JP2002141388A - Evaluation method for semiconductor device and evaluation device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately predict and provide the reliability guarantee life of a gate insulation film and the distribution. SOLUTION: The change at the time of the temperature acceleration test of a dielectric breakdown life distribution is adopted as the temperature change of the gradient of a distribution function in the cumulative percent defective. In a method for predicting and obtaining the dielectric breakdown life distribution at a low temperature from the temperature change of the dielectric breakdown life distribution at a high temperature, the dielectric breakdown life distribution at the high temperature is indicated by the distribution function of the cumulative defective percent, the gradient is Arrhenius-plotted and the gradient of the distribution function of the cumulative defective percent is obtained for the dielectric breakdown life distribution at the low temperature from the extrapolation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a semiconductor device evaluation method and a semiconductor device evaluation apparatus. In particular, the present invention relates to an evaluation technique for improving a prediction of a dielectric breakdown life at a low guaranteed temperature from a high temperature accelerated test of a dielectric breakdown life of a thin insulating film such as a gate oxide film.

[0002]

^2. Description of the Related Art Logic, E ² PROM, DRAM,
In semiconductor devices such as LSIs such as SRAMs, reliability assurance is extremely important. In particular, the dielectric breakdown life of the gate insulating film depends on the operation of the device, and therefore needs to be accurately predicted. Further, in recent years, the thickness of the gate insulating film has been reduced, and the dielectric breakdown life has been prolonged. Therefore, it has been practically impossible to actually measure the reliability guaranteed life under the guaranteed temperature and the guaranteed electric field. Therefore, an acceleration test at a temperature higher than the guaranteed temperature and an acceleration test under an electric field higher than the guaranteed electric field have become indispensable evaluation methods.

FIG. 9 shows a flowchart of a conventional method for estimating the dielectric breakdown lifetime of a gate insulating film.

First, as shown in FIG. 9 (a), with respect to the manufactured device, the substrate temperature is higher than room temperature at at least two points, for example, 125 ° C. and 150 ° C., and under a certain stress condition, for example. The dielectric breakdown lifetime is measured at Eox (magnitude of an electric field applied to the oxide film) = 11.0 MV / cm.

Next, as shown in FIG. 9 (b), several points of the cumulative failure rate, for example, 98%, in the obtained measurement data are obtained.
At the failure rate of%, 50%, and 1%, the measured data (dielectric breakdown life) is plotted as a reciprocal of temperature, that is, an Arrhenius plot is performed, and the temperature acceleration constant of each failure rate is obtained.

Next, as shown in FIG. 9C, at one temperature, for example, 125 ° C., the stress intensity is changed by at least two kinds of stress conditions, for example, Eox = 10.5.
At MV / cm and 12.0 MV / cm, the dielectric breakdown lifetime is measured in the same manner. Using the data obtained by this,
Also, as shown in FIG. 9D, a guaranteed temperature under each electric field, for example, a dielectric breakdown lifetime distribution at 85 ° C. is predicted from an Arrhenius plot using a temperature acceleration coefficient obtained by a temperature acceleration test in advance.

Next, as shown in FIG. 9E, in the predicted breakdown life distribution, a guaranteed failure rate, for example, 1
The breakdown lifetime in ppm is estimated and plotted against, for example, the electric field (E). The points thus obtained, ie Eox 10.5 MV / cm, 11.0 MV
/ Cm and 12.0 MV / cm are extrapolated from the least squares method to obtain a life under a guaranteed electric field, for example, Eox = 4 MV / cm, and this is defined as a reliability guaranteed life.

[0008]

However, in this conventional method for estimating the dielectric breakdown life, the temperature dependence of the dielectric breakdown life distribution during the temperature accelerated test is not taken into consideration, so that the temperature under the expected guaranteed temperature is not considered. It was found that the distribution of the dielectric breakdown life in the above did not match the actual distribution of the dielectric breakdown life.

