Semiconductor Device Having a Double-Layer Interconnection Structure

1. Field of The Invention

This invention relates to electronic devices and more particularly, to semiconductor devices having a super LSI multilayer interconnection structure.

2. Description of The Prior Art

As is known in the art, the high degree of integration of super large scale integration circuits invariably requires multilayer or multilevel interconnections where the interconnection structure has two or more layers built up therein. Semiconductor devices having such an interconnection structure include a protective layer for preventing moisture and/or impurities from entering. This protective layer is to cover the multilayer interconnection therewith and has usually a structure which includes a phosphosilicate glass layer (P-containing SiO.sub.2) or a non-doped silicate glass layer (made of SiO.sub.2 alone) and a silicon nitride film formed by plasma-enhanced CVD. A typical example of a two-layer electrode or wiring structure is illustrated with reference to FIG. 9. In the figure, there is generally shown a semiconductor device S which includes a substrate 20 having a transistor region (not shown) covered with an insulating film 21 having an opening 21a. A first metal film 22 serving as a first electrode is formed on the substrate 20 as shown. Moreover, an insulating film 23 having an opening 23a is formed over the first metal film 22 and a second metal film 24 serving as a second electrode is provided on the insulating film 23. The second metal film 24 is in contact, as C, with the first metal film 22 through the opening 23a. Finally, a protective layer 25 having a double-layered structure of a phosphosilicate glass layer-or non-doped silicate glass layer and a silicon nitride layer (both not shown) is formed to cover the entire surface of the semiconductor device, thereby completing a two-layer interconnection structure. This interconnection structure is fundamentally composed of three lines including a line of the first electrode 22, a line of the second electrode 24 and the contact portion C where the first electrode 22 and the second electrode 24 are electrically contacted.

The degree of integration of large scale integrated circuits becomes higher with a finer size of the metal films or electrodes. When an electric current is applied to such a circuit, there will arise the problem that the breakage of the fine electrode takes place. This is called electromigration (which may be hereinafter referred to often as EM) failure. This EM failure takes place due to the fact that Al atoms ordinarily used as the metal film are moved along the direction of movement of electrons generated by application of the current and a portion of the film where the movement is violent is broken. To avoid this, it has been proposed that elements other than Al, e.g. Cu, Ti and the like, are added to the Al film so that the number of vacancies present in Al grain boundaries are decreased to reduce the movement of Al atoms (P. B. Gate, Solid State Technology, Vol. 3, pp. 113-120, (1983)).

Another problem involved in the fine electrode of the super large scale integrated circuits is a phenomenon where when kept merely at high temperatures, the electrodes are broken. This is called stress migration (which may be referred to often as SM). With regard to the SM phenomenon, it has been reported that the breakage frequently takes place when the circuit is maintained at a temperature of approximately 150.degree. C. This will be overcome to an extent by addition of Cu or the like element (J. Kelma et al, The 22nd Annual Proceeding International Reliability Physics Symposium, pp. 1-5 (1984)).

The above two reports have been made to solve the problems on the single-layer electrode line. With LSI circuits including a double-layer or multilayer interconnection structure, the problems involved in the EM and SM phenomena at contact portions between the interconnected electrodes have to be solved, like the single-layer wiring line.

Few reports have been made on the EM and SM phenomena at the contact portion. It is empirically assumed that the EM and SM characteristics will be improved by tapering the insulating film at the contact portion to improve the aluminium coverage of the second wiring. This tapered technique has now been in use. However, it has been reported that this technique has the following disadvantage (H. Tomioka et al, The 27th Annual Proceeding International Reliability Physics Symposium, p. 57 (1989)). The second aluminium film is usually formed by sputtering at relatively high sputtering power. In this condition, when the contact portion is tapered, the substrate made generally of SiO.sub.2 is apt to be sputtered at the tapered contact portion as well. This results in SiO.sub.2 formed between the first and second aluminium films, thereby causing the EM characteristics to be degraded. Thus, care should be paid to the formation of SiO.sub.2 at the contact interface.

