JP7078525B2 - Method for predicting stress tolerance of thin film transistors - Google Patents

Description

The present invention relates to a method for predicting stress tolerance of a thin film transistor (TFT) used in a display device such as a liquid crystal display or an organic EL display.

Flat panel displays (hereinafter referred to as "FPDs") are required to have higher definition, higher display frequency, and lower power consumption with the spread of televisions and smartphones. Along with this, the TFT used in the circuit for driving the display is required to have high-speed responsiveness, that is, high mobility as a semiconductor characteristic.

As a semiconductor thin film material forming a TFT, an amorphous oxide semiconductor thin film (hereinafter referred to as "oxide semiconductor thin film") has been attracting attention in recent years, and in particular, In-Ga-Zn-O ₄ (hereinafter referred to as "IGZO"). Is being studied as a powerful material. IGZO has higher mobility than conventionally used amorphous silicon, can achieve high definition, and can suppress leakage current, so that it has advantages such as contributing to low power consumption of FPD. Therefore, IGZO is expected to be applied to next-generation displays that require large size, high resolution, and high-speed drive.

However, the oxide semiconductor thin film may have electrical defects introduced into the film due to compositional deviation due to its pluralism or structural fluctuation due to amorphous. In particular, in oxide semiconductor thin films, the carrier concentration that governs the TFT characteristics changes significantly due to lattice defects generated in the film formation process and hydrogen in the film, and the electronic state changes due to the subsequent heat treatment. It is known to affect the quality of TFTs. Therefore, the threshold voltage ( _Vth : Thrashold Voltage) before and after stress application shifts due to the variation in mobility due to the film quality and the negative bias stress under light irradiation, and the switching characteristics change, which affects the TFT characteristics. Is a problem. For example, a TFT using IGZO incorporated in an FPD has a problem that its switching characteristics deteriorate due to stress caused by light exposed during use and applied voltage during standby. Further, in an FPD using an OLED (Organic Light Emitting Diode), there is a problem that Vth shifts due to the influence of a positive drive voltage that causes the OLED to emit light. Since the TFT characteristics are caused by the electronic state of the oxide semiconductor thin film, it is considered that the deterioration of the switching characteristics due to stress is also caused by the change in the electronic state of the oxide semiconductor thin film.

Therefore, in the process of manufacturing an oxide semiconductor thin film, the electronic state in the oxide semiconductor thin film is grasped, the influence of the manufacturing process on the electronic state is evaluated, and the result is fed back to adjust the manufacturing conditions of the TFT. Quality control is important from the viewpoint of improving productivity.

As a test method for measuring the difference ΔV _th (hereinafter, sometimes referred to as “threshold shift ΔV _th ”) of the change in the threshold voltage that affects the switching characteristics, a negative gate voltage (negative) is applied to the TFT in the standby state. The LNBTS (Light Negative Bias Voltage Threshold) test, which is an acceleration test simulating a state in which light irradiation by the stray light of the backlight is continuously applied while the bias) is applied, is adopted. Further, a PBTS (Positive Bias Temperature Stress) test is adopted as an acceleration test simulating a state in which a positive gate voltage (positive bias) is applied to a TFT in a standby state. The LNBTS test and PBTS test measure the amount of change in the threshold voltage before and after stress is applied, and the smaller the threshold shift ΔV _th calculated based on the test results, the better the stress tolerance. , It can be evaluated that it has excellent switching characteristics in practice. The LNBTS test and the PBTS test are widely used as highly reliable evaluation methods, but in order to perform these tests, it is necessary to actually manufacture a TFT with electrodes, which requires time and cost, so that it is easier to use. Moreover, a method capable of accurately evaluating stress tolerance has been desired.

On the other hand, as a method for evaluating the stress resistance of an oxide semiconductor thin film by a non-contact method without attaching an electrode, for example, Patent Document 1 and Patent Document 2 describe a microwave photoconductivity attenuation method (μ). -PCD method) is being viewed. This μ-PCD method is a method that utilizes the fact that when an oxide semiconductor thin film is irradiated with excitation light and microwaves, the reflected wave of the microwave from the oxide semiconductor thin film changes due to the irradiation of the excitation light. be. Then, in Patent Document 1, the lifetime value is obtained from the time-dependent change in the reflection intensity of the slow microwave wave observed in about 1 μs after the irradiation of the excitation light is stopped, and the stress tolerance is evaluated. Further, in Patent Document 2, after the irradiation of the excitation light is stopped, the reflectance of the microwave is measured for each elapsed time, the product of the elapsed time and the reflectance is calculated, and the time constant at which the value of this product becomes the maximum is used. Evaluating stress tolerance.