FIG. 1 shows the results of experiments conducted by the inventors. FIG. 1 shows a conventional method in which the breakdown life at a guaranteed temperature of 85 ° C. was predicted by an Arrhenius plot from the breakdown life at 98%, 50%, and 1% of the breakdown life distribution at 125 ° C. and 150 ° C. The result of comparing the result with the actually measured value of the dielectric breakdown life at the guaranteed temperature of 85 ° C is shown.

In FIG. 1, what is plotted with ● is the expected value, and ○ (defective rate 1 ppm), □ (defective rate 1%, Δ
The values plotted with (defective rate 50%) and ◇ (defective rate 98%) are the actually measured values.

As can be seen from FIG. 1, the dielectric breakdown life estimated at each defect rate predicted by the conventional method is longer than the actually measured dielectric breakdown life. The difference is increasing. This means that the reliability guarantee life predicted by the conventional method is overestimated than the actual guarantee life.

SUMMARY OF THE INVENTION The present invention has been made in order to solve such a problem, and an object of the present invention is to provide a semiconductor device evaluation method and an evaluation device for predicting and obtaining an accurate reliability guaranteed life and its distribution. And

[0013]

In order to achieve the above object, the present invention provides a method for evaluating a semiconductor device for determining the distribution of dielectric breakdown lifetime of an insulating film formed on a semiconductor substrate.
The step of measuring the breakdown life distribution at high temperature and the extrapolation of the slope when the breakdown life distribution at high temperature is represented as a distribution function of the cumulative failure rate show the cumulative failure rate in the breakdown life distribution at low temperature. Determining the slope of the distribution function of
A step of predicting the dielectric breakdown life at at least one point of the cumulative failure rate at the low temperature from the Arrhenius plot of the dielectric breakdown life at at least one point of the cumulative failure rate at the high temperature, ,
Extrapolating the dielectric breakdown life at at least one point of the cumulative defect rate at the predicted low temperature to obtain the dielectric breakdown life distribution at the low temperature. .

At this time, it is preferable that the measurement of the dielectric breakdown lifetime distribution at the high temperature is performed at at least two different temperatures.

The distribution function when the dielectric breakdown lifetime distribution at a high temperature is represented as a distribution function of a cumulative failure rate is as follows:
It is preferably a Weibull distribution or a lognormal distribution.

Further, the distribution of the breakdown life at the high temperature is represented by a distribution function of the cumulative failure rate, and the slope thereof is plotted by an Arrhenius plot. Is preferably obtained.

The present invention is also directed to a semiconductor device evaluation apparatus for determining a dielectric breakdown lifetime distribution of an insulating film formed on a semiconductor substrate, comprising: means for measuring a dielectric breakdown lifetime distribution at a high temperature; Means for determining the slope of the distribution function of the cumulative failure rate in the breakdown life distribution at low temperature from the extrapolation of the slope when the breakdown life distribution is expressed as the distribution function of the cumulative failure rate; A means for predicting and obtaining a dielectric breakdown life at least at one point of the cumulative failure rate at a low temperature from an Arrhenius plot of a dielectric breakdown life at at least a certain point of the cumulative failure rate; Means for obtaining the distribution of the dielectric breakdown life at a low temperature by extrapolating through the dielectric breakdown life at at least one cumulative failure rate. To provide an evaluation device of the semiconductor device.

The present invention is characterized in that the change in the dielectric breakdown lifetime distribution during the temperature accelerated test is taken as the temperature change of the gradient of the distribution function in the cumulative failure rate. This makes it possible to accurately predict the dielectric breakdown lifetime distribution at the guaranteed temperature from a temperature accelerated test at a temperature higher than the guaranteed temperature.

According to the present invention, when estimating a low-temperature dielectric breakdown life from a temperature-accelerated test at a high temperature, a highly accurate prediction can be realized in a short time in a simple method. Also, when calculating the reliability guarantee life in conjunction with the electric field acceleration test, the error with the actual distribution function can be made smaller than in the past, so that the guarantee life can be predicted and calculated with high accuracy.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS First, the present inventors set the temperature of a silicon substrate on which a MOS type semiconductor device is formed to 50 ° C.
It heated to 125 degreeC, 225 degreeC, and each temperature of 300 degreeC, and measured the dielectric breakdown lifetime distribution. Each of the obtained results was normalized using the dielectric breakdown life at which the cumulative failure rate became 50%, and the standard deviation indicating the spread of the distribution was calculated.

