WO1999023688A1

WO1999023688A1 - Cleaning dried residue of a chemical-mechanical polishing slurry

Info

Publication number: WO1999023688A1
Application number: PCT/US1998/014257
Authority: WO
Inventors: Anita M. Thurman-Gonzalez; Peter J. Beckage
Original assignee: Advanced Micro Devices, Inc.
Priority date: 1997-10-30
Filing date: 1998-07-07
Publication date: 1999-05-14

Abstract

Using an aqueous solution of hydrogen chloride and a surfactant, residue from a tungsten slurry is cleaned without laborious, time-consuming abrasive pressure. Many surfaces, such as polypropylene plastic may be cleaned without noticeable deterioration to the surface. The solution is applied to a surface containing dried slurry residues, is allowed to soak for a period of time, and then is removed, typically by rinsing in water. Such a procedure removes the tell-tale orange stains typically present on surfaces which have come into contact with iron-containing slurries, particularly those used to polish tungsten. This is accomplished in a chemical reaction without laborious manually-applied abrasive scrubbing. Since dried slurry residues contribute to particulate contaminants within the slurry, more frequent and complete residue removal reduces the number of particulates within the slurry. Consequently, semiconductor surface scratching is reduced and the yields are necessarily increased. Moreover, damage to the CMP tool surfaces from particulate scratching is also reduced. As a bonus, the cosmetic appearance of a CMP area within a fabrication facility is markedly improved as a result of easy, frequent residue cleanings.

Description

CLEANING DRIED RESIDUE OF A CHEMICAL-MECHANICAL POLISHING SLURRY

TECHNICAL FIELD

The present invention relates to polishing slurry, and more particularly to removing residue of a chemical-mechanical polishing slurry for polishing semiconductor wafers.

Background Art

In the manufacture of integrated circuits, the planarization of semiconductor wafers is becoming increasingly important as the number of layers used to form integrated circuits increases. For instance, metallization layers formed to provide interconnects between various devices may result in nonuniform surfaces. The surface nonuniformities may interfere with the optical resolution of subsequent lithographic steps, leading to difficulty with printing high resolution patterns. The surface nonuniformities may also interfere with step coverage of subsequently deposited metal layers and possibly cause open or shorted circuits.

Various techniques have been developed to planarize the top surface of a semiconductor wafer. One such approach involves polishing the wafer using a polishing slurry that includes abrasive particles mixed in a suspension agent. With this approach, the wafer is mounted in a wafer holder, a polishing pad has its polishing surface coated with the slurry, the pad and the wafer are rotated such that the wafer provides a planetary motion with respect to the pad, the polishing surface is pressed against an exposed surface of the wafer, and the slurry is used as a hydrodynamic layer between the polishing surface and the wafer. The polishing chemically and mechanically erodes the wafer surface, and the process continues until the wafer surface topography is largely planarized.

In chemical-mechanical polishing (CMP), the abrasive particles provide friction while oxidants and/or etchants cause a chemical reaction at the wafer surface. Additives can also be added to enhance the removal rate, uniformity, selectivity, etc., and dilution by deionized water is also practiced.

CMP is becoming a preferred method of planarizing tungsten interconnects, vias and contacts.

Tungsten CMP slurries typically include abrasive particles such as alumina, a ferric salt oxidizer such as ferric nitrate, a suspension agent such as propylene glycol, and deionized water. With proper process parameters, CMP tungsten processing has shown significantly improved process windows and defect levels over standard tungsten dry etching. One significant advantage of CMP tungsten processing is that it has a highly selective polish rate for tungsten as compared to the dielectric. This selectivity allows for over-polishing while still achieving a flat tungsten stud. When overetching occurs using dry etching, the contact or via becomes further recessed, which creates a serious disadvantage since overetching is frequently required to remove defects. The advantages of CMP, however, can be offset by the creation of significant defects during polishing, such as scratches. The prior art teaches that scratching can be controlled by the proper manufacturing, size control and filtering of the abrasive particles. The prior art also teaches that the proper mixing sequence of the abrasive particles with the suspension agent leads to lower defects.

Unfortunately, use of typical tungsten CMP slurries results in dried residues which remain on work surfaces of tubing, tooling, mixing stations, floors, and any other surface coming into contact with such a slurry which is not flushed clean. These residues frequently result in orange discolorations (from the ferric ion) on what are usually white or semi-transparent plastic surfaces, and have been historically very difficult to clean from such surfaces. In fact, many people avoid using the ferric slurries for tungsten because of the difficulty of cleaning the residues. If not removed, the residues may contribute even more particles to the slurry which, if they make their way onto the polishing pad, results in even greater wafer scratching. Even if repolishing a wafer is effective to remove previous scratches, the lost time due to analysis of the scratches and, of course, the re-polish itself adds significantly to the cost of such wafers. Moreover, the presence of such "rust-colored" stains in an otherwise pristine wafer fabrication facility is aesthetically unpleasing to both casual observers and to ISO-9000 inspectors alike.

