Nitrogen Plasma Treatment to Prevent Field Device Leakage in VLSI Processing

(1) Field of the Invention

The present invention relates to method for making multiple layer metallurgy such as double metal, spin-on-glass multilevel metallurgy device structures which prevents field device leakage in VLSI structures.

(2) Description of the Prior Art

Multiple layer metallurgy such as double metal, spin-on-glass multilevel metallurgy is now used in high density integrated circuits for electrical connection to the device elements located in and on semiconductor substrate. Spin-on-glass is a very desirable material to be used in such methods and resulting structures to overcome the irregularity or substantially nonplanar surfaces of the first level of metallurgy. This irregular or nonplanar surface problem causes loss of resolution in the lithographic masking processing. The problem increases with higher level of metallurgy.

These problems have been recognized in the prior art and attempts have been made to overcome these topographical problems principally in the one micron and above feature dimensions. These techniques can be generally grouped in categories of planarization either involving etchback or nonetchback techniques.

In the case of etchback processing, typically a coating is formed on top of the irregular surfaces by an means of spin-on-glass or thermoplastic deposition techniques which produces a planar surface. A plasma, reactive or chemical wet etching is used uniformly across the planar surface to remove the deposited layer and the tops of the irregular humps or the like to produce a planar surface at the desired depth.

In the early nonetchback processing, the usual technique was to heat the dielectric layer, which was typically glass until the flow of the glass reduced the irregularities.

More recently the nonetchback planarization using a spin-on-glass sandwich has found interest at the about one micron feature dimension integrated circuit processing. This technique is described in U.S. Pat. No. 5,003,062 to Daniel L. Yen and assigned to the same assignee of the present invention. In this patent a sandwich dielectric structure is formed and used in one micron or less processing of spin-on-glass sandwiched between two silicon oxide layers. The spin-on-glass layer functions as the means for planarizing the irregularities.

In the Yen U.S. Pat. No. 5,003,062, spin-on-glass that has been cured still remains in the final product. However, workers in the field such as A. Malazgirt et al U.S. Pat. No. 4,986,878 has found that spin-on-glass even after curing by conventional techniques causes reliability problems, such as the presence therein of mobile ions, e.g. sodium to the point that they use the spin-on-glass in an etchback process for planarization and then completely remove the remaining material. Then a conventional insulator is deposited.

In the paper entitled "FIELD INVERSION IN CMOS DOUBLE METAL CIRCUITS DUE TO CARBON BASED SOGS" by D. Pramanik et al, a further discussion of the reliability problems in the use of spin-on-glass (SOG) is considered. They conclude that the use of SOG must be heavily restricted to overcome field inversion, that is a positive charge build up between the SOG layer and the other dielectric layers. They state that SOG must be purely inorganic phosphosilicate glass, since organic based SOG emanates hydrogen. Further, the require that etchback be used to reduce the thickness of the SOG to the bare minimum in the final product. Still further, they require that the layer covering the SOG be formed of a material that does not liberate hydrogen, such as silicon dioxide. Normal processes for depositing silicon nitride or oxynitride do liberate hydrogen.

In the U.S. patent application Ser. No. 07/743,779 dated Aug. 12, 1991 by L. S. Tsai et al entitled "Method and Resulting Device for Field Inversion Free Multiple Layer Metallurgy VLSI Processing" it is proposed that the inversion problem can be overcome by the incorporation of a specific dielectric between the Via 1 dielectric layer and the spin-on-glass (SOG) layer. This is a successful technique.

It is a principal object of the present invention to provide a method which does not have the restrictions for the use of SOG that the prior art require while solving the charge problem.

It is a further object of the present invention to provide methods involving the nitrogen plasma treatment of the via dielectric barrier layer below the SOG composite layer which allows the use of organic based SOG materials and makes the use of any etchback of the SOG unnecessary.