FIG. 2 shows the result of an Arrhenius plot of the standard deviation of the normalized breakdown life distribution versus the reciprocal of temperature.

From FIG. 2, there is a correlation between the spread of the standard deviation of the dielectric breakdown lifetime distribution standardized and the substrate temperature at the time of measurement. The higher the temperature is, the smaller the standard deviation is, that is, the distribution of the dielectric breakdown lifetime is reduced. It turns out that it is steep. In the conventional life prediction method, the prediction has been made on the assumption that the dielectric breakdown life distribution has no temperature dependency, so that the above-described error has occurred.

Therefore, in the present invention, in order to incorporate the temperature dependence of the dielectric breakdown lifetime distribution into the lifetime prediction method,
By incorporating the temperature dependency into the slope of the distribution function of the dielectric breakdown life, it is possible to accurately predict the dielectric breakdown life at the guaranteed temperature.

In FIG. 3, the horizontal axis represents the standard deviation of the normalized breakdown life distribution, and the vertical axis represents the slope when the normalized breakdown life distribution is represented by a Weibull distribution function. FIG. 6 is a diagram showing a relationship at the time. The slope of the Weibull distribution function is ln (-ln (1-P)) = β · ln (t
_bd ) + β ₀ corresponds to the value of β. Here, P is the cumulative failure rate, t _bd is the dielectric breakdown lifetime, and β ₀ is a coefficient.

FIG. 3 shows that there is a correlation between the standard deviation of the breakdown life distribution and the slope of the Weibull distribution of the breakdown life. Therefore, in the present invention, an accurate dielectric breakdown life at the guaranteed temperature is predicted by incorporating the temperature dependency into the slope of the Weibull distribution of the dielectric breakdown life.

FIG. 4 is a diagram showing the correlation between the slope of the Weibull distribution of the dielectric breakdown life and the substrate temperature at the time of measurement from FIGS. 2 and 3. Although it is also a feature of the present invention, when calculating the slope of the Weibull distribution at a low temperature from the slope of the Weibull distribution of the dielectric breakdown lifetime at a high temperature, a correlation between the slope of the Weibull distribution and the temperature is derived by an Arrhenius plot. I have.

In FIG. 4, .largecircle. Indicates the slope of the Weibull distribution obtained from the dielectric breakdown lifetime distribution measured at 85.degree. C., 125.degree. C. and 150.degree.
This is an extrapolation of the slope of the Weibull distribution obtained from the dielectric breakdown lifetime distribution measured at ° C. According to this, the slope of the Weibull distribution at 85 ° C. (1000 / T is about 2.8 in FIG. 4) in the dotted line and the slope of the Weibull distribution calculated from the measured value of the dielectric breakdown lifetime distribution at 85 ° C. match well. You can see that.

That is, the dielectric breakdown lifetime distribution is obtained by an accelerated test at a high temperature of 125 ° C. or 150 ° C., and the slope of the Weibull distribution is extrapolated by Arrhenius plot.
This shows that the slope of the Weibull distribution at a low temperature of 85 ° C. can be accurately obtained.

As an example, a high temperature of 150 ° C. and a temperature of 85 ° C.
This shows that the slope of the dielectric breakdown lifetime distribution at an intermediate temperature of 125 ° C. can be calculated from the measured value at a low temperature. Further, as an example, it is shown that the dielectric breakdown lifetime distribution at a high temperature of 150 ° C. can be predicted from measured data of the dielectric breakdown lifetime distribution at a low temperature of 85 ° C. and an intermediate temperature of 125 ° C. That is, by performing an Arrhenius plot at least at two temperatures, the slope of the Weibull distribution at an arbitrary temperature can be accurately obtained.

FIG. 5 is a diagram showing the results of comparing the Weibull distribution of the dielectric breakdown life obtained by the conventional life distribution prediction method and the Weibull distribution of the dielectric breakdown life obtained by the present invention with the actually measured values.

