JPH10214966A - Method for managing process - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate the long term reliability lifetime of a damaged device in a short time by storing the relationship of the total quantity of charges to be implanted before breakdown of a semiconductor device, the time to be elapsed before breakdown of the semiconductor device, the field acceleration coefficient and the activation energy in a data base and utilizing the data base. SOLUTION: Constant current TDDB evaluation is performed for an MOS capacitor and the total quantity of charges Qbd to be implanted before breakdown is measured. A data base 1 storing a relationship between the time tdb to be elapsed before breakdown of the semiconductor device and ΔQdb is than prepared through a constant voltage TDDB test followed by generation of a data base 2 storing the relationship of the ΔQdb, the field acceleration coefficient and the activation energy. Based on the data base 1, tdb(rA) in the constant voltage TDDB measurement of ΔQdb(rA) is found and employed, along with the tbd(rA) and the field acceleration coefficient and the activation energy in the data base 2, for finding the long term reliability lifetime according to a formula; τ=Ae×p(-Ea/kT)×10βE.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for managing a process in a MOS semiconductor, and more particularly to a method for managing a process in a MOS semiconductor.
It is an object of the present invention to provide a process management system capable of monitoring insulation film damage during the manufacture of a semiconductor device and performing screening in the course of the process based on the long-term reliability life estimated in the initial stage of the manufacture.

[0002]

2. Description of the Related Art In recent years, high integration of semiconductor integrated circuits has been greatly advanced.
With the miniaturization of transistor elements, the thickness of a gate insulating film has been reduced. As a result, specifically, 0.2
Under the 5 μm rule, a thin gate insulating film of about 6 to 8 nm is being used.

When manufacturing a semiconductor device having a thin gate insulating film as described above, there is a problem of damage to the gate insulating film occurring during the manufacturing process. This problem becomes more serious with the miniaturization of the device structure in which the wiring structure will have a high aspect ratio in the future. The damage to the oxide film as described above is expected to affect the long-term reliability life.

Therefore, conventionally, screening has been carried out as a countermeasure against the above problem in order to prevent the occurrence of market defects due to the deterioration of the reliability life of the oxide film.

[0005]

However, the above-described screening is mainly a test for excluding a defect generated for an initial or accidental reason, and is a test for deteriorating the long-term reliability life of an oxide film. It is not the target and it is difficult to perform sufficient screening.

Therefore, the following two methods can be considered for solving the long-term reliability life deterioration due to process damage.

First, the amount of damage accumulated in each process is minimized by optimizing the process conditions in each process. Second, long-term reliability life evaluation is performed in a sampling inspection after the last process. , Screening. Therefore, these two methods will be described in detail below.

First, in optimizing the process conditions shown as the first method, a technique for quantitatively evaluating the damage of the oxide film is very important. As an example of this method, when evaluating the dry etching damage of the wiring, a T-type antenna having an antenna wiring as shown in FIG.
Using EG (Test Element Group),
Evaluation using a sample in which the antenna length of the antenna TEG is changed has been performed. The following two methods can be cited as methods for evaluating such a sample.

One is a constant current TDDB (Time Dep.
end Dielectric Breakdown
n) Evaluation of the total injected charge amount Qbd in the test, and another evaluation of the breakdown time tbd in the constant voltage TDDB test.

[0010] The Qbd evaluation in the constant current test has the advantages that the evaluation can be performed in a relatively short time and the process damage can be accurately evaluated.
Since the relationship between d and the actual device life, which is the life at a constant voltage, is unknown, there is a problem that the index for performing optimization is not clear.

On the other hand, in the evaluation of tbd at a constant voltage, although it is possible to evaluate long-term reliability by using an acceleration coefficient, it has been reported that tbd has an area where process damage cannot be evaluated accurately. Problems arise when giving feedback to In addition, it is reported that there is a phenomenon that an acceleration coefficient such as an electric field acceleration coefficient or an activation energy depends on damage. Therefore, even when tbd is used, the long-term reliability life under actual operating conditions cannot be accurately evaluated unless the damage dependency of the acceleration coefficient is considered.

Next, as shown in the above-mentioned second method, when the long-term reliability life evaluation after the final process is performed, for example, damage is relatively large in a relatively early process, and the life is greatly deteriorated. In this case as well, manufacturing must be performed up to the final process, which may result in waste of manufacturing costs. Further, since long-term reliability life evaluation is performed after the final process, quick feedback to the problem process becomes difficult. .

In view of the above problems, the present invention provides a method for estimating a long-term reliability life under an accurate actual operating condition in a short period of time even for a device damaged in the course of a process. It is intended to provide a process management method for providing feedback.

[0014]

In order to achieve the above object, a process management method according to the present invention uses a constant current TDDB test to determine a total injected charge amount Qb until a semiconductor device is destroyed.
A configuration in which the relationship between d and the time tdb until the semiconductor device is destroyed by the constant voltage TDDB test and the relationship between Qbd, the electric field acceleration coefficient, and the activation energy are stored in a database, and screening is performed during semiconductor manufacturing based on the database. It has become.

With the above configuration, accurate life estimation can be performed, and quick feedback to the process can be performed.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A process management method according to an embodiment of the present invention will be described below with reference to the drawings.

In the present embodiment, an example of a pre-screening flow for the dry etching step of the first layer wiring of a semiconductor element having a multilayer wiring structure will be described. First, FIG. 9 shows a flow of a process management method in the embodiment.

