Fixed disc flatness control method in semiconductor silicon wafer grinding

Abstract

The invention discloses a fixed disc flatness control method in semiconductor silicon wafer grinding, which comprises the following steps of: before each grinding liquid cycle is started, measuring the current flatness delta n of a fixed disc; according to the value range of the delta n, selecting a correction strategy of the current period; and when the threshold value is T, two-stage control is executed: in the first stage, grinding is continuously carried out in a single rotation direction until the flatness approaches a zero point; in the second stage, alternate forward and reverse rotation grinding is switched until the period is finished; when delta n is smaller than or equal to T, alternative forward and reverse rotation grinding is executed in the whole process. Wherein the rotation direction selection follows: [delta] ngt; when 0, the first stage is forward rotation, delta nlt; when 0, the first stage is reversal. According to the technical scheme, the technical problem that fluctuation of flatness of the fixed disc is large can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and particularly to a method for controlling the flatness of a platen in the grinding of semiconductor wafers. Background Art

[0002] In the semiconductor wafer industry, generally after a grinding fluid cycle is completed, a correction gear is used to correct the flatness of the platen, and then the rotation direction (clockwise or counterclockwise) of the wafer in the current cycle is set according to the flatness of the lower platen measured after the end of the previous cycle. Thus, the effect of dynamically correcting the flatness of the platen is achieved.

[0003] At present, the flatness of the platen can be corrected during the rotation of the wafer. According to experience, "clockwise rotation corrects convexity, and counterclockwise rotation corrects concavity", that is, if the rotation method of the wafer is clockwise (positive), the lower platen will become concave, and vice versa, the lower platen will become convex. For example: if the flatness of the platen in the previous cycle is +12um, then the rotation in this cycle is clockwise; if the flatness of the platen in the previous cycle is -5um, then the rotation in this cycle is counterclockwise; if the flatness of the platen in the previous cycle is 0, then the rotation in this cycle is alternately clockwise and counterclockwise. However, if the rotation in an entire cycle is always clockwise or counterclockwise, it will cause overcorrection during processing, resulting in large fluctuations in the flatness of the platen, and a more refined control method is needed to narrow the fluctuation range. Summary of the Invention

[0004] The main object of the present invention is to propose a method for controlling the flatness of a platen in the grinding of semiconductor wafers, aiming to at least solve the technical problem of large fluctuations in the flatness of the platen in the related art.

[0005] To achieve the above object, a method for controlling the flatness of a platen in the grinding of semiconductor wafers proposed by the present invention includes the following steps:

[0006] Step S1. Before starting each grinding fluid cycle, measure the current flatness Δn of the platen;

[0007] Step S2. Select the correction strategy for this cycle according to the numerical range of Δn:

[0008] When |Δn| > threshold T, perform two-stage control:

[0009] The first stage: Grind continuously in a single rotation direction until the flatness approaches zero;

[0010] The second stage: Switch to alternate clockwise and counterclockwise grinding until the end of the cycle;

[0011] When |Δn| ≤ T, perform alternate clockwise and counterclockwise grinding throughout the process;

[0012] Among them, the selection of the rotation direction follows: in the first stage, it is clockwise when Δn > 0, and it is counterclockwise when Δn < 0.

[0013] Further defined, the value range of the threshold T is 8 - 12 μm, preferably 10 μm.

[0014] Further defined, the duration of the first stage is calculated by the following formula:

[0015]

[0016] Wherein, HS is the number of times of maintaining single - direction grinding, R is the correction efficiency per time, with a value range of 0.4 - 0.6 μm / time, and ceil() is the ceiling function.

[0017] Further defined, the correction efficiency R per time is dynamically calibrated through historical data, and the calibration formula is:

[0018]

[0019] Wherein, α is the smoothing coefficient (0 < α < 1), Δn prev is the previous measured flatness, and HS actual is the actual number of executions.

[0020] Further defined, the second - stage alternate forward and reverse grinding adopts one of the following modes:

[0021] a) Fixed - interval alternation: Switch the rotation direction every N times, where N is 2 - 5 times;

[0022] b) Random alternation: Switch the direction at the beginning of each time with a probability P, where P ranges from 30% to 70%;

[0023] c) Progressive alternation: The alternation frequency increases with the increase of the remaining grinding time.

[0024] Further defined, an on - line flatness detection device is embedded during the grinding process. When the real - time flatness reaches Δ < ±2 μm, the second - stage mode switching is triggered in advance.

