Method for predicting service life of metal oxide semiconductor type gas sensor

Abstract

The invention discloses a method for predicting the service life of a metal oxide semiconductor type gas sensor, which comprises the following steps of: determining a normal stress (the working state of a rated local gas sensitive material of the sensor) temperature Top, and selecting at least three accelerated stress temperatures Tstressi higher than the Top; the method comprises the following steps: establishing an accelerated aging test system, placing a sensor sample in a constant-temperature environment box, and adjusting the voltage / power of a heating component through an independent programmable power supply, so that the temperature of a gas sensitive material is stabilized at each Tstressi; periodically monitoring and recording sensitivity, baseline resistance and a failure threshold value under a standard condition; for each Tstressi, calculating a median life L50i, and after converting the median life L50i into a Kelvin temperature, fitting based on an Arrhenius equation by taking 1 / Tstressi as an independent variable and ln (L50i) as a dependent variable to obtain failure activation energy Ea; and substituting Ea and Top into an Arrhenius equation, and extrapolating a predicted life Lpred under a normal use condition. The method can accurately and efficiently predict the life, and the effect is good.

Description

Technical Field

[0001] This invention relates to a method for predicting the lifetime of a metal oxide semiconductor gas sensor. Background Technology

[0002] Metal-oxide-semiconductor (MOS) gas sensors are widely used in environmental monitoring, industrial safety, and smart homes due to their high sensitivity, low cost, and long lifespan. Their core working principle involves heating the sensitive material (such as SnO2 or ZnO) to hundreds of degrees Celsius (typically 200°C to 400°C) using an internal miniature heating element. At this temperature, the target gas undergoes a redox reaction with the surface of the sensitive material, causing a change in the material's resistance, thereby detecting the gas concentration. However, this necessary high-temperature operating environment leads to a slow degradation of sensor performance, primarily manifested as decreased sensitivity and baseline resistance drift. This degradation is due to physicochemical changes in the sensitive material under prolonged high temperatures, such as sintering, grain growth, and catalyst deactivation. Therefore, accurately predicting the lifespan of MOS sensors is crucial for assessing system reliability and developing preventative maintenance strategies. Accelerated life testing (ALT) is a key technology for evaluating the reliability of long-life products. Its core principle is to apply enhanced stress to induce product failure within a short period, and then extrapolate the lifespan to normal operating conditions using a physical model. The Arrhenius equation is a classic model describing the relationship between temperature stress and chemical reaction rate, i.e., aging rate in this case, and is widely used in the ALT of electronic components.

[0003] Currently, the ALT method in existing technologies involves high-temperature aging in a constant-temperature environment chamber. This method places multiple MOS sensor samples in a high-temperature environmental test chamber, for example, setting the ambient temperature to 85℃, 105℃, 125℃, etc., and maintaining the sensors at their rated operating voltage continuously or intermittently, periodically monitoring their performance parameters until failure. This method suffers from inaccurate stress localization and low acceleration efficiency. The aging driving force of the MOS sensor is mainly the operating temperature of the local sensitive material area, with the ambient temperature having only a minor impact. Existing technologies indirectly affect the temperature of the gas-sensitive material's heating zone by increasing the ambient temperature. However, this method has a limited temperature rise, and the actual temperature of the gas-sensitive material's heating zone is difficult to precisely control and quantify, resulting in extremely low acceleration efficiency and a still very long testing cycle. Furthermore, it introduces confounding variables, leading to poor model accuracy. In this approach, the actual temperature of the sensor's gas-sensitive material is the superposition of "ambient temperature" and "self-heating," making it an uncontrollable variable. This makes it impossible to establish a pure "temperature-lifetime" relationship. The Arrhenius model fitted based on this data has low accuracy, and the extrapolated predicted lifetime results are unreliable. It may also introduce atypical failure mechanisms: excessively high ambient temperatures may cause non-core components such as sensor packaging materials and pin solder joints to fail before the sensitive materials. This is not a typical failure mode in actual product use, rendering lifetime prediction meaningless. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for predicting the lifespan of metal oxide semiconductor gas sensors. This method can accurately and efficiently predict the lifespan of MOS gas sensors and has good performance.

