Sampling scheme evaluation method for steady-state life test of image sensor GSENSE2020BSI

Abstract

The invention provides a sampling scheme evaluation method for a steady-state life test of an image sensor GSENSE2020BSI, and the method comprises the following steps: 1, determining the steady-state failure rate and the standard deviation of the image sensor GSENSE2020BSI stored at normal temperature based on a technology similarity principle; 2, calculating a temperature influence factor; 3, determining the steady-state failure rate and the standard deviation thereof under the black box condition; 4, determining the steady-state failure rate and the standard deviation thereof after the aging test data are fused; 5, calculating the confidence upper limit of the failure rate as the real failure rate of the image sensor GSENSE2020BSI under the steady state condition; step 6, calculating the reliability of the image sensor GSENSE2020BSI under the'zero defect 'sampling method; and 7, evaluating a'zero defect 'sampling scheme. Through the steps, the sampling scheme of the GSENSE2020BSI steady-state life test of the image sensor can be quantitatively evaluated, and the feasibility of the test sampling scheme is determined.

Description

Technical field:

[0001] The invention provides a sampling scheme evaluation method for a steady-state life test of an image sensor GSENSE2020BSI, which relates to a test sampling scheme evaluation method for an image sensor and belongs to the field of reliability evaluation of electronic components. Background technology:

[0002] Image sensors are mainly divided into two categories: CMOS (Complementary Metal Oxide Semiconductor) image sensors and CCD (Charge Coupled Device) image sensors. They can convert received light signals into electrical signals, amplify and denoise the electrical signals, and then output them in the form of digital signals. They are the core hardware equipment of camera technology and are widely used in mobile photography, vehicle-mounted imaging, medical and other fields.

[0003] Compared with CCD, CMOS image sensors have high integration, lower power consumption and faster processing speed. With the continuous breakthroughs in the design level and manufacturing process of integrated circuits in the semiconductor industry, the defects of CMOS image sensors such as high noise and low sensitivity have been overcome one by one, and the image quality is no less than CCD. Therefore, the demand for CMOS image sensors in the aerospace and military fields continues to expand, and they are widely used in weapons and equipment and model missions.

[0004] In terms of quality control, the "zero defect" sampling scheme is widely used in the inspection of semiconductor integrated circuits. This scheme requires that no defects are allowed in the inspection batch. Once a defect is detected, the entire batch is judged to be unqualified. This sampling scheme is based on the binomial distribution formula, suitable for the inspection of continuous and isolated batches, and emphasizes quality control in the production process. Among them, GJB 597 "General Specifications for Semiconductor Integrated Circuits" and GJB 5968 "General Specifications for Charge Coupled Imaging Devices" both adopt the "zero defect" sampling scheme.

[0005] In the sampling method of image sensors, the "zero defect" sampling scheme in GJB 5968-2007 is currently mainly used. This standard stipulates three quality assurance levels of "C1", "C2" and "C3" for CCD devices, and stipulates the sampling schemes for screening, identification and quality consistency inspection for different quality levels. The sampling scheme specified in this specification can provide a reference for the assessment of CMOS image sensors, but due to the differences in the process and imaging principles of CCD and CMOS image sensors, some sampling schemes in the specification are not applicable to the evaluation of CMOS image sensors, and their applicability analysis is needed.

[0006] GSENSE2020BSI is a scientific-grade, back-illuminated CMOS image sensor with a resolution of 4 megapixels. The present invention provides a sampling scheme evaluation method for a steady-state life test of the image sensor GSENSE2020BSI, quantifies the feasibility of the sampling scheme, and then performs a sampling scheme analysis on the steady-state life test of the device to determine the applicability of the sampling scheme. Summary of the invention:

[0007] 1. Purpose: The purpose of the present invention is to provide a sampling scheme evaluation method for the steady-state life test of the image sensor GSENSE2020BSI, so as to provide a basis for the sampling scheme evaluation of the steady-state life test of the CMOS image sensor.