Japanese Patent No. 6152348 Japanese Patent No. 6250855

By the way, in a general method for manufacturing a TFT, for example, as shown in FIG. 1, an oxide semiconductor thin film is formed after forming a gate electrode, followed by pre-annealing, film formation and patterning of a protective film, and a source / drain electrode. It goes through each process of film formation and patterning, film formation of protective film, contact hole etching, post-annealing, etc. During this series of steps, the oxide semiconductor thin film is exposed to various processing conditions, and its characteristics also change in various ways. In particular, pre-annealing with heat treatment, film formation of a protective film, post-annealing, etc. have a great influence on the change in the characteristics of the oxide semiconductor thin film.

However, in Patent Document 1 and Patent Document 2, the stress resistance method by the μ-PCD method is evaluated by any of the steps after the oxide semiconductor thin film is formed. Further, although some parameters are obtained by the μ-PCD method, the evaluation is performed by individual parameters, and the evaluation is not performed by combining the parameters. Therefore, it cannot be said that the evaluation by individual parameters in one process is reflected in other processes, and the accuracy of the prediction of ΔVth may be insufficient.

The present invention has been made in view of the above circumstances, and an object of the present invention is to more accurately predict ΔVth by the μ-PCD method.

The method for predicting the stress tolerance of a thin film transistor in the present invention that has solved the above problems is
The oxide semiconductor thin film is irradiated with the excitation light and the microwave by the microwave photoelectric attenuation method, and the reflection intensity of the microwave changed by the irradiation of the excitation light is measured to measure the stress of the thin film including the oxide semiconductor thin film. It ’s a way to predict tolerance,
In the film forming step of forming the oxide semiconductor thin film at the time of manufacturing the thin film transistor and the first step of obtaining various parameters for each step by measuring by the microwave photoelectric attenuation method in the post-step of the film forming step. ,
The second step of measuring the fluctuation ΔVth of the threshold voltage before and after applying the light irradiation / bias voltage of the manufactured thin film transistor, and
The third step of performing multiple regression analysis using the various parameters obtained in the first step as explanatory variables and the ΔVth obtained in the second step as the objective variable to obtain a regression equation, and the third step.
When a new oxide semiconductor thin film is formed, the parameters of the same type as the regression equation are obtained by the microwave photoelectric attenuation method, and the ΔVth of the thin film transistor having the new oxide semiconductor coating is predicted from the regression equation. With a process,
The various parameters obtained in the first step are characterized by including at least the following parameters (a), (b) and (c) in each step accompanied by heat treatment.
(A) Peak value of the reflection intensity
(B) Lifetime value after about 1 μs of excitation light irradiation stop
(C) Time constant at which the E value calculated by the following formula becomes the maximum when the value obtained by measuring the reflection intensity of the microwave for each elapsed time after the irradiation of the excitation light is stopped is used as the signal value.
E = signal value x elapsed time of signal value

In particular, it is effective in predicting the stress tolerance of a thin film transistor including the oxide semiconductor thin film containing at least one element selected from the group consisting of In, Ga, Zn and Sn.

In the present invention, the oxide semiconductor thin film is measured by the μ-PCD method in the film forming step and the step after the film forming step, and the parameters in two or more steps are combined into an explanatory variable, and the ΔVth of the actually measured TFT is used. The regression equation is obtained by performing multiple regression analysis with. Then, when a new oxide semiconductor thin film is formed, the same type of parameters used in the regression equation are measured to predict ΔVth of the TFT provided with the new oxide semiconductor thin film. Therefore, as in the conventional case, more accurate prediction than the prediction of ΔVth of the TFT based on the parameters obtained by the μ-PCD method in a single step becomes possible.