The cost of cleaning such CMP slurry residues has been considerable, in lost production time of both equipment and operators, as well as in consumables. Expensive production tools are taken out-of-service to clean such residues, and current cleaning techniques require lengthy (e.g., 45-60 minutes) downtime. Moreover, considerable operator attention and mechanical action (i.e., manual "elbow grease") has been typically required, further adding to the human cost of removing the residues. In addition, existing cleaning techniques require numerous consumables, such as abrasive pads (e.g., 3M™ pads), cleaning wipes, and mildly-abrasive scrubbing solutions (e.g., SOFT SCRUB™ cleanser). Isopropyl alcohol (IPA), citrus acid (e.g., lime juice) as well as various bleaches and detergents have also been traditionally used to clean such residues, but without significant advantages nor reduction in time and effort required to clean an item.

Such cleaning problems have been vexing for many years. CMP techniques have been known for many years, and were discussed in the technical literature at least as long ago as 1990. Nonetheless, CMP processing areas within state-of the-art wafer fabrication facilities are frequently distinguished as the "dirtiest looking" areas within an otherwise spotless facility, because of the difficulty and time-consuming nature of removing slurry residues, particularly those slurries used to polish tungsten. Improved methods and solutions are clearly still needed.

DISCLOSURE OF INVENTION

Using a solution of hydrogen chloride, residue from a tungsten slurry is cleaned without laborious, time-consuming abrasive pressure. Many surfaces, such as polypropylene plastic may be cleaned without noticeable deterioration to the surface. The solution is applied to a surface containing dried slurry residues, is allowed to soak for a period of time, and then is removed, preferably by rinsing in water. Such a procedure removes the tell-tale orange stains typically present on surfaces which have come into contact with iron- containing slurries, particularly those used to polish tungsten. This is accomplished in a chemical reaction without laborious manually-applied abrasive scrubbing.

The advantages of this invention are numerous and significant. Because the cleaning time is so much faster than previous techniques, substantially less operator time is required and tool availability is dramatically increased. Consequently, production time is increased. Moreover, because frequent cleanings may be performed with little interruption in production schedules, there is less need to schedule downtime for preventative maintenance to perform major cleaning of a CMP tool.

Since dried slurry residues contribute to particulate contaminants within the slurry and onto the pad, more frequent and complete residue removal reduces the number of particulates within the slurry, on the sides of the polisher, on the polishing pads, and at other harmful locations. Consequently, semiconductor surface scratching is reduced and the yields are necessarily increased. Moreover, damage to the CMP tool surfaces from particulate scratching is also reduced. As a bonus, the cosmetic appearance of a CMP area within a fabrication facility is markedly improved as a result of easy, frequent residue cleanings.

As an additional advantage, cleaning slurry residues using the present invention is less expensive than previous cleaning solutions and the consumables necessary to apply those previous cleaning solutions, even those that don't even work as well.

These and other advantages have the dramatic effect of reducing the cost of semiconductor production, reducing the time required for cleaning of related fabrication equipment, and increasing the yields of semiconductor wafers undergoing a CMP process operation.

Commercially available products have been marketed for a number of years for removing various lime, scale, and rust stains, particularly from common bathroom surfaces (e.g., sinks, bath tub, toilet bowl, shower tile and grout, etc.). Such products have never been contemplated, however, as suitable solutions for use within a semiconductor fabrication facility. One potential reason is the perception that such products don't achieve the high purity requirements of a semiconductor fabrication facility. Another potential reason for this may be that these products frequently include constituent chemicals (e.g., strong acids) which are usually used in a semiconductor fabrication facility only in a well-contained environment, such as within a reactor chamber or under a well ventilated exhaust hood. Surprisingly, certain of these commercially available products actually perform very well in removing dried slurry residues from workpieces coming into contact with a tungsten slurry.

In one embodiment of the present invention suitable for a semiconductor manufacturing environment, a method of cleaning chemical-mechanical polishing (CMP) slurry residue from a surface of an item includes providing a solution including hydrogen chloride, applying the solution to the surface of the item, letting the solution remain on the surface of the item for a period of time, and removing the solution from the surface of the item.

These and other objects, features, and advantages of the invention will be further described and more readily apparent from a review of the description of the preferred embodiments which follows.

BRIEF DESCRIPTION OF DRAWINGS

The following description of the preferred embodiments may be better understood by referencing the accompanying drawings.