[0005] To achieve the above objectives, the present invention provides a method for predicting the lifetime of a metal-oxide-semiconductor gas sensor, comprising the following steps: Step 1, determine the accelerated stress temperature level: First, determine the rated local gas-sensitive material operating temperature T_op of the metal oxide semiconductor gas sensor, and then select at least three accelerated stress temperature levels T_stress_i (i=1,2,3...) that are higher than T_op. Step 2: The accelerated aging test system is constructed by placing all the sensor samples in a constant temperature chamber to stabilize the ambient temperature of the chamber at a preset fixed value. Each sensor sample is connected to an independent programmable power supply. By adjusting the voltage or power of the heating components inside the sensor sample, the local temperature of the gas-sensitive material of each sensor sample is precisely stabilized at each T_stress_i set in Step 1. Step 3: Perform periodic performance monitoring and failure judgment. After completing the initial performance test, allow the sensor to work periodically under standard test conditions. After the cycle time, return it to normal working state. Measure and record the sensitivity, baseline resistance value and failure threshold of each test sensor before and after accelerated aging. Step 4: Establish an accelerated model and calculate the activation energy. For each T_stress_i, calculate the median lifetime L50_i based on the failure time data of all sensor samples at that temperature. Convert each T_stress_i to Kelvin absolute temperature T(K) = T(°C) + 273.15. With 1 / T_stress_i as the independent variable and ln(L50_i) as the dependent variable, perform linear regression fitting based on the Arrhenius equation to calculate the failure activation energy Ea of the sensor aging. Step 5: Predict the lifetime under normal operating conditions. Substitute the Ea and T_op obtained in Step 4 into the Arrhenius equation L_pred=L50_stress×exp[(Ea / k_B)×(1 / T_op-1 / T_stress)]; where L50_stress is the median lifetime corresponding to a certain T_stress, and T_stress is the absolute temperature of the accelerated stress temperature. Calculate the predicted lifetime L_pred of the sensor under normal operating conditions.

[0006] The beneficial effects of this setup are as follows: The ambient temperature of the constant-temperature chamber is stabilized at a preset fixed value, which refers to a temperature and humidity value close to normal operating conditions. This method directly uses the temperature of the local gas-sensitive material inside the sensor as the sole accelerating stress, accurately locating the fundamental driving force behind the high-temperature aging of the sensitive material. This avoids interference from the environmental and self-heating superposition of traditional ambient temperature acceleration, significantly improving acceleration efficiency and fully compressing the detection cycle. Relying on the constant-temperature chamber to isolate environmental fluctuations and with independent programmable power supply temperature control, precise quantitative control of the temperature in the heating zone of the gas-sensitive material is achieved, ensuring a pure relationship between temperature and its service life, providing high-precision data support for subsequent model fitting. Periodically switching to unified standard conditions for performance monitoring eliminates parameter errors caused by differences in the test environment. Combined with a clearly defined failure threshold, accurate failure judgment is ensured, accelerating only the aging of the sensitive material, consistent with actual failure modes. Based on measured data, an Arrhenius model is fitted and activation energy is calculated. Theoretical and experimental extrapolation of normal lifespan is combined, resulting in reliable predictions. Ultimately, this provides accurate basis for sensor reliability assessment and preventative maintenance strategy formulation, effectively improving system reliability in environmental monitoring, industrial safety, and other scenarios.

[0007] As a further feature of the present invention, the working cycle in step three is set to Q; when the number of sensor sample failures under a certain accelerated stress temperature level reaches 50% of the total number of samples in the group, the time interval is shortened to 1 / 2 Q.

[0008] The benefits of this setup are as follows: Assuming an initial 48-hour working cycle, this setup ensures the validity of performance monitoring data while avoiding increased operating costs due to excessively frequent testing. It also reduces unnecessary losses in the sensor during accelerated aging and switching between standard tests. When the failure rate of samples reaches 50% under a certain accelerated stress, it indicates that the remaining samples have entered the failure concentration stage. At this point, reducing the interval to 24 hours allows for more intensive tracking of the performance degradation process, significantly reducing the deviation in failure time recording, accurately capturing the failure nodes of the remaining samples, and thus improving the accuracy of median lifetime calculation at that temperature level. This provides more reliable data support for subsequent Arrhenius model fitting and accurate calculation of activation energy and normal lifetime.

[0009] As a further provision of the present invention, the standard test conditions in step three include the standard gas-sensitive material heating zone temperature T_test, the standard ambient temperature, the standard target gas concentration, and the standard target gas concentration flow rate.

[0010] The benefits of this setup are as follows: By limiting the local heating temperature T_test of the gas-sensitive material, the ambient temperature, and the target gas concentration and flow rate to a uniform standard, the interference of fluctuations in test conditions on performance parameters (such as sensitivity and baseline resistance) can be completely avoided. This ensures that the parameter benchmarks are consistent for each monitoring session, accurately reflecting the performance degradation of the sensor due to accelerated aging, rather than spurious deviations caused by changes in environmental or gas conditions. Simultaneously, the unified standard makes the performance data of different accelerated stress temperature groups comparable, providing accurate data support for subsequent failure time determination and median lifetime calculation, further improving the fitting accuracy of the Arrhenius model and the reliability of normal lifetime prediction.

[0011] As a further feature of the present invention, in step 3, the performance failure threshold is 50% ± 2% of the initial sensitivity.

