Deep corrosion method for back cavity of MEMS pressure sensor

Abstract

The invention belongs to the technical field of MEMS (Micro Electro Mechanical System) manufacturing, and particularly relates to a deep corrosion method for a back cavity of an MEMS pressure sensor. Comprising the following steps: constructing a'preset reference-gradient rough corrosion-real-time correction-precise fine corrosion 'closed-loop control system: firstly, establishing a precise temperature-corrosion rate corresponding relation through a QC wafer experiment; according to the target depth of the back cavity, multiple rounds of gradient rate coarse corrosion processes are planned, the corrosion depths of all rounds are distributed according to the proportion from large to small, and the corrosion rate is decreased progressively along with the rounds through temperature regulation and control to rapidly approach 80%-95% of the target depth; the depth is detected immediately after each round of corrosion, and targeted correction of'parameter fine adjustment 'or'deviation compensation' is realized based on a double-threshold criterion; and finally, turning to multiple rounds of low-rate fine corrosion, continuing a gradient rate and a real-time correction strategy, and precisely closing a loop to a target depth. According to the invention, submicron level control of the depth of the back cavity is realized, the process stability and adaptability are high, and the performance consistency and the large-scale production yield of the MEMS pressure sensor are effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of microelectromechanical systems (MEMS) manufacturing technology, and specifically relates to a method for deep etching of the back cavity of a MEMS pressure sensor, which is particularly suitable for anisotropic etching processes using tetramethylammonium hydroxide (TMAH) or potassium hydroxide (KOH) etching solutions. Background Technology

[0002] MEMS pressure sensors, with their advantages of small size, high sensitivity, and low cost, are widely used in automotive electronics, medical monitoring, industrial control, and consumer electronics. The core mechanical sensing structure of the sensor is typically a thin film or beam structure, beneath which a cavity of a specific depth needs to be fabricated to form a pressure-sensing cavity. The cavity depth is one of the key geometric parameters determining the sensor's sensitivity, linearity, range, and resonant frequency; its accuracy directly affects the sensor's final performance and consistency.

[0003] Currently, the fabrication of the back cavity of silicon-based MEMS pressure sensors mainly relies on wet anisotropic etching processes. Among these, TMAH and KOH solutions have become the most commonly used etching solutions in the industry due to their excellent crystal orientation-selective etching characteristics for silicon, mature processes, and relatively low costs. However, existing TMAH / KOH-based cavity etching processes still face significant challenges in depth precision control, mainly in the following aspects: First, the process control is open-loop, with depth detection only performed after the entire etching process is completed. There is a lack of real-time monitoring and immediate correction mechanisms, causing small depth deviations from a single etching cycle to accumulate and eventually exceed the correction capabilities of subsequent processes, directly leading to product defects and scrap. Second, the transition between the coarse and fine etching stages lacks quantitative judgment standards, generally employing a simple two-stage strategy of "single coarse etching + single fine etching." This fails to scientifically allocate the etching amount and rate for each stage based on the target cavity depth, often resulting in excessive correction burden in the fine etching stage, leading to loss of precision control or low process efficiency. Third, the process adaptability is poor. For different target cavity depths, numerous parameter debugging experiments need to be conducted again, lacking a universal process control framework, significantly increasing the R&D cycle and process transfer costs for large-scale production.

[0004] Therefore, there is an urgent need in this field for an etching method that can achieve submicron-level precision control of the back cavity depth, has real-time process feedback and deviation self-compensation capabilities, and can flexibly adapt to different depth requirements, so as to improve the performance consistency, production yield and market competitiveness of MEMS pressure sensors. Summary of the Invention

[0005] The purpose of this invention is to provide a method for etching the back cavity depth of MEMS pressure sensors based on gradient rate control and real-time deviation correction, aiming to solve the technical bottlenecks of low back cavity depth control accuracy and poor yield due to deviation accumulation in existing TMAH / KOH wet etching processes. This invention establishes a precise process benchmark through pre-set experiments, uses multi-round coarse etching at a gradient rate to rapidly approach the target depth, and combines real-time deviation correction and single-round deviation compensation. Finally, it achieves closed-loop control of accuracy through multi-round fine etching at a low rate. Through gradient rate planning, real-time feedback adjustment based on dual thresholds (δa / δb), single-round deviation compensation, and clear stage transition (E) and final accuracy (ε) standards, a high-precision, highly stable, and highly adaptive back cavity depth control system is constructed.

[0006] To address the aforementioned technical problems, this invention provides a method for deep etching of the back cavity of a MEMS pressure sensor, based on gradient rate control and real-time deviation correction; it constructs a closed-loop control system of "preset benchmark - gradient coarse etching - real-time correction - precise fine etching"; including the following steps:

[0007] Step S1: Pre-set experiment; Establish a precise temperature-corrosion rate correspondence through QC sheet experiments;

[0008] Step S2: Coarse corrosion stage; Based on the target depth of the back cavity, plan a multi-round gradient rate coarse corrosion process, allocate the corrosion depth of each round in descending proportion, and reduce the corrosion rate with each round through temperature control to quickly approach 80% to 95% of the target depth; After each round of corrosion, detect the depth immediately, and make targeted corrections for parameter fine-tuning or deviation compensation based on dual threshold criteria.

[0009] Step S3: Fine erosion stage; transition to multiple rounds of low-rate fine erosion, continue the gradient rate and real-time correction strategy, and accurately close the loop to the target depth.

[0010] Preferably, the preset experiment specifically includes: preparing a QC sheet with the same material as the MEMS pressure sensor substrate, and establishing a temperature-corrosion rate correspondence table based on the corrosion rate corresponding to different test temperatures in a TMAH or KOH etching solution system; wherein the test temperature range is 60℃~95℃, the temperature gradient is 2℃~5℃, and the corrosion rate of at least 3 QC sheets is tested at each temperature point, and the average value is taken as the corrosion rate corresponding to that temperature.

[0011] Preferably, the concentration of the TMAH or KOH etching solution is 10%wt to 30%wt, and magnetic stirring is used during the etching process at a stirring rate of 100r / min to 300r / min.

[0012] Preferably, in the coarse corrosion stage, multiple rounds of coarse corrosion are performed based on the target cavity depth H to approximate P% of the target depth H, where P is 80% to 95%, and dynamically adjusted according to the target depth H; specifically, it includes:

[0013] Step S21: Determine the total number of roughing corrosion cycles N and the theoretical depth ha for each round of roughing corrosion. i (i = 1, 2, ..., N): Distribute the theoretical depth ha of each round of coarse corrosion in descending order of proportion. i The sum of theoretical depths of N rounds of coarse corrosion ∑ha N P% of the target depth H, and satisfying the theoretical depth ha of the i-th round of coarse corrosion. i >The theoretical depth of the (i+1)th round of coarse corrosion, ha i+1 (i = 1, 2, ..., N-1);

[0014] Step S22: Adjust the temperature of each round of coarse corrosion: Based on the temperature-corrosion rate correspondence table established in step S1, match the corresponding corrosion temperature for each round of coarse corrosion, so that the coarse corrosion rate V of the i-th round is... i >Rate of coarse corrosion in round i+1 V i+1 (i = 1, 2, ..., N-1);

[0015] Step S23: Perform multiple rounds of coarse corrosion and correct in real time: Perform the first to Nth rounds of coarse corrosion sequentially; after each i-th round of coarse corrosion, immediately detect the cumulative actual corrosion depth ∑Ha. i And calculate the actual corrosion depth Ha in the i-th round. i ; Calculate the depth deviation Δha of a single wheel i =Ha i -ha i ;

[0016] If |Δha i If |≥δa, then perform instantaneous parameter fine-tuning: when Δha i When Δha > 0, shorten the corrosion time of the (i+1)th round or reduce the corrosion temperature of the (i+1)th round; when Δha i When δa < 0, extend the corrosion time of the (i+1)th round or increase the corrosion temperature of the (i+1)th round; where δa is the single-round deviation adjustment threshold for coarse corrosion;

[0017] If |Δha i If |<δa, then the preset time and temperature parameters for the (i+1)th round will not be adjusted, but the deviation Δha of that round will be changed. i The amount is included in the compensation for the next round, and is used to compensate when planning the theoretical depth to be achieved in the (i+1)th round.