In accordance with the present invention, a method for forming multiple layer metallurgy, spin-on-glass multilayer metallurgy for a one micrometer or less feature size integrated circuit with substantially free field inversion, that is the positive charge between the first via layer and the SOG is described. A semiconductor substrate having a pattern of field effect device source/drain regions therein with a pattern of gate dielectric and gate electrode structures associated therewith and a pattern of field isolation structures at least partially within said semiconductor substrate electrically separating certain of these source/drain regions from one another are provided. A passivation layer is formed over the surfaces of said patterns. Then the multilayer metallurgy is formed thereover by opening a pattern of openings through the passivation layer to at least some of the source/drain regions, depositing and patterning a first metallurgy layer in contact with the pattern of openings, forming a first via dielectric layer over the pattern of first metallurgy layer, exposing the first silicon oxide via dielectric layer to a nitrogen plasma, forming a spin-on-glass layer over the via dielectric layer and curing the layer, forming a second via dielectric layer over the spin-on-glass layer, forming a pattern of openings in the second via layer, the spin-on-glass layer, and the first via layer, and depositing and patterning a second metallurgy layer in contact with the pattern of openings to make electrical contact with the first metallurgy layer wherein the multilevel metallurgy integrated circuit with substantially free field inversion is completed.

The important nitrogen plasma treatment step for the first silicon oxide dielectric layer is responsible for the removal of the charge problem in the final product. The plasma process is preferably at a temperature of less than about 450.degree. C. and for a time of more than about 30 seconds. The plasma uses a source of nitrogen, such as nitrogen gas that does not include hydrogen ions.

FIGS. 1, 2 and 3 are schematic, vertical cross-sectional representations of steps in the formation of the double metal, spin-on-glass device of the present invention.

FIGS. 4 and 5 are graphical representations which show the current voltage curves for devices made by the Examples of the present invention.

Referring now more particularly to FIG. 1, there is shown an illustration of a partially completed, single field effect transistor (FET). The first series of steps involve the formation of the dielectric isolation regions for isolating semiconductor surface regions from other such regions in the semiconductor substrate 10. The semiconductor substrate is preferably composed of silicon having a (100) crystallographic orientation. In an effort to simplify the description and the drawings the dielectric isolation between devices has been only partially shown and will not be described in detail, because they are conventional. For example, one method is described by E. Kooi in his U.S. Pat. No. 3,970,486 wherein certain selected surface portions of a silicon semiconductor substrate is masked against oxidation, and then the exposed unmasked surface is oxidized to grow a thermal oxide which in effect sinks into the silicon surface at the unmasked areas. The masked silicon remains as a mesa surrounded by the sunken silicon dioxide or Field OXide pattern, FOX 8. Then semiconductor devices can be provided in the silicon mesas according to the following processes.

The surface of the silicon substrate 10 is thermally oxidized to form the desired gate oxide 11 thickness. The preferred thickness is between about 70 to 200 Angstroms. The polysilicon layer 12 is blanket deposited by LPCVD (Low Pressure Chemical Vapor Deposition) method. The preferred thickness of the polysilicon layer 12 is between about 3000 to 4400 Angstroms. The polysilicon layer 12 is ion implanted with phosphorous or arsenic ions by conventional methods and dosages to render the polysilicon layer conductive or doped with phosphorus oxychloride at a temperature about 900.degree. C. The surface of the layer is either thermally oxidized or a chemical vapor deposition process to form silicon oxide layer 13. The layers 11, 12 and 13 are patterned by conventional lithography and anisotropic etching techniques as are conventional in the art to provide a desired pattern of gate electrodes and structure on the FOX 8 surfaces or elsewhere as seen in FIG. 1.

The source/drain structure of the MOS FET may now be formed by the following steps. The FIG. 1 illustrates the formation of an N channel FET integrated circuit device wherein the substrate or well is doped as P-. However, it is well understood by those skilled in the art that a P channel FET integrated circuit device could also be formed by simply substituting opposite polarities to those given for the N channel embodiment. Also, a CMOS FET could in a similar way be formed by making both N channel and P channel devices upon the same substrate.

FIG. 1, for example shows the device regions, typically source/drain regions 14 in the substrate as of N+ dopants. Lithographic masks may be required to protect the areas not to be subjected to that particular N ion implantation. The formation of the lithographic masks are done by conventional lithography and etching techniques. Although not illustrated for simplicity sake, lightly doped drain structures can be formed using the conventional and well known techniques.

The N+ source/drain ion implantation uses typically Arsenic, As75 with a dose of between about 1 E 15 to 6 E 15 atoms/cm..sup.2 and energy of between about 70 to 100 Kev. to complete the source/drain regions 14 of the N channel MOS FET integrated circuits device as seen in the FIG. 1.