Another approach is to make the first electrode layer 22 which has a multi-layered structure including a first sub-layer 22a, a second sub-layer 22b and a third sub-layer 22c as is shown in FIG. 9. Likewise, the second electrode layer 24 is arranged as having a multilayer structure which includes a first sub-layer 24a, a second sub-layer 24b and a third sub-layer 24c as is shown in FIG. 9. The sub-layers 22a, 24b are each made of Al or an Al alloy. By this, the movement of Al atoms toward the substrate 20 and the first electrode layer can be prevented to an extent by the presence of the sub-layer 22a or 24a. More particularly, the EM and SM phenomena can be suppressed to an extent on comparison with the case where the layers 22 and 24 are made only of Al or an Al alloy.

The status of the double-layer electrode arrangements can be summarized in the following table.

As will be apparent from the above table, although EM and SM characteristics of the first and second metal films and the problems involved in the contact portion have been individually reported, we have not found any attempt to match the respective films and the contact portion when the double-layer electrode structure is applied to semiconductor devices.

It is accordingly an object of the invention to provide a semiconductor device having a double-layer interconnection structure with improved electromigration and stress migration characteristics wherein the respective metal films, each having a multi-layer structure, and a contact portion of the films are so matched as to improve the electromigration characteristics particularly at the contact portion.

It is another object of the invention to provide a semiconductor device having a double-layer interconnection structure wherein a metal film covered with a protective layer and formed at a contact portion of first and second multi-layered metal films is reduced in stress from the protective layer.

It is a further object of the invention to provide a semiconductor device having-a double-layer interconnection structure whose electromigration life is improved by one order of magnitude over known interconnection structures.

The above objects can be achieved, according to the invention, by a semiconductor device of the type which comprises a first metal film formed on a semiconductor substrate, an insulating film having openings and formed on the first metal film, a second metal film provided in electric connection with the first metal film through the openings of the insulating film, and a protective layer made of a film selected from the group consisting of phosphosilicate glass and non-doped silicate glass and a silicon nitride film formed on the second metal film in this order wherein the first and second metal films have, respectively, a multi-layer structure including a first sub-layer made of a member selected from Ti, TiN, TiW, Co and W, a second sub-layer made of a metal selected from Al and Al alloys containing Si, Cu, Ti, Pd or mixtures thereof in an amount of up to 5 wt % in total, and a third sub-layer made of a member as used in the first sub-layer, i.e. Ti, TiN, TiW, Co or W. In the semiconductor device, the improvement resides in that a value obtained by multiplying a thickness of the silicon nitride film formed on the second metal film layer at the openings by a stress of the silicon nitride film is not larger than 2/5 of a value of a thickness of the silicon nitride film formed on the second metal film on portions other than the openings by a stress of the silicon nitride film on the portions.

The openings may be tapered or not. If tapered, an angle of the taper is preferably in the range of not less than 75.degree..

The first and third sub-layers of both the first and second metal films may have a multi-layer structure including two or more layers made of different elements or compounds defined with respect to the first and third sub-layers.

Figs. 1a to 1c are, respectively, a semiconductor device according to an embodiment of the invention wherein FIG. 1 a is a schematic sectional view of a contact chain portion used to measure the EM life of a multilayer interconnection, FIG. 1b is a schematic sectional view of a first metal film having a multi-layered structure, and FIG. 1c is a schematic sectional view of a second metal film having a multi-layered structure;

FIG. 2 is a schematic sectional view showing a contact portion between the first metal film and the second metal film according to the invention;

FIGS. 3a to 3c are, respectively, graphical representations wherein FIG. 3a is a graph showing the relation between the coverage rate of metal films and the angle of taper at a contact portion, FIG. 3b is a graph showing the relation between the thickness of a SiN protective film and the angle of taper at the contact portion, and FIG. 3c is a graph showing the relation between the average failure time and the angle of taper at the contact portion;

FIG. 4 is a graphical representation of the relation between the thickness of a SiN protective film and the average failure time;

FIG. 5 is a graphical representation of the relation between the total stress and the thickness of a SiN protective film;

FIG. 6 is a graphical representation of the relation between the average failure time and the total stress;

FIG. 7a is a graphical representation of the relation between the cumulative failure rate and the test time for semiconductor devices for comparison and of the invention, and FIG. 7b is a graphical representation of the relation between the electrode life and the test time;

FIG. 8 is a graphical representation of the relation between the percent defective and the retention time at 150.degree. C. for a stress migration characteristic of a contact chain;

FIG. 9 is an illustrative sectional view of a double-layer interconnection in a known semiconductor device;

FIG. 10 is a schematic plan view of a contact portion of a double-layer interconnection of a semiconductor device according to the invention; and

FIG. 11 is a graphical representation of the relation between the cumulative failure rate and the test time for illustrating an electromigration characteristic of a first metal film, a second metal film and a contact chain between the first and second metal films, respectively .