[0018] Step S24: Stage transition judgment: When the cumulative actual depth ∑Ha iWhen the transition condition is met, the coarse corrosion stage ends and the fine corrosion stage begins; the transition condition is: the cumulative actual depth ∑Ha i The absolute deviation from the target value H×P% shall not exceed the allowable transfer deviation E.

[0019] Preferably, in step S21, the total number of coarse corrosion cycles N is 2 to 5, the ratio of coarse corrosion depth in each cycle is N:N-1:...:1, and the ratio of coarse corrosion depth between two adjacent cycles is 1.2 to 2.0.

[0020] Preferably, in the fine etching stage, multiple rounds of low-rate fine etching are performed to accurately reach the target depth H; specifically, this includes:

[0021] Step S31: Calculate the depth of the target to be corroded, ΔH = H - ∑Ha N Determine the total number of fine corrosion cycles M and the theoretical depth hb for each round of fine corrosion. j (j=1,2,...,M): The theoretical depth hb of each round of fine corrosion is allocated in descending order of proportion. j The sum of the theoretical depths of M-cycle fine corrosion ∑hb M It equals ΔH and satisfies the theoretical depth hb of the j-th round of fine corrosion. j >The theoretical depth of the (j+1)th round of fine corrosion is hb j+1 (j = 1, 2, ..., M-1);

[0022] Step S32: Adjust the temperature of each round of fine etching: Based on the temperature-etching rate correspondence table established in step S1, match the corresponding etching temperature for each round of fine etching, so that the fine etching rate V of the j-th round is... j >The fine corrosion rate V in the (j+1)th round j+1 Furthermore, all fine corrosion rates were less than the final round of coarse corrosion rate V. N ;

[0023] Step S33: Perform multiple rounds of fine etching and correct in real time: Perform the first to the Mth rounds of fine etching sequentially; after each jth round of fine etching, immediately detect the cumulative actual etching depth ∑Hb. j And calculate the actual corrosion depth Hb in the j-th round. j ; Calculate the single-wheel depth deviation Δhb j =Hb j -hb j ;

[0024] If |Δhb j If |≥δb, then perform instantaneous parameter fine-tuning: when Δhb j When Δhb > 0, shorten the corrosion time of the (j+1)th round or decrease the corrosion temperature of the (j+1)th round; when Δhb jWhen δb < 0, extend the corrosion time of the (j+1)th round or increase the corrosion temperature of the (j+1)th round; where δb is the single-round deviation adjustment threshold for fine corrosion;

[0025] If |Δhb j If |<δb, then the preset time and temperature parameters for the (j+1)th round will not be adjusted, but the deviation Δhb of that round will be reduced. j The amount is included in the compensation for the next round, and is used to compensate when planning the theoretical depth to be achieved in the (j+1)th round.

[0026] Step S34: Process completion judgment: When the cumulative actual depth satisfies |H-(∑Ha) N +∑Hb M When |≤ε, the process is complete, where ε is the final allowable error.

[0027] Preferably, in step S23, when performing time fine-tuning, the adjustment amount ΔT i+1 It is proportional to the absolute value of the deviation of a single wheel, that is: ΔT i+1 =k1×|Δha i | where k1 = 0.2 min / μm ~ 0.5 min / μm;

[0028] In step S33, when performing time fine-tuning, the adjustment amount Δt j+1 It is proportional to the absolute value of the single-wheel deviation, i.e.: Δt j+1 =k2×|Δhb j | where k2 = 0.1 min / μm ~ 0.3 min / μm.

[0029] Preferably, in steps S22 and S32, the corrosion rate is adjusted by a decrease of 0.5 μm / min to 2.0 μm / min per round, and the minimum corrosion rate in the fine corrosion stage is not less than 0.1 μm / min.

[0030] Preferably, in steps S23 and S33, a laser interferometric thickness gauge or a step gauge is used to detect the actual corrosion depth, with a detection accuracy of not less than 0.01 μm.

[0031] Preferably, in step S23, the value range of the coarse corrosion single-wheel deviation adjustment threshold δa is 0.2μm to 0.5μm;

[0032] In step S24, the allowable transfer deviation E is the larger of (0.5% to 2%) × H or 0.5 μm to 3.0 μm;

[0033] In step S33, the value range of the fine corrosion single-wheel deviation adjustment threshold δb is 0.1μm~0.3μm;

[0034] In step S34, the final allowable error ε is 0.5 μm.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. Submicron-level depth control accuracy and significantly improved performance consistency: By setting a strict deviation threshold (δb) of 0.1 to 0.3 μm and a final tolerance (ε) standard of 0.5 μm in the fine etching stage, combined with a real-time correction mechanism after each round of etching, the back cavity depth deviation can be stably controlled within ±0.5 μm, which is far superior to the deviation level of several micrometers in traditional processes, effectively ensuring the batch consistency of core performance such as sensor sensitivity and linearity.

[0037] 2. Closed-loop management throughout the entire process significantly enhances process stability: Abandoning the traditional open-loop experience-based operation mode, a real-time feedback control system is constructed with "temperature-corrosion rate benchmark table" as the data foundation, "depth detection per round" as the feedback basis, and "dual threshold criterion (δa / δb)" as the decision core. This achieves closed-loop management throughout the entire process of "monitoring-decision-execution," effectively suppressing rate fluctuations in single-round corrosion, avoiding the accumulation of deviations from the root cause, and significantly improving the process repeatability within and between batches.

[0038] 3. Innovative deviation compensation mechanism, breaking through the limitations of traditional correction: For minor deviations that do not trigger immediate fine-tuning, an innovative "single-round deviation compensation" strategy is proposed. Instead of ignoring minor deviations, it quantifies and incorporates them into the theoretical corrosion depth of the next round, achieving proactive correction of minor deviations, avoiding the invisible accumulation of minor deviations, providing double protection for the final depth accuracy, and significantly reducing the risk of finished product deviations.

[0039] 4. Quantitative stage transition standard, balancing efficiency and accuracy: The "transition deviation E" quantitative judgment standard is introduced (E is the larger value between (0.5%~2%)×H and 0.5μm~3.0μm), which clarifies the objective conditions for the transition from coarse corrosion to fine corrosion. This ensures that the residual corrosion amount at the beginning of fine corrosion is always within the stable processing window of the low-rate process, solving the problem of efficiency and accuracy imbalance caused by the ambiguity of the coarse-fine transition in traditional processes, and achieving efficient synergy of "rapid approximation + precise correction".