A passivation or insulating layer 16 is now formed over the surfaces of the patterns. This layer may be composed of multilayers such as a thin layer of silicon oxide and a much thicker layer of borophophosilicate glass, phosphosilicate glass or similar insulating layer. The operational thicknesses of these layers are between about 800 to 1400 Angstroms for the oxide layer and between about 2000 to 8000 or more Angstroms for the glasseous layer. These layers are typically deposited by chemical vapor deposition in low pressure or atmospheric pressure, or in a plasma enhanced reactive chamber.

The contact windows are openings are formed through the insulating structure to the source/drain regions 14. Conventional lithography and etching techniques are used to form this pattern of openings.

The first metallurgy contact layer 20 is now deposited over the surface of the structure and within the pattern of contact openings. The metallurgy is preferably Aluminum/silicon/copper having a thickness of between about 5000 to 10,000 Angstroms. However, other possible metallurgy include titanium, titanium/tungsten, titanium nitride or the like. The preferred method of depositing this metallurgy is by sputtering under the processing conditions of about 2 to 8 kilowatts of power, 25.degree. to 450.degree. C., and 0.1 to 2 Pascal. However, other possible methods of deposition include chemical vapor deposition. The metallurgy is now patterned into the desired conductive lines by conventional lithography and etching techniques to form the pattern of metal layer 20 in FIG. 1.

The critical processing now begins to form the important planarization and dielectric layered structure between the first metallurgy layer pattern 20 and the second metallurgy layer pattern 36 which is shown in FIG. 3.

The usual first dielectric silicon oxide layer 22 of the spin-on-glass sandwich planarization structure is now formed above the first metallurgy pattern 20. It is typically in the range of between about 1000 to 4000 Angstroms in thickness. This layer is deposited either by the silane method or by the TEOS method. The following is the silane method process conditions of pressure of about 2.95 Torr, power of about 180 watts, temperature of about 400.degree. C., spacing of about 270 mils, silane of about 32 ccm, and dinitrogen oxide (N.sub.2 O) of 900 ccm. However, TEOS method is preferred to give the best step coverage over the surfaces being deposited thereon.

The important nitrogen plasma treatment step for the first silicon oxide dielectric layer is now done to the structure seen in FIG. 2. This critical step is responsible for the removal of the charge problem in the final product. The structure with the silicon oxide via layer is placed in a plasma vacuum chamber. The plasma process is preferably maintained at a temperature of less than about 450.degree. C. and for a time of more than about 30 seconds. The plasma uses a source of nitrogen, such as nitrogen gas that does not include hydrogen ions. The preferred vacuum is less than about 20 torr. Surprisingly, the presence of hydrogen ion form, for example ammonia in the plasma will not cause the improvement obtained with nitrogen gas alone. Other nitrogen ion sources can be used, but the use of gases such as nitrogen difluoride should be avoided due to the undesirable fluoride ions in semiconductor processing.

The spin-on-glass materials and how they are processed are critical to the success of the process for planarization of integrated circuits as described in the above cited in the Daniel L. Yen Patent. The most useful materials are silicates-Si (OH).sub.4 and siloxanes -(RO)nSi(OH).sub.4-n. These types of materials are generally known and available. Examples of the silicate type is OCD Type 2 made by Tokyo Okha Corp. and siloxane type is OCD Type 6 made by Tokyo Okha Corp. It is possible to use multiple coating of spin-on-glass. Each spin-on-glass coating is less than about 0.3 micrometers and preferably between about 0.08 and 0.2 micrometers. The more coatings that are used, the better the planarity. The material to be applied is thoroughly mixed in a suitable solvent which is usually a combination of a high boiling point solvent and a low boiling point solvent.

The preferred low boiling point solvents are methanol, ethanol, and propanol. The middle boiling point solvents are buthanol, penthanol, hexanol and methyl cellosolve. The high boiling point solvents are butyl cellosolve, propylene glycol, diethylene glycol and Carbindol. Other potential vehicles or solvents are NMP, HMPA, N.N-dimethylacetoamide, acetyl acetone, and malonic acid diethylester and the like.

The spin-on-glass material suspended in the vehicle or solvent is deposited onto the semiconductor wafer surface and uniformly spread thereover by the action of spinning the wafer. The material fills the indentations in the integrated circuit wafer surface, that is planarization. Most of the vehicle or solvent is driven off by a low temperature baking step. At this point, the critical vacuum degassing step is accomplished by subjecting the wafer to a vacuum of less than about 100 mtorr and 350 degrees C. This last step removes chemical materials which could during latter processing cause cracking and corrosion of the next level conductor material. Other coatings of the spin-on-glass material are applied, baked and vacuum degassed until the desired spin-on-glass layer is formed.