Initially, the reason why we attempted to improve the EM characteristic at contact portions between first and second metal films is described with reference to FIGS. 1 and 10 to 1 wherein like reference numerals indicate like parts.

The sizes of the first and second metal films and the contact portions usually employed in semiconductor devices are first described. For instance, when the width of the first metal film is set at a minimum level for microprocessors where a great load is applied to the metal film, e.g. 1.2 .mu.m, the width of the second metal film is determined at a minimum level of 1.2 .mu.m. The contact portion is determined to have a maximum size at a portion where the first and second are crossed. The area of the contact portion becomes 1.2 micrometer square. This is particularly shown in FIG. 10. In the figure, the alignment margin extending from first and second metal films 12, 14 and provided around contact portions C is an allowance in anticipation of an alignment shear in lithography. Thus, the sizes of the first and second metal films 12, 14 and the contact portion C used to make a double-layer interconnection structure are determined such that the first and second metal films 12, 14 have the same width and the contact portion C is a quadrilateral with one side corresponding to the width.

The EM characteristic of the double-layer interconnection structure as shown in FIG. 10 is measured. The double-layer interconnection structure consists of the first and second metal films and the contact portion and the sections of these portions are shown particularly in FIGS. 1a to 1c.

The first and second metal films in the semiconductor device according to the invention have, respectively, a multi-layered structure. For the measurement, the first metal film 12 has a sub-layer 12a which has a two-layer structure of TiN and Ti layers, a sub-layer 12b made of an Al alloy containing 1% of Si and 0.5% of Cu, and a sub-layer 12c made of TiN and serving as a protective coating in lithography. Likewise, the second metal film 14 has a sub-layer 14a made of Ti, a sub-layer 14b made of the Al alloy as used for the sub-layer 12b, and a sub-layer 14c made of TiN. 400 contact portions C of the first and second metal films 12, 14 with 1.2 micrometer square are connected in series (hereinafter referred to as a contact chain) as shown in FIG. 1a. The device of FIG. 1a will be described in detail hereinafter.

Since a step coverage of the metals for the multi-layered structures by sputtering is relatively poor, each contact portion is tapered for convenience' sake at about 55.degree. in a series of experiments made by us only for confirmation.

In addition to the double-layer interconnection shown in FIG. 1a, an electrode structure including the first metal film 12 having a width of 1.2 .mu.m and a thickness of 0.855 .mu.m is shown in FIG. 1b. In this connection, the first metal film has a thickness of 115 nm for the sub-layer 12a wherein the TiN layer has a thickness of 100 nm and the Ti layer has a thickness of 15 nm, 700 nm for the sub-layer 12b and 40 nm for the sub-layer 12c. Likewise, an electrode structure including the second metal film 14 having a width of 1.2 .mu.m and a thickness of 0.865 .mu.m is shown in FIG. 1c wherein the thickness is a total one including 25 nm for the sub-layer 14a, 800 nm for the sub-layer 14b and 40 nm for the sub-layer 14c. The lives of the respective structures are determined by EM measurement.

The EM measurement is made in the following manner.

After determination of the sectional areas of the first and second metal films and the contact portions, the respective structures are kept at 150.degree. C. and applied with an electric current with a current density of 2.times.10.sup.6 A/cm.sup.2 to determine a time before it fails. This is usually called EM measurement. It will be noted that the EM measurement is an acceleration test wherein a life test corresponding to 10 years is effected in about 100 hours. The upper limit of actual use conditions of LSI circuits include those conditions of a temperature of 80.degree. C. and a DC current density of 1.times.10.sup.5 A/cm.sup.2 whereupon a pulse current is passed through a 1.2 .mu.m square contact portion. The life of the interconnection structure is estimated by keeping a sample at 150.degree. C. and applying a DC current with a current density of 2.times.10.sup.6 A/cm.sup.2. With the contact portion, when the time before it fails is over 12 hours, its actual service life corresponds to over 10 years. With the first and second films, the time over 50 hours corresponds to an actual service life of over 10 years.

