electrode is appropriately biased.

[0024] Refer to FIG. 11 for the waveforms of the modulation signal (SM) and the demodulation signal (±SV) (or similar ones as shown in FIG. 11). In one embodiment shown in FIG. 11, the demodulation frequency (96 kHz) of the demodulation signal (±SV) may be half the modulation frequency (e.g., 192 kHz) of the modulation signal (SM). Specifically, the pulse polarity of the modulation signal (SM) is such that for a constant voltage, one operating cycle time (T CYIt alternates / toggles twice during ). Assuming that at a specific point in time, the demodulation signal (+SV) includes a first pulse having a first polarity for a constant / average voltage and the demodulation signal (-SV) includes a second pulse having a second polarity for a constant / average voltage, the first and second polarities are opposite, but the first and second pulses may have the same amplitude. The pulse polarity of the demodulation signal (+SV or -SV) is for one operating cycle time (T) for a constant / average voltage. CY It is alternated / toggled once during ). As a result, the flaps (501a and 503a) form an opening (512a) with an ultrasonic opening rate of 192 kHz, and the AFG device (500a) with an ultrasonic pulse rate of 192 kHz. f Pulse It generates air pulses. Ultrasonic carrier frequency f UC The operating cycle time (T CY ) is the ultrasonic pulse rate f Pulse It can be the reciprocal of, that is, T CY =1 / f Pulse am.

[0025] In practice, differential mode movement (demodulation) and common mode movement (modulation) may not occur in a time-division manner. Instead, at a given time point, the common mode displacement and differential mode displacement can be combined through the aforementioned wiring method to produce a net movement of the flap (501a or 503a). For example, FIG. 12 shows time point t 11 from t 17 An example of the (symmetrical movement) of the flap pair (502a) up to is illustrated, and the bottom corner of FIG. 11 is at the corresponding time point t 11 from t 17 An enlarged view of the top corner of Figure 11 up to is shown.

[0026] In Fig. 12, at time t 14 from t 17Until then, since flap (501a) moves upward and flap (503a) moves downward, VV (512a) is at time t 17 It is considered to be in an "open" state (i.e., the opening (512a) is formed) (and remains open thereafter). Similarly, the opening (512a) is at time t 11 It exists in the flap pair (502a) (and prior to that). This t 14 ~t 17 During the time period (or at time t) 11 The common mode movement of the flaps (501a and 503a) is effectively "made to disappear".

[0027] In Fig. 12, at time t 11 from t 14 Until then, since flap (501a) moves downward and flap (503a) moves upward, VV (512a) is considered to be in a "closed" state, which means that flaps (501a and 503a) are at this t 11 ~t 14 This means that it can be treated as a continuous membrane within the time period and behaves like a single (complete membrane) in terms of membrane movement. When VV (512a) is in the "closed" state, the difference in displacement between the free ends of the flaps (501a and 503a) is less than or equal to the thickness of the film structure (504a).

[0028] The "closed" state of VV (512a) occurs during the transition of the differential mode movement of the flaps (501a and 503a). Specifically, at the (first) transition time (e.g., t 11 ~t 17 At ), the flap (501a) driven by the demodulation signal (+SV) transitions motion from top to bottom; (second) transition time (e.g., t 11 ~t 17In ), the flap (503a) driven by the demodulation signal (-SV) transitions motion from bottom to top. In other words, VV (512a) is the transition time of the flaps (501a and 503a) (e.g., t 11 ~t 17 ) Transition time of the demodulated signal (-SV and +SV) (e.g., t 11 ~t 17 A subinterval within ) (e.g., t 13 ~t 15 It remains closed during ) - that is, the flaps (501a and 503a) move in opposite directions and the demodulation signals (-SV and +SV) increase / decrease in opposite directions. In short, when VV (512a) is closed, the flaps (501a and 503a) are in a moving state.

[0029] The direction of the forward airflow (500aF) can be controlled by adjusting the phase between the modulation signal (SM) and the demodulation signal (±SV). For example, in FIG. 12, the first transition time (t) of the demodulation signal (+SV) 11 ~t 17 ) occurs when the modulation signal (SM) is low. In this case, the AFG device (500a) can generate airflow (500aF) in one direction. When the demodulation signal (±SV) shifts and the transition time of the demodulation signal (±SV) matches the time interval when the modulation signal (SM) is high, the AFG device (500a) can instead generate airflow (500aF) in the opposite direction.

