JPH04212423A - Semiconductor device - Google Patents

Abstract

PURPOSE:To obtain a semiconductor device wherein EM characteristics at the contact part of a two-layered wiring are improved. CONSTITUTION:The product of the thickness and the stress amount of a silicon nitride film 15b formed on a second metal wiring 14 formed in a contact part is set to be less than or equal to 2/5 of the product of the thickness and the stress amount of an SiN film 15b formed so as to cover the second metal wiring 14. By controlling the stress of the SiN film 15b and the thickness of the SiN film formed in the contact part, the stress applied on the aluminum wiring in the contact part, which stress is caused by a protective film, especially by the SiN film 15b, is restrained to be low. Thereby the electromigration life of the contact part affected by the stress is improved approximately by one digit.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor device having a multilayer wiring structure of a very large scale integrated circuit (VLSI).

0002

2. Description of the Related Art Conventionally, with the increasing integration of VLSIs, a multilayer wiring structure in which wiring structures are stacked in two or more layers has been adopted, and semiconductor devices in which the wirings are formed have been designed to prevent the penetration of moisture and impurities. Therefore, SiO2 containing P is used as a protective film.
(Hereinafter referred to as PSG (phosphosilicate glass))
A structure in which the substrate is covered with a silicon nitride film (hereinafter referred to as SiN) formed by plasma CVD is used.

FIG. 9 shows an example of a cross-sectional structure of such a two-layer wiring structure. In the same figure, the insulating film 1 is partially opened.
A first layer wiring 12 is formed on a substrate 11 covered with P0 and a transistor region is formed thereon, a second layer wiring 14 is provided thereon via an interlayer insulating film 13, and a final protective film 15 is made of P.
Two-layer wiring is completed by forming SG and SiN. This wiring structure basically has three parts: a line of the first layer wiring 12, a line of the second layer wiring 14, and a line of the first layer wiring 12 and the second layer wiring 14.
It consists of a portion 13a (hereinafter referred to as a contact portion) that electrically connects the layer wiring 14.

By the way, as the integration of LSIs progresses and the size of wiring becomes finer, a problem arises in that just by passing a current through the wiring, a disconnection failure occurs in the wiring. This phenomenon is called an electromigration (hereinafter referred to as EM) defect, and the reason for this is that Al atoms move in the direction in which electrons flow due to current, and the portions where the movement is intense are disconnected. As a countermeasure to prevent this, a method has been taken to reduce the number of vacancies existing at aluminum grain boundaries and reduce the movement of Al atoms by adding other elements (e.g., Cu, Ti, etc.) to the aluminum film. (P.B.
．． Gate, Solid State Tech
ology, Vol. 3, pp. 113-120,19
83).

[0005] Another problem is that the wiring may break just by keeping it at a high temperature. This is called stress migration (hereinafter referred to as SM). S
Regarding the M phenomenon, it has been reported that the frequency of wire breakage increases when the temperature is maintained at around 150°C, and this can be improved by adding Cu, etc. (J. Kelma et al., 22nd Ann.
ual Proceeding International
ional Reliability Physi
cs Symposium, pp. 1-5, 1984
).

The above two problems have been reported as problems and solutions for single-layer wiring lines. However, in two-layer wiring due to high integration, E
M, SM problem needs to be solved.

[0007] There are few reports of EM and SM phenomena regarding this contact part, and it has been empirically shown that EM and SM characteristics can be improved by tapering the interlayer insulating film of the contact part and improving the coverage of the aluminum of the second wiring. It has been predicted and implemented. However, if the contact part is tapered, when the aluminum of the second wiring is formed by sputtering, the SiO2 of the contact part of the substrate is tapered and easily sputtered when the input power for sputtering is large. It has been reported that care must be taken when forming SiO2 at the contact interface, as SiO2 is formed between the contact interface and the second aluminum layer, which causes deterioration of EM characteristics (
H. Tomioka et al., 27th Annual
Proceedings International
Reliability Physics S
ymposium, p. 57, 1989).

[0008]

However, the current state of the structure of such two-layer wiring is shown in Table 1.