Figure 1 is a flowchart illustrating one embodiment of a residue removal sequence in accordance with the present invention.

The use of the same reference symbols in different drawings indicates similar or identical items.

MODE(S) FOR CARRYING OUT THE INVENTION

We have discovered that a diluted solution of a household product marketed to remove stains from residential toilet bowls is extremely effective at removing dried residue from a polishing slurry used in a semiconductor fabrication facility for chemical-mechanical polishing of tungsten. A solution including a concentration of LIME-A-WAY® brand Stain and Rust Dissolving Toilet Bowl Cleaner (hereinafter LIME-AWAY® brand cleaner) may be applied to a work surface containing dried tungsten slurry residue, allowed to soak for a short period of time, and removed by rinsing or washing with water. The tell-tale orange stains typically present on such work surfaces which have come into contact with such a tungsten slurry are removed with no time-consuming abrasive rubbing. The solution rinses completely with water and has no apparent residuals to cause particle contamination.

- 3 -

LIME-A-WAY® brand cleaner is distributed by Benckiser Consumer Products, Inc., Danbury Connecticut 06810, and is marketed solely "for toilet bowls only. Not for use on toilet seats, lids, or any other surface." A telephone call placed to their toll-free Consumer Relations telephone number reinforced the recommendation for use only on porcelain toilet bowls, as the agent instructed us that Benckiser "would not accept liability for any other use." In particular, no use on any plastic surface was endorsed, even though the product is labeled as safe for plumbing and septic systems.

Experiment 1

A stained workpiece from a production CMP tool (brushes on a SpeedFam™ wheel) was initially used to test the effectiveness of LIME-A-WAY® brand cleaner in removing slurry residue stains. The solution was applied to the workpiece, was allowed to soak for a particular time, and then removed by rinsing with clean water, as shown in Table 1 below.

TABLE 1

The LIME-A-WAY® brand cleaner is stated (in its Material Safety Data Sheet, "MSDS") as containing 14.5% Hydrogen Chloride, has a pH < 1.0, and has a specific gravity of 1.070 +/- 0.010 g/cc. Measurements, however, indicated a pH of about 2. The HCl is believed to be the active agent responsible for breaking away the rust-colored stains from the workpiece surface. Moreover, the product is believed to contain a surfactant (also sometimes referred to as a suspension agent or a dispersion agent) for suspending the stain particles in solution until rinsed away. The product also apparently contains a perfume, for the MSDS states the product as having a "sweet wintergreen odor." Any such perfume is not believed to be contributory to the effectiveness of the solution in removing slurry residue stains. A proper hydrogen chloride concentration is believed to be important for satisfactory slurry residue removal, without damage to underlying workpiece surfaces.

Of significant note, the MSDS for LIME-A-WAY® brand cleaner warns against mixing the product with bleach, mildew stain removers, other household products, chlorinated detergents, or sanitizers. These are some the very chemicals previously used in attempting to clean slurry residue stains, and suggests the unlikelihood of attempting to even try to use LIME-A-WAY® brand cleaner in an application where these other chemicals were previously in use.

It is an important consideration for use within a semiconductor wafer fabrication facility that any cleaning solution be free of phosphates. Other commercially available products, such as LIME-SHINE™ cleaner, available from National Laboratories, L&F Products, located in Montvale, New Jersey, may perform satisfactorily in its ability to remove residue stains, but whose use is nonetheless highly undesirable because of the phosphate content.

Experiment 2

Other stained workpieces (e.g., an index table for a SpeedFam CMP tool) were used to test the effectiveness of various cleaning solution concentrations. In each case, the solution was prepared and applied to the workpiece, was allowed to soak for a particular time, and then removed by rinsing with clean water, as shown in Figure 1. A litmus test is used to confirm adequate rinsing to remove the solution. Several concentrations of LIME-A-WAY® brand cleaner mixed with water were tested for effectiveness and for potential deleterious effect to the workpiece. Results of this testing are shown in Table 2 below.

TABLE 2

The best results were achieved with the 50% solution (row II above). The 75% solution (row I above) was observed to be stronger than necessary in most cases, except for severely stained areas, where a full strength solution provided good results. A "black wipe" test confirmed the absence of remaining particles or polypropylene workpiece surface deterioration. Additional materials tested which also showed no deterioration included a variety of other parts from CMP tools, including tubing and fittings, tie wraps, connectors, a load/unload cup lifter, and a slurry distributor.

Oxide slurry residues (which are typically white rather than orange) were not as easily removed by using the solutions described above. Moreover, some apparent damage to TEFLON™ tape was observed, although only outer portions and not within the pipe threads of the fittings.

