Thinning device and method

Abstract

The invention relates to a thinning device and method, and the device comprises a grinding disc which is used for bearing a wafer, and the grinding disc can rotate around a shaft; the grinding wheel is arranged above the grinding disc and is used for feeding and acting on the wafer in the vertical direction so as to grind the wafer; the magnetic fluid layer is positioned below the grinding disc and is connected to the grinding disc; establishing a damage function model of wafer damage; and based on the collected vibration acceleration signal a (t) and the damage function model, the rotating speed w of the grinding wheel, the flow velocity vt of the grinding fluid, the supply flow Qt of the grinding fluid, the spraying angle theta j of the grinding fluid, the feeding pressure Pg of the grinding wheel, the rotating speed wcarrier of the grinding disc, the adjusting pressure Pm of the magnetic fluid and the supporting rigidity Km of the magnetic fluid are determined and applied to next wafer machining. The wafer damage layer is reduced by adjusting parameters, and the wafer quality is improved.

Description

Technical Field

[0001] The present application relates to the field of semiconductor wafer W processing, and in particular to a thinning device and method. Background Art

[0002] In the semiconductor manufacturing process, wafer W thinning is a crucial link, which directly affects the smooth progress of subsequent processes such as packaging and interconnection, as well as the performance of the final product.

[0003] In the prior art, during the thinning process, the wafer W is thinned by grinding with the grinding wheel 200. However, the grinding wheel 200 acting on the wafer W will cause unavoidable vibration, and the vibration will cause a position deviation between the wafer W and the grinding wheel 200. It is difficult to offset the effect of the vibration on the wafer W by controlling and adjusting the grinding wheel 200 or the grinding disc 100 alone, resulting in a thicker processing damage layer on the wafer W and poor thinning quality of the wafer W.

[0004] Therefore, the technical problem of the prior art is that the quality of the wafer W is poor. Summary of the Invention

[0005] The present application provides a thinning device and method, which can reduce the wafer damage layer and improve the wafer quality by adjusting parameters.

[0006] On the one hand, the present application provides a thinning method, which adopts the following technical solution:

[0007] A thinning method, applicable to a thinning device, the thinning device comprising:

[0008] A grinding disc, which is used to carry the wafer and can rotate around an axis;

[0009] A grinding wheel is disposed above the grinding disc and is used to feed the wafer vertically to grind the wafer; and

[0010] a magnetic fluid layer, the magnetic fluid layer being located below the grinding disc and connected to the grinding disc;

[0011] The thinning method comprises:

[0012] Establish vibration acceleration signal a(t) and eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m Damage function model for wafer damage;

[0013] The eddy current influence factor Γ v Including grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θj ;

[0014] The magnetic fluid influence factor Φ m Including grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0015] Based on the collected vibration acceleration signal a(t) and the damage function model, the grinding wheel speed w and the grinding fluid flow rate v are determined. t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m , and applied to the next wafer processing.

[0016] As a preference, the grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m "include:

[0017] Collect vibration acceleration signal a(t);

[0018] Adjust the grinding wheel speed w and grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0019] Minimize the damage function model iteratively until it converges to a threshold;

[0020] Determine the grinding wheel speed w and grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m .

[0021] Preferably, the damage function model is:

[0022]

[0023] Where k1 is the eddy current influence factor Γ v The control coefficient of

[0024] k2 is the grinding wheel feed pressure P g The control coefficient of

[0025] k3 is the magnetic fluid influence factor Φ m The control coefficient of .

[0026] As an advantage, it also includes:

[0027] Establish the grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j About the Eddy Current Influence Factor Γ v Eddy current function model;

[0028] Based on the collected grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j And the eddy current function model determines the eddy current influence factor Γ v .

[0029] Preferably, the eddy current function model is:

[0030]

[0031] Where ρ is the grinding fluid density;

[0032] η is the grinding fluid viscosity; η0 is the reference viscosity;

[0033] R g is the grinding wheel radius;

[0034] θ j The angle between the grinding fluid spray direction and the wafer normal.

[0035] As an advantage, it also includes:

[0036] Establish grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m About the influence factor Φ of magnetic fluid m The magnetic fluid function model;

[0037] Based on the grinding wheel speed wcarrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m And the magnetic fluid function model determines the magnetic fluid influence factor Φ m .

[0038] Preferably, the magnetic fluid function model is:

[0039]

[0040] Among them, K m is the support stiffness of the magnetic fluid layer;

[0041] R p is the radius of the grinding disc;

[0042] ρ m is the density of magnetic fluid.