[0040] 5. Universal process framework to adapt to multiple depth requirements: By dynamically adjusting the target ratio of coarse etching (P = 80% to 95%), the round allocation (N = 2 to 5 rounds, M = 1 to 3 rounds), and the threshold parameters, the same process framework can be adapted to the preparation of back cavity with different target depths of 250μm to 500μm and above. There is no need to carry out a large number of parameter debugging experiments for different depths, which significantly reduces the R&D cost and process transfer cycle of multiple product lines. Attached Figure Description

[0041] Figure 1This is a flowchart of a high-precision etching method for the back cavity depth of a MEMS pressure sensor based on gradient rate control and real-time deviation correction, provided by the present invention. Detailed Implementation

[0042] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become clearer from the following description. It should be noted that the drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0043] like Figure 1 As shown, this embodiment of the invention specifically provides a method for deep etching of the back cavity of a MEMS pressure sensor based on gradient rate control and real-time deviation correction, which specifically includes the following steps:

[0044] Step 1: Pre-set QC sheet experiments to establish a temperature-corrosion rate benchmark; prepare a quality control sheet with the same material and crystal orientation as the substrate of the MEMS pressure sensor to be corroded, and place it in a TMAH or KOH etching solution of a set concentration. Within a temperature range of 60℃ to 95℃, with a temperature gradient of 2℃ to 5℃, test the corrosion rate of the QC sheet at each temperature point. Test at least 3 sheets at each temperature point, and use a laser interferometric thickness gauge or a profilometer to measure the thickness change before and after corrosion, calculate the corrosion rate and take the average value, and finally establish a temperature-corrosion rate correspondence table for the system to provide data support for subsequent dynamic rate adjustment.

[0045] Step 2: Perform multiple rounds of gradient rate coarse corrosion and approach the target through real-time feedback; first, determine the target depth H of the back cavity, set the total coarse corrosion target depth to P% of H (P = 80%~95%, dynamically adjusted according to the H value), and plan the coarse corrosion process.

[0046] S21. Parameter Planning: Determine the total number of rough corrosion cycles N (2-5 cycles). Allocate the theoretical corrosion depth ha for each cycle according to a descending ratio. i The ratio is (N:N-1:…:1), ensuring that ha i >ha i+1 Furthermore, the depth ratio between adjacent rounds is between 1.2 and 2.0.

[0047] S22. Rate Matching: Based on the correspondence table established in step 1, match the corresponding corrosion temperature for each round to make the corrosion rate V... i Decrease with increasing round number (V) i >V i+1 The rate reduction was controlled within 0.5–2.0 μm / min.

[0048] S23. Execution and Real-time Correction: Execute each round of erosion according to the plan. After each round, immediately check the current cumulative depth ∑Ha. i And calculate the actual depth Ha of a single wheel. i Calculate the single-wheel depth deviation Δha i =Ha i -ha i The threshold δa (0.2~0.5μm) is adjusted based on the coarse corrosion deviation for judgment and processing.

[0049] If |Δha i |≥δa, perform instantaneous parameter fine-tuning: when Δha i When the value is greater than 0, the time for the next corrosion cycle is shortened (the amount of shortening is ΔT). i+1 =k1×Δha i (k1 = 0.2–0.5 min / μm) or reduce the temperature by 2–3 °C; when Δha i If the value is less than 0, extend the next round of time or increase the temperature by 2-3°C. The theoretical depth for the next round will be executed according to the original planned value.

[0050] If |Δha i If |<δa, then the corrosion time and temperature parameters for the next round will not be adjusted, but the deviation Δhai from the current round will be compensated for in the theoretical depth of the next round. The theoretical depth to be executed in the next round is ha′. i+1 Adjusted to: ha′ i+1 =ha i+1 -Δha i This single-round deviation compensation mechanism can instantly correct minor deviations and prevent deviations from accumulating.

[0051] S24, Phase Transition: When the cumulative actual depth ∑Ha i The coarse corrosion stage ends when the transition condition is met, i.e., the absolute deviation from the target value H×P% does not exceed the allowable transition deviation E. The value of E is the larger of (0.5%~2%)×H and (0.5~3.0)μm, ensuring a balance between high efficiency and controllability in the process.

[0052] Step 3: Perform multiple rounds of low-rate fine corrosion to achieve precise closed-loop control; after the coarse corrosion stage, calculate the remaining precise depth to be corroded ΔH = H - ∑Ha N It then enters the fine corrosion stage.

[0053] S31. Parameter Planning: Determine the total number of fine corrosion cycles M (1-3 cycles). Allocate the theoretical corrosion depth hb for each cycle according to a descending ratio. j The ratio is (M:M-1:…:1), ensuring hb j hb j+1 Furthermore, the depth ratio between adjacent rounds is between 1.5 and 3.0.

[0054] S32, Rate Matching: Based on the correspondence table in step 1, a lower corrosion temperature is matched for each round to improve the fine corrosion rate V. j Decrease with increasing round number (V) j >V j+1 ), and all V j All are less than the rate V of the last round of coarse corrosion. N The minimum speed is not less than 0.1 μm / min.

[0055] S33. Execution and Real-time Correction: Execute each round of fine erosion according to the plan. After each round, immediately check the current cumulative depth ∑Hb. j And calculate the actual depth Hb of a single wheel. j Calculate the depth deviation Δhb of a single wheel. j =Hb j -hb j The threshold δb (0.1~0.3μm) is adjusted based on the fine corrosion deviation for judgment and processing.

[0056] If |Δhb j |≥δb, perform instantaneous parameter fine-tuning: when Δhb j When the value is greater than 0, the time for the next corrosion cycle is shortened (the amount of shortening is Δt). j+1 =k2×Δhb j (k2 = 0.1–0.3 min / μm) or reduce the temperature by 1–2 °C; when Δhb j If the value is less than 0, extend the next round of time or increase the temperature by 1-2°C. The theoretical depth for the next round will be executed according to the original planned value.

[0057] If |Δhb j If |<δb, then the corrosion time and temperature parameters for the next round will not be adjusted, but the deviation Δhbj from the current round will be compensated for in the theoretical depth of the next round. The theoretical depth hb′ to be executed in the next round. j+1 Adjusted to: hb′ j+1 =hb j+1 -Δhb j .

[0058] S34. Process Completion: When the final back cavity depth satisfies |H-(∑Ha) N +∑Hb M When |≤ε, the process is complete. The final allowable error ε is 0.5μm, meaning the depth deviation is precisely controlled within ±0.5μm.

[0059] The present invention will be further described in detail below with reference to specific embodiments. Each embodiment adopts the TMAH or KOH anisotropic wet etching system and strictly follows the core technical logic of "pre-set QC sheet experimental table establishment → multi-round gradient coarse etching (including real-time correction in each round) → multi-round low-rate fine etching (including real-time correction in each round) → depth target achieved". In order to centrally verify the technical advantages of gradient rate control and deviation correction, each embodiment uniformly sets the coarse etching target depth to 95% of the total target depth H of the back cavity (i.e., P = 95%), so that the fine etching stage undertakes the task of precise correction of 5% × H. In each embodiment, the coarse etching deviation threshold δa = 0.3 μm, the fine etching deviation threshold δb = 0.2 μm, the transfer deviation E is taken as the larger value between (0.5%~2%) × H and 0.5~3.0 μm, and the final tolerance ε = 0.5 μm.

[0060] Example 1

[0061] This invention specifically provides a method for etching the back cavity depth of a MEMS pressure sensor based on gradient rate control and real-time deviation correction. The target depth in this embodiment is 300 μm (using 20% ​​KOH etching solution, with time correction as the primary method), and specifically includes:

[0062] Step 1: Pre-set QC wafer experiment to establish temperature-etching rate benchmark; Select a P-type (100) crystal orientation single crystal silicon wafer with the same material and crystal orientation as the substrate of the MEMS pressure sensor to be etched as the QC wafer (quality control wafer). Pour 20% KOH etching solution into the constant temperature etching bath, magnetically stir at 200 r / min, turn on the constant temperature control system, and set the test temperature range to 65℃~90℃ with a temperature gradient of 5℃. Place 3 QC wafers in parallel at each temperature point and uniformly etch for 10 min; After etching, immediately remove the QC wafers, rinse and dry them quickly. Measure 5 points on each QC wafer using a laser interferometer, take the average value, and finally establish the temperature-etching rate correspondence table as shown in Table 1 below.