[0025] Further defined, it is allowed to introduce rotational speed modulation during the alternate forward and reverse grinding stage. When rotating forward, the rotational speed is ω1, and when rotating backward, the rotational speed is ω2, and 1.2 ≤ ω1 / ω2 ≤ 0.8 is satisfied.

[0026] Further defined, it is allowed to apply a transition buffer process when switching between the first stage and the second stage, including:

[0027] Perform 1 - 3 times of speed - down grinding before direction switching, and the rotational speed linearly drops to 50% - 70% of the reference value;

[0028] Perform 1 - 3 times of speed - up grinding after switching, and gradually recover to the reference rotational speed.

[0029] Further defined, a closed-loop feedback mechanism is established, including:

[0030] Record the actual flatness change amount Δn+1 in each cycle, calculate the correction error E = Δn - (HSgR), and dynamically adjust the R value in subsequent cycles accordingly.

[0031] The present invention has at least the following beneficial effects:

[0032] By introducing a threshold trigger mechanism and a phased control logic, the problem of "overcorrection" caused by traditional full-cycle unidirectional correction is effectively solved. When a large flatness deviation (|Δn| > T) is detected, first use high-precision unidirectional correction to quickly converge to near zero (±2μm interval), and then switch to an alternating forward and reverse grinding mode, reducing the flatness fluctuation amplitude by 40%-60% compared with the traditional method. The measured fluctuation range is compressed from the traditional ±20μm to within ±10μm, which can effectively meet the stringent requirements of advanced processes for TTV (total thickness variation). Furthermore, it effectively improves the technical problem of large fluctuations in the platen flatness during the grinding of semiconductor wafers. Description of the Drawings

[0033] Figure 1 It is a schematic flow chart of an embodiment of a method for controlling the platen flatness in the grinding of semiconductor wafers provided by the present invention;

[0034] Figure 2 It is a diagram showing the effects of the prior art and the original solution of a method for controlling the platen flatness in the grinding of semiconductor wafers provided by the present invention. Detailed Embodiments

[0035] Next, the technical solutions in the embodiments of the present application will be clearly described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present application.

[0036] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such used data can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. generally belong to the same category, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / " generally indicates an "or" relationship between the associated objects.

[0037] The following will combine the accompanying drawings and, through specific embodiments and their application scenarios, elaborate in detail on a method for controlling the flatness of a platen in semiconductor wafer grinding provided by the embodiments of the present application.

[0038] Please refer to Figure 1 and Figure 2 , the present invention provides a method for controlling the flatness of a platen in semiconductor wafer grinding, which can be used to improve the technical problem of large fluctuations in the flatness of the platen in current semiconductor wafer grinding.

[0039] Specifically, the method for controlling the flatness of a platen in semiconductor wafer grinding disclosed by the present invention can at least include the following steps.

[0040] First, perform step S1. Before starting each grinding fluid cycle, measure the current flatness Δn of the platen. By measuring the flatness Δn of the platen, initial data of the platen can be obtained. It can be understood that the flatness Δn of the platen represents the height difference between the highest point and the lowest point on the platen surface. In the actual measurement process, a laser interferometer can be used for measurement, but it is not limited thereto, and it can be determined according to actual needs.

[0041] Secondly, perform step S2, and select the correction strategy for this cycle according to the numerical range of Δn.

[0042] In step S2, different Δn values correspond to different correction strategies. Specifically, in this embodiment, when |Δn| > threshold T, two-stage control is performed. Among them, the two-stage control includes a first stage and a second stage.

[0043] Specifically, the first stage represents continuous grinding in a single rotation direction until the flatness approaches zero. It should also be noted that in the first stage, the single rotation direction is selected according to the polarity of Δn. When Δn > 0, clockwise grinding is used, and when Δn < 0, counterclockwise grinding is used. Continuously perform the HS back-grinding operation until the predicted flatness enters the [-2μm, +2μm] interval.

[0044] The second stage represents switching to alternating forward and reverse grinding until the end of the cycle. Among them, the alternating frequency is not less than one direction change every 5 turns.

[0045] When |Δn| ≤ T, alternating forward and reverse grinding is performed throughout the process.

[0046] It should be noted that in this technical solution, the selection of the rotation direction follows: in the first stage, it is forward rotation when Δn > 0, and reverse rotation when Δn < 0.