[0008] 2. Technical solution:

[0009] The present invention proposes a sampling scheme evaluation method for a steady-state life test of an image sensor GSENSE2020BSI, which comprises the following steps:

[0010] Step 1: Based on the principle of process similarity, determine the steady-state failure rate and standard deviation of the image sensor GSENSE2020BSI under room temperature storage. GSENSE2020BSI is a CMOS image sensor, and the manufacturing process of CMOS is very similar to that of dynamic random access memory (DRAM). Therefore, the failure rate of DRAM is used instead of the failure rate of CMOS image sensor. The steady-state failure rate λ of the image sensor GSENSE2020BSI under room temperature storage is G and its standard deviation σ G The calculation formula is as follows:

[0011] λ G =exp[1.56+0.151lnB+0.00287(lnB) 2 ]

[0012] σ G =λ G [0.371-0.0186lnB+0.00652(lnB) 2 ]

[0013] Where B is the number of pixels.

[0014] Step 2: Calculate the temperature impact factor. Considering that the failure rate of the device is different under different temperature conditions, combined with the Arrhenius model, the temperature impact factor π T Calculated by the following formula:

[0015]

[0016] Where, T0 is the reference temperature 40℃(313K), T1 is the operating temperature, E a is the activation energy, k is the Boltzmann constant, k = 8.62 × 10 -5 eV / K.

[0017] Step 3: Determine the steady-state failure rate and its standard deviation under black-box conditions. The black-box method uses the steady-state failure rate of the device under room temperature storage as a reference, and predicts the failure rate of the device under different steady-state conditions by changing parameters such as quality grade, electrical stress and temperature.

[0018] The black box method assumes that the device failure rate follows a gamma distribution, and the steady-state failure rate of the device under black box conditions is λ BB for:

[0019] λ BB =λ G π Q π S π T

[0020] The standard deviation of the steady-state failure rate of the device under black box conditions is σ BB for:

[0021] σ BB =σ G π Q π S π T

[0022] Among them, π Q is the mass coefficient, π S is the electric stress influencing factor.

[0023] Step 4: Determine the steady-state failure rate and its standard deviation after integrating the aging test data. When the device is subjected to the aging test, the data obtained from the aging test can be combined with the black box method to obtain the steady-state failure rate λ after integrating the aging test data. SS and its standard deviation σ SS The calculation formula is:

[0024]

[0025] Where n is the number of failed samples in the aging test, and A is the fusion factor, which can be calculated by the following formula:

[0026] A=2+4×10 -6 ×N0Wλ G π Q

[0027] Where N0 is the total number of test samples (i.e. the number of samples), and W is the time factor. W can be calculated by the following formula:

[0028] t1=ta A L

[0029] t e =A b,d t b,d

[0030]

[0031] Where t1 is the effective test time, t a is the actual test time, A L is the temperature acceleration factor, t e is the total effective time of the aging test, A b，d is the aging test temperature acceleration factor, t b，d It is the aging test time.

[0032] Step 5: Calculate the upper confidence limit of the failure rate as the true failure rate of the image sensor GSENSE2020BSI under steady-state conditions. Considering that the device undergoes other screening tests, such as temperature cycling, stability baking and other environmental tests, before the steady-state life test, in addition to the aging screening test, the calculated device failure rate may cause uncertainty and changes. Therefore, a conservative estimate of the predicted failure rate is required.

[0033] Considering that the failure rate of the device follows the gamma distribution, when the failure rate mean λ and standard deviation σ are known, the scale parameter K and shape parameter θ of the gamma distribution can be obtained as follows:

[0034] K=(λ / σ) 2

[0035] θ=σ 2 / λ

[0036] The upper confidence limit λ of the confidence level P% of the gamma distribution P％UCL It can be obtained by the following formula:

[0037] λ P％UCL =Ga -1 (P / 100,K,θ)

[0038] Step 6: Calculate the reliability of the image sensor GSENSE2020BSI under the "zero defect" sampling method. Convert the failure rate of the device to the reliability, and the reliability R(t) can be calculated by the following formula:

[0039] R(t)=e -λt

[0040] For a certain test time, a certain reliability R can be obtained, and then the reliability R is used to indicate the degree to which the image sensor GSENSE2020BSI is accepted:

[0041]

[0042] Combined with the "zero defect" sampling method, let n = 0, and the expression of the reliability of the image sensor GSENSE2020BSI is obtained as follows:

[0043]

[0044] Step 7: Evaluation of the “zero defect” sampling plan. Calculate the production risk α corresponding to the sampling plan:

[0045]

[0046] It can be seen from the producer risk that as the sampling number N0 increases, the producer risk also increases. The industry usually sets the critical value of producer risk at 5%. By calculating the producer risk corresponding to the sampling plan, it can be determined whether the sampling plan for the steady-state life test of the image sensor GSENSE2020BSI is feasible.

[0047] 3. Advantages and effects:

[0048] The present invention provides a sampling scheme evaluation method for a steady-state life test of an image sensor GSENSE2020BSI, and the advantages of the present invention are:

[0049] (1) This method predicts the failure rate of the image sensor GSENSE2020BSI under steady-state life test conditions based on the black box method and calculates the upper confidence limit of the failure rate. GSENSE2020BSI is a CMOS image sensor, which provides a basis for determining the failure rate of CMOS image sensors;

[0050] (2) This method calculates the failure rate of the image sensor GSENSE2020BSI to obtain the reliability of the device, and then obtains the acceptance degree of the image sensor GSENSE2020BSI. Finally, the production risk corresponding to a certain sampling scheme is obtained, which provides a quantitative indicator for the feasibility of the sampling scheme and provides a method reference for the evaluation of the sampling scheme for the steady-state life test of CMOS image sensors.

[0051] (3) This method can obtain the calculation formula of the manufacturer's risk of the product. By determining the manufacturer's risk, the minimum sampling sample size of the sampling plan can be determined, a suitable sampling plan can be formulated, the test cost can be saved, and the reliability of the image sensor GSENSE2020BSI can be better evaluated. Description of the drawings:

[0052] In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings.

[0053] Figure 1 It is a flow chart of the method of the present invention. Specific implementation method:

[0054] The present invention is further described below in conjunction with specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0055] The present invention provides a sampling scheme evaluation method for the steady-state life test of the image sensor GSENSE2020BSI. Specifically, taking the "zero defect" sampling scheme of the steady-state life test in GJB 5968 as an example, 5 devices of this model are selected to carry out the steady-state life test. After the test, no product failure should occur. The test environment temperature is 85°C and the time is 1000h. Before the identification inspection, the samples have 100% passed all screening tests, including the aging test (the aging test environment temperature is 85°C and the time is 240h). The implementation flow chart is as follows: Figure 1 As shown, the specific implementation steps are as follows:

[0056] Step 1: Based on the principle of process similarity, determine the steady-state failure rate and standard deviation of the image sensor GSENSE2020BSI under room temperature storage. GSENSE2020BSI is a CMOS image sensor, and the manufacturing process of CMOS is very similar to that of dynamic random access memory (DRAM). Therefore, the failure rate of DRAM is used instead of the failure rate of CMOS image sensor. The steady-state failure rate λ of the image sensor GSENSE2020BSI under room temperature storage is G and its standard deviation σ G The calculation formula is as follows:

[0057] λ G =exp[1.56+0.151lnB+0.00287(lnB) 2 ]

[0058] σ G =λ G [0.371-0.0186lnB+0.00652(lnB) 2 ]

[0059] Where B is the number of pixels.

[0060] According to the device manual, GSENSE2020BSI is a 4-megapixel resolution CMOS image sensor. The steady-state failure rate λ of the device under general conditions can be obtained. G is 5.9 Fits, and the standard deviation of the steady-state failure rate σ G It is 2.1 Fits.

[0061] Step 2: Calculate the temperature impact factor. Considering that the failure rate of the device is different under different temperature conditions, combined with the Arrhenius model, the temperature impact factor π T Calculated by the following formula:

[0062]

[0063] Where, T0 is the reference temperature 40℃(313K), T1 is the operating temperature, E a is the activation energy (E a =0.45eV), k is the Boltzmann constant, k = 8.62×10 -5 eV / K.