The results are shown in FIG. 11. As will be apparent from the figure, the contact chain is shorter in the time by one order of magnitude or over than the other parts.

It has been reported that the EM characteristic at the contact chain is degraded due to the presence of a thin SiO.sub.2 layer at the contact portion-between the first and second metal films. More particularly, when the second metal film is deposited by sputtering for the multi-layer structure, SiO.sub.2 at the tapered portion is sputtered by means of Ar ions and re-deposited on the first metal film exposed at the opening. To avoid this, the sputtering for the second film is performed at a small power level so that SiO.sub.2 is not sputtered.

For confirmation of the absence of SiO.sub.2 from a physical aspect, the section of the 1.2 .mu.m contact portion has been subjected to observation through a transmission electron microscope, with the result that no SiO.sub.2 is found.

In view of the above results, we made further studies to solve the problem that the EM characteristic at the contact portions between the first and second metal films is poor irrespective of the absence of SiO.sub.2 at the interface and thus to provide a semiconductor device with an improved EM characteristic at the contact portions of the double-layer interconnection.

Reference is now made to the accompanying drawings and particularly, to FIGS. 1a to 1c.

In FIG. 1a, there is shown a semiconductor device S which includes a semiconductor substrate 10 having a SiO.sub.2 film 11 thereon, a first metal film 12 formed on the SiO.sub.2 film 11. An insulating film 13 having openings 13a is formed on the first metal film 12, on which a second metal film 14 is formed as connected with the first metal film 12 through each opening 13a at a contact portion C with an area of about 1.2 micrometer square. Further, a protective film 15 consisting of a phosphosilicate glass film 15a and an SiN film 15b is formed over the entire surface. The semiconductor substrate 10 has a semiconductor region (not shown). The first and second metal films 12, 14 have, respectively, a first sub-layer 12a or 14a formed directly on the SiO.sub.2 film 11, a second sub-layer 12b or 14b, and a third sub-layer 12c or 14c as shown in FIG. 1a. These sub-layers are all electrically conductive in nature and are, respectively, made of metals or metal compounds defined hereinbefore. The sub-layer 12a is provided to prevent Al atoms in the sub-layer 12b from diffusion into the substrate. The sub-layer 12c serves as an anti-reflective coating in lithography. The sub-layer 12a may be in the form of a multi-layer such as a TiN layer and a Ti layer as used hereinbefore with respect to FIG. 10. In this case, the TiN layer is provided to prevent Ti atoms from conversion into TiSi.sub.2 on contact with the Si substrate. Likewise, the sub-layer 14a is provided to prevent mutual diffusion of Al atoms. The sub-layer 14c is provided as an anti-reflective coating like the sub-layer 12c. The sub-layers 12c, 14c may have a multi-layer structure like the sub-layers 12, 14a.

FIG. 1b is a schematic-view of a single-layer structure including the first metal film 12 alone. For instance, the first metal film 12 is 1.2 .mu.m in width and 0.855 .mu.m in thickness. Likewise, FIG. 1c is a schematic view of a single-layer structure including the second metal film 14 having, for example, a width of 1.2 .mu.m and a thickness of 0.865 .mu.m.

In the present invention, the EM characteristic at the contact portion C as shown in FIG. 1a is improved. In order to clarify the influence of the SiN film 15b, a taper or inclination angle at the contact portions as shown in FIG. 2 is changed. Then, the average failure time of the resulting semiconductor device is checked in relation to the coverage rate of the second film with the multi-layered structure and the thickness of the SiN film.