[0030] Alternatively, the direction of the forward airflow (500aF) may vary depending on the modulation signal (SM). Specifically, the modulation signal (SM) may be an input signal (S) that includes an alternating current (AC) component or a non-zero direct current (DC) voltage / offset. INIt can be generated according to ). The polarity of the DC offset may be related to the direction of the forward airflow (500aF). For example, FIG. 13 is a schematic diagram of an air pulse (AP) according to one embodiment of the present invention. During a time interval (T1), the air pulse (AP1) generated by the AFG device (500a) can generate a continuous (first) forward airflow in the (first) direction (D1) in response to the DC offset being positive. On the other hand, during a time interval (T2), the air pulse (AP2) generated by the AFG device (500a) can generate a continuous second forward airflow in the second direction (D2) opposite to the first direction (D1) in response to the DC offset being negative.

[0031] In other words, the AFG device (500a) can generate a unidirectional forward airflow (e.g., 500aF). Alternatively, the AFG device (500a) can change the direction of the airflow (500aF). However, the time interval (T1 or T2) (e.g., 0.5 seconds) is the operating cycle time (T CY ) or longer than the reciprocal of the minimum audible frequency (e.g., 10 Hz), and thus the (first or second) forward airflow generated by the air pulse (AP1 or AP2) can be considered continuous in a single direction (D1 or D2).

[0032] The strength of the net airflow (500aF) is controllable. Specifically, the strength of the net airflow (500aF) can be influenced by the magnitude of the modulation signal (SM). For example, the strength of the net airflow (500aF) can be a function of the DC offset. The strength of the net airflow (500aF) is the amplitude of individual air pulses (e.g., AP1 or AP2) (e.g., the peak value in FIG. 13). p 1, p 3, or p It can be determined by 5). The amplitude of the air pulse (AP1 or AP2) may vary from pulse to pulse or remain consistent throughout the pulse.

[0033] In FIG. 13, one operating cycle time (T CY The air pulses (e.g., AP1 or AP2) within are asymmetric. The degree of asymmetry is p 2nd generation p It can be evaluated at a ratio of 1, where p 1 > p It is 2. Here, p 1 represents the peak value of a first half-cycle pulse having a first polarity with respect to a reference level, and p 2 represents the peak value of a second half-cycle pulse having a second polarity relative to a reference level. This reference level may correspond to ambient conditions (e.g., ambient pressure or zero airflow).

[0034] Asymmetry of the air pulse (e.g., AP1 or AP2) may indicate the presence of low-frequency component(s) of the air pulse (AP1 or AP2) generated by the AFG device (500a). The greater the asymmetry, the stronger the baseband spectral component(s) of the air pulse (AP1 or AP2).

[0035] The AFG device (500a) can generate an asymmetric air pulse (AP1 or AP2) by aligning the opening timing of VV (512a) (in response to the demodulation signal (±SV)) with the acceleration timing of the common mode movement of the flaps (501a and 503a) (in response to the modulation-driven signal (SM)). Specifically, converting the symmetric UAW generated through the modulation operation into an asymmetric air pulse (e.g., AP1 or AP2) is the demodulation operation of the AFG device (500a). When the "open" period of VV (512a) overlaps with the time interval of one of the two polarities of the acceleration of the common mode flap movement, the AFG device (500a) must generate a single-ended (SE) or SE-like air pulse. Thus, as illustrated in FIG. 12, the transition time of the demodulation signal (±SV) may not coincide with the transition time of the modulation signal (SM). In other words, pulse asymmetry depends on the appropriate timing of opening VV (512a).

[0036] The opening of VV (512a) does not determine the strength of the air pulse (AP1 or AP2), but it does affect how strong the "net-zero pressure" effect is. When the opening of VV (512a) is wide, the "net-zero pressure" effect is more pronounced, auto-neutralization is completed, and asymmetry becomes more distinct, resulting in a significant baseband signal.

[0037] The AFG device (500a) can be configured / manufactured using different technologies depending on the application requirements. In FIG. 5, a chamber (515a) is defined between the cap structure (511a) and the film structure (504a). The film structure (504a), supported by the support structure (521a), can be manufactured using a Micro Electro Mechanical Systems (MEMS) fabrication process. A silicon (Si) substrate with a thickness of 250 to 500 micrometers can be etched to form the support structure (521a). On top of this Si substrate, a thin layer, typically 3 to 6 micrometers thick, made of silicon on an insulator (SOI) or poly on an insulator (POI), can be etched to form flaps (501a and 503a). A layer of piezoelectric material such as lead zirconate titanate (PZT) can be deposited on a pair of flaps (502a) to form actuators (501aA and (503aA).