[0009]

[Table 1] As is clear from (Table 1), although there are reports on the EM and SM characteristics of the first and second wirings themselves, as well as individual reports on problems in the contact area, element(
There is no matching of each part when used in (Table 1) medium device).

[0010] The purpose of the present invention is to improve the EM characteristics of the contact portion in particular, in order to realize a two-layer wiring with good EM and SM characteristics that match each part of the wiring, and to also determine the SM characteristics at the same time. I will explain the technical background that led to this.

First, the relationship between the sizes of each part of wiring used in a semiconductor device is as follows. For example, if the first wiring width is 1.2 .mu.m (the minimum size for a microprocessor in which wiring has a large load), then the second layer wiring width is also the minimum size, 1.2 .mu.m. The contact portion at this time is determined so that the size at the point where the first wiring and the second wiring intersect is maximized, and in this case, it is made to be 1.2 microns square. This situation is shown in FIG. The figure shows a planar state, and the spread of the first layer wiring and the second layer wiring around the contact (described as alignment margin in the figure) is a margin that takes into account misalignment in photolithography. From this, when making a two-layer wiring, the size of each part is such that the wiring widths of the first layer and the second layer are the same size, and the contact size is a rectangle with the wiring width as one side.

FIG. 11 is the result of measuring the EM characteristics of each part of the wiring as described above. (The SM characteristics will be described in the examples, but there were no problems.) The cross-sectional structure of each part used in this measurement is shown in FIG. In Figure 1, 1% Si and 0.
Aluminum containing 5% Cu (hereinafter referred to as aluminum) was used for the first and second wiring metals. A series connection (hereinafter referred to as a contact chain) of 400 contacts with a contact area of 1.2 microns square between the first wiring and the second wiring was made with the structure shown in Figure 1(a), and EM measurements were performed. .

In FIG. 1A, a first metal wiring 12 and an SiO2 film 10 are formed on a substrate 11 on which a SiO2 film 10 is formed.
2 and a second metal wiring 1 connected to the first metal wiring 12 via a contact portion 13a.
4 is formed. On top of this, PSG is used as a protective film 15.
A film 15a and a SiN film 15b are formed. Because the step coverage of aluminum by sputtering is poor,
In order to improve this, the contact part is approximately 55mm.
° taper processing is applied.

[0014] Also, when creating the structure shown in Figure (a), at the same time,
0.8μm with a width of 1.2μm electrically insulated from the second wiring
A part with only the thick first wiring (Figure 1(b)) and a part with only the second wiring with a width of 1.2 μm and a thickness of 1.0 μm electrically insulated from the first wiring (Figure 1(c)) were created. , each wiring life was measured by EM measurement.

The measurement conditions were as follows: After determining the cross-sectional area of each part of the first wiring, second wiring, and contact, a current was applied at 150° C. so that the current density was 2×10 6 A/cm 2 .
The occurrence time of the disconnection failure was investigated (this is generally referred to as EM measurement). The results are shown in FIG. As is clear from the figure, the time required for the contact chain to become defective in EM characteristics is more than one order of magnitude shorter than that of other types.

In the conventional example, the cause of the deterioration of the EM characteristics of this contact chain is that a thin S
It has been reported that iO2 exists (this occurs because when depositing aluminum for the second wiring by sputtering, SiO2 in the tapered part is sputtered with Ar ions and re-deposited on the first wiring that has an opening at the contact). has been done. In order to eliminate this factor, in the present invention, the aluminum sputtering of the second wiring was performed in a high vacuum state and at a low sputtering power so that SiO2 would not be sputtered with Ar ions.

Furthermore, the absence of SiO2 at the interface was confirmed as follows. If a thin layer of SiO2 exists at the interface, the contact chain resistance will vary greatly after the two-layer wiring is formed, and the resistance value will change significantly after heat treatment at about 400°C. The results of this confirmation are shown in FIG.

The figure shows 1 when the taper angle is 55°.
0000 contact chain resistance 450℃150
Changes before and after heat treatment for 30 minutes are shown. Set the contact size to 0.
8, 1.0, and 1.2 micron squares, and 108 samples each were measured. The resistance variation is 0.8 micron square.
No significant change in resistance was observed before and after heat treatment, except for a slight change. This means that thin Si is used in the contact area.
This means that there is no O2 present. For further physical confirmation, the cross section of the 1.2 micron square contact section was observed using a transmission electron microscope, and it was confirmed that there was no SiO2 and aluminum was alloyed in the contact section.