[0063] Table 1: Relationship between Temperature and Corrosion Rate in 20% KOH Corrosion Solution

[0064] Temperature (°C) Corrosion rate (μm / min) 65 1.2 70 1.8 75 2.5 80 3.2 85 3.8 90 4.5

[0065] Step 2: Multi-round gradient rate coarse erosion (N=3 rounds, 3:2:1 ratio);

[0066] S21. Core parameter planning: Target depth of back cavity H = 300 μm, target depth of coarse corrosion ∑ha = 300 × 95% = 285.00 μm; total number of coarse corrosion cycles N = 3 cycles, with theoretical depths of each cycle allocated in a 3:2:1 ratio: ha1 = 285.00 × (3 / 6) = 142.50 μm; ha2 = 285.00 × (2 / 6) = 95.00 μm; ha3 = 285.00 × (1 / 6) = 47.50 μm.

[0067] S22. Rate and temperature matching: Set the coarse corrosion rate gradient to decrease by 1.0 μm / min per round (meeting the adjustment range requirement of 0.5~2.0 μm / min). Match the temperature of each round according to Table 1: V1=3.2 μm / min (corresponding to 80℃); V2=2.2 μm / min (interpolation calculation corresponds to 73℃); V3=1.2 μm / min (corresponding to 65℃).

[0068] The theoretical corrosion time is calculated as follows: T1 = 142.50 ÷ 3.2 ≈ 44.53 min; T2 = 95.00 ÷ 2.2 ≈ 43.18 min; T3 = 47.50 ÷ 1.2 ≈ 39.58 min.

[0069] S23. Implementation and Real-Time Deviation Correction:

[0070] First round (80℃, 44.53 min): Corrosion time was 44.53 min; the actual depth was detected as Ha1 = 143.27 μm. The single-round depth deviation Δha1 was calculated as: Δha1 = Ha1 - ha1 = 143.27 - 142.50 = +0.77 μm. Since |Δha1| = 0.77 μm ≥ δa = 0.3 μm, real-time parameter fine-tuning and compensation for the next round are required.

[0071] ① Immediate fine-tuning: Δha1>0 (actual corrosion excess), adopt the method of shortening the corrosion time of the next round, take k1=0.3min / μm, and the time reduction ΔT2=k1×|Δha1|=0.3×0.77=0.23min;

[0072] ② Determination of the compensation amount for the next round: Due to excessive corrosion in this round, the corresponding corrosion depth needs to be reduced in the next round. Therefore, the compensation amount for the next round = -Δha1 = -0.77μm;

[0073] ③ Corrected parameters: The original theoretical depth ha2 in the second round was 95.00 μm, and the corrected ha2' was ha2 + compensation for the next round = 95.00 + (-0.77) = 94.23 μm; the original theoretical time T2 was 43.18 min, and the adjusted actual corrosion time was T2 - ΔT2 = 43.18 - 0.23 = 42.95 min;

[0074] Second round (73℃, 42.95 min): Corrosion time was 42.95 min; the actual depth Ha2 was measured to be 94.76 μm, and the calculated deviation Δha2 = Ha2 - ha2' = 94.76 - 94.23 = +0.53 μm. Since |Δha2| = 0.53 μm ≥ δa = 0.3 μm, further real-time parameter fine-tuning and compensation in the next round are required.

[0075] ① Immediate fine-tuning: Δha2>0 (still excessive corrosion), continue to shorten the corrosion time for the next round, ΔT3=k1×|Δha2|=0.3×0.53=0.159min≈0.16min;

[0076] ② Determination of the next round of compensation: The next round of compensation = -Δha2 = -0.53μm;

[0077] ③ Corrected parameters: The original theoretical depth of the third round ha3=47.50μm, the corrected ha3'=ha3+compensation amount of the next round=47.50+(-0.53)=46.97μm; the original theoretical time T3=39.58min, the adjusted actual corrosion time=T3-ΔT3=39.58-0.16=39.42min;

[0078] Third round (65℃, 39.42 min): Corrosion time was 39.42 min; the actual depth Ha3 was measured to be 46.89 μm, and the calculated deviation Δha3 = Ha3 - ha3' = 46.89 - 46.97 = -0.08 μm. Since |Δha3| = 0.08 μm < δa = 0.3 μm, no immediate parameter fine-tuning or subsequent compensation is required. The cumulative coarse corrosion depth ∑Ha3 = 143.27 + 94.76 + 46.89 = 284.92 μm, and the calculated deviation from the target coarse corrosion value (285.00 μm) is 284.92 - 285.00 = -0.08 μm. Since the deviation E = max((0.5%~2%)×300μm, 0.5~3.0μm) = max(1.5~6μm, 0.5~3.0μm) = 6μm (calculated using the upper limit), and the current deviation -0.08μm≤E, the transition condition is met, and the coarse corrosion stage ends.

[0079] Step 3: Multiple rounds of low-rate fine etching (M = 2 rounds, 2:1 ratio);

[0080] S31. Core parameter planning: Remaining depth to be etched ΔH = 300.00 - 284.92 = 15.08 μm; Total number of fine etching cycles M = 2 cycles, with theoretical depths allocated in a 2:1 ratio: hb1 = 15.08 × (2 / 3) ≈ 10.05 μm, hb2 = 15.08 × (1 / 3) ≈ 5.03 μm.

[0081] S32. Rate and temperature matching: The fine corrosion rate is less than the last round of coarse corrosion rate V3 = 1.2 μm / min. The rate gradient is set to decrease by 0.3 μm / min per round: V1 = 0.8 μm / min (interpolation corresponding to 66℃), V2 = 0.5 μm / min (interpolation corresponding to 62℃); The theoretical time is calculated as: t1 = 10.05 ÷ 0.8 ≈ 12.56 min, t2 = 5.03 ÷ 0.5 ≈ 10.06 min.

[0082] S33. Implementation and Real-Time Deviation Correction:

[0083] First round (66℃, 12.56 min): Corrosion time is 12.56 min; measured actual depth Hb1 = 9.90 μm, calculated single-round deviation Δhb1 = Hb1 - hb1 = 9.90 - 10.05 = -0.15 μm. Since |Δhb1| = 0.15 μm < δb = 0.2 μm, no immediate parameter fine-tuning is needed, but the compensation amount for the next round needs to be set: next round compensation amount = -Δhb1 = +0.15 μm; corrected parameters: second round original theoretical depth hb2 = 5.03 μm, corrected hb2′ = hb2 + next round compensation amount = 5.03 + 0.15 = 5.18 μm; original theoretical time t2 = 10.06 min, corrected corrosion time t2′ = hb2′ ÷ v2 = 5.18 ÷ 0.5 = 10.36 min;

[0084] Second round (62℃, 10.36 min): Corrosion time was 10.36 min; the actual depth Hb2 was measured to be 5.21 μm, and the calculated deviation Δhb2 = Hb2 - hb2′ = 5.21 - 5.18 = +0.03 μm. Since |Δhb2| = 0.03 μm < δb = 0.2 μm, no further correction is needed. The cumulative fine corrosion depth ∑Hb2 = 9.90 + 5.21 = 15.11 μm;

[0085] S34. Final Depth Verification: Final back cavity depth = cumulative coarse corrosion depth + cumulative fine corrosion depth = 284.92 + 15.11 = 300.03 μm, which deviates from the target depth of 300.00 μm by +0.03 μm. This meets the process completion requirement of |H-(∑Ha3+∑Hb2)|≤ε=0.5 μm, and no additional corrosion is required.