It is a process diagram which shows the manufacturing process of a general TFT. It is a graph which shows an example of the time-dependent change of the reflection intensity of the microwave of the oxide semiconductor thin film obtained by the μ-PCD method. It is a schematic cross-sectional structure diagram which shows the TFT produced in an Example. It is a graph which shows the correlation between the predicted ΔVth from only the parameter C of post-annealing, and the measured ΔVth. It is a graph which shows the correlation between the predicted ΔVth and the actually measured ΔVth when the parameter C of each process of the patterning of the source electrode and the drain electrode, the film formation of the final protective film, the contact hole etching, and the post-annealing is combined. A graph showing the correlation between the predicted ΔVth and the measured ΔVth when the slope at ¹ / e and the slope at 1 / e2 in the time curve of the post-annealing parameter C, the peak value, and the microwave reflection intensity are combined. be. Parameter C of each step of patterning of source electrode and drain electrode, film formation of final protective film, contact hole etching, post-annealing, peak value, slope at 1 / e and 1 / e in the time curve of the reflection intensity of microwaves. It is a graph which shows the correlation between the predicted ΔVth and the actually measured ΔVth when the slopes in ² are combined.

Hereinafter, embodiments for carrying out the present invention will be described in detail. The present invention is not limited to the embodiments described below.

In the present invention, first, a TFT provided with an oxide semiconductor thin film is actually manufactured, and various types are used by the microwave photoelectric attenuation method (μ-PCD method) in the film forming step of the oxide semiconductor thin film and the post-step of the film forming step. Find the parameters of. At the same time, the fluctuation (ΔVth) of the threshold value before and after the light irradiation and the application of the bias voltage is measured for the manufactured TFT. Then, multiple regression analysis is performed with the parameters of two or more steps as explanatory variables and the measured ΔVth as the objective variable, and the regression equation is calculated.

When manufacturing a TFT, for example, as shown in the process diagram shown in FIG. 1, first, a gate electrode and a gate insulating film are sequentially formed on a glass substrate, then an oxide semiconductor thin film is formed and patterning is performed. Form a semiconductor layer.

After forming the semiconductor layer, as a post-process, first, pre-annealing is performed, a protective film (ESL) is formed, and patterning is performed. Next, a source electrode and a drain electrode are formed into a film, and patterning (SD) is performed. Then, after forming a final protective film (PV) and patterning it, a contact hole is formed by etching and post-annealed (PA) to obtain a TFT.

In the present invention, measurement is performed by the μ-PCD method for each step of the film forming step and the post-step, and various parameters are obtained from the measured values. In the μ-PCD method, after the irradiation of the excitation light is stopped, the reflection intensity of the microwave of the oxide semiconductor thin film is measured, and various parameters are obtained from the change with time of the reflection intensity. FIG. 2 shows a representative example of the change over time in the reflected intensity of microwaves in the oxide semiconductor thin film obtained by the μ-PCD method. When the excitation light is irradiated, the reflection intensity of the microwave of the oxide semiconductor thin film rises sharply and shows a peak value, but after the irradiation of the excitation light is stopped, the reflection intensity gradually attenuates with the passage of time. Shows a curve.

As the measuring device based on the μ-PCD method, for example, the measuring device shown in FIG. 1 of JP2012-33857 can be used. Details of the apparatus are described in the same publication, and are omitted here.

Then, in Patent Document 1, the reflection intensity after about 1 μs of irradiation stop of the excitation light is measured over time, and each lifetime value shown below is calculated based on the measurement result.
(A) The reciprocal of the slope obtained by logarithmically converting the reflection intensity from the maximum value ¹ / e to the maximum value 1 / e2 (B) The reflection intensity (y) of the microwave is expressed by the following equation. Represented by (1), lifetime value τ ₂ when parameter fitting
y = n ₁ exp (-t / τ ₁ ) + n ₂ exp (-t / τ ₂ ) ... (1)
(In the equation, t is the elapsed time (seconds), n ₁ and n ₂ are constants, τ ₁ is the lifetime of the carrier with a short time constant, and τ ₂ is the lifetime value of the carrier with a long time constant.)
(C) The reflection intensity (y) of the microwave is expressed by the following equation (2), and the lifetime value τ ₂ when parameter fitting is performed.
y = n ₁ exp (-t / τ ₁ ) + n ₂ exp (-t / τ ₂ ) ^β ... (2)
(In the equation, t is the elapsed time (seconds), n ₁ and n ₂ are constants, τ ₁ is the lifetime value of the carrier with a short time constant, τ ₂ is the lifetime value of the carrier with a long time constant, β is It is a relaxation factor.)
(D) The reflection intensity (y) of the microwave is represented by the following equation (3), and the lifetime value (parameter B) when parameter fitting is performed.
y = A × exp ^{(-x / B)} ... (3)
(In the formula, x is the elapsed time.)
(E) The reflection intensity (y) of the microwave is represented by the following equation (4), and the lifetime value (parameter C) when parameter fitting is performed.
y = A × x ^c ... (4)
(In the formula, x is the elapsed time.)