The coverage rate of the second film and the thickness of th SiN layer are determined through observation of a section of the contact portion with a scanning-type electron microscope. Further, samples with different coverage rates of the second film are provided wherein 500 contact portions are connected in series in each sample. The same level of electric current is applied to all the samples in such a way that the current density is set at 2.times.10.sup.6 A/cm.sup.2 at a minimum sectional area of the contact portion of each sample with a taper angle of 85.degree. to measure an EM value, thereby determining an average failure time. It will be noted that the average failure time is one at which a cumulative failure reaches 50% when the EM value is measured as shown in FIG. 11. The results of the above measurement are shown in FIGS. 3a to 3c.

As will be seen from FIG. 3a, a higher taper angle, .theta., at the contact portion results in a lower coverage rate of the multi-layered second film. Similarly, FIG. 3b reveals that a higher angle results in a lower thickness of the SiN film. In contrast, the average failure time increases with an increase of the angle, .theta., as is seen from FIG. 3c. In order to obtain the EM life of the contact chain at the same level as the EM life of the second film, the taper angle at the contact portion should preferably be in the range of not less than 75.degree., more preferably not less than 80.degree.. It will be noted that in the practice of the invention, it is not necessarily intended to define the taper angle at the contact portion, but physical factors which contribute to limitation of the EM life of the contact chain are important. In view of the coverage rate of the second film, the results of FIG. 3 are contrary to the generally accepted assumption that an increase of the current density leads to degradation of the EM characteristic. Accordingly, the influence of the SiN film has been investigated in the following manner.

There are fabricated double-layer interconnection structures of the type as shown in FIG. 2 where any SiN film is not formed in one case, and SiN films are formed with different thicknesses at the same stress level, and a SiN film is formed whose quantity of stress is reduced to 1/2 of the above-mentioned SiN films. In all cases, the taper angle is set at 85.degree.. The EM characteristic of each structure is measured to determine an average failure time. The results are shown in FIG. 4. In the figure, the mark ".smallcircle." is for the SiN films with the same level of stress. In this case, it will be found that the average failure time increases with a decrease of tile SiN film thickness. In the absence of the SiN film, the average failure time will be over 100 hours and is thus significantly improved with respect to the EM life. This means that the EM life of the contact chain has close relation with the SiN film. The mark ".circle. " in the figure is for the SiN film whose stress is half the stress of the SiN films indicated by the mark ".smallcircle.". From the results of FIG. 4, it will be seen that the EM characteristic at the contact portion is dependent not only on the thickness of the SiN film, but also on the stress of the SiN film. Thus, the EM life of the double-layer interconnection structure at the contact portion is greatly influenced by the thickness and the stress of the protective SiN layer. A smaller thickness results in a better EM characteristic and a smaller stress results in a better EM characteristic. The EM characteristic at the contact portion depends similarly on the thickness and the stress of the SiN film. This is because a degree of deformation of the multi-layered structure such as by warpage, compression or the like is expressed by a value obtained by multiplying the thickness of SiN film by the stress (dynes/cm.sup.2). This value will be hereinafter referred to as total stress.

In order to numerically express the above relationship, the stress is measured for different thicknesses of the SiN films by the use of the AUTOSORT MARK II 150 instrument, available from GCA Corporation. The stress of the SiN film at the contact portion is measured by depositing on a Si substrate by plasma-enhanced CVD a SiN film with the same thickness as the SiN film thickness at the contact portion determined by the SEM technique as in FIG. 3b. It will be noted that although the protective layer includes the PSG film, the stress of the PSG or non-doped silicate glass film is approximately 10.sup.8 dynes/cm.sup.2 and is negligible. The relation between the SiN film thickness and the stress is shown in FIG. 5. The film quality is changed by changing the deposition conditions of the plasma-enhanced CVD. In the figure, the mark ".smallcircle." indicates a 500 nm thick SiN film having a stress of 3.times.10.sup.9 dynes/cm.sup.2 and the mark ".circle. " indicates a 50 nm thick SiN film having a stress of 1.5.times.10.sup.9 dynes/cm.sup.2.

The relation between the measured total stress and the average failure time obtained from the EM value is shown in FIG. 6. As will be apparent from the figure, the life of the second film at the contact portion is inversely proportional to the total stress. In order to improve the EM life by one order of magnitude for the reason set out hereinlater using the protective layer having such a stress as mentioned above, consideration should be made to scatterings in measurement of the stress, the SiN film thickness and the life at the contact portion.