[0047] Furthermore, the value range of the threshold T is 8 - 12μm, and preferably 10μm.

[0048] By introducing a threshold trigger mechanism and a phased control logic, the problem of "overcorrection" caused by traditional full-cycle unidirectional correction is effectively solved. When a large flatness deviation (|Δn| > T) is detected, high-precision unidirectional correction is first used to quickly converge to near zero (±2μm range), and then it is switched to an alternating forward and reverse grinding mode, reducing the flatness fluctuation amplitude by 40%-60% compared with the traditional method. The measured fluctuation range is compressed from the traditional ±20μm to within ±10μm, which can effectively meet the stringent requirements of advanced processes for TTV (total thickness variation).

[0049] Furthermore, in some embodiments, the duration of the first stage in step S2 needs to be strictly controlled.

[0050] In this embodiment, the duration of the first stage is obtained in the following manner, which includes:

[0051]

[0052] Where HS is the number of times of maintaining single-direction grinding, R is the correction efficiency per time, with a value range of 0.4 - 0.6μm / time, and ceil() is the ceiling function.

[0053] It should be noted that the correction efficiency R per time is dynamically calibrated through historical data, and the calibration formula is:

[0054]

[0055] Where α is the smoothing coefficient (0 < α < 1), Δn prev is the previous measured flatness, and HS actual is the actual number of executions.

[0056] It can be understood that the correction efficiency R per time dynamically calibrated based on historical data, combined with a real-time feedback mechanism, constructs a control system with self-learning ability. This system can automatically compensate for time-varying factors such as equipment wear and the attenuation of grinding fluid performance, ensuring long-term stability of the correction efficiency.

[0057] Among them, in one embodiment, a closed-loop feedback mechanism is also established. It includes: recording the actual flatness change amount Δn+1 in each cycle, calculating the correction error E = Δn - (HS×R), and dynamically adjusting the R value in subsequent cycles accordingly.

[0058] In some embodiments, the second-stage alternating forward and reverse grinding adopts one of the following modes:

[0059] a) Fixed-interval alternation: Switch the rotation direction every N times, where N is 2 - 5 times.

[0060] It is understandable that by switching at fixed intervals (such as switching directions every 3 times), a regular correction rhythm is formed to avoid the uncertainty brought about by random switching. And to ensure that each area of the surface plate is subjected to uniform mechanical loads, reducing the risk of local overheating.

[0061] b) Random alternation: At the beginning of each cycle, the direction is switched with a probability P, where the value range of P is 30% - 70%.

[0062] It is understandable that random alternation can break the fixed switching rhythm and effectively suppress vibration harmonics of specific frequencies. And through probabilistic switching, the residual errors of unidirectional correction are averaged, effectively improving the flatness standard deviation.

[0063] c) Progressive alternation: The alternation frequency increases as the remaining grinding time increases.

[0064] It is understandable that by performing low-frequency switching at the initial stage (such as 5 times per cycle) to ensure the correction efficiency, and high-frequency switching at the later stage (1 time per cycle) to improve the accuracy. And dynamically adjust the switching frequency during the grinding process to comprehensively reduce energy consumption. At the same time, the high-frequency switching at the end stage forms a "micro-vibration" effect, effectively suppressing the overshoot phenomenon near the zero point.

[0065] It should be noted that during the switching process between the first stage and the second stage, it is allowed to apply a transition buffer during the stage switch. The transition buffer includes performing 1 - 3 times of speed reduction grinding before the direction switch, and the rotational speed linearly decreases to 50% - 70% of the reference value. And performing 1 - 3 times of speed increase grinding after the switch, gradually restoring to the reference rotational speed. By setting the transition buffer, the grinding effect can be effectively improved.

[0066] In order to improve the actual use effect of this device during actual use, it is allowed to embed an on-line flatness detection device during the grinding process. When the real-time flatness reaches Δ < ±2μm, the second stage mode switch is triggered in advance. By triggering the second stage mode in advance, the grinding efficiency can be improved.

[0067] At the same time, it is also allowed to introduce rotational speed modulation during the alternating forward and reverse grinding stage. When rotating forward, the rotational speed is ω1, and when rotating backward, the rotational speed is ω2, and 1.2 ≤ ω1 / ω2 ≤ 0.8 is satisfied. By introducing rotational speed modulation, the stability of the surface plate during the grinding process can be effectively ensured.