[0043] On the other hand, the present application provides a thinning device, which adopts the following technical solution:

[0044] A thinning device, comprising:

[0045] A grinding disc, which is used to carry the wafer and can rotate around an axis;

[0046] A grinding wheel is disposed above the grinding disc and is used to feed the wafer vertically to grind the wafer; and

[0047] a magnetic fluid layer, the magnetic fluid layer being located below the grinding disc and connected to the grinding disc;

[0048] Also includes:

[0049] Function building module, used to establish vibration acceleration signal a(t), eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m Damage function model for wafer damage;

[0050] The eddy current influence factor Γ v Including grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j ;

[0051] The magnetic fluid influence factor Φ m Including grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0052] Parameter confirmation module, used to determine the grinding wheel speed w and grinding fluid flow rate v based on the collected vibration acceleration signal a(t) and damage function model t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m .

[0053] Preferably, the function establishment module further includes:

[0054] The first function establishment module is used to establish the grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j About the Eddy Current Influence Factor Γ v Eddy current function model;

[0055] The second function establishment module is used to establish the grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m About the influence factor Φ of magnetic fluid m The magnetic fluid function model.

[0056] Preferably, the parameter confirmation module further includes:

[0057] The first parameter confirmation module is used to confirm the grinding wheel speed w and grinding fluid flow rate v based on the collected data. t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j And the eddy current function model determines the eddy current influence factor Γ v To confirm the grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j ;

[0058] The second parameter confirmation module is used to confirm the grinding wheel feed pressure P g ;as well as

[0059] The third parameter confirmation module is used to confirm the rotation speed of the grinding wheel based on the carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m And the magnetic fluid function model determines the magnetic fluid influence factor Φ m , to confirm the grinding wheel speed w carrier, Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m .

[0060] In summary, this application includes at least one of the following beneficial technical effects:

[0061] By taking the eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m By introducing the damage function F and combining it with the orthogonal experimental calibration coefficient and gradient descent optimization, the process parameters can be dynamically controlled, which can significantly reduce the sub-surface damage of the wafer and improve the quality of the thinned wafer. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a first schematic diagram of the thinning device described in this application;

[0063] Figure 2 It is a schematic flow chart of the thinning method described in this application;

[0064] Figure 3 It is a method for determining the eddy current influence factor in the thinning method described in this application;

[0065] Figure 4 It is a method for determining the magnetic fluid influence factor in the thinning method described in this application;

[0066] Figure 5 This is the second schematic diagram of the thinning device described in this application.

[0067] Description of reference numerals:

[0068] 100, grinding disc; 200, grinding wheel; 300, magnetic fluid layer; 400, function establishment module; 410, first function establishment module; 420, second function establishment module; 500, parameter confirmation module; 510, first parameter confirmation module; 520, second parameter confirmation module; 530, third parameter confirmation module; W, wafer. DETAILED DESCRIPTION

[0069] The serial numbers assigned to the components herein, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning. The "connection" and "coupling" mentioned in this application include direct and indirect connections (couplings) unless otherwise specified. In the description of this application, it should be understood that the orientations or positional relationships indicated by the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", etc. are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to this application.

[0070] In this application, unless otherwise expressly specified or limited, when a first feature is "above" or "below" a second feature, it may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Furthermore, when a first feature is "above," "above," or "above" a second feature, it may mean that the first feature is directly above or diagonally above the second feature, or simply means that the first feature is at a higher level than the second feature. When a first feature is "below," "below," or "below" a second feature, it may mean that the first feature is directly below or diagonally below the second feature, or simply means that the first feature is at a lower level than the second feature.

[0071] The embodiments of the present application provide a thinning device and method, which can reduce the damaged layer of a wafer W and improve the quality of the wafer W by adjusting parameters.

[0072] In order to better understand the above technical solution, the following will be described in detail with reference to the accompanying drawings and specific implementation methods. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0073] In the wafer W thinning process, grinding with a grinding wheel 200 is a key step in achieving ultra-thin wafer W. However, the grinding wheel 200 spindle inevitably vibrates during operation due to factors such as bearing play, dynamic balancing errors, and motor electromagnetic pulsation. This can lead to dynamic errors between the grinding wheel 200 and the wafer W. This often requires adjusting the relative position of the grinding wheel 200 and wafer W to compensate for this.