Further, in Patent Document 2, after the irradiation of the excitation light is stopped, the reflection intensity of the microwave is measured every elapsed time, and a time constant that maximizes the value (E value) calculated by the following formula is adopted. The signal value in the equation is the reflection intensity at each elapsed time.
E = signal value x elapsed time of signal value

Also in the present invention, each lifetime value of Patent Document 1 and the time constant of Patent Document 2 can be adopted as parameters.

Further, in the above equation (4) of (E), the exponent C (hereinafter referred to as “parameter C”) can be adopted as a parameter.

By the way, in Patent Document 1 and Patent Document 2, the measurement by the μ-PCD method is performed in any step after the film formation of the oxide semiconductor thin film, and the ΔVth of the obtained TFT is predicted. However, as shown in Examples described later, after post-annealing, which is the final step, parameter C is obtained by the μ-PCD method, regression analysis is performed with ΔVth (actual measurement ΔVth) of the actually manufactured TFT, and based on the regression equation. When the correlation between the predicted ΔVth (predicted ΔVth) of the TFT and the actually measured ΔVth was obtained, it was found that the correlation was not sufficient.

Therefore, similarly, from the reflection intensity of the microwave after post-annealing, the peak value and the slope of the curve when it becomes 1 / e of the peak value in the region about 1 μs after the irradiation stop of the excitation light are obtained, and the weight with the measured ΔVth. When the regression equation was obtained by performing regression analysis and the correlation between the predicted ΔVth based on the regression equation and the actually measured ΔVth was obtained, it was found that the correlation was higher than the predicted ΔVth from the parameter C alone.

Furthermore, not only in post-annealing, but also in other steps (for example, patterning of source and drain electrodes, film formation of final protective film, contact hole etching) with parameter C, peak value, 1 / e by the μ-PCD method. The slope of and the slope at 1 / e ² were obtained, and the regression equation was obtained by performing multiple regression analysis with the measured ΔVth. It was found that the correlation was higher than the prediction ΔVth from.

In this way, by combining various parameters by the μ-PCD method in a plurality of steps, more accurate prediction of ΔVth of the TFT becomes possible.

Therefore, regression is performed when a new oxide semiconductor thin film is formed based on the regression equation obtained by performing multiple regression analysis using the parameters obtained in a plurality of steps as explanatory variables and the measured ΔVth of the TFT as the objective variable. By predicting ΔVth using the same parameters as the equation, more accurate prediction becomes possible.

In selecting the parameters, it is preferable to perform measurement by the μ-PCD method in all the steps, but each step of pre-annealing, which is a step involving heat treatment, film formation of a protective film, film formation of a final protective film, and post-annealing. It is preferable to measure by the μ-PCD method. Since the characteristics of the oxide semiconductor thin film change significantly due to the heat treatment, it is preferable to adopt the parameters in the process involving the heat treatment.

The present invention is based on such findings and will be further described in the following examples.

(Manufacturing of TFT)
FIG. 3 is a schematic cross-sectional structural view showing the manufactured etch stopper (ESL) type TFT. First, a Mo thin film was sequentially formed as a gate electrode 42 on a glass substrate 20a having a thickness of 6 inches and a thickness of 0.7 mm, and a SiO ₂ gate insulating film 43 was sequentially formed with a film thickness of 200 nm. The gate electrode 42 was formed by a DC sputtering method using a pure Mo sputtering target. The sputtering conditions were substrate temperature: room temperature and gas pressure: 2 mTorr. Further, the gate insulating film 43 uses a plasma CVD method, and a carrier gas: a mixed gas of SiH ₄ and N ₂ O is mixed with N ₂ O = 100 sccm, SiH ₄ = 4 sccm, N ₂ = 36 sccm, and a film forming power: 200 W or 300 W. The film forming temperature was 320 ° C.

Next, IGZO was formed as an oxide semiconductor thin film 20b by a sputtering method. The film forming conditions are as follows.
-Sputtering device: "CS-200" manufactured by ULVAC, Inc.
-Sputtering target composition: InGaZnO ₄ [In: Ga: Zn = 1: 1: 1 (atomic ratio)]
・ Substrate temperature: Room temperature ・ Film thickness of oxide semiconductor thin film: 40 nm
・ Gas pressure: 1mTorr
-Oxygen addition amount (volume ratio): O ₂ / (Ar + O ₂ ) = 0% to 60%

Then, patterning was performed by photolithography and wet etching. As the wet etchant liquid, "ITO-07N" manufactured by Kanto Chemical Co., Inc. was used.