As described above, the problem to be solved by the present invention is that when an optimal two-layer wiring is created, Si is deposited at the interface between the first-layer wiring and the second-layer wiring.
This is to analyze defects in the EM characteristics, which determine the life of a semiconductor device, even in the absence of O2. That is, an object of the present invention is to provide a semiconductor device with improved EM characteristics in a contact portion of a two-layer wiring.

[0020]

[Means for Solving the Problems] In order to solve the above-mentioned problems, the present invention clarifies that the EM life of the contact part is determined by the stress of the protective film formed on the contact part. The value obtained by multiplying the thickness of the silicon nitride film formed on the second metal wiring formed on the second metal wiring by the amount of stress is the value obtained by multiplying the thickness of the silicon nitride film formed to cover the second metal wiring and the amount of stress. 2/5 of the multiplied value
As described below, the present invention is provided with a configuration for controlling the stress of the silicon nitride film and the thickness of the silicon nitride film formed in the contact portion.

[0021]

[Operation] With the above-described structure, the present invention can improve the EM life of the contact portion by nearly one order of magnitude by suppressing the stress applied to the aluminum wiring of the contact portion due to the protective film, especially the silicon nitride film. It becomes possible.

[0022]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a cross-sectional structural diagram of a semiconductor device according to the present invention. This figure shows the structure of the two-layer wiring broken down into the first wiring, the second wiring, and the contact portion.

FIG. 1(a) shows a contact chain section, in which a first metal wiring 12, an interlayer insulating film 13 made of SiO2 are formed on a substrate 11 on which a SiO2 film 10 is formed, 1.
A second metal wiring 14 is formed which connects to the first metal wiring 12 via a 2 micron square contact portion 13a. On top of this, a PSG film 15a and a SiN film 15 are formed as a protective film 15.
A film 15b is formed.

FIG. 1(b) shows the first wiring section, which is a portion consisting only of the first wiring, which is electrically insulated from the second wiring and has a width of 1.2 μm and a thickness of 0.8 μm. FIG. 1C shows only a second wiring having a width of 1.2 μm and a thickness of 1.0 μm, which is electrically insulated from the first wiring.

As mentioned in the problem section, the EM characteristics of the contact part shown in FIG. As shown, the relationship between the coverage ratio of the second layer aluminum, the SiN film thickness, and the mean time to failure was investigated by changing the slope θ of the contact.

The aluminum coverage and the SiN film thickness were determined by observing the cross section of the contact portion using a scanning electron microscope. Furthermore, 500 contacts were connected in series, and the same current (set to be 2 x 106 A/cm2 at the minimum aluminum cross-sectional area of the contact part of the sample at 85°) was applied to all samples. Measurement was carried out, and the mean failure time (according to the EM measurement results shown in Figure 11, the cumulative failure rate was 50%).
% is called the mean failure time). The results are shown in FIG.

In FIG. 2A, the coverage ratio of aluminum decreases as the contact angle θ increases. Similarly, as shown in FIG. 4B, the SiN film thickness at the contact portion is also reduced. On the other hand, as shown in FIG. 4(c), the mean time to failure improves and becomes higher as θ increases. In particular, in order to obtain the EM life of the contact chain that is on the same level as the EM life of the second wiring, which is the object of the present invention, this can be achieved by setting the taper angle of the contact to 80° or more. However, the present invention does not attempt to prescribe the angle, but rather identifies the physical factors that limit the EM life of the contact chain.
It stipulates that Considering the coverage ratio of aluminum, FIG. 3 contradicts the fact that it is usually said that as the current density increases, the EM characteristics deteriorate. For that purpose, Si
Focusing on the N film, we did the following.