[0074] To compensate for vibration errors, the equipment needs to adjust the relative position of the grinding wheel 200 and the wafer W in real time. However, during the dynamic adjustment process, the fluctuation of the distance between the grinding wheel 200 and the wafer W will significantly affect the flow state of the grinding fluid. In the scenario where the wafer W rotates, the grinding fluid is prone to form vortices, and their intensity and distribution are directly controlled by the distance between the grinding wheel 200 and the wafer W. When the distance decreases, the dynamic pressure of the fluid in the grinding area increases to form a stagnation zone, resulting in a decrease in local flow velocity and an increase in temperature, which weakens the cooling and chip removal efficiency. When the distance increases, although the flow velocity increases, the increase in vortex intensity will cause uneven force on the surface of the wafer W, forming a coupled disturbance of vibration and flow field. Furthermore, when adjusting the position of the grinding wheel 200, the grinding wheel speed must be adjusted. However, the speed is closely related to the disturbance effect of the grinding fluid, which will also have a significant impact on the vortex situation of the grinding fluid.

[0075] The closed-loop effect of "vibration-flow field-quality" is amplified step by step like a butterfly effect: the initial vibration changes the flow pattern of the grinding fluid, causing distortion of the local temperature and stress fields, ultimately leading to larger surface defects or thickness deviations on wafer W.

[0076] This application proposes a thinning method, such as Figure 1 As shown, it is suitable for a thinning device, which includes a grinding disc 100, a grinding wheel 200 and a magnetic fluid layer 300. The grinding disc 100 is used to carry the wafer W, and the grinding disc 100 can rotate around the axis; the grinding wheel 200 is arranged above the grinding disc 100, and the grinding wheel 200 is used to feed the wafer W in the vertical direction to grind the wafer W; the magnetic fluid layer 300 is located below the grinding disc 100 and is connected to the grinding disc 100. It can be understood that the wafer W is arranged on the grinding disk 100, and the grinding wheel 200 located above acts on the wafer W. The grinding wheel 200 grinds the wafer W to perform a thinning operation. During the thinning process, the grinding wheel 200 can feed the wafer W axially; the magnetic fluid layer 300 is arranged below the grinding disk 100. The magnetic fluid layer 300 has a magnetic fluid structure, and the state of the magnetic fluid layer 300 can be adjusted by controlling the magnetic field, so that the force of the magnetic fluid layer 300 acting on the grinding disk 100 can be changed, that is, fine and precise position adjustment can be achieved between the grinding disk 100 and the grinding wheel 200.

[0077] Furthermore, the thinning device further includes a nozzle for supplying grinding fluid to the grinding disc 100 or the wafer W, and the angle between the spray direction of the grinding fluid and the tangential direction of the grinding wheel 200 is θ j , which is defined as the grinding fluid injection angle; an angle encoder can be set on the nozzle to detect θ j ; Grinding fluid velocity v can be measured on the nozzle t , Grinding fluid supply flow Q t .

[0078] A grinding force measuring sensor is provided on the main shaft of the grinding wheel 200 to detect the grinding wheel feed pressure Pg ; A magnetoelectric sensor or a Hall effect sensor is provided on the main shaft of the grinding wheel 200 for detecting the grinding wheel speed w; a piezoelectric sensor is provided on the main shaft of the grinding wheel 200 for collecting the vibration acceleration signal a(t).

[0079] A pressure sensor is provided between the magnetic fluid layer 300 and the grinding disc 100 to detect the magnetic fluid regulating pressure P of the magnetic fluid layer 300 on the grinding disc 100. m The rotating shaft of the grinding disc 100 is provided with a rotary encoder, a magnetoelectric sensor or a Hall effect sensor for detecting the rotation speed w of the grinding disc 100 carrier .

[0080] like Figure 2 As shown, the thinning methods include:

[0081] S1.1: Vibration acceleration signal a(t), eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m Damage function model for wafer W damage;

[0082] Among them, the eddy current influence factor Γ v Including grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j ;Magnetic fluid influence factor Φ m Including grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0083] S1.2: Determine the grinding wheel speed w and grinding fluid flow rate v based on the collected vibration acceleration signal a(t) and the damage function model t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0084] S1.3: And applied to the next wafer W processing.