In FIG. 5, .largecircle. Indicates the measured value at 85.degree. C., and .quadrature. And the dotted line indicate the Weibull distribution at 85.degree. C. predicted from the dielectric breakdown lifetime distribution measured at 125.degree. C. and 150.degree. The solid line is the value 8 predicted from the distribution of the dielectric breakdown life measured at 125 ° C. and 150 ° C. according to the present invention.
The Weibull distribution at 5 ° C. is shown.

In the present invention, as shown in FIG.
The slope of the Weibull distribution at 85 ° C. was determined from the extrapolation of the slope of the Weibull distribution at the dielectric breakdown life at 150 ° C. and 150 ° C., and the insulation failure with a cumulative failure rate of 50% in the dielectric breakdown life distribution measured at 125 ° C. and 150 ° C. The fracture life was determined, and this was determined at 85 ° C. by an Arrhenius plot.
Is obtained, and the Weibull distribution function having the slope obtained as described above is obtained through this point.

As shown in FIG. 5, it can be seen that the Weibull distribution at 85 ° C. predicted by the conventional method is steeper than the actually measured value, and a large error occurs in the dielectric breakdown life especially at a low cumulative failure rate. On the other hand, in the prediction method incorporating the temperature dependence of the breakdown life distribution according to the present invention,
It can be seen that the measured value at 85 ° C. is accurately reproduced.

Here, in obtaining the Weibull distribution at 85 ° C. in the present invention, the dielectric breakdown life at a cumulative failure rate of 50% is predicted and obtained as a distribution function passing through it. However, the present invention is not limited to this. If necessary, the dielectric breakdown life at each failure rate may be calculated, and this may be used to obtain a Weibull distribution function at 85 ° C.
Although the prediction of the dielectric breakdown lifetime distribution at 85 ° C. is made based on the results of the accelerated test at 50 ° C., the substrate temperature is not limited to this and can be applied in any temperature range.

FIG. 6 shows a device diagram according to the semiconductor device evaluation method of the present invention. The sample 1 to be measured is fixed to the stage 2 of the measuring device 3 with a vacuum chuck or the like. Here, the stage 2 is movable in a three-dimensional direction by a stepping motor or the like, and can be automatically moved by an external signal. The probes 4a and 4b are brought into contact with the semiconductor device formed on the sample 1. A voltage source 6 a and an ammeter 6 b are connected to the probe 4 b by a cable 5. Further, a current source may be used instead of the voltage source, and a voltmeter may be used instead of the ammeter. For example, a voltage is applied from a voltage source and the current is measured by an ammeter, or the current is applied by a current source and the voltage is measured.

These measuring device 3, voltage source 6a, ammeter 6
“b” is connected to an evaluation computer 8 by a signal cable 7 such as GP-IB. The evaluation computer can automatically set the operation of the stage 2 of the measuring device 3 and the measurement parameters of the voltage source 6a by a program, read the current value from the ammeter 6b at the time of measurement, and monitor the monitor 9 Can be displayed.

The evaluation computer 8 has a built-in program for evaluating the dielectric breakdown lifetime distribution as shown in FIG. 7, which will be described later, so that measurement and analysis of measured data can be performed. ing. In addition, the analysis result is
It is also possible to display the evaluation result on the.

FIG. 7 shows a flowchart of a method for predicting the life under the guaranteed condition from the acceleration test, the temperature acceleration test, and the electric field acceleration test in the semiconductor device evaluation method of the present invention.

First, with respect to the semiconductor device installed in the measuring instrument, a certain stress electric field, for example, Eox = 11.0MV, at at least two or more temperatures higher than the guaranteed temperature, such as 125 ° C. and 150 ° C. / C
The dielectric breakdown lifetime distribution is measured under the condition of m (FIG. 7-a).

At this time, in the same semiconductor device, at least 30
Measure the measurement points above the point. Here, the reason why the score is 30 or more will be described.