[0064] According to the steady-state life test parameters, the operating temperature T1 is 85°C, and the temperature influence factor π can be calculated. T It is 8.14.

[0065] Step 3: Determine the steady-state failure rate and its standard deviation under black-box conditions. The black-box method uses the steady-state failure rate of the device under room temperature storage as a reference, and predicts the failure rate of the device under different steady-state conditions by changing parameters such as quality grade, electrical stress and temperature.

[0066] The black box method assumes that the device failure rate follows a gamma distribution, and the steady-state failure rate of the device under black box conditions is λ BB for:

[0067] λ BB =λ G π Q π S π T

[0068] The standard deviation of the steady-state failure rate of the device under black box conditions is σ BB for:

[0069] σ BB =σ G π Q π S π T

[0070] Among them, π Q is the mass coefficient, π S is the electric stress influencing factor.

[0071] Check the manual to get the quality grade of GSENSE2020BSI Q is 1, and the electric stress influence factor is π Sis 10.6. Combining the results calculated in step 1 and step 2, we can get the steady-state failure rate λ of GSENSE2020BSI under black box conditions: BB is 508.9 Fits, and the standard deviation of the steady-state failure rate under black box conditions is σ BB It is 181.1 Fits.

[0072] Step 4: Determine the steady-state failure rate and its standard deviation after integrating the aging test data. When the device is subjected to the aging test, the data obtained from the aging test can be combined with the black box method to obtain the steady-state failure rate λ after integrating the aging test data. SS and its standard deviation σ SS The calculation formula is:

[0073]

[0074] Where n is the number of failed samples in the aging test, and A is the fusion factor, which can be calculated by the following formula:

[0075] A=2+4×10 -6 ×N0Wλ G π Q

[0076] Where N0 is the total number of test samples (i.e. the number of samples), and W is the time factor. W can be calculated by the following formula:

[0077] t1=t a A L

[0078] t e =A b,d t b,d

[0079]

[0080] Where t1 is the effective test time, t a is the actual test time, A L is the temperature acceleration factor, t e is the total effective time of the aging test, A b，d is the aging test temperature acceleration factor, t b，d It is the aging test time.

[0081] According to the steady-state life test conditions of the image sensor GSENSE2020BSI, the actual test time t a The aging test time is 1000h. b，d The temperature acceleration factor A is obtained by querying the parameters. L And the aging test temperature acceleration factor A b，dThe values ​​are both 6.4. Based on the above parameters, the time factor W can be obtained to be 3.18, and the fusion factor A can be obtained to be 2.00. Substituting the fusion factor, the number of failed device samples, and the device steady-state failure rate and the standard deviation of the steady-state failure rate under the black box condition obtained in step 3 into the above formula, the steady-state failure rate λ after the fusion of the aging test data can be obtained. SS is 508.8 Fits, and the standard deviation of the steady-state failure rate is σ SS It is 359.7 Fits.

[0082] Step 5: Calculate the upper confidence limit of the failure rate as the true failure rate of the image sensor GSENSE2020BSI under steady-state conditions. Considering that the device undergoes other screening tests, such as temperature cycling, stability baking and other environmental tests, before the steady-state life test, in addition to the aging screening test, the calculated device failure rate may cause uncertainty and changes. Therefore, a conservative estimate of the predicted failure rate is required.

[0083] Considering that the failure rate of the device follows the gamma distribution, when the failure rate mean λ and standard deviation σ are known, the scale parameter K and shape parameter θ of the gamma distribution can be obtained as follows:

[0084] K=(λ / σ) 2

[0085] θ=σ 2 / λ

[0086] The upper confidence limit λ of the confidence level P% of the gamma distribution P％UCL It can be obtained by the following formula:

[0087] λ P％UCL =Ga -1 (P / 100,K,θ)

[0088] Combining the steady-state failure rate and the standard deviation of the steady-state failure rate after the fusion aging test data obtained in step 4, the scale parameter K of the gamma distribution is obtained to be 2.00, and the shape parameter θ is 254.33.