[0085] Specifically, first establish a damage function model for wafer W, which includes the grinding wheel speed w, the grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j, Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0086] The vibration acceleration signal a(t) of the main shaft of the grinding wheel 200 during the machining process is collected in real time by the sensor;

[0087] The gradient descent algorithm is used to iteratively minimize the damage function and dynamically adjust the grinding wheel speed w and the grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0088] When the damage function converges to the threshold, the current parameters of the grinding wheel speed w and the grinding fluid flow rate v are set. t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ; and apply the currently set parameters to the processing of the next wafer W.

[0089] The damage during wafer W thinning is primarily due to the nonlinear coupling effect of mechanical vibration energy and process parameters. The damage function is designed to quantify instantaneous damage accumulation and reflect the total damage during the entire processing cycle through integration. The basis for this function is the coupling of vibration energy and process parameters. Vibration energy is the square of the vibration acceleration signal a(t), reflecting the instantaneous mechanical impact energy and is positively correlated with the crack growth rate. Process parameter coupling refers to the eddy current influence factor Γ. v , Grinding wheel feed pressure P g , magnetic fluid influence factor Φ m , representing the contributions of fluid disturbance, mechanical load and magnetorheological effect respectively.

[0090] Specifically, the damage function model adopts the form of time domain integration:

[0091]

[0092] Where k1 is the eddy current influence factor Γ v The control coefficient of

[0093] k2 is the grinding wheel feed pressure P g The control coefficient of

[0094] k3 is the magnetic fluid influence factor Φ m The control coefficient of

[0095] This application captures the damage accumulation effect within the entire processing time [t1, t2] rather than the instantaneous value; the vibration acceleration signal a(t) is coupled with the parameter (1+Σk i · parameter) to characterize the nonlinear relationship between vibration energy and process parameter modulation; through the weighting coefficient k i Quantified eddy current influence factor Γ v , Grinding wheel feed pressure P g , magnetic fluid influence factor Φ m contribution.

[0096] Wherein, a(t) is the vibration acceleration signal, which can be measured by a piezoelectric accelerometer (model PCB 356A15) installed on the spindle flange of the grinding wheel 200;

[0097] Γ v As the eddy current influencing factor, a mathematical model to quantify the damage of grinding fluid flow to wafer W is established, and the formula is:

[0098]

[0099] P g is the grinding wheel feed pressure, which can be measured by a pressure sensor (Honeywell26PC) integrated into the hydraulic circuit of the grinding wheel 200 feed mechanism;

[0100] Φ m is the influence factor of magnetic fluid, combined with the grinding disk speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m , the formula is:

[0101] Furthermore, the eddy current influence factor Γ v The control coefficient k1, grinding wheel feed pressure P g The control coefficient k2 and the magnetic fluid influence factor Φ m The control coefficient k3 can be obtained through calibration.

[0102] Specifically, a five-factor four-level orthogonal experiment was designed to cover the following parameter ranges:

[0103] Γ v ∈[0.5,2.0],

[0104] P g∈[0.1,0.3]MPa,

[0105] Φ m ∈[0.8,2.5],

[0106] Q l ∈[1.0,5.0]L / min,

[0107] θ j ∈[0°,60°];

[0108]

[0109] Number of experimental groups: L 32 (4 5 ) orthogonal table, a total of 32 groups of experiments, covering the combined effects of all parameters.

[0110] A 12-inch single-crystal silicon wafer was used, and each set of experiments was repeated three times to eliminate random errors.

[0111] The subsurface damage depth d was measured using a laser confocal microscope (Keyence VK-X1000). ssd , take the average value of 5 points in the radial direction of wafer W (center, R / 4, R / 2, 3R / 4, edge);

[0112] Experimental data:

[0113]

[0114]

[0115] Build ssd With Γ v 、P g and Φ m Quantitative relationship, extracting eddy current influence factor Γ v The control coefficient k1, grinding wheel feed pressure P g The control coefficient k2 and the magnetic fluid influence factor Φ m The control coefficient k3.

[0116] Build a linear regression model:

[0117] d ssd =k1Γ v +k2P g +k3Φ m +ξ

[0118] Among them, ξ is the random error term;

[0119] Matrix form:

[0120] d=Xk+ξ

[0121] Where d is the 32×1 damage thickness vector;

[0122] X is a 32×3 design matrix (columns corresponding to Γ v 、P g and Φ m );

[0123] k is a 3×1 coefficient vector [k1, k2, k3] T ;

[0124] Least squares solution:

[0125] k=(X T X) -1 X T d

[0126] Input experimental data matrix X and observation vector d; calculate X T X and X T d; Find the inverse matrix (X T X) -1 ; Solve to get:

[0127] k1=0.15±0.02, k2=0.08±0.01, k3=0.12±0.02, (R 2 =0.96).