This is more particularly explained with regard to FIG. 6. The SiN film thickness deposited on the second film is 500 nm and the total stress at the second film is approximately 1.5.times.10.sup.5 dynes/cm. The relationship between the total stress at the contact portion and the average failure time may be represented by the solid line in the figure. In addition, the average failure time of the EM life of the second film is in the range of 10 to 200 hours as shown in the figure. In order to prolong the EM life at the contact chain substantially at the same level as that of the second film or to improve the life by one order of magnitude over that of a prior art counterpart, the total stress has to be not larger than approximately 6 to 8.times.10.sup.4 dynes/cm at the contact portion. This value corresponds to about 2/5 of the total stress of the SiN film formed on the second film at portions other than the contact portion. More particularly, the total stress of the SiN film formed on the second film in the opening should be not larger than 2/5 of the total stress of the SiN film formed on the second film at portions other than the opening.

The EM characteristic of a prior art double-layer interconnection structure of the type as shown in FIG. 9 having a contact portion with an area of 1.2 micrometer square and a double-layer interconnection structure of the invention having the same area of the contact portion as mentioned above where the total stress is not larger than 2/5 of the total stress of the protective film on a non-contact area of the multi-layered film is shown in FIG. 7a. From this, it will be seen that the EM characteristic of the invention is improved by not less than one order of magnitude over that of the prior art.

The reason why the EM life has to be improved by approximately one order of magnitude is ascribed to the guaranteed life of semiconductor products. The life of LSI semiconductor products is generally determined by the current density and the use temperature, under which it is guaranteed over approximately ten years. The measurement of the EM characteristic is made under accelerating conditions. To this end, the EM measuring conditions in this test include a temperature of 150.degree. C. and current density of 2.times.10.sup.6 A/cm.sup.2. The actual use conditions include a temperature of 80.degree. C. and a current density of 1.times.10.sup.5 A/cm.sup.2 for substantially all types of products except for specific applications. The cumulative failure of about 0.3% in FIG. 7a corresponds to 2.83 test hours, which in turn corresponds to a life of the metal film wirings of ten years under actual use conditions. This relationship is shown in FIG. 7b, revealing that the guaranteed life of longer than ten years can be achieved according to the invention.

In the foregoing, the second film including an Al sub-layer containing Si and Cu has been described. Pure Al with or without Si, Ti and/or Pd may be likewise used as the sub-layer with similar EM characteristic-improving effect.

Then, a stress migration characteristic which is another factor for determining the film life is described. 400 contact chains formed thereover with a protective layer made of a 300 nm thick PSG film and a 500 nm thick SiN film are connected in series and have a taper angle of 55.degree. for comparison and 85.degree. used in the present invention. The sample is maintained at 150.degree. C. to determined a percent defective of film breakage in relation to the time. The results are shown in FIG. 8. In the figure, the mark ".DELTA." indicates a prior art case where the contact size is 1.0 micrometer square, the mark " " indicates a prior art case where the contact size is 1.2 micrometer square, the mark ".quadrature." indicates a case of the invention where the contact size is 1.0 micrometer square, and the mark " " indicates a case of the invention where the contact size is 1.2 micrometer square. No percent defective takes place up to a time of 1000 hours in both the prior art and the invention. At 2000 hours for the 1.0 micrometer square contact portion, the percent defective takes place in the prior art. However, any problem will not be involved in the semiconductor products when the time of 1000 hours is ensured. The SiN film at the contact portion is not considered to significantly influence the stress migration.

As will be apparent from the foregoing, the stress of the SiN film and the thickness of the SiN film at the contact portion are properly controlled so that the value obtained by multifying the thickness of the silicon nitride formed on the second metal film having a multilayered structure at the contact portion is not larger than 2/5 of the value obtained by multifying the thickness of the silicon nitride film formed on the second metal film at the non-contact portion by the stress of the silicon nitride at the non-contact portion. By this, the stress exerted on the second metal film can be suppressed to a low level. The electromigration life of the contact portion can be improved by about one order of magnitude. The SiN film is usually formed in a thickness of from 300 to 1500 nm on portions of the second multi-layered film other than in openings. In openings, the thickness is in the range of from 120 to 600 nm.