[0086] Example 2

[0087] This invention specifically provides a method for etching the back cavity of a MEMS pressure sensor based on gradient rate control and real-time deviation correction. The target depth in this embodiment is 360 μm (20% KOH etching solution, mainly temperature correction), and specifically includes:

[0088] Step 1: The temperature-corrosion rate relationship is directly referenced from Table 1 of Example 1, without the need to repeat the QC sheet experiment.

[0089] Step 2: Multi-round gradient rate coarse erosion (N=4 rounds, 4:3:2:1 ratio)

[0090] S21. Core parameter planning: Target depth of back cavity H = 360 μm, target depth of coarse corrosion ∑ha = 360 × 95% = 342.00 μm; total number of coarse corrosion cycles N = 4 cycles, with theoretical depths allocated to each cycle in a 4:3:2:1 ratio: ha1 = 342.00 × (4 / 10) = 136.80 μm; ha2 = 342.00 × (3 / 10) = 102.60 μm; ha3 = 342.00 × (2 / 10) = 68.40 μm; ha4 = 342.00 × (1 / 10) = 34.20 μm.

[0091] S22. Rate and temperature matching: Set the coarse corrosion rate gradient to decrease by 1.0 μm / min per round (meeting the adjustment range requirement of 0.5~2.0 μm / min). Match the rate and temperature of each round according to Table 1: V1=3.8 μm / min (corresponding to 85℃); V2=2.8 μm / min (interpolation calculation corresponds to 82℃); V3=1.8 μm / min (corresponding to 70℃); V4=0.8 μm / min (interpolation calculation corresponds to 63℃).

[0092] The theoretical corrosion time is calculated as follows: T1 = 136.80 ÷ 3.8 ≈ 36.00 min; T2 = 102.60 ÷ 2.8 ≈ 36.64 min; T3 = 68.40 ÷ 1.8 = 38.00 min; T4 = 34.20 ÷ 0.8 = 42.75 min.

[0093] S23. Implementation and Real-Time Deviation Correction:

[0094] First round (85℃, 36.00 min): Corrosion time is 36.00 min; the actual depth detected is Ha1 = 136.15 μm, and the single-round depth deviation Δha1 = Ha1 - ha1 = 136.15 - 136.80 = -0.65 μm is calculated. Since |Δha1| = 0.65 μm ≥ δa = 0.3 μm, immediate parameter fine-tuning + compensation for the next round is required: ① Immediate fine-tuning: Δha1 < 0 (actual corrosion is insufficient), the corrosion temperature for the next round is increased (temperature correction is the main method), with an increase of 2℃; ② Determination of compensation for the next round: Because the corrosion in this round is insufficient, the corresponding corrosion depth needs to be increased in the next round, so the compensation for the next round = -Δha1 = +0.65 μm; ③ Corrected parameters: the original theoretical depth for the second round. ha2 = 102.60 μm, after correction ha2' = ha2 + next round compensation amount = 102.60 + 0.65 = 103.25 μm; the original matching temperature was 82℃, after adjustment the actual corrosion temperature = 82℃ + 2℃ = 84℃, the corresponding corrosion rate is calculated by interpolation to be 3.0 μm / min, the original theoretical time T2 = 36.64 min, after correction the corrosion time = ha2' ÷ 3.0 = 103.25 ÷ 3.0 ≈ 34.42 min;

[0095] Second round (84℃, 34.42min): Corrosion time was 34.42min; the actual depth Ha2 was measured to be 102.28μm, and the calculated deviation Δha2 = Ha2 - ha2' = 102.28 - 103.25 = -0.97μm. Since |Δha2|=0.97μm≥δa=0.3μm, further real-time parameter fine-tuning and compensation in the next round are required: ① Real-time fine-tuning: Δha2<0 (still insufficient corrosion), continue to increase the corrosion temperature in the next round by 3℃; ② Determine the compensation amount for the next round: compensation amount for the next round = -Δha2=+0.97μm; ③ Corrected parameters: the original theoretical depth ha3=68.40μm in the third round, and the corrected ha3'=ha3+compensation amount for the next round =68.40+0.97=69.37μm; the original matching temperature was 70℃, and the adjusted actual corrosion temperature =70℃+3℃=73℃, with a corresponding corrosion rate calculated by interpolation as 2.5μm / min; the original theoretical time T3=38.00min, and the corrected corrosion time = ha3'÷2.5=69.37÷2.5≈27.75min;

[0096] Third round (73℃, 27.75min): Corrosion time was 27.75min; the actual depth Ha3 was measured to be 68.52μm, and the deviation Δha3 was calculated to be Ha3-ha3'=68.52-69.37=-0.85μm. Since |Δha3|=0.85μm≥δa=0.3μm, immediate parameter fine-tuning + next round compensation is required: ① Immediate fine-tuning: Δha3<0 (still insufficient corrosion), increase the next round corrosion temperature by 2℃; ② Determine the next round compensation amount: next round compensation amount=-Δha3=+0.85μm; ③ Corrected parameters: the original theoretical depth ha4=34.20μm in the fourth round, the corrected ha4'=ha4+next round compensation amount=34.20+0.85=35.05μm; the original matching temperature was 63℃, the adjusted actual corrosion temperature=63℃+2℃=65℃, the corresponding corrosion rate is 1.2μm / min (refer to Table 1), the original theoretical time T4=42.75min, the corrected corrosion time=ha4'÷1.2=35.05÷1.2≈29.21min;

[0097] Fourth round (65℃, 29.21 min): Corrosion time was 29.21 min; the actual depth Ha4 was measured to be 35.11 μm, and the calculated deviation Δha4 = Ha4 - ha4' = 35.11 - 35.05 = +0.06 μm. Since |Δha4| = 0.06 μm < δa = 0.3 μm, no immediate parameter fine-tuning or subsequent compensation is required. The cumulative coarse corrosion depth ∑Ha4 = 136.15 + 102.28 + 68.52 + 35.11 = 342.06 μm, and the calculated deviation from the coarse corrosion target value (342.00 μm) is 342.06 - 342.00 = +0.06 μm. Since the deviation E = max((0.5%~2%)×360μm, 0.5~3.0μm) = max(1.8~7.2μm, 0.5~3.0μm) = 7.2μm (calculated using the upper limit), and the current deviation + 0.06μm ≤ E, the transition condition is met, and the coarse corrosion stage ends.

[0098] Step 3: Multiple rounds of low-rate fine etching (M = 2 rounds, 2:1 ratio)

[0099] S31. Core parameter planning: Remaining depth to be etched ΔH = 360.00 - 342.06 = 17.94 μm; Total number of fine etching cycles M = 2 cycles, with theoretical depths allocated in a 2:1 ratio: hb1 = 17.94 × (2 / 3) = 11.96 μm, hb2 = 17.94 × (1 / 3) = 5.98 μm.

[0100] S32. Rate and temperature matching: The fine corrosion rate is less than the last round of coarse corrosion rate V4 = 0.8 μm / min. The rate gradient is set to decrease by 0.3 μm / min per round: v1 = 0.7 μm / min (interpolation calculation corresponds to 62℃), v2 = 0.4 μm / min (interpolation calculation corresponds to 58℃); Theoretical calculation time: t1 = 11.96 ÷ 0.7 ≈ 17.09 min, t2 = 5.98 ÷ 0.4 = 14.95 min.