[0128] k1=0.15 represents the eddy current influence factor Γ v For every 1 unit increase, the damage thickness increases by 0.15 μm, which reflects the effect of the grinding fluid eddy current on the damage of wafer W.

[0129] k2=0.08 represents the grinding wheel feed pressure P g For every 0.1MPa increase, the damage thickness increases by 0.08μm, reflecting the effect of mechanical stress (from grinding wheel 200) on the damage of wafer W;

[0130] k3=0.12 represents the magnetic fluid influence factor Φ m For every 1 unit increase, the damage thickness increases by 0.12μm; the effect of magnetorheological fluid on wafer W damage.

[0131] Further, such as Figure 3 As shown, the thinning method proposed in this application also includes:

[0132] S2.1: Establish grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j About the Eddy Current Influence Factor Γ v Eddy current function model;

[0133] S2.2: Based on the collected grinding wheel speed w and grinding fluid flow rate v t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j And the eddy current function model determines the eddy current influence factor Γ v .

[0134] The eddy current function model is:

[0135]

[0136] Where ρ is the grinding fluid density;

[0137] η is the grinding fluid viscosity; η0 is the reference viscosity;

[0138] R g The radius of the grinding wheel is 200;

[0139] θ j The angle between the grinding fluid jet direction and the tangent direction of the grinding wheel 200.

[0140] This embodiment establishes a mathematical model to quantify the damage caused by grinding fluid flow to wafer W, linking controllable process parameters (rotational speed, flow rate, spray angle, etc.) with the surface damage mechanism of wafer W. During the thinning process, the rotation of the grinding wheel 200 drives the centrifugal diffusion of the grinding fluid, forming a dynamic flow field. Shearing and impact momentum transfer occur between the grinding fluid and the grinding wheel 200 / wafer W. Grinding fluid viscosity affects flow resistance and energy dissipation. By correlating these relevant process parameters, the wafer W damage is determined.

[0141] The eddy current function model is:

[0142]

[0143] Density Corrected Centrifuge-Ejection Term , quantify the combined effect of centrifugal flow field and jet impact:

[0144] Centrifugal flow rate: the centrifugal acceleration a generated by the grinding wheel speed w c =w 2 R g , drives the grinding fluid to diffuse outward, the flow rate v cent ∝wR g ;

[0145] Jet momentum: grinding fluid velocity v t and grinding fluid supply flow rate Q t The product Q t ·v t Characterizes fluid momentum; sinθ j is the effective perpendicular component of the spray direction and the normal to the wafer W;

[0146] Density correction: Grinding fluid density ρ determines inertial force and centrifugal impact force F impact ∝w·ρ·Q t ·v t sinθ j ;

[0147] Normalization: Divide by the square of the grinding wheel radius 200 , eliminating the size effect.

[0148] Shear momentum term In the figure, the damage mechanism reflects the combined shear force and fluid momentum:

[0149] Shear rate: Shear rate at the interface between the grinding wheel 200 and the grinding fluid Where h is the thickness of the liquid film;

[0150] Momentum coupling: Experiments show that shearing and momentum transfer have a nonlinear relationship. Comprehensively characterize the synergistic effect of grinding wheel speed and fluid velocity.

[0151] Viscosity drag term quantify the effect of viscosity on flow resistance and energy loss:

[0152] Viscous dissipation: The increase in grinding fluid viscosity η leads to an increase in flow resistance and dissipated power

[0153] Experimental fitting: Viscosity resistance and w in actual process 1.5 Related (boundary layer effect simplification);

[0154] Dimensionless: Eliminating the dimension, η0 is the reference viscosity (such as the viscosity of water).

[0155]

[0156] A high-speed camera (Phantom V2512) is used to capture the flow of the grinding fluid, and the vortex field is extracted using PIV (particle imaging velocimetry) technology. It should be noted that in order to accurately measure the eddy current influence factor, the grinding wheel 200 is made of sapphire material to allow light to penetrate and observe the flow, so that the high-speed camera can accurately capture the flow of the grinding fluid. Alternatively, the position and shooting angle of the high-speed camera can be adjusted so that the high-speed camera can shoot sideways or tilted.