The final step in the making of the spin-on-glass layer is curing. Curing is a high temperature heating step to cause the breakdown of the silicate or siloxane material to a silicon dioxide like cross linked material. Water in the form of steam is the major reaction product of this reaction. The silicate spin-on-glass is cured at about 390.degree. to 420.degree. C. and siloxane spin-on-glass is cured at about 450.degree. C. in nitrogen. The spin-on-glass layer 24 is now complete.

The second via dielectric layer 26 is now formed on top of the SOG layer 24. This layer is formed in a similar manner as was the first via layer 22. This layer can be deposited by either the TEOS or the silane method. The preferred thickness of this layer is between about 2000 to 6000 Angstroms.

Openings are now made through the second via layer 26, the SOG layer 24, barrier layer 23, the first via layer 22 to the first metallurgy layer 20. These openings are made using conventional lithography and with the preferred etching conditions at a pressure of 800 mtorrs, power of 750 watts, gap of 0.8 cm., argon flow of 400 ccm, carbon tetrafluoride flow of 23 ccm, and CHF.sub.3 flow of 12 ccm.

The second metallurgy contact layer 36 is now deposited over the surface of the structure and within the pattern of contact openings to the first metallurgy layer 20. The metallurgy is preferably aluminum/silicon/copper having a thickness of between about 5000 to 11,000 Angstroms. However, other possible metallurgy include titanium, titanium/tungsten, titanium nitride or the like. The preferred method of depositing this metallurgy is by sputtering under the same processing conditions as given for the first metallurgy layer 20. However, other possible methods of deposition include chemical vapor deposition as is known in the art. The metallurgy is now patterned into the desired conductive lines by conventional lithography and etching techniques to form the pattern of metal layer 36 in FIG. 3.

The integrated circuit may now be completed with higher levels of metallurgy and passivation as is known to those skilled in the art or similar additional multilevel metallurgy layers as described above are repeated to make triple or higher metallurgy levels on the integrated circuit structure.

The following Examples are given to show the important features of the invention and to aid in the understanding thereof and variations may be made by one skilled in the art without departing from the spirit and scope of the invention.

The Examples were made using MOS field effect transistor test devices. The conditions for the silicon oxide layers--Normal silane; TEOS; and silicon-rich either TEOS or silane types--were those given in the description above. The nitrogen treatment conditions were: time of 2 minutes, temperature of 400.degree. C., power of 500 watts, spacing of 480 mils, pressure of 5 mtorr., and nitrogen flow of 500 ccm. The ammonia treatment conditions were: time of 2 minutes, temperature of 400.degree. C., power of 500 watts, spacing of 480 mils, pressure of 5 mtorr., and ammonia flow of 70 ccm. The SOG material used was the silicate type. The testing was done by forcing 5 volts on drain regions and increasing the voltage and measured the drain current, ID until 10.sup.-7 amps. current was detected.

The following table shows the split conditions and results to illustrate the critical effect of the nitrogen plasma treatment of the silicon oxide first via dielectric layer the important improvement of the fixed charge.

For VIA I split conditions, the VTFPO,N (poly field device N channel threshold voltage, VTH) test results show degraded devices unless there was a nitrogen plasma treatment of the silicon oxide dielectric layer before depositing the SOG layer. The abnormally low VTFPO,N (less than 10) of Examples 2, 4, 5 and 6 will result in chip yields of 0%. Both the NORMAL silane oxide process in Examples 2 which uses TEOS oxide, and the use of ammonia plasma treatment of Examples 4 and 5 failed and would show 0% yield. Only Examples 1 and 3 which used nitrogen gas plasma would have significant chip yield.

The FIG. 4 shows the I-V (current leakage to gate voltage) with the plasma nitrogen treatment of the Example 1 given in the lower curve and the NORMAL TEOS without treatment given as the upper curve. The lower curve is a very acceptable one, while the upper curve is not acceptable. FIG. 5 shows the I-V curves wherein the topmost curve is Example 4 and the next highest is Example 5. Both of these curves are unsatisfactory. The best and satisfactory curves are the lowest which is Example 1 and the next lowest of Example 3. Our failure analysis of the zero yield wafers demonstrate to be due to leakage which agree well with the I-V curves in these FIGS. 4 and 5.

While the invention has been particularly shown and described with reference to the preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made without departing from the spirit and scope of the invention.