[0101] S33. Implementation and Real-Time Deviation Correction:

[0102] First round (62℃, 17.09min): Corrosion time was 17.09min; the actual depth Hb1 was measured to be 12.28μm, and the single-round deviation Δhb1 was calculated as Hb1-hb1=12.28-11.96=+0.32μm. Since |Δhb1|=0.32μm≥δb=0.2μm, immediate parameter fine-tuning + next round compensation is required: ① Immediate fine-tuning: Δhb1>0 (actual corrosion excess), the next round corrosion temperature is reduced by 1℃; ② Next round compensation amount determined: Next round compensation amount=-Δhb1=-0.28μm; ③ Corrected parameters: The original theoretical depth hb2=5.98μm in the second round, the corrected hb2'=hb2+next round compensation amount=5.98-0.28=5.70μm; the original matching temperature was 58℃, the adjusted actual corrosion temperature=58℃-1℃=57℃, the corresponding corrosion rate is calculated by interpolation to be 0.38μm / min, the original theoretical time t2=14.95min, the corrected corrosion time=hb2'÷0.38=5.70÷0.38≈15.00min;

[0103] Second round (57℃, 15.00 min): Corrosion time was 15.00 min; the actual depth Hb2 was measured to be 5.75 μm, and the calculated deviation Δhb2 = Hb2 - hb2' = 5.75 - 5.70 = +0.05 μm. Since |Δhb2| = 0.05 μm < δb = 0.2 μm, no further correction is needed. The cumulative fine corrosion depth ∑Hb2 = 12.28 + 5.75 = 18.03 μm;

[0104] S34. Final Depth Verification: Final back cavity depth = cumulative coarse corrosion depth + cumulative fine corrosion depth = 342.06 + 18.03 = 360.09 μm, which deviates from the target depth of 360.00 μm by +0.09 μm. This meets the process completion requirement of |H-(∑Ha4+∑Hb2)|≤ε=0.5 μm, and no additional corrosion is required.

[0105] Example 3

[0106] This invention specifically provides a method for etching the back cavity depth of a MEMS pressure sensor based on gradient rate control and real-time deviation correction. The target depth in this embodiment is 390 μm (25% TMAH etching solution, multiple rounds of stable correction), and specifically includes:

[0107] Step 1: Pre-set QC sheet experiment to establish temperature-corrosion rate benchmark;

[0108] The etching solution was 25% TMAH, the test temperature range was set to 70-95℃, the temperature gradient was 5℃, and all other steps were the same as step one in Example 1. The temperature-corrosion rate correspondence table is shown in Table 2 below.

[0109] Table 2: Temperature-Corrosion Rate Correlation of 25% TMAH Etching Solution

[0110] Temperature (°C) Corrosion rate (μm / min) 70 0.8 75 1.3 80 1.9 85 2.5 90 3.1 95 3.7

[0111] Step 2: Multi-round gradient rate coarse erosion (N=5 rounds, 5:4:3:2:1 ratio);

[0112] S21. Core Parameter Planning: Target depth of back cavity H = 420 μm, target depth of coarse corrosion ∑ha = 420 × 95% = 399.00 μm; total number of coarse corrosion cycles N = 5 cycles, with theoretical depths allocated to each cycle in a ratio of 5:4:3:2:1: ha1 = 399.00 × (5 / 15) = 133.00 μm; ha2 = 399.00 × (4 / 15) = 106.40 μm; ha3 = 399.00 × (3 / 15) = 79.80 μm; ha4 = 399.00 × (2 / 15) = 53.20 μm; ha5 = 399.00 × (1 / 15) = 26.60 μm.

[0113] S22. Rate and temperature matching: Set the coarse corrosion rate gradient to decrease by 0.6 μm / min per round (meeting the adjustment range requirement of 0.5~2.0 μm / min). Match the rate and temperature of each round according to Table 2: V1=3.1 μm / min (corresponding to 90℃); V2=2.5 μm / min (corresponding to 85℃); V3=1.9 μm / min (corresponding to 80℃); V4=1.3 μm / min (corresponding to 75℃); V5=0.7 μm / min (interpolation calculation corresponds to 68℃). The theoretical corrosion time is calculated as follows: T1 = 133.00 ÷ 3.1 ≈ 42.90 min; T2 = 106.40 ÷ 2.5 ≈ 42.56 min; T3 = 79.80 ÷ 1.9 ≈ 42.00 min; T4 = 53.20 ÷ 1.3 ≈ 40.92 min; T5 = 26.60 ÷ 0.7 ≈ 38.00 min.

[0114] S23. Implementation and Real-Time Deviation Correction:

[0115] First round (90℃, 42.90min): Corrosion time is 42.90min; the actual depth detected is Ha1=132.25μm, and the single-round depth deviation Δha1=Ha1-ha1=132.25-133.00=-0.75μm is calculated. Since |Δha1|=0.75μm≥δa=0.3μm, real-time parameter fine-tuning (time + temperature joint correction) + compensation for the next round is required: ① Real-time fine-tuning: Δha1<0 (actual corrosion is insufficient), the corrosion temperature for the next round is increased by 1℃ and the corrosion time is extended, taking k1=0.4min / μm, and the time extension ΔT2=k1×|Δha1|=0.4×0.75=0.30min; ② Determination of compensation for the next round: compensation for the next round=-Δha1=+0 0.75μm; ③ Corrected parameters: The original theoretical depth ha2 in the second round was 106.40μm, and the corrected ha2' was ha2 + compensation for the next round = 106.40 + 0.75 = 107.15μm; The original matching temperature was 85℃, and the adjusted actual corrosion temperature was 85℃ + 1℃ = 86℃, with the corresponding corrosion rate calculated by interpolation to be 2.6μm / min; The original theoretical time T2 was 42.56min, and the corrected corrosion time was 42.56 + 0.30 = 42.86min;

[0116] Second round (86℃, 42.86min): Corrosion time was 42.86min; the actual depth Ha2 was measured to be 107.83μm, and the deviation Δha2 = Ha2 - ha2' = 107.83 - 107.15 = +0.68μm. Since |Δha2| = 0.68μm ≥ δa = 0.3μm, further real-time parameter fine-tuning (time + temperature combined correction) and compensation for the next round are required: ① Real-time fine-tuning: Δha2 > 0 (actual corrosion overload), the corrosion temperature for the next round is reduced by 1℃ and the corrosion time is shortened. The time reduction ΔT3 = k1 × |Δha2| = 0.4 × 0.68 = 0.27min; ② Determination of compensation for the next round: Compensation for the next round = -Δha2 = -0.68μm ③ Corrected parameters: The original theoretical depth ha3 in the third round was 79.80 μm, and the corrected ha3' was ha3 + compensation for the next round = 79.80 - 0.68 = 79.12 μm; the original matching temperature was 80℃, and the adjusted actual corrosion temperature was 80℃ - 1℃ = 79℃, with the corresponding corrosion rate calculated by interpolation to be 1.7 μm / min; the original theoretical time T3 was 42.00 min, and the corrected corrosion time was 42.00 - 0.27 = 41.73 min;

[0117] Third round (79℃, 41.73min): Corrosion time was 41.73min; the actual depth Ha3 was measured to be 78.45μm, and the calculated deviation Δha3 = Ha3 - ha3' = 78.45 - 79.12 = -0.67μm. Since |Δha3| = 0.67μm ≥ δa = 0.3μm, immediate parameter fine-tuning (time + temperature joint correction) + compensation for the next round is required: ① Immediate fine-tuning: Δha3 < 0 (actual corrosion is insufficient), increase the corrosion temperature of the next round by 2℃ and extend the corrosion time. The time extension ΔT4 = k1 × |Δha3| = 0.4 × 0.67 = 0.27min; ② Determine the compensation amount for the next round: the compensation amount for the next round = -Δha3 = +0.67μm; ③ Correction Corrected parameters: The original theoretical depth ha4 in the fourth round was 53.20 μm, and the corrected ha4' was ha4 + compensation for the next round = 53.20 + 0.67 = 53.87 μm; the original matching temperature was 75℃, and the adjusted actual corrosion temperature was 75℃ + 2℃ = 77℃, with the corresponding corrosion rate calculated by interpolation to be 1.5 μm / min; the original theoretical time T4 was 40.92 min, and the corrected corrosion time was 40.92 + 0.27 = 41.19 min.