[0157] Orthogonal experimental method is used to control the grinding wheel speed w and grinding fluid flow rate v t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j and grinding fluid viscosity η, measure Γ vThe experimental matrix covers the parameter range of

[0158] v t ∈[1.0,5.0]m / s,w∈[100,500]rpm

[0159] θ j ∈[0°,60°], η∈[1.0,10.0]cP

[0160] After multiple nonlinear regression, the experimental data is substituted into the model:

[0161]

[0162] Using the least squares method to solve α, β, and γ, we get:

[0163] α=0.023±0.002, β=0.15±0.01, γ=0.08±0.005, R 2 =0.94.

[0164] Further, such as Figure 4 As shown, the thinning method proposed in this application also includes:

[0165] S3.1: Establish grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m About the influence factor Φ of magnetic fluid m The magnetic fluid function model;

[0166] S3.2: Based on the grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m And the magnetic fluid function model determines the magnetic fluid influence factor Φ m .

[0167] The function module of the magnetic fluid is:

[0168]

[0169] Among them, K m Provides support stiffness for the magnetic fluid layer 300;

[0170] R p is the radius of the grinding disc 100;

[0171] ρ m is the density of magnetic fluid.

[0172] The magnetic fluid function needs to quantify the influence of magnetorheological effect on the dynamic characteristics of the grinding disc 100. The key parameters include the magnetic fluid support stiffness K m, Magnetic fluid regulating pressure P m And the grinding wheel speed w carrier , the magnetofluid function model includes the magnetofluid pressure term and the centrifugal inertia term:

[0173] The magnetic fluid pressure term refers to the relationship between the supporting pressure of the magnetic fluid on the grinding disc 100 and its density. The magnetic fluid adjusts the pressure P m and the magnetic fluid density ρ m Inversely proportional, reflecting the effective support under unit density;

[0174] The centrifugal inertia term refers to the centrifugal force generated by the rotation of the grinding disc 100, which is proportional to the square of the rotation speed and the square of the radius, and affects the dynamic distribution and stability of the magnetic fluid.

[0175] The support pressure of magnetorheological fluid under the action of magnetic field can be expressed as:

[0176]

[0177] Among them, P m is the magnetic fluid regulating pressure, ρ m is the density of magnetic fluid.

[0178] When the grinding disc 100 rotates, the inertial pressure generated by the centrifugal force is:

[0179]

[0180] Preserve geometric and kinematic parameters after simplification:

[0181] Magnetic fluid influence factor Φ m The sum of the magnetic fluid pressure term and the centrifugal inertia term, multiplied by the magnetic fluid support stiffness K m :

[0182]

[0183] K m To amplify the comprehensive mechanical effect of the magnetic fluid layer 300;

[0184] Reflects the static support capacity of magnetic field regulation;

[0185] Quantify the contribution of dynamic centrifugal inertial forces.

[0186] The magnetic fluid of the magnetic fluid layer 300 adjusts the pressure P m It can be dynamically adjusted by applying an external magnetic field strength. Specifically, the density of the magnetic fluid ρ m and the magnetic fluid support stiffness K m Calibration to determine the magnetic fluid density ρ m and the magnetic fluid support stiffness Km :

[0187] Specifically, the fixed grinding wheel speed w carrier =0, adjust the magnetic fluid pressure P m , measure the displacement δ of the magnetic fluid layer 300 (laser displacement sensor); fix the magnetic fluid to adjust the pressure P m , adjust the grinding disc speed w carrier , measuring the displacement increment caused by centrifugal force.

[0188] Experimental parameter range:

[0189] P m ∈[0.1,0.5]MPa,w carrier ∈[100,500]rpm,R p =150mm

[0190] According to the static experimental data (w carrier =0), fitting magnetic fluid density ρ m and the magnetic fluid support stiffness K m :

[0191]

[0192] Through linear regression, we get:

[0193] K m =1.2×10 3 N / m, ρ m =850kg / m 3 (R 2 =0.98)

[0194] Dynamic experiment verifies centrifugal items:

[0195]

[0196] The eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m Introducing the damage function, the damage function F is the time domain integral of the coupling of vibration energy and process parameters, and is defined as:

[0197]

[0198] Wherein, a(t) is the vibration acceleration signal;

[0199] Γ v is the eddy current influence factor;

[0200] P g is the grinding wheel feed pressure;

[0201] Φm is the magnetic fluid impact factor;

[0202] k1 is the eddy current influence factor Γ v The control coefficient of

[0203] k2 is the grinding wheel feed pressure P g The control coefficient of

[0204] k3 is the magnetic fluid influence factor Φ m The control coefficient of .