FIG. 8 is a graph in which the horizontal axis represents the number of measurement points and the vertical axis represents the slope β of the Weibull distribution. When measurement points smaller than 30 are measured in this way, the slope β of the obtained Weibull distribution is very low, which indicates that the probability of including an error increases. If the probability of including the error increases, the state of the distribution of the dielectric breakdown life cannot be obtained with high accuracy. Therefore, in the present embodiment, the measurement was performed at at least 30 measurement points.

The obtained dielectric breakdown life data is
At each temperature, for example, a Weibull distribution function is obtained, and its slope is calculated (FIG. 7B).

Next, the correlation between the determined slope of the Weibull distribution function and the measured temperature is represented by an Arrhenius plot (FIG. 7-).
c) Extrapolate it by the least squares method or the like, and calculate a gradient at a guaranteed temperature, for example, 85 ° C. (FIG. 7-d).

Next, in the insulation life distribution at, for example, 125 ° C. and 150 ° C., at least one point, for example, a dielectric breakdown life distribution with a cumulative failure rate of 50% is obtained (FIG. 7-e). From this, the temperature acceleration coefficient is determined by an Arrhenius plot (FIG. 7-f). Then, at least one point at a guaranteed temperature of 85 ° C., for example, a dielectric breakdown life at a cumulative failure rate of 50% is predicted and calculated (FIG. 7-g).

Next, the insulation lifetime distribution is obtained by extrapolating through the insulation lifetime.

Next, at a certain temperature, for example, 125 ° C.
The stress electric field is applied to at least two points including the stress condition of the temperature acceleration test, for example, 10.5 MV / cm,
The dielectric breakdown lifetime distribution is measured at 11.0 MV / cm and 12.0 MV / cm (FIG. 7-h). Next, the obtained breakdown life distribution is expressed by, for example, a Weibull distribution function,
The inclination is calculated (FIG. 7-i).

Next, the correlation between the slope of the Weibull distribution obtained by the temperature acceleration test and the substrate temperature, and the failure rate of 50%
Using the temperature acceleration coefficient calculated from the Arrhenius plot of the dielectric breakdown life (FIG. 7-j) (FIG. 7-k), the Weibull distribution of the dielectric breakdown life at the guaranteed temperature of 85 ° C. under each stress condition is calculated. (FIG. 7-l).

Next, in the expected Weibull distribution in each electric field, the dielectric breakdown life at which the failure rate becomes 1 ppm, for example, is calculated (FIG. 7-m), and this is plotted against the electric field (E), for example. From the reliability guaranteed life of the gate insulating film under the guaranteed condition of the defect rate of 1 ppm
(FIG. 7-n).

Thus, accurate prediction can be made.

In the present invention, the dielectric breakdown life of a silicon thermal oxide film has been described as an example of a gate insulating film. However, the present invention is not limited to this. For example, an oxide film containing nitrogen, a nitride film, This is a method for calculating the dielectric breakdown life which can be applied to other high dielectric films.

Further, as an example of the temperature acceleration test, an example is given in which the dielectric breakdown lifetime distribution at a constant temperature is predicted by using the dielectric breakdown lifetime distribution at two or more high temperatures. However, the present invention is not limited to this. It is also possible to predict and obtain the dielectric breakdown lifetime distribution at a high temperature from the data of the dielectric breakdown lifetime distribution.

In addition, various modifications can be made without departing from the scope of the present invention.

[0053]

According to the present invention, it is possible to accurately predict and calculate the life expectancy of a gate insulating film by a simple method.

[Brief description of the drawings]

FIG. 1 is a characteristic diagram showing the temperature dependence of a breakdown life distribution.

FIG. 2 is a characteristic diagram showing the temperature dependence of the standard deviation of the normalized breakdown life distribution of the present invention.

FIG. 3 is a characteristic diagram showing a correlation between a standard deviation of a normalized breakdown life distribution and a slope of a Weibull distribution in the normalized breakdown life according to the present invention.

FIG. 4 is a characteristic diagram showing the temperature dependence of the slope of the Weibull distribution according to the present invention.

FIG. 5 is a characteristic diagram showing a dielectric breakdown life distribution at 85 ° C. predicted in consideration of the temperature dependence of the dielectric breakdown life distribution and a dielectric breakdown life distribution at 85 ° C. predicted by a conventional method according to the present invention.