First, the one without SiN film (taper angle 8
5°) and one in which the SiN film thickness was increased and the SiN stress amount was halved (taper angle 85°), and the mean failure time was determined. The resulting mean failure time versus SiN film thickness is shown in FIG. The circles in the figure indicate SiN films with the same amount of stress, and it can be seen that the mean failure time increases as the SiN film thickness becomes thinner, and the EM life increases. Without the SiN film, the mean time to failure is greatly improved (more than 100 arbitrary hours). This is because the EM life of the contact chain is determined by
Indicates that it is related to SiN film. Also, the □ mark in the same figure is S
The stress amount of the iN film is 1/2 of that of the SiN film marked with a circle. This means that the E
The M characteristics do not depend only on the SiN film thickness;
This shows that it is also related to the amount of stress on the N film.

As is clear from FIGS. 3(c) and 4, the EM life of the contact portion in the two-layer wiring is strongly influenced by the thickness of the protective film covering the contact portion and the amount of stress. The thinner the film, the better the EM characteristics.
When the amount of stress is reduced, the EM life of the contact is improved. In this way, the EM characteristics of the contact portion depend on the thickness of the SiN film and the amount of stress in the same manner. This means that the amount of deformation that causes aluminum to warp, compress, etc. is the SiN film thickness multiplied by the amount of stress (dynes/cm2) (
This is because this is expressed as total stress (hereinafter referred to as total stress).

Further, in order to quantify this relationship, Si
The amount of stress was actually measured by varying the thickness of the N film. In order to find out the amount of stress on the protective film of the contact part, a protective film with the same thickness as the protective film thickness of the contact part determined from the cross-sectional SEM shown in FIG. The stress amount of PSG film is 108 dynes
/cm2, so the measured stress amount can ignore the PSG film), and the stress amount was measured. The relationship between the SiN film thickness and the stress amount at this time is shown in FIG. In the figure, the SiN film deposition method (plasma CVD
The film quality was changed by changing the deposition conditions of 500n
The amount of stress calculated based on the thickness of m is 3 x 109 dyne.
s/cm2, □ mark is 1.5 x 109 dynes/cm
2. This actual measured total stress and EM
Figure 6 shows the relationship with the mean failure time determined from measurements. As is clear from the figure, the wiring life of the contact portion is inversely proportional to the total amount of stress.

[0031] Normally, the amount of stress on the SiN film is approximately 1 to
500nm to 800 at 4×109dyenes/cm2
A nm thick protective film is used. In order to improve the EM life of a protective film under such stress by about one order of magnitude (the reason for this will be explained later), it is necessary to take into account variations in stress amount, SiN film thickness, contact life measurement, etc.

To explain this with reference to FIG. 6, since the thickness of the SiN film deposited on the second interconnect is 500 nm, the total stress in the second interconnect is 1.5×105 dyn.
It will be about es/cm. Further, the relationship between the total stress of the contact portion and the mean time to failure is generally a solid line, but when variations are taken into account, it becomes a broken line. Further, as shown in the figure, the average failure time (=EM life) of the second wiring is 10 to 20 (arbitrary value). In order to achieve the same EM life of the contact chain (improved by about one order of magnitude compared to the conventional one), the total stress in the contact portion must be about 6×10 4 dynes/cm or less from the broken line. This value is approximately 2/5 of the total stress value of the second wiring section.

As described above, the EM characteristics of a 1.2 micron square contact part in the conventional example and when the total stress in the contact part is set to 2/5 or less of the total stress in the protective film of the second metal wiring are as follows. It is shown in FIG. 7(a). According to the present invention, the EM life is improved by more than one order of magnitude compared to the conventional example.

The significance of improving the EM life by one order of magnitude comes from the product life guarantee. The product life of an LSI is usually determined by the operating current density and operating temperature, and under these conditions, the product is guaranteed for about 10 years. EM measurement is performed by accelerating this. For this reason, the EM measurement conditions for this test were 150°C and a current density of 2 x 106 A/cm2, which is the actual product usage condition (
Current density is 1 x 105 A/cm2 at 80°C. (This is the level of most product usage except for special applications), the straight line of cumulative failure rate shown in Figure 7(a) is approximately 0.3%.
The time required for this is 2.83 test hours, and this time corresponds to 10 years of wiring life. This relationship is shown in Figure 7(b).
Shown below. As is clear from the figure, the present invention makes it possible to achieve a 10-year product warranty.