[0118] Fourth round (77℃, 41.19min): Corrosion time was 41.19min; the actual depth Ha4 was measured to be 54.52μm, and the calculated deviation Δha4 = Ha4 - ha4' = 54.52 - 53.87 = +0.65μm. Since |Δha4| = 0.65μm ≥ δa = 0.3μm, immediate parameter fine-tuning (time + temperature joint correction) + compensation for the next round is required: ① Immediate fine-tuning: Δha4 > 0 (actual corrosion excess), reduce the corrosion temperature of the next round by 2℃ and shorten the corrosion time. The time reduction ΔT5 = k1 × |Δha4| = 0.4 × 0.65 = 0.26min; ② Determine the compensation amount for the next round: the compensation amount for the next round = -Δha4 = -0.65μm; ③ Correction Corrected parameters: The original theoretical depth ha5 in the fifth round was 26.60 μm, and the corrected ha5' was ha5 + compensation for the next round = 26.60 - 0.65 = 25.95 μm; the original matching temperature was 68℃, and the adjusted actual corrosion temperature was 68℃ - 2℃ = 66℃, with the corresponding corrosion rate calculated by interpolation to be 0.6 μm / min; the original theoretical time T5 was 38.00 min, and the corrected corrosion time was 38.00 - 0.26 = 37.74 min.

[0119] Fifth round (66℃, 37.74 min): Corrosion time was 37.74 min; the actual depth Ha5 was measured to be 25.88 μm, and the calculated deviation Δha5 = Ha5 - ha5' = 25.88 - 25.95 = -0.07 μm. Since |Δha5| = 0.07 μm < δa = 0.3 μm, no immediate parameter fine-tuning or subsequent compensation is required. The cumulative coarse corrosion depth ∑Ha5 = 132.25 + 107.83 + 78.45 + 54.52 + 25.88 = 398.93 μm, and the calculated deviation from the coarse corrosion target value (399.00 μm) is 398.93 - 399.00 = -0.07 μm. Since the deviation E = max((0.5%~2%)×420μm, 0.5~3.0μm) = max(2.1~8.4μm, 0.5~3.0μm) = 8.4μm (calculated using the upper limit), and the current deviation -0.07μm≤E, the transition condition is met, and the coarse corrosion stage ends.

[0120] Step 3: Multiple rounds of low-rate fine etching (M = 3 rounds, 3:2:1 ratio);

[0121] S31. Core parameter planning: Remaining depth to be etched ΔH = 420.00 - 398.93 = 21.07 μm; Total number of fine etching cycles M = 3 cycles, with theoretical depths allocated in a 3:2:1 ratio: hb1 = 21.07 × (3 / 6) ≈ 10.53 μm, hb2 = 21.07 × (2 / 6) ≈ 7.02 μm, hb3 = 21.07 × (1 / 6) ≈ 3.52 μm.

[0122] S32. Rate and temperature matching: The fine corrosion rate is less than the last round of coarse corrosion rate V5 = 0.6 μm / min. The rate gradient is set to decrease by 0.2 μm / min per round: V1 = 0.5 μm / min (interpolation calculation corresponds to 65℃), V2 = 0.3 μm / min (interpolation calculation corresponds to 61℃), V3 = 0.1 μm / min (interpolation calculation corresponds to 57℃); Theoretical calculation time: t1 = 10.53 ÷ 0.5 ≈ 21.06 min, t2 = 7.02 ÷ 0.3 ≈ 23.40 min, t3 = 3.52 ÷ 0.1 = 35.20 min.

[0123] S33. Implementation and Real-Time Deviation Correction:

[0124] First round (65℃, 21.06min): Corrosion time was 21.06min; the actual depth Hb1 was measured to be 10.78μm, and the single-round deviation Δhb1 was calculated to be Hb1-hb1=10.68-10.53=+0.25μm. Since |Δhb1|=0.15μm≥δb=0.2μm, immediate parameter fine-tuning + next round compensation is required: ① Immediate fine-tuning: Δhb1>0 (actual corrosion excess), shorten the next round corrosion time by taking k2=0.2min / μm, the time reduction Δt2=k2×|Δhb1|=0.2×0.25μm=0.05min; ② Determine the next round compensation amount: next round compensation amount=-Δhb1=-0.25μm; ③ Corrected parameters: the original theoretical depth hb2=7.02μm in the second round, the corrected hb2'=hb2+next round compensation amount=7.02-0.25=6.77μm; the original theoretical time t2=23.40min, the adjusted corrosion time=t2-Δt2=23.40-0.05=23.35min;

[0125] Second round (61℃, 23.35min): Corrosion time was 23.35min; the actual depth Hb2 was measured to be 6.69μm, and the calculated deviation Δhb2 = Hb2 - hb2' = 6.69 - 6.77 = -0.08μm. Since |Δhb2|=0.08μm<δb=0.2μm, no immediate parameter fine-tuning is required. Set the compensation amount for the next round: next round compensation amount=-Δhb2=+0.08μm; corrected parameters: the original theoretical depth hb3=3.52μm for the third round, and the corrected hb3'=hb3+next round compensation amount=3.52+0.08=3.60μm; the original theoretical time t3=35.20min. Since the deviation did not trigger immediate fine-tuning, no time adjustment is required, but the actual corrosion time needs to be calculated based on the corrected corrosion depth, i.e., t3'=hb3'÷v3=3.60÷0.1=36.00min;

[0126] Third round (57℃, 36.00 min): Corrosion time was 36.00 min; the actual depth Hb3 was measured to be 3.62 μm, and the calculated deviation Δhb3 = Hb3 - hb3' = 3.62 - 3.60 = +0.02 μm. Since |Δhb3| = 0.02 μm < δb = 0.2 μm, no further correction is needed. The cumulative fine corrosion depth ∑Hb3 = 10.68 + 6.79 + 3.62 = 21.09 μm;

[0127] S34. Final Depth Verification: Final back cavity depth = cumulative coarse corrosion depth + cumulative fine corrosion depth = 398.93 + 21.09 = 420.02 μm, which deviates from the target depth of 420.00 μm by +0.02 μm. This meets the process completion requirement of |H-(∑Ha5+∑Hb3)|≤ε=0.5 μm, and no additional corrosion is required.

[0128] The above description is merely a description of preferred embodiments of the present invention and is not intended to limit the scope of the present invention in any way. Any changes or modifications made by those skilled in the art based on the above disclosure shall fall within the protection scope of the claims.

Claims

1. A method for deep etching of the back cavity of a MEMS pressure sensor, based on gradient rate control and real-time deviation correction; characterized in that, By constructing a closed-loop control system of "preset benchmark - gradient coarse corrosion - real-time correction - precise fine corrosion", the following steps are included: Step S1: Pre-set experiment; Establish a precise temperature-corrosion rate correspondence through QC sheet experiments; Step S2: Coarse corrosion stage; Based on the target depth of the back cavity, plan a multi-round gradient rate coarse corrosion process, allocate the corrosion depth of each round in descending proportion, and reduce the corrosion rate with each round through temperature control to quickly approach 80% to 95% of the target depth; After each round of corrosion, detect the depth immediately, and make targeted corrections for parameter fine-tuning or deviation compensation based on dual threshold criteria. Step S3: Fine erosion stage; transition to multiple rounds of low-rate fine erosion, continue the gradient rate and real-time correction strategy, and accurately close the loop to the target depth.

2. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 1, characterized in that, The pre-set experiment specifically includes: preparing QC sheets with the same material as the MEMS pressure sensor substrate, and establishing a temperature-corrosion rate correspondence table based on the corrosion rate corresponding to different test temperatures in a TMAH or KOH etching solution system; wherein the test temperature range is 60℃~95℃, the temperature gradient is 2℃~5℃, and the corrosion rate of at least 3 QC sheets is tested at each temperature point, and the average value is taken as the corrosion rate corresponding to that temperature.

3. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 2, characterized in that, The concentration of the TMAH or KOH etching solution is 10%wt to 30%wt, and magnetic stirring is used during the etching process at a stirring rate of 100r / min to 300r / min.

4. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 1, characterized in that, In the coarse corrosion stage, multiple rounds of coarse corrosion are performed according to the target depth H of the back cavity to approach P% of the target depth H, where P is 80% to 95%, and is dynamically adjusted according to the target depth H. Specifically, it includes: Step S21: Determine the total number of roughing corrosion cycles N and the theoretical depth ha for each round of roughing corrosion. i (i = 1, 2, ..., N): Distribute the theoretical depth ha of each round of coarse corrosion in descending order of proportion. i The sum of theoretical depths of N rounds of coarse corrosion ∑ha N P% of the target depth H, and satisfying the theoretical depth ha of the i-th round of coarse corrosion. i >The theoretical depth of the (i+1)th round of coarse corrosion, ha i+1 (i = 1, 2, ..., N-1); Step S22: Adjust the temperature of each round of coarse corrosion: Based on the temperature-corrosion rate correspondence table established in step S1, match the corresponding corrosion temperature for each round of coarse corrosion, so that the coarse corrosion rate V of the i-th round is... i >Rate of coarse corrosion in round i+1 V i+1 (i = 1, 2, ..., N-1); Step S23: Perform multiple rounds of coarse corrosion and correct in real time: Perform the first to Nth rounds of coarse corrosion sequentially; after each i-th round of coarse corrosion, immediately detect the cumulative actual corrosion depth ∑Ha. i And calculate the actual corrosion depth Ha in the i-th round. i ; Calculate the depth deviation Δha of a single wheel i =Ha i -ha i ; If |Δha i If |≥δa, then perform instantaneous parameter fine-tuning: when Δha i When Δha > 0, shorten the corrosion time of the (i+1)th round or reduce the corrosion temperature of the (i+1)th round; when Δha i When δa < 0, extend the corrosion time of the (i+1)th round or increase the corrosion temperature of the (i+1)th round; where δa is the single-round deviation adjustment threshold for coarse corrosion; If |Δha i If |<δa, then the preset time and temperature parameters for the (i+1)th round will not be adjusted, but the deviation Δha of that round will be changed. i The amount is included in the compensation for the next round, and is used to compensate when planning the theoretical depth to be achieved in the (i+1)th round. Step S24: Stage transition judgment: When the cumulative actual depth ∑Ha i When the transition condition is met, the coarse corrosion stage ends and the fine corrosion stage begins; the transition condition is: the cumulative actual depth ∑Ha i The absolute deviation from the target value H×P% shall not exceed the allowable transfer deviation E.

5. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 4, characterized in that, In step S21, the total number of coarse corrosion cycles N is 2 to 5, the ratio of coarse corrosion depth in each cycle is N:N-1:...:1, and the ratio of coarse corrosion depth between two adjacent cycles is 1.2 to 2.

0.

6. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 4, characterized in that, In the fine etching stage, multiple rounds of low-rate fine etching are performed to accurately reach the target depth H; Specifically, it includes: Step S31: Calculate the depth of the target to be corroded, ΔH = H - ∑Ha N Determine the total number of fine corrosion cycles M and the theoretical depth hb for each round of fine corrosion. j (j=1,2,...,M): The theoretical depth hb of each round of fine corrosion is allocated in descending order of proportion. j The sum of the theoretical depths of M-cycle fine corrosion ∑hb M It equals ΔH and satisfies the theoretical depth hb of the j-th round of fine corrosion. j >The theoretical depth of the (j+1)th round of fine corrosion is hb j+1 (j = 1, 2, ..., M-1); Step S32: Adjust the temperature of each round of fine etching: Based on the temperature-etching rate correspondence table established in step S1, match the corresponding etching temperature for each round of fine etching, so that the fine etching rate V of the j-th round is... j >The fine corrosion rate V in the (j+1)th round j+1 Furthermore, all fine corrosion rates were less than the final round of coarse corrosion rate V. N ; Step S33: Perform multiple rounds of fine etching and correct in real time: Perform the first to the Mth rounds of fine etching sequentially; after each jth round of fine etching, immediately detect the cumulative actual etching depth ∑Hb. j And calculate the actual corrosion depth Hb in the j-th round. j ; Calculate the single-wheel depth deviation Δhb j =Hb j -hb j ; If |Δhb j If |≥δb, then perform instantaneous parameter fine-tuning: when Δhb j When Δhb > 0, shorten the corrosion time of the (j+1)th round or decrease the corrosion temperature of the (j+1)th round; when Δhb j When δb < 0, extend the corrosion time of the (j+1)th round or increase the corrosion temperature of the (j+1)th round; where δb is the single-round deviation adjustment threshold for fine corrosion; If |Δhb j If |<δb, then the preset time and temperature parameters for the (j+1)th round will not be adjusted, but the deviation Δhb of that round will be reduced. j The amount is included in the compensation for the next round, and is used to compensate when planning the theoretical depth to be achieved in the (j+1)th round. Step S34: Process completion judgment: When the cumulative actual depth satisfies |H-(∑Ha) N +∑Hb M When |≤ε, the process is complete, where ε is the final allowable error.

7. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 6, characterized in that, In step S23, when performing time fine-tuning, the adjustment amount ΔT i+1 It is proportional to the absolute value of the deviation of a single wheel, that is: ΔT i+1 =k1×|Δha i | where k1 = 0.2 min / μm ~ 0.5 min / μm; In step S33, when performing time fine-tuning, the adjustment amount Δt j+1 It is proportional to the absolute value of the single-wheel deviation, i.e.: Δt j+1 =k2×|Δhb j | where k2 = 0.1 min / μm ~ 0.3 min / μm.

8. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 6, characterized in that, In steps S22 and S32, the corrosion rate is adjusted by decreasing by 0.5 μm / min to 2.0 μm / min per round, and the minimum corrosion rate in the fine corrosion stage is not less than 0.1 μm / min.

9. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 6, characterized in that, In steps S23 and S33, a laser interferometric thickness gauge or a step gauge is used to detect the actual corrosion depth, with a detection accuracy of not less than 0.01 μm.

10. The method for deep etching of the back cavity of a MEMS pressure sensor as described in claim 6, characterized in that, In step S23, the value range of the coarse corrosion single-wheel deviation adjustment threshold δa is 0.2μm~0.5μm; In step S24, the allowable transfer deviation E is the larger of (0.5% to 2%) × H or 0.5 μm to 3.0 μm; In step S33, the value range of the fine corrosion single-wheel deviation adjustment threshold δb is 0.1μm~0.3μm; In step S34, the final allowable error ε is 0.5 μm.

Images