FIG. 6 is a schematic diagram of a semiconductor evaluation apparatus provided with a method for calculating a dielectric breakdown life in consideration of the temperature dependence of the dielectric breakdown life of the present invention.

FIG. 7 is a flowchart of a method for calculating a dielectric breakdown life in consideration of the temperature dependence of the dielectric breakdown life of the present invention.

FIG. 8 is a characteristic diagram according to the measurement points of the dielectric breakdown lifetime of the present invention.

FIG. 9 is a flowchart of a conventional method for calculating a dielectric breakdown life.

Continued on the front page (72) Inventor Hitoshi Ito 8 Shinsugita-cho, Isogo-ku, Yokohama-shi, Kanagawa Prefecture Inside the Toshiba Yokohama Office (72) Inventor Yasushi Nakazaki 8 Shinsugita-cho, Isogo-ku, Yokohama-shi, Kanagawa Toshiba Yokohama Co., Ltd. Inside the office (72) Inventor Naoki Yasuda 8 Shinsugita-cho, Isogo-ku, Yokohama-shi, Kanagawa Prefecture Inside the Toshiba Yokohama office (72) Inventor Akiko Nara 8 Shinsugita-cho, Isogo-ku, Yokohama-shi, Kanagawa Toshiba Yokohama office (72) Inventor Masato Koyama 8 Shinsugita-cho, Isogo-ku, Yokohama-shi, Kanagawa Prefecture F-term in the Toshiba Yokohama office (reference) 4M106 AA01 AA20 AB02 BA01 BA14 BA20 CA14 CA31 CA56 CA60 DH01 DH46 DJ02 DJ04 DJ05 DJ20 DJ23

Claims (5)

[Claims] An evaluation method of a semiconductor device for determining a dielectric breakdown lifetime distribution of an insulating film formed on a semiconductor substrate, comprising: a step of measuring a dielectric breakdown lifetime distribution at a high temperature; From the extrapolation of the slope when expressed as a distribution function of the cumulative failure rate, a step of obtaining the slope of the distribution function of the cumulative failure rate in the breakdown life distribution at a low temperature, and at least in the breakdown life distribution at the high temperature. A step of predicting and obtaining a dielectric breakdown life at at least one point of the cumulative failure rate at the low temperature from an Arrhenius plot of the dielectric breakdown life at a point of the cumulative defective rate; Extrapolating the breakdown life at the cumulative failure rate to obtain the breakdown life distribution at the low temperature. Method. 2. The method for evaluating a semiconductor device according to claim 1, wherein the measurement of the dielectric breakdown lifetime distribution at a high temperature is performed at at least two different temperatures. 3. The semiconductor device according to claim 1, wherein the distribution function when expressing the breakdown life distribution at a high temperature as a distribution function of a cumulative failure rate is a Weibull distribution or a lognormal distribution. Evaluation method. 4. A distribution function of a cumulative failure rate in the dielectric breakdown life distribution at a low temperature based on an Arrhenius plot of the distribution of the breakdown life at a high temperature represented by a distribution function of the cumulative failure rate. 2. The method for evaluating a semiconductor device according to claim 1, wherein an inclination of the semiconductor device is obtained. 5. A semiconductor device evaluation apparatus for determining a dielectric breakdown lifetime distribution of an insulating film formed on a semiconductor substrate, comprising: means for measuring a dielectric breakdown lifetime distribution at a high temperature; Means for determining the slope of the distribution function of the cumulative failure rate in the breakdown life distribution at low temperature from the extrapolation of the slope when expressed as the distribution function of the cumulative failure rate, and at least in the breakdown life distribution at the high temperature. Means for predicting and obtaining a dielectric breakdown life at at least one point of the cumulative failure rate at the low temperature from an Arrhenius plot of the dielectric breakdown life at one point of the cumulative failure rate; Means for obtaining the distribution of the dielectric breakdown lifetime at the low temperature by extrapolating through the dielectric breakdown life at the cumulative failure rate and extrapolating the same. Evaluation device.