After the film formation, first, a pre-annealing treatment was performed under the conditions of an atmospheric atmosphere and an atmospheric pressure of 350 ° C. to 600 ° C. for 1 hour.

Next, an etch stop layer was formed as a protective film 45 on the oxide semiconductor thin film 20b under the following conditions.
・ Gas pressure: 75Pa-250Pa
-Film film power: 50W to 250W
-Film film temperature: 160 ° C to 230 ° C
-Film thickness: 100 nm
-Raw material gas: N ₂ O = 25 sccm to 250 sccm, SiH ₄ / N ₂ = 1/9 sccm to 10/90 sccm, SiH ₄ / N ₂ O ratio = 0.024 to 0.12

After forming the protective film 45, patterning was performed by photolithography and wet etching. As the wet etchant liquid, "ITO-07N" manufactured by Kanto Chemical Co., Inc. was used.

Next, using pure Mo, a film was formed so as to have a film thickness of 100 nm by a DC sputtering method, and then patterning was performed to form a source electrode 46a and a drain electrode 46b. The film forming method and the patterning method of the pure Mo film were the same as those of the above-mentioned gate electrode, and the channel length of the TFT was 10 μm and the channel width was 200 μm.

Then, as the final protective film 47, a laminated film of SiO ₂ having a film thickness of 200 nm and SiN having a film thickness of 200 nm was formed. The final protective film 47 was formed by using "PD-220NL" manufactured by SAMCO and using a plasma CVD method. At that time, after plasma treatment with N ₂ O gas, SiO ₂ and SiN were sequentially formed under the following conditions. A mixed gas of SiH ₄ , N ₂ , and N ₂ O was used for forming SiO ₂ , and a mixed gas of SiH ₄ , N ₂ , and NH ₃ was used for forming SiN. In each case, the film forming power was 100 W and the film forming temperature was 150 ° C.
(First layer): SiO ₂
-Carrier gas: SiH ₄ / N ₂ = 4/36 sccm, N ₂ O = 100 sccm
・ Gas pressure: 133Pa
・ Film formation power: 100W
-Film film temperature: 150 ° C
(Second layer): SiN
-Carrier gas: SiH ₄ = 12.5 sccm, N ₂ = 297.5 sccm, NH ₃ = 6 sccm
・ Gas pressure: 133Pa
・ Film formation power: 100W Film formation temperature: 150 ° C

Next, a contact hole 48 for a probe for evaluating transistor characteristics was formed in the final protective film 47 by photolithography and dry etching to obtain an ESL type TFT.

In addition, a BCE type TFT was obtained by the following procedure. After forming the oxide semiconductor layer, the source electrode and the drain electrode were formed without providing the etch stop layer. Of the final protective film formation, heat treatment was performed at 200 ° C. to 600 ° C. for 1 hour in an air atmosphere, a nitrogen atmosphere, or a water vapor atmosphere before and after the formation of the first layer SiO ₂ layer. Finally, a second layer SiN was formed and a contact hole was formed in the same manner as the ESL type TFT.

After forming the contact hole, a TFT sample for stress tolerance measurement was prepared by performing a heat treatment for 30 minutes at 200 ° C. to 300 ° C. under atmospheric pressure in an atmospheric atmosphere or a nitrogen atmosphere as post-annealing.

(Measurement of stress tolerance)
A stress application test was conducted in which light was applied while applying a negative bias to the gate electrodes, simulating the stress environment during actual panel drive. The stress application conditions are as follows. The wavelength of the light was set to about 400 nm, which is close to the band gap of the oxide semiconductor and the transistor characteristics are liable to fluctuate.
・ Gate voltage: -20V
-Substrate temperature: 60 ° C
-Light source: White light source-Intensity of light applied to the TFT as illuminance: 25,000 NIT
-Light irradiation device: YSM-1410 manufactured by Yang Electronics
・ Stress application time: 2 hours

(Correlation between predicted ΔVth and measured ΔVth)
In the above TFT manufacturing process, after each step of patterning the source electrode and drain electrode (SD), forming a final protective film (PV), contact hole etching (CO) and post-annealing (PA), the μ-PCD method is performed. The reflected intensity of the microwave was measured by the above method, and the slope of the curve that became 1 / e of the parameter C, the peak value, and the peak value in each step, and the slope of the curve that became 1 / e ² of the peak value were calculated.