[0205] Eddy current influence factor Γ v : The vortex formed by the rotation of the grinding wheel 200 and the spraying of the grinding fluid will cause local flow field turbulence, induce temperature gradient and stress concentration, and aggravate the sub-surface damage of the wafer W;

[0206] Grinding wheel feed pressure P g : The increase in grinding wheel feed pressure directly increases the mechanical stress on the surface of wafer W, which may cause microcracks to propagate;

[0207] Magnetic fluid influence factor Φ m : The magnetic fluid layer 300 suppresses or amplifies the vibration of the grinding disc 100 by adjusting the support stiffness and pressure, thereby indirectly affecting the processing stability.

[0208] Furthermore, the eddy current related parameters are the grinding wheel speed w, the grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j ;

[0209] The relevant parameters of magnetic fluid are the grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0210] Sensor data: vibration acceleration signal a(t), grinding wheel feed pressure P g .

[0211] Eddy current influence factor Γ v for:

[0212]

[0213] Magnetic fluid influence factor Φ m for:

[0214]

[0215] The damage function is calculated using a discretized integral (such as the gradient law):

[0216]

[0217] N is the number of time sampling points; △t is the sampling interval, for example 1ms.

[0218] Using L 32 (4 5 ) orthogonal table, covering Γ v 、P g , Φ m , Q t ,θ j Parameter combination; 12-inch single crystal silicon wafers were used, and each group of experiments was repeated 3 times; measurement indicators: subsurface damage depth d ssd (Laser confocal microscopy).

[0219] Establish a linear regression model Establish a linear regression model:

[0220] d ssd =k1Γ v +k2P g +k3Φ m +ξ

[0221] Among them, ξ is the random error term;

[0222] The coefficients are obtained by the least squares method:

[0223] k1=0.15±0.02, k2=0.08±0.01, k3=0.12±0.02, (R 2 =0.96).

[0224] Furthermore, minimize the damage function:

[0225]

[0226] Initialize process parameters;

[0227] Calculate F under the current parameters;

[0228] Calculating gradients

[0229] Update parameters along the negative gradient direction:

[0230] (κ is the learning rate, for example, κ = 0.01)

[0231] Iterate until F converges to a threshold (e.g. F < 10 -3 ).

[0232] By taking the eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ mBy introducing the damage function F and combining it with the orthogonal experimental calibration coefficient and gradient descent optimization, the process parameters can be dynamically controlled, significantly reducing the sub-surface damage of the wafer W and improving the quality of the thinned wafer W; the coupled analysis of the "vibration-flow field-magnetic fluid" multi-physics field is realized, providing a theoretical basis and real-time control strategy for the ultra-precision thinning process.

[0233] This application also proposes a thinning device, such as Figure 1 As shown, it includes a grinding disc 100, a grinding wheel 200 and a magnetic fluid layer 300; the grinding disc 100 is used to carry the wafer W and can rotate around an axis; the grinding wheel 200 is arranged above the grinding disc 100 and is used to feed the wafer W in a vertical direction to grind the wafer W; the magnetic fluid layer 300 is located below the grinding disc 100 and is connected to the grinding disc 100;

[0234] like Figure 5 As shown, it also includes: a function establishment module 400 for establishing the vibration acceleration signal a(t), the eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m Damage function model for wafer W damage; eddy current influence factor Γ v Including grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j ;Magnetic fluid influence factor Φ m Including grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ;

[0235] The parameter confirmation module 500 is used to determine the grinding wheel speed w and the grinding fluid flow rate v based on the collected vibration acceleration signal a(t) and the damage function model. t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m .

[0236] The function building module 400 also includes:

[0237] The first function establishment module 410 is used to establish the grinding wheel speed w, the grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j About the Eddy Current Influence Factor Γv Eddy current function model;

[0238] The second function establishment module 420 is used to establish the grinding disc speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m About the influence factor Φ of magnetic fluid m The magnetic fluid function model.

[0239] The parameter confirmation module 500 further includes:

[0240] The first parameter confirmation module 510 is used to determine the grinding wheel speed w and the grinding fluid flow rate v based on the collected parameters. t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j And the eddy current function model determines the eddy current influence factor Γ v To confirm the grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j ;

[0241] The second parameter confirmation module 520 is used to confirm the grinding wheel feed pressure P g ;as well as

[0242] The third parameter confirmation module 530 is used to determine the value of the grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m And the magnetic fluid function model determines the magnetic fluid influence factor Φ m , to confirm the grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m .