In this example, aluminum wiring containing Si and Cu was described, but even if the wiring material is pure aluminum, Si
The same effect is obtained even with aluminum containing Ti or Pd.
It is not limited by the wiring material.

Further, the stress migration characteristic, which is another factor determining the wiring life, was measured and is shown in FIG. The figure shows the conventional example (taper angle 55°) and the present invention (taper angle 85°) samples of wiring in which 400 contacts on which a protective film was formed were connected in series, and were held at 150°C, and the disconnection versus holding time. It shows the defective rate. Contact sizes of 1.2 and 1.0 micron square were investigated, respectively. In both the conventional example and the present invention, no defects occur after holding for 1000 hours, but defects occur in both the conventional example and the present invention after 2100 hours of 1.0 micron square. If the holding time is 1,000 hours, there is no problem with the product, and if the holding time is 2,100 hours, both the conventional example and the present invention will be defective, so the difference in the SiN film in the contact area may be due to stress migration. It won't have a big impact.

[0037]

As is clear from the above description, according to the present invention, the value obtained by multiplying the thickness of the silicon nitride film formed on the second metal wiring formed in the contact portion by the amount of stress is: S formed to cover the second metal wiring
By controlling the stress of the SiN film and the thickness of the SiN film formed in the contact area so that the stress of the SiN film is 2/5 or less of the value obtained by multiplying the iN film thickness and the amount of stress, the S
By suppressing the stress applied to the aluminum wiring of the contact portion by the iN film, the electromigration life of the contact portion due to the influence of stress is improved by nearly one order of magnitude.

[Brief explanation of the drawing]

FIG. 1(a) is a cross-sectional structural diagram of a contact chain portion for EM life measurement of wiring used in an embodiment of the present invention. (b) is a cross-sectional structural diagram of the first layer wiring. (c) is the second
FIG. 3 is a cross-sectional structural diagram of layer wiring.

FIG. 2 is a cross-sectional structural diagram showing a contact portion of the present invention.

FIG. 3(a) is a coverage characteristic diagram of aluminum in a contact portion. (b) is a characteristic diagram of the SiN film thickness of the protective film in the contact portion. (c) is a characteristic diagram of the mean failure time in the contact portion.

FIG. 4 is a diagram showing the relationship between the SiN film thickness of the contact portion and the mean failure time.

FIG. 5 is a diagram showing the relationship between SiN film thickness and total stress.

FIG. 6 is a diagram showing the relationship between total stress and wiring life.

FIG. 7(a) is an electromigration characteristic diagram of a contact portion of a semiconductor device according to the present invention. (b)
is a diagram showing the relationship between test time and wiring life.

FIG. 8 is a stress migration characteristic diagram of a contact chain.

FIG. 9 is a cross-sectional structural diagram showing two-layer wiring in a semiconductor device.

FIG. 10 is a plan view showing a contact of two-layer wiring of a semiconductor device according to the present invention.

FIG. 11 is an electromigration characteristic diagram of each part of a two-layer wiring in a conventional example.

FIG. 12 is a contact chain resistance characteristic diagram before and after heat treatment.

[Explanation of symbols]

10 SiO2 film 11 Semiconductor substrate 12 First metal wiring 13 Interlayer insulation film 13a Contact part 14 Second metal wiring 15a PSG film 15b Silicon nitride film

Claims (2)

[Claims] 1. A first metal wiring formed on a semiconductor substrate, a second metal wiring electrically connected to the first metal wiring via an insulating film having a partially opened opening, and a second metal wiring formed on the semiconductor substrate. In a semiconductor device in which SiO2 containing phosphorus and a silicon nitride film are formed on a second metal wiring, the thickness and stress amount of the silicon nitride film formed on the second metal wiring formed in the opening are A semiconductor device characterized in that the multiplied value is 2/5 or less of the value multiplied by the stress amount and the thickness of the silicon nitride film formed to cover the second metal wiring. 2. A semiconductor device characterized in that a value obtained by multiplying the thickness of the silicon nitride film according to claim 1 by the amount of stress is 6×10 4 dynes/cm or less.