The μ-PCD method was performed under the following conditions.
-Laser wavelength: 349 nm (ultraviolet light)
・ Laser energy: 10μJ
・ Pulse width: 5ns
・ Beam diameter: 1.5 mmφ
・ Number of pulses in one measurement: 256 shots ・ Device: LTA-1610SP (K) (manufactured by Kobelco Kaken Co., Ltd.)

(1) Correlation between the predicted ΔVth from the PA parameter C only and the measured ΔVth Multiple regression analysis was performed between the PA parameter C and the measured ΔVth, and the regression equation was obtained. The regression equation is as follows. Then, the prediction ΔVth based on this regression equation was calculated, and the relationship with the actually measured ΔVth was investigated. FIG. 4 shows a graph in which the predicted ΔVth and the measured ΔVth are plotted.
Regression equation: C (PA) × -0.8984 + (C (PA) -1.3) ² × 2.631 + 4.859
(In the equation, C (PA) is the parameter C of PA.)

Then, the ^R2 value indicating the degree of correlation between the predicted ΔVth and the measured ΔVth was 0.028.

(2) Correlation between the predicted ΔVth from the parameter C of each process of SD, PV, CO, and PA and the measured ΔVth In addition to the PA, the weight of the parameter C of each process of SD, PV, and CO and the measured ΔVth. Regression analysis was performed and the regression equation was obtained. The regression equation is as follows. Then, the prediction ΔVth based on this regression equation was calculated, and the relationship with the actually measured ΔVth was investigated. FIG. 5 shows a graph in which the predicted ΔVth and the measured ΔVth are plotted.
Regression equation: C (PA) × -2.938 + C (CO) × 2.428 + C (PV) × -0.6623 + C (SD) × 0.7456 + 5
(In the equation, C (PA) is the parameter C of PA, C (CO) is the parameter C of CO, C (PV) is the parameter C of PV, and C (SD) is the parameter C of SD.)

Then, the ^R2 value indicating the degree of correlation between the predicted ΔVth and the actually measured ΔVth is 0.032, and the correlation is enhanced by combining the parameter C of another process in addition to the PA.

(3) Correlation between the predicted ΔVth from the PA (parameter C, peak value, slope at 1 / e) and the measured ΔVth In addition to the PA parameter C, the peak value of the reflected intensity of the microwave and the excitation light The slope of the curve when it became 1 / e of the peak value in the region 1 μs after the stop was obtained, and a multiple regression analysis was performed between these values and the actually measured ΔVth to obtain the regression equation. The regression equation is as follows. Then, the prediction ΔVth based on this regression equation was calculated, and the relationship with the actually measured ΔVth was investigated. FIG. 6 shows a graph in which the predicted ΔVth and the measured ΔVth are plotted.
Regression equation: C (PA) x Paak (PA) x -0.002032 + C (PA) / Log1 / e (PA) x 0.05898 + C (PA) / Peak (PA) x-1240.9 + (C (PA)- 1.3) ² / Peak (PA) x 1781.0 + 5.489
(In the equation, C (PA) is the parameter C of PA, Peak (PA) is the peak value of PA, and Log1 / e (PA) is the logarithmic conversion value of the slope which is 1 / e of PA.)

Then, when the ^R2 value indicating the degree of correlation between the predicted ΔVth and the actually measured ΔVth is obtained, it is 0.048, which is higher than that of the parameter C alone.