[0012] The benefits of this setting are as follows: it avoids premature false failures due to overly strict thresholds, or overlooking early performance degradation risks due to overly lenient thresholds; and it unifies failure judgment standards across different test scenarios and by different operators through precise control with a certain degree of fluctuation, reducing subjective errors. At the same time, a precise failure threshold ensures more accurate failure time recording, providing reliable data for subsequent median lifetime calculations and Arrhenius model fitting, further improving the credibility of normal lifetime prediction results. Detailed Implementation

[0013] This invention provides an embodiment of a method for predicting the lifetime of a metal oxide semiconductor gas sensor. Taking the H2SMEMS semiconductor gas sensor from Wenzhou Zhigan Technology Co., Ltd. as an example, the lifetime is tested at the rated operating temperature of the gas-sensitive material of 300°C, including the following steps: Step 1: Set T_op = 300℃. Select three acceleration stress levels: T_stress1 = 350℃, T_stress2 = 400℃, T_stress3 = 450℃; Step two: Divide the 12 sensor samples into 3 groups of 4. Place all samples in a constant temperature chamber at 25°C. Using precise power control, stabilize the temperature of the gas-sensitive material in the first group at 350°C, the second group at 400°C, and the third group at 450°C. Step 3: After continuous heating, all sensors underwent initial performance testing at their normal operating temperature of 300℃. Then, all samples were switched to standard test conditions (gas-sensitive material temperature 350℃ / 400℃ / 450℃, 100ppm hydrogen sulfide), maintained for 48 hours of heating, and then returned to the normal operating temperature of 300℃ until stable. Finally, their sensitivity was measured; the failure threshold was defined as 50% of the initial sensitivity. The tests showed that at 450℃, the failure times for the four samples were 300h, 320h, 280h, and 340h, with a median lifetime L50_450 = 310 hours; at 400℃, the median lifetime L50_400 = 800 hours; and at 350℃, the median lifetime L50_350 = 2000 hours.

[0014] Step 4: Convert the temperatures to Kelvin: 1 / 723K, 1 / 673K, 1 / 623K; corresponding lifetime logarithms: ln(310)=5.74, ln(800)=6.68, ln(2000)=7.60; perform linear fitting, obtaining a slope of approximately 11500; calculate the activation energy Ea = slope * k_B ≈ 11500 * 8.617 × 10 -5 e≈0.99eV.

[0015] Step 5, substitute T_op=573K, T_stress=673K (400℃), L50_stress=800h, and Ea=0.99eV into the formula: L_pred=800*exp[(0.99 / 8.617e-5)*(1 / 573-1 / 673); L_pred≈800*exp[11491*(0.001745-0.001486)]≈800*exp[2.97]≈15600h The median lifetime of the sensor is predicted to be approximately 15,600 hours under continuous operation at 300°C.

[0016] The above examples are merely one preferred embodiment of the present invention. Ordinary variations and substitutions made by those skilled in the art within the scope of the technical solution of the present invention are all included within the protection scope of the present invention.

Claims

1. A method for predicting the lifetime of a metal oxide semiconductor gas sensor, characterized by: The method comprises the following steps, Step one, determining the accelerated stress temperature levels: first, determining the rated local gas-sensitive material operating temperature T_op of a metal-oxide semiconductor gas sensor, and then selecting at least three accelerated stress temperature levels T_stress_i (i=1, 2, 3...) higher than the T_op; Step two, building an accelerated aging test system, placing all the sensor samples in a constant temperature environment box, stabilizing the environmental temperature of the constant temperature environment box at a preset fixed value, connecting an independent programmable power supply for each sensor sample, and precisely stabilizing the local temperature of the gas-sensitive material of each sensor sample at each T_stress_i set in step one by adjusting the voltage or power of the internal heating components of the sensor sample; Step three, performing periodic performance monitoring and failure judgment, after completing the initial performance test, allowing the sensor to work periodically under standard test conditions, and then returning to the normal working state after the period of time, measuring and recording the sensitivity, baseline resistance value and failure threshold of each test sensor before and after accelerated aging; Step four, establishing an acceleration model and calculating the activation energy, for each T_stress_i, calculating the median life L50_i according to the failure time data of all sensor samples at the temperature, converting each T_stress_i to Kelvin absolute temperature T(K)=T(℃)+273.15, taking 1 / T_stress_i as the independent variable and ln(L50_i) as the dependent variable, and performing linear regression fitting based on the Arrhenius equation to calculate the failure activation energy Ea of the sensor aging; Step five, predicting the life under normal use conditions, substituting the Ea and T_op obtained in step four into the Arrhenius equation L_pred=L50_stress×exp[(Ea / k_B)×(1 / T_op-1 / T_stress)], wherein L50_stress is the median life corresponding to a certain T_stress, T_stress is the absolute temperature of the accelerated stress temperature, and the predicted life L_pred of the sensor under normal use conditions is calculated.

2. The method for predicting the life of a metal oxide semiconductor gas sensor according to claim 1, characterized by: The working period in step three is set to Q; when the number of failed sensor samples under a certain accelerated stress temperature level reaches 50% of the total number of samples in this group, the time interval is shortened to 1 / 2 Q.

3. The method for predicting the life of a metal oxide semiconductor gas sensor according to claim 1, characterized by: The standard test conditions in step three include the standard gas-sensitive material heating zone temperature T_test, the standard environmental temperature, the standard target gas concentration and the standard target gas concentration flow.

4. The method for predicting the life of a metal oxide semiconductor gas sensor according to claim 1, characterized by: In step 3, the performance failure threshold is 50%±2% of the initial sensitivity.