[0089] The failure rate confidence level of 90% is taken as the true failure rate of the image sensor GSENSE2020BSI under steady-state conditions, and the true failure rate λ = 987.44Fits of the image sensor GSENSE2020BSI is finally solved.

[0090] Step 6: Calculate the reliability of the image sensor GSENSE2020BSI under the "zero defect" sampling method. Convert the failure rate of the device to the reliability, and the reliability R(t) can be calculated by the following formula:

[0091] R(t)=e -λt

[0092] For a certain test time, a certain reliability R can be obtained, and then the reliability R is used to indicate the degree of acceptance of the image sensor GSENSE2020BSI:

[0093]

[0094] Combined with the "zero defect" sampling method, let n = 0, and the expression of the reliability of the image sensor GSENSE2020BSI is obtained as follows:

[0095]

[0096] According to the device failure rate calculated in step five, the reliability R of the device can be obtained as 0.999.

[0097] Step 7: Evaluation of the “zero defect” sampling plan. Calculate the production risk α corresponding to the sampling plan:

[0098]

[0099] It can be seen from the producer risk that as the sampling number N0 increases, the producer risk also increases. The industry usually sets the critical value of producer risk at 5%. By calculating the producer risk corresponding to the sampling plan, it can be determined whether the sampling plan for the steady-state life test of the image sensor GSENSE2020BSI is feasible.

[0100] Taking the sampling scheme of the steady-state life test in GJB 5968 as an example, when the number of pixels of the product is greater than 1 million, the number of samples (acceptance number) is 5 (0), that is, none of the 5 products selected can fail after the steady-state life test.

[0101] Substituting N0=5 into the device failure rate, we can obtain:

[0102] α=0.5%

[0103] The risk borne by the manufacturer is 0.5%, which is less than the 5% usually set in the industry and is within the acceptable range of the manufacturer. Therefore, it is determined that the sampling plan for the steady-state life test in GJB 5968 can be used in the identification of the image sensor GSENSE2020BSI.

Claims

1. A sampling scheme evaluation method for a steady-state life test of an image sensor GSENSE2020BSI, characterized in that: The following steps are involved: Step 1: Based on the principle of process similarity, determine the steady-state failure rate λ of the image sensor GSENSE2020BSI under room temperature storage G is 5.9 Fits, and its standard deviation σ G 2.1 Fits; Step 2: Determine the temperature influence factor π when the operating temperature T1 is 85°C T is 8.14, and the steady-state failure rate λ of the image sensor GSENSE2020BSI under black box conditions is obtained BB is 508.9 Fits, and its standard deviation σ BB 181.1 Fits; Step 3: According to the aging test conditions, the time factor W can be obtained by the following formula: t1=t a From L t e =A b,d t b,d Where t1 is the effective test time, t a is the actual test time, A L is the temperature acceleration factor, t e is the total effective time of the aging test, A b，d is the aging test temperature acceleration factor, t b，d It is the time for aging test; According to the aging test time (t b，d = 250h) and actual test time (t a =1000h), the time factor W is determined to be 3.18, and the time factor A of the fusion aging test data is determined to be: A=2+4×10 -6 ×3.18×N0λ G p Q Where, N0 is the total number of test samples; Step 4: Steady-state failure rate λ after integrating aging test data SS and its standard deviation σ SS : Where n is the number of failed samples in the aging test; Step 5: Take the upper confidence limit of the failure rate as the true failure rate λ of the device under steady-state conditions, and then calculate the reliability R of the device. The device reliability R is used to express the degree of acceptance of the image sensor GSENSE2020BSI: Combined with the "zero defect" sampling method, let n = 0, and the expression of the image sensor GSENSE2020BSI reliability L(R) is obtained as follows: Step 6: Calculate the producer risk α corresponding to the sampling plan: Step 7: Evaluation of the "zero defect" sampling scheme. It is usually set in the industry that when the manufacturer's risk α is less than 5%, the sampling scheme for the steady-state life test of the image sensor GSENSE2020BSI is feasible; when the manufacturer's risk α is greater than 5%, the sampling scheme for the steady-state life test of the image sensor GSENSE2020BSI is not feasible.

Images