[0038] The number of AFG device(s) can be adjusted for different scenarios. For example, FIG. 2 is a schematic diagram of a heat dissipation structure (20) according to one embodiment of the present invention. The heat dissipation structure (20) may include only one AFG device (200) located substantially close to one end of the duct (250) of the heat dissipation structure (20). The number of AFG devices (200) (i.e., 1) is less than the number of memory chips (291) of the memory module (290) (i.e., 4).

[0039] Even a single AFG device (i.e., 200) helps remove excess heat from the memory chip (291). Specifically, the AFG device (200) is operated to generate air pulses capable of continuously generating a net airflow (e.g., 600 bF) in the -Z or +Z direction. Thus, the AFG device (200) can draw cold air into the duct (250) or push hot air out of the duct (250). Consequently, the AFG device (200) placed directly on the duct (250) can move the heated air within the duct (250).

[0040] FIG. 6 is a schematic cross-sectional view of a heat dissipation structure (60) according to one embodiment of the present invention. The cross-section of the heat dissipation structure (60) may be taken along the cross-sectional line (B-B') shown in FIG. 2. The AFG device (600) or duct (650) of the heat dissipation structure (60) may be used to implement an AFG device (200) or duct (250).

[0041] A channel (655) formed in a duct (650) has openings (656a and 656b) to create an airflow path. The channel (655) is connected to an AFG device (600) through the opening (656b), so that a net airflow (600bF) driven by the AFG device (600) can flow out of (or into) the channel (655) through the opening (656b). Since the AFG device (600) is located near the end (652) of the duct (650), the end (652) and the opposite end (651) of the duct (650) can be designed to be closed and open, respectively. In response to the airflow (600bF), an airflow (600aF) can flow into (or out of) the channel (655) through the opening (656a) near the end (651).

[0042] The position of the AFG device can be adjusted for different scenarios. For example, FIG. 3 (a) and (b) are schematic diagrams of a heat dissipation structure (30), respectively. The heat dissipation structure (30) may include only one AFG device (300) positioned substantially close to the center of the duct (350) of the heat dissipation structure (30) to establish a center-pull configuration.

[0043] The center-draw configuration can induce airflow in different directions within the duct (350). For example, to discharge hot air from the duct (350), the AFG device (300) generates an air pulse capable of generating a continuous net airflow in the +Z direction. This net airflow forms an airflow (300bF') substantially directed from the AFG device (300) toward the periphery in the +Z direction and creates local low pressure inside the duct (350). This low pressure draws in cold ambient air through openings (356a and 356b) located at opposite ends (351 and 352) of the duct (350) to form airflows (300aF and 300cF). Thus, the air within the duct (350) can move substantially in the +X and -X directions across the channels of the duct (350) to absorb heat from the memory chip (391) of the memory module (390).

[0044] In FIG. 3, the AFG device (300) is top-firing, which means that the cap structure of the AFG device (300) (e.g., 611 in FIG. 3) has an opening formed at the top of the cap structure (e.g., 611p in FIG. 3). The top-firing AFG device (300) generates a net airflow in the +Z or -Z direction of the AFG device (300) where the flap moves substantially. In another embodiment, the AFG device may be side-firing, which means that the cap structure of the AFG device has an opening formed on the side wall of the cap structure.

[0045] For example, FIG. 7 (a) and (b) are schematic diagrams of a heat dissipation structure (70), respectively. FIG. 8 (a) and (b) are schematic cross-sectional diagrams of a heat dissipation structure (70) taken along the cross-sectional lines (CC' and D-D') shown in FIG. 7, respectively. The heat dissipation structure (70) may include a side-spraying type AFG device (700) and a duct (750).

[0046] As with the top-spraying AFG device (300) illustrated in FIG. 3, in FIG. 7, the direction of the airflow (700bF') toward the periphery from the side-spraying AFG device (700) is perpendicular to the direction of the airflow (700aF or 700cF) from the opening (756a or 756b) of the duct (750). For example, the AFG device (700) generates an air pulse capable of generating a continuous net airflow (700bF) in the +Z direction. This net airflow (700bF) forms an airflow (700bF') substantially directed toward the periphery from the AFG device (700) in the +Y direction. The local low pressure inside the AFG device (700) caused by the net airflow (700bF) draws hot air from the duct (750) into the AFG device (700). As a result, cold ambient air is forced into the duct (750) through the openings (756a and 756b) to form a substantially airflow (700aF or 700cF) in the ±X direction perpendicular to both the +Z and +Y directions.