In FIG. 1, first, in step 100, a database 1 having a relationship between ΔQbd and tbd as shown in FIG. 2 is created. This database 1
For example, using a TEG having an antenna pattern as shown in FIG. 6, Qbd at several antenna ratios (r)
(R) and tbd (r) are evaluated, and the antenna ratio is 0
Difference from Qbd (0) at the time of ΔQbd = Qbd (0) −
By performing Qbd (r), ΔQ as shown in FIG.
The relationship between bd and tbd is determined and created.

Next, in step 101, a database 2 is created as shown in FIG. 3 which has a relationship between ΔQbd, electric field acceleration coefficient and activation energy. This database 2 can be obtained, for example, by evaluating both Qbd (r) and tbd (r) at several antenna ratios (r) using a TEG having an antenna pattern as shown in FIG. At that time, for the electric field acceleration coefficient, both Qbd and tbd are measured for several electric fields, and for the activation energy, both Qbd and tbd are measured for several temperatures. ΔQ as shown in FIG.
The relationship between bd and the acceleration coefficient can be obtained.

After performing the above two database creation operations, in step 102, for a product to be prescreened, an isolation region is formed on a silicon substrate serving as a substrate by a known method, and then gate oxidation is performed. Formation of a film and deposition of, for example, polycrystalline silicon as a gate electrode material are performed. Subsequently, in step 103, by patterning the polycrystalline silicon, a gate electrode required for circuit operation, a MOS having an oxide film 201 having an area of 1 μm ² and a pad section 200 having an area of 40 μm × 40 μm as shown in FIG. The formation of the capacitor is performed simultaneously.

Thereafter, actual evaluation is started. First, in step 104, a constant current TDDB evaluation is performed on the MOS capacitor, and the total injected charge amount Q until the destruction is reached.
Measure bd1. Then, in step 105, after forming source / drain regions, interlayer films, contacts, and the like by ion implantation, CVD, dry etching, and the like by a known method, a metal wiring material such as an Al film is formed. Perform deposition. Subsequently, in step 106, a metal wiring material is patterned (process 1) to simultaneously form a wiring pattern required for circuit operation and a MOS capacitor having antenna ratios r1 and r2 as shown in FIG. After that, in step 107, after performing the above process 1, the MOS capacitors having the antenna ratios r1 and r2 are subjected to constant current TDDB evaluation, respectively, and the total injected charge amounts Qbd2 (r1) and Qbd2 (r2) until breakdown are obtained. Is measured.

Next, at step 108, Qbd1
And Qbd2 (r1) and Qbd2 (r2), ΔQbd (rA) of the antenna ratio rA in the actual device is ΔQb
It is determined using the equation d (r) = Ar ^B. Here, A = ΔQbd2 (r1) · r− ^B , B = (log
((ΔQbd2 (r1) / ΔQbd2 (r2))) / l
og (r1 / r2).

Thereafter, in step 109, tbd (rA) in the constant voltage TDDB measurement of ΔQbd (rA) is obtained from the database 1, and in step 110, the obtained tbd (rA), the electric field acceleration coefficient of the database 2, From the activation energy, the equation τ = Aexp (−
Obtaining long-term reliability life with Ea / kT) × 10β ^E.

Finally, in step 111, only those having a long-term reliability life of 10 years or more are subjected to processes such as interlayer film formation and wiring formation by using a well-known method, thereby completing device fabrication. Note that a value stored in a database can be used for Qbd1.

As described above, according to the present invention, the yield at the time of manufacturing a semiconductor device can be improved by feeding back the result of the pre-screening to the process 1. At this time, screening is performed after the final step as needed, and the data is fed back to the databases 1 and 2, so that more accurate pre-screening can be performed.

By using the acceleration coefficient in consideration of the process damage, the long-term reliability life can be accurately estimated during the manufacturing. FIG. 5 shows life estimation using an electric field acceleration coefficient that does not consider plasma damage and life estimation using an electric field acceleration coefficient that considers plasma damage. From this, it can be seen that there is a difference of about one digit in the estimated value of the life in the actual use area.

Further, according to the present invention, by preliminarily performing pre-screening on wafers to be removed in the final screening step, manufacturing costs in steps after the pre-screening can be reduced. Furthermore, according to the present embodiment, the yield is improved because feedback is given to the process in consideration of the long-term reliability life at an early stage.

[0028]

As described above, the present invention evaluates Qbd by the constant current TDDB method, and uses a database consisting of Qbd and tbd, an electric field acceleration coefficient and activation energy to provide a long-term reliability life under actual operating conditions. Can be accurately calculated, and screening in the middle of the process can be performed based on the result.

[Brief description of the drawings]

FIG. 1 is a process diagram of a process management method according to an embodiment of the present invention.

FIG. 2 is a diagram showing the contents of a database based on the relationship between ΔQbd and tbd.

FIG. 3 is a diagram showing contents of a database including a relationship between ΔQbd, an electric field acceleration coefficient, and an activation energy.

FIG. 4 is a schematic view showing a structure of a TEG in a gate electrode forming step.

FIG. 5 is a diagram showing long-term reliability life estimation based on a difference in acceleration coefficient.

FIG. 6 is a perspective view showing the structure of a TEG in a wiring step.

[Explanation of symbols]

 Reference Signs List 200 pad 201 nMOS capacitor 202 activation area 203 resist 204 interlayer film 205 LOCOS 206 gate electrode 207 gate oxide film 208 silicon substrate

Claims (1)

[Claims] 1. A constant current TDDB test shows a total injected charge amount Qbd and a constant voltage TDDB until a semiconductor device is destroyed.
A database is created for a relationship between the time tdb until the semiconductor device is destroyed by the test and a relationship between the Qbd, the electric field acceleration coefficient, and the activation energy, and screening is performed during semiconductor manufacturing based on the database. Process management method.