[0068] In summary, the present invention discloses a method for controlling the flatness of a platen in semiconductor wafer grinding. By introducing a threshold trigger mechanism and a phased control logic, the problem of "overcorrection" caused by traditional full-cycle unidirectional correction is effectively solved. When a large flatness deviation (|Δn| > T) is detected, first, high-precision unidirectional correction is used to quickly converge to near zero (±2μm range), and then it is switched to an alternating forward and reverse grinding mode, reducing the flatness fluctuation amplitude by 40%-60% compared with the traditional method. The measured fluctuation range is compressed from the traditional ±20μm to within ±10μm, which can effectively meet the stringent requirements of advanced processes for TTV (Total Thickness Variation). Furthermore, the technical problem of large fluctuations in the flatness of the platen in current semiconductor wafer grinding is effectively improved.

[0069] It should be noted that in this article, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or device including that element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed. It may also include performing functions in a substantially simultaneous manner or in the reverse order according to the functions involved. For example, the described method may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, the features described with reference to certain examples may be combined in other examples.

[0070] The embodiments of the present application have been described above in conjunction with the accompanying drawings. However, the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims, and all of them belong to the protection scope of the present application.

Claims

1. A method for controlling the flatness of a platen during semiconductor silicon wafer polishing, characterized in that: The following steps are involved: Step S1. Before starting each polishing liquid cycle, measure the current flatness Δn of the fixed plate; Step S2. Select the current cycle correction strategy according to the numerical range of Δn: When |Δn|>threshold T, two-stage control is performed: The first stage: grinding in a single rotation direction until the flatness approaches zero; The second stage: switch to alternating forward and reverse grinding until the cycle ends; When |Δn|≤T, the whole process is performed with alternating forward and reverse grinding; The rotation direction is selected as follows: when Δn>0, the first stage is forward rotation, and when Δn<0, the first stage is reverse rotation.

2. The method for controlling the flatness of a semiconductor wafer during polishing according to claim 1, wherein: The threshold T has a value range of 8-12 μm, preferably 10 μm.

3. The method for controlling the flatness of a semiconductor wafer during polishing according to claim 1, wherein: The duration of the first phase is calculated by the following formula: Among them, HS is the number of times that single-direction grinding needs to be maintained, R is the single-pass correction efficiency, with a value range of 0.4-0.6μm / pass, and ceil() is the upward rounding function.

4. The method for controlling the flatness of a platen during semiconductor silicon wafer polishing according to claim 3, characterized in that: The single-pass correction efficiency R is dynamically calibrated through historical data, and the calibration formula is: Among them, α is the smoothing coefficient (0<α<1), Δn prev is the flatness measured last time, HS actual The actual number of executions.

5. The method for controlling the flatness of a platen during semiconductor silicon wafer polishing according to claim 1, wherein: The second stage of alternating forward and reverse grinding adopts one of the following modes: a) Fixed interval alternation: the rotation direction is switched every N times, N is 2-5 times; b) Random alternation: At the beginning of each round, the direction is switched with probability P, and the value range of P is 30%-70%; c) Progressive alternation: The alternation frequency increases as the remaining grinding time increases.

6. The method for controlling the flatness of a platen during semiconductor silicon wafer polishing according to claim 5, characterized in that: An online flatness detection device is embedded in the grinding process, and when the real-time flatness reaches Δ<±2μm, the second-stage mode switching is triggered in advance.

7. The method for controlling the flatness of a platen during semiconductor silicon wafer polishing according to claim 6, wherein: It is permissible to introduce speed modulation in the alternating forward and reverse grinding stages, wherein the speed during forward rotation is ω1, the speed during reverse rotation is ω2, and 1.2≤ω1 / ω2≤0.8 is satisfied.

8. The method for controlling the flatness of a platen during semiconductor silicon wafer polishing according to claim 1, characterized in that: It is possible to apply a transition buffer process when switching between the first and second stages, including: Before switching direction, perform 1-3 times of speed reduction grinding, and the speed linearly drops to 50%-70% of the reference value; After switching, perform 1-3 speed increase grinding and gradually return to the reference speed.

9. The method for controlling the flatness of a platen during semiconductor silicon wafer polishing according to claim 4, characterized in that: Establish a closed-loop feedback mechanism, including: The actual flatness change Δn+1 of each cycle is recorded, and the correction error E=Δn-(HSgR) is calculated, and the R value of the subsequent cycle is dynamically adjusted accordingly.

Images