[0243] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0244] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A thinning method, characterized in that: Applicable to a thinning device, the thinning device comprising: A grinding disc (100), the grinding disc (100) is used to carry a wafer (W), and the grinding disc (100) is rotatable around an axis; A grinding wheel (200), the grinding wheel (200) is arranged above the grinding disc (100), and the grinding wheel (200) is used for vertically feeding and acting on the wafer (W) to grind the wafer (W); and a magnetic fluid layer (300), the magnetic fluid layer (300) being located below the grinding disc (100) and connected to the grinding disc (100); The thinning method comprises: Establish vibration acceleration signal a(t) and eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m Damage function model for wafer (W) damage; The eddy current influence factor Γ v Including grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j ; The magnetic fluid influence factor Φ m Including grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ; Based on the collected vibration acceleration signal a(t) and the damage function model, the grinding wheel speed w and the grinding fluid flow rate v are determined. t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m , and applied to the next wafer (W) processing; Also includes: Establish the grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j About the Eddy Current Influence Factor Γ v Eddy current function model; Based on the collected grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j And the eddy current function model determines the eddy current influence factor Γ v .

2. A thinning method according to claim 1, characterized in that: The eddy current function model is: Where ρ is the grinding fluid density; η is the grinding fluid viscosity; η0 is the reference viscosity; R g is the radius of the grinding wheel (200); θ j The angle between the grinding fluid spray direction and the wafer (W) normal.

3. A thinning method according to claim 2, characterized in that: Also includes: Establish grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m About the influence factor Φ of magnetic fluid m The magnetic fluid function model; Based on the grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m And the magnetic fluid function model determines the magnetic fluid influence factor Φ m .

4. A thinning method according to claim 3, characterized in that: The magnetic fluid function model is: Among them, K m is the support stiffness of the magnetic fluid layer (300); R p is the radius of the grinding disc (100); ρ m is the density of magnetic fluid.

5. A thinning device, characterized in that: include: A grinding disc (100), the grinding disc (100) is used to carry a wafer (W), and the grinding disc (100) is rotatable around an axis; A grinding wheel (200), the grinding wheel (200) is arranged above the grinding disc (100), and the grinding wheel (200) is used for vertically feeding and acting on the wafer (W) to grind the wafer (W); and a magnetic fluid layer (300), the magnetic fluid layer (300) being located below the grinding disc (100) and connected to the grinding disc (100); Also includes: Function establishment module (400), used to establish vibration acceleration signal a(t), eddy current influence factor Γ v , Grinding wheel feed pressure P g And the magnetic fluid influence factor Φ m Damage function model for wafer (W) damage; The eddy current influence factor Γ v Including grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j ; The magnetic fluid influence factor Φ m Including grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m ; The parameter confirmation module (500) is used to determine the grinding wheel speed w and the grinding fluid flow rate v based on the collected vibration acceleration signal a(t) and the damage function model. t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j , Grinding wheel feed pressure P g , grinding disc speed w carrier , Magnetic fluid regulating pressure P m , and the magnetic fluid support stiffness K m .

6. A thinning device according to claim 5, characterized in that: The function establishment module (400) further includes: The first function establishment module (410) is used to establish the grinding wheel speed w, the grinding fluid flow rate v t , Grinding fluid supply flow Q t and the grinding fluid injection angle θ j About the Eddy Current Influence Factor Γ v Eddy current function model; The second function establishment module (420) is used to establish the grinding disc speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m About the influence factor Φ of magnetic fluid m The magnetic fluid function model.

7. A thinning device according to claim 6, characterized in that: The parameter confirmation module (500) further includes: The first parameter confirmation module (510) is used to determine the grinding wheel speed w and the grinding fluid flow rate v based on the collected data. t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j And the eddy current function model determines the eddy current influence factor Γ v To confirm the grinding wheel speed w, grinding fluid flow rate v t , Grinding fluid supply flow Q t , Grinding fluid jet angle θ j ; The second parameter confirmation module (520) is used to confirm the grinding wheel feed pressure P g ;as well as The third parameter confirmation module (530) is used to determine the value of the grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m And the magnetic fluid function model determines the magnetic fluid influence factor Φ m , to confirm the grinding wheel speed w carrier , Magnetic fluid regulating pressure P m , magnetic fluid support stiffness K m .

Images