[0047] In FIG. 8, the duct (750) has walls (W1 to W6) of approximately uniform thickness, but is not limited thereto. For example, wall (W3) may be thicker than wall (W4); otherwise, wall (W1 or W2) may be thinner than wall (W5 or W6).

[0048] In FIG. 8, the length (LN7) of the duct (750) in the X direction is greater than the width (WD7) in the Y direction or the thickness (TH7) in the Z direction. The ±X direction is the direction in which the memory chips (791) are aligned to form an array, and the ±Z direction is the main direction of movement of the flaps (701 and 703) of the AFG device (700). The ±X direction in which the airflow (700aF and 700bF) flows out through the openings (756a and 756b) is perpendicular to the +Y direction in which the airflow (700bF') can flow out through the opening (711p) of the AFG device (700).

[0049] Optionally, the length (LN7), width (WD7), or thickness (TH7) is the ultrasonic carrier frequency f UC Wavelength corresponding to l UC Multiples of, wavelength l UC An odd multiple of half, or wavelength l UC It is substantially different from an odd multiple of one-fourth of the. Optionally, the slit (712) between flap (701) and flap (703) is positioned so as not to align with the antinode(s) or node(s) of the resonance of the duct (750). Optionally, when not in operation (or in the “closed” state), the film structure (704) of the AFG device (700) is symmetric with respect to the symmetry plane XZ or YZ of the duct (750).

[0050] A single heat dissipation structure may be placed next to one or more memory modules. For example, FIG. 4(a) and (b) are schematic side cross-sectional views of a heat dissipation structure (40a and 40b) according to an embodiment of the present invention. A duct (450) of the heat dissipation structure (40a or 40b) is located between two memory modules (490a, 490b or 490c) that are opposite each other. An AFG device (400) of the heat dissipation structure (40a or 40b) is located between two adjacent memory chips (491b and 491c) on memory modules (490b and 490c). The surface of the duct (450) may come into contact with a PCB (492a, 492b, or 492c) (as shown in FIG. 4(a)) or a memory chip (491a, 491b, or 491c) (as shown in FIG. 4(b)).

[0051] A memory module may be placed next to one or more heat dissipation structures. For example, FIG. 9(a) is a side schematic view of heat dissipation structures (90a and 90c) according to one embodiment of the present invention. A memory module (990) may be sandwiched between heat dissipation structures (90a) and heat dissipation structures (90c).

[0052] AFG device(s) of a specific configuration (e.g., 900a and 900c) can add much more airflow to the memory module (990) based on the principles of fluid dynamics. For example, an airflow (900aF') induced by an air pulse from the AFG device (900a) of the heat dissipation structure (90a) and continuously directed in the +Z direction can draw in air from behind and entrain air from the side; similarly, an airflow (900cF') induced by an air pulse from the AFG device (900c) of the heat dissipation structure (90c) and also continuously directed in the +Z direction can push air from the front and entrain air from the side. In other words, the airflows (900aF' and 900cF') can induce an airflow (900eF) in the +Z direction, which further cools the area around the memory module (990). Likewise, the airflow (900bF' and 900dF') generated by the AFG device (900b) of the heat dissipation structure (90a) and the AFG device (900d) of the heat dissipation structure (90c) can induce an airflow (900fF) in the -Z direction. The induced airflow (900eF and 900fF) helps to dissipate heat.

[0053] The configuration of the AFG device(s) and duct can be modified according to the application scenario. For example, FIG. 9(b) is a side schematic diagram of a heat dissipation structure (90g) according to one embodiment of the present invention.

[0054] The duct (950g) of the heat dissipation structure (90g) and the AFG devices (900g, 900h, and 900i) are closely interconnected and mutually influence each other. For example, whether the end (951g or 952g) of the duct (950g) is closed may depend on the location of the nearby AFG device or the direction of the airflow of the AFG device. In one embodiment, the end (952g) may be open; thus, heated air may exit the heat dissipation structure (90g) through the adjacent AFG device (900i) or the opening of the duct (950g) (e.g., the opening of the end (952g)). Alternatively, the direction of the airflow caused by the AFG device (e.g., 900g) may be related to the location(s) of the AFG device(s) (e.g., 900g, 900h, or 900i) of the heat dissipation structure (90g), or the location of the opening(s) of the duct (950g) (e.g., the opening of the end (952g)).