(4) Correlation between predicted ΔVth and measured ΔVth from each step of SD, PV, CO, PA (parameter C, peak value, slope at ¹ / e, slope at 1 / e2) SD, PV, The parameters C of each step of CO and PA, the peak value of the reflected intensity of the microwave, and the slope of the curve when it becomes 1 / e of the peak value in the region 1 μs after the stop of the excitation light are obtained, and these values and the measured ΔVth Multiple regression analysis was performed with and the regression equation was obtained. The regression equation is as follows. Then, the prediction ΔVth based on this regression equation was calculated, and the relationship with the actually measured ΔVth was investigated. FIG. 7 shows a graph in which the predicted ΔVth and the measured ΔVth are plotted.
Regression formula: 1 / e (PA) x-494.9 + Log1 / e (PA) x-447.5 + C (PA) x Log1 / e ² (PA) x 183.0 + C (PA) x Peak (PA) x- 0.01413 + C (PA) / Peak (PA) × -2029.4 + C (CO) × Log1 / e (CO) × 232.8 + C (CO) × Peak (CO) × 0.01382 + C (CO) / Peak (CO) X-1701.5 + C (CO) x-12.14 + C (PV) x-33.27 + Log1 / e ² (PV) x-145.6 + C (PV) x Log1 / e ² (PV) x 232.2 + C (PV) / Log1 / e ² (PV) x 0.6791 + 1 / e (SD) x -118.0 + Peak (PV) x -0.008818 + Peak (CO) x -0.01036 + Peak (PA) x 0.01254 + (C (PA) -1.3) ² / Peak (PA) x 3723.0 + 51.72
(In the equation, I / e (PA) is the slope of PA 1 / e, Log1 / e (PA) is the logarithmic conversion value of the slope of PA I / e, and C (PA) is the parameter C of PA. Log1 / e ² (PA) is the logarithmic conversion value of the slope that becomes I / e ² of PA, Peak (PA) is the peak value of PA, C (CO) is the parameter C of CO, and Log1 / e (CO) is CO. 1 / e of the slope, Peak (CO) is the peak value of CO, C (PV) is the parameter C of PA, Log1 / e ² (PV) is the logarithmic conversion value of the slope of I / e ² of PV. 1 / e (SD) is the slope that becomes 1 / e of SD, and Pear (PV) is the peak value of PV.)

Then, when the ^R2 value indicating the degree of correlation between the predicted ΔVth and the actually measured ΔVth is obtained, it is 0.330, and the correlation is higher.

As described above, it can be seen that the correlation with the actually measured ΔVth is enhanced by forming the oxide semiconductor thin film and using the parameters obtained by the μ-PCD method in a plurality of steps. Therefore, the regression equation is obtained by multiple regression analysis with the parameters in each step as the explanatory variables and the measured ΔVth as the objective variable, and the μ-PCD method is used for the new oxide semiconductor thin film under the conditions that imitate each step. By performing the measurement and using the same kind of parameters as the regression equation, it is possible to accurately predict the ΔVth of the TFT when the TFT is made of a new oxide semiconductor thin film.

20a Glass substrate 20b Oxide semiconductor thin film 42 Gate electrode 43 Gate insulating film 45 Etch stop layer or protective film 46a Source electrode 46b Drain electrode 47 Final protective film 48 Contact hole

Claims (2)

The oxide semiconductor thin film is irradiated with the excitation light and the microwave by the microwave photoelectric attenuation method, and the reflection intensity of the microwave changed by the irradiation of the excitation light is measured to measure the stress of the thin film including the oxide semiconductor thin film. It ’s a way to predict tolerance,
In the film forming step of forming the oxide semiconductor thin film at the time of manufacturing the thin film transistor and the first step of obtaining various parameters for each step by measuring by the microwave photoelectric attenuation method in the post-step of the film forming step. ,
The second step of measuring the fluctuation ΔVth of the threshold voltage before and after applying the light irradiation / bias voltage of the manufactured thin film transistor, and
The third step of performing multiple regression analysis using the various parameters obtained in the first step as explanatory variables and the ΔVth obtained in the second step as the objective variable to obtain a regression equation, and the third step.
When a new oxide semiconductor thin film is formed, the parameters of the same type as the regression equation are obtained by the microwave photoelectric attenuation method, and the ΔVth of the thin film transistor having the new oxide semiconductor coating is predicted from the regression equation. With a process,
A method for predicting stress tolerance of a thin film transistor, characterized in that at least the following parameters (a), (b) and (c) in each step accompanied by heat treatment are included in the various parameters obtained in the first step .
(A) Peak value of the reflection intensity
(B) Lifetime value after about 1 μs of excitation light irradiation stop
(C) Time constant at which the E value calculated by the following formula becomes the maximum when the value obtained by measuring the reflection intensity of the microwave for each elapsed time after the irradiation of the excitation light is stopped is used as the signal value.
E = signal value x elapsed time of signal value The method for predicting the stress resistance of a thin film transistor according to claim 1 , wherein the oxide semiconductor thin film contains at least one element selected from the group consisting of In, Ga, Zn and Sn.

Images