Experiment 3

Various surfaces within a production CMP tool (e.g., an index table, clean station, platen tubs, underside of platens, load station, unload cascade, and unload track) were used to test the effectiveness of LIME-A-WAY® brand cleaner in removing slurry residue stains, as well as to determine if such a solution caused any negative effects on the tool or upon wafers processed after the cleaning. The solution was applied to the workpiece, was allowed to soak for a particular time, and then removed by rinsing with clean water, as shown in Table 3 below.

TABLE 3

Specifically, this solution was used to clean the unload track portion of a two different polishers. Wafers were then unloaded through the tracks to determine if contaminants within the cleaning solution, and not removed by rinsing the unload track, were transmitted to the wafers. A first wafer was merely loaded and then unloaded through the track, while a second wafer was loaded, polished, and then unloaded through the track. A control wafer, which was not processed on the unload track, was also measured to baseline the measurements. A total of 6 wafers (2 from each polisher, along with a control wafer from each polisher) were analyzed using Vapor-Phase Decomposition-Inductively Coupled Plasma Mass Spectroscopy (VPDICPMS). Wafers processed after cleaning yielded well. Testing showed fewer particulates and fewer wafer surface scratches.

The experimental procedure and equipment for Experiment 3 were as follows:

Equipment: IPEC 472 Avanti Polisher

Wafer Carrier: Standard design.

Polish Pad: Industry standard.

Test Wafer: 8000 A of plasma TEOS (CVD deposited) on prime silicon wafer.

Metrology Tool: Vapor-Phase Decomposition-Inductively Coupled Plasma Mass Spectroscopy

Process Cycle: 1st Wafer: Load; then unload into the unload track.

2nd Wafer: Load as normal; polish for 30 seconds at 5 psi, 25 rpm carrier,

100 rpm platen; then unload per usual.

A trace metals analysis of the 50% solution of LIME-A-WAY® brand cleaner reveals the following elements, as shown in Table 4 below. TABLE 4

The measurements on the test wafers for the same contaminants is shown in Table 5 for the first polisher (POLISHER 10) and in Table 6 for the second polisher (POLISHER 1 1). The contaminant measurements for the control wafer, not processed through the respective polisher, are also shown in the respective two Tables.

TABLE 5

TABLE 6

It is encouraging to note that potassium (K) and sodium (Na), which are both present in the LIME-AWAY® brand cleaner solution, and which are harmful to semiconductor transistors, are virtually lacking on the wafers processed through the unload track. The measured amounts of aluminum (Al) and iron (Fe) are due to their respective presence in the slurry. The measured amounts of titanium (Ti) correspond to its presence on the processed silicon wafers. Curiously, while the measured amount of Calcium (Ca) in the LIME-A- WAY® brand cleaner solution was fairly high, it is unknown why the amount of calcium varies on the wafers measured. The sodium and calcium are believed to be part of a surfactant within the LIME-A-WAY® brand cleaner. Experiment 4

An additional series of experiments was conducted using solutions of hydrogen chloride mixed with various commercially available surfactants. The results of these experiments are shown in Table 7 below. The experimental procedure and equipment for Experiment 4 were as follows:

Test material: Index table brushes from SpeedFam™ polisher.

Process Cycle: All test material devices placed into beaker of the respective solution; soaked for 5-10 minutes; then removed from the beaker and rinsed for 30 seconds; soaked in H₂0 bath for 5 minutes; then air-dried. TABLE 7

The Dodecyl Sulfate, Na salt 99% (surfactant IV) is available from Acros Organics, located in New Jersey. Another surfactant in the above table (surfactant I) is a commercially available aqueous mixture of propylene glycol and methyl paraben from Universal Photonics, Inc., sold under the trade name "Everflo White".

As can be seen from the above table, increasing amount of HCl does not necessarily lead to more effective stain removal. The surface deterioration in the case of just HCl and water solution is a chemical attack of plastic and is believed to be caused by lack of a surfactant present in the solution. Moreover, solutions including more than one surfactant show no particular advantage. Conclusion

While the invention has been described with respect to the embodiments set forth above, the invention is not necessarily limited to these embodiments. For example, one or more of the solutions described above for use with iron-containingslurries such as tungsten slurries are also believed to be useful for removing residue stains of other slurries, such as those containing aluminum or copper. Moreover, the invention may be used for cleaning slurry residues from a wide variety of different surfaces, including, for example, floors, walls, tools, cleanup stations, slurry mixing stations, as well CMP tool workpieces. The description above uses the term "surfactant," but one skilled in the art will appreciate that "dispersion agenf and "suspension agenf may be equally well used. Accordingly, other embodiments, variations, and improvements not described herein are not necessarily excluded from the scope of the invention