[0055] The number, temperature, or power density of memory chips (e.g., 991) of a memory module (e.g., 990) can be a decisive factor. For example, the number (e.g., 3) of AFG devices (e.g., 900g, 900h, and 900i) of a heat dissipation structure (90g) may be a function of the number or power density of memory chips (e.g., 991) of a memory module (e.g., 990) placed adjacent to the heat dissipation structure (90g). The distance between two adjacent AFG devices (e.g., 900g and 900h) may be a function of the number or power density of memory chips located between them. The intensity of the airflow generated by the AFG device (e.g., 900g) may be controllable according to the temperature of the memory module; alternatively, the air pulse generated by the AFG device (e.g., 900g) may flow as a continuous or periodic flow to regulate the temperature of the memory module.

[0056] It should be noted that the chips described as memory chips above are for illustrative purposes only and are not limited to memory. Some of the chip(s) described above may be replaced with other types of (semiconductor) chip(s), and this is also within the scope of the present invention. For example, some of the chip(s) may be CPU / GPU (center / graphic processing unit), controller, DRAM, NAND flash, etc. The heat dissipation structure of the present invention may be placed on, adjacent to, or above such chip(s) to dissipate heat.

[0057] In FIG. 5, the structure and operation of the AFG device (500b) are similar to, but not limited to, the structure and operation of the AFG device (500a). For example, FIG. 9(c) is a schematic top view of a film structure (904m) and a film structure (904n) for a duct (950) of a heat dissipation structure (90) according to one embodiment of the present invention. As shown in FIG. 9(c), the film structures (904m and 904n) placed over the openings (956m and 956n) of the duct (950) have different configurations. This can help reduce resonance.

[0058] In FIG. 9(c), the orientation of the flap pair (e.g., 906m) of the AFG device (900m) is different from the orientation of the flap pair (e.g., 906n) of the AFG device (900n). For example, the plane of symmetry of the flap (e.g., 907m) of the AFG device (900m) is perpendicular to the plane of symmetry of the flap (e.g., 907n) of the AFG device (900n). Aside from the orientation, the structure and operation of the AFG devices (900m and 900n) may be the same or different from each other.

[0059] In one embodiment, the structure and operation of two adjacent pairs of flaps may be identical. For example, to form the pair of flaps (902m) of the AFG device (900m), two flaps (901m and 903m) opposite each other are actuated to move in opposite directions to create a VV between them. Similarly, an adjacent pair of flaps (906m) may also be actuated to form a VV between the flaps (907m and 905m), with the flap (905m) positioned next to the flap (903m) without a slit. Due to this similarity, all VVs of the AFG device (900m) may be closed simultaneously and likewise opened simultaneously. Because the adjacent flaps (903m and 905m) of the two adjacent flap pairs (902m and 906m) move in opposite directions while their bottom electrodes are electrically connected, current flows between the two adjacent flap pairs (902m and 906m), which contributes to a reduction in overall power consumption.

[0060] In one embodiment, the structure and operation of two adjacent flap pairs may differ from each other. For example, the flap pair (902n) of the AFG device (900n) may generate a (first) air pulse toward the opening (956n) in response to a demodulation signal and a modulation signal, whereas the flap pair (906n) of the AFG device (900n) may generate a (second) air pulse toward the same opening (956n) in response to different demodulation signals and different modulation signals. The demodulation signal for the flap of the flap pair (902n) (e.g., 901n) may be a delayed version of the demodulation signal for the flap of the flap pair (906n) (e.g., 905n) (e.g., operation cycle time (T CY T, which is half of ) CY(Delayed by / 2). Furthermore, the modulation signal of the flap pair (902n) can be considered as the inverse or polarity-inverted version of the modulation signal of the flap pair (906n). Accordingly, the first air pulse and the second air pulse can be interleaved in time with each other to increase (e.g., double) the pulse rate.

[0061] In one embodiment, the operation of two flap pairs in different AFG devices may differ. For example, the demodulation or modulation signal for the flap (e.g., 901m) of flap pair (902m) may be a delayed version of the demodulation or modulation signal for the flap (e.g., 901n) of flap pair (902n).

[0062] FIG. 14 or FIG. 15 is a schematic diagram of a heat dissipation structure (A0 / B0) according to an embodiment of the present invention. The heat dissipation structure (A0 / B0) includes a duct (A50 / B50) and an AFG device (A00 / B00). A channel (or air passage) is formed inside the duct (A50 / B50). The AFG device (A00 / B00) generates an airflow that flows through the channel. Specifically, the AFG device (A00 / B00) draws in cold air through the channel formed within the duct (A50 / B50) and discharges hot or warm air.

[0063] The heat dissipation structure (A0 / B0) may be disposed on the semiconductor device (A90 / B90), on the side of the semiconductor device (A90 / B90), or on the top of the semiconductor device (A90 / B90). The airflow generated by the AFG device (A00 / B00) is configured to dissipate heat generated from the semiconductor device (A90 / B90). In one embodiment, the semiconductor device (A90 / B90) may include a Flip Chip-Ball Grid Array (FCBGA) package, but is not limited thereto.

[0064] FIG. 16 is a schematic diagram of a heat dissipation structure (C0) according to an embodiment of the present invention. The heat dissipation structure (C0) includes an AFG device (C00) disposed on a semiconductor device (C90) which may also include an FCBGA package, on a side of the semiconductor device (C90), or on top of the semiconductor device (C90). The duct of the heat dissipation structure (C0) is omitted in FIG. 16. In FIG. 16, the AFG device (C00) is molded onto the top of the semiconductor device (C90) through a molding material (C02).

[0065] In FIGS. 14 to 16, the front surface of the AFG device (A00 / B00 / C00) is positioned toward the first direction (-Z) and / or the semiconductor device (A90 / B90 / C90), and the rear surface of the AFG device (A00 / B00 / C00) is positioned toward the second direction (+Z), which is opposite to the first direction. Additionally, the AFG device can discharge hot air in a direction parallel to the +Z direction or in a direction perpendicular to the +Z direction, where the Z direction is perpendicular to the front / rear surface (parallel to the XY plane) of the AFG device.

[0066] FIG. 17 is a schematic top view of a semiconductor device (D90) according to an embodiment of the present invention. The semiconductor device (D90) may be an SSD (solid state drive, which can also be considered as a type of memory module) comprising a controller (D92) and a flash memory (D94), or may include the same. The flash memory (D94) may be a NAND flash memory. Optionally, the SSD (D90) may include a DRAM (D96) (DRAM: dynamic random access memory).

[0067] The heat dissipation structure of the present invention may be placed on a semiconductor device or SSD (D90), on a side of a semiconductor device or SSD (D90), or on top of a semiconductor device or SSD (D90) to dissipate heat generated by a controller (D92), a DRAM (D96), or a flash memory (D94), and this will facilitate the operation of the SSD (D90).

[0068] The use of ordinal terms such as "first" and "second" does not, in itself, imply that one element has priority, precedence, or order over another, the temporal order in which actions of a method are performed, or the necessity for all elements to exist simultaneously; these terms are simply used as labels to distinguish one element with a specific name from another element with the same name.

[0069] The term "substantially" generally implies that small deviations may or may not exist. For example, the terms "substantially parallel" or "substantially aligned" indicate that the angle between two components may be below a specific threshold (e.g., 5, 1, or 0.1 degrees). The term "substantially aligned" indicates that the deviation between two components may be below a specific threshold (e.g., 1 or 0.1 micrometers or milliseconds). The term "substantially identical" indicates that the deviation is within a specific percentage (e.g., 5%, 1%, or 0.1%).

[0070] The technical features described in the embodiments of the present invention may be mixed or combined in various ways as long as they do not conflict with one another.

[0071] The heat dissipation structure of the present invention can facilitate direct-to-chip cooling by using micro-ducting for cold air intake or hot air exhaust. Furthermore, the heat dissipation structure can improve heat removal from heat sinks, spreaders, or vapor chambers. Due to its high heat dissipation capability, the present invention can help reduce reliance on expensive materials and avoid performance throttling.

[0072] In summary, the AFG device for the heat dissipation structure of the present invention is compact and can generate significant airflow. By effectively moving air across a single component or chip at the device level, the present invention has the potential to change conventional approaches to thermal management.

[0073] Those skilled in the art will readily observe that numerous modifications and changes to the device and method may be made while maintaining the teachings of the present invention. Accordingly, the disclosure should be interpreted as being limited only to the spirit and scope of the appended claims.

Claims

Claim 1 A heat dissipation structure comprising: a duct—in which a channel is formed within the duct—; and an airflow generating device, wherein the heat dissipation structure is disposed on a semiconductor device, or on the side of a semiconductor device, or on the top of a semiconductor device, and is configured to dissipate heat generated from the semiconductor device, wherein the airflow generating device is configured to generate an airflow flowing through a channel within the duct to dissipate heat generated from the semiconductor device, wherein the airflow generating device includes a pair of flaps, wherein the pair of flaps is configured to receive a modulation signal and perform a common mode movement, and wherein the pair of flaps is configured to receive a demodulation signal and perform a differential mode movement. Claim 2 A heat dissipation structure according to claim 1, wherein the length of the duct in the first direction is greater than the width or thickness of the duct, and the air within the duct moves substantially in the first direction. Claim 3 In claim 1, the semiconductor device comprises a heat dissipation structure including a memory module. Claim 4 In paragraph 3, the duct has a first surface and a second surface opposite to the first surface, the airflow generating device is placed directly on the first surface, and the second surface is in contact with the board or memory chip of the memory module, a heat dissipation structure. Claim 5 In paragraph 3, the duct is a heat dissipation structure located between two memory modules positioned on opposite sides. Claim 6 In claim 1, the semiconductor device includes a controller and a flash memory, and the duct is a heat dissipation structure that overlaps with the controller or the flash memory. Claim 7 In claim 1, the semiconductor device further includes a DRAM (dynamic random-access memory), and the duct is a heat dissipation structure that overlaps with the DRAM. Claim 8 In claim 1, the semiconductor device comprises a plurality of chips, and the duct overlaps with the plurality of chips, forming a heat dissipation structure. Claim 9 In claim 1, the semiconductor device comprises a heat dissipation structure including an FCBGA (Flip Chip-Ball Grid Array) package. Claim 10 In claim 1, the semiconductor device comprises a heat dissipation structure including a solid state drive (SSD). Claim 11 A heat dissipation structure comprising a plurality of airflow generating devices in claim 1. Claim 12 A heat dissipation structure according to claim 1, comprising: a first airflow generating device configured to generate a first airflow toward the channel; and a second airflow generating device configured to generate a second airflow away from the channel. Claim 13 In claim 1, the airflow generating device comprises a film structure, and the film structure is configured to operate to generate a plurality of air pulses at an ultrasonic pulse rate, a heat dissipation structure. Claim 14 In claim 13, the film structure comprises a pair of flaps, wherein the pair of flaps comprises a first flap and a second flap opposite each other, and the pair of flaps is configured to perform the differential mode movement and form a virtual valve or an opening at an ultrasonic opening rate synchronized with the ultrasonic pulse rate, a heat dissipation structure. Claim 15 In claim 14, the virtual valve is a heat dissipation structure that closes within a period corresponding to the first transition time of the first flap and the second transition time of the second flap. Claim 16 A heat dissipation structure according to claim 1, wherein the modulation frequency of the modulation signal is twice the demodulation frequency of the demodulation signal. Claim 17 In claim 1, the above duct is a heat dissipation structure that is a thermally conductive duct. Claim 18 In claim 1, the duct is a heat dissipation structure including a first opening. Claim 19 In claim 1, the airflow generating device is a heat dissipation structure molded onto the upper part of the semiconductor device through a molding material. Claim 20 A heat dissipation structure according to claim 1, wherein the front surface of the airflow generating device is positioned toward a first direction and toward the semiconductor device, and the rear surface of the airflow generating device is positioned toward a second direction opposite to the first direction. Claim 21 In claim 1, the airflow generating device is a heat dissipation structure that discharges air in a direction perpendicular or parallel to the front of the airflow generating device. Claim 22 A heat dissipation method comprising: a step of forming a channel on top of a semiconductor device; and a step of an airflow generating device generating an airflow through the channel, wherein the airflow dissipates heat generated from the semiconductor device, and wherein the airflow generating device comprises a pair of flaps, wherein the pair of flaps receives a modulation signal and performs a common mode movement, and the pair of flaps receives a demodulation signal and performs a differential mode movement. Claim 23 A heat dissipation method according to claim 22, comprising the step of forming a duct, wherein the channel is formed within the duct. Claim 24 In paragraph 23, the above duct is a heat-conducting duct, a heat dissipation method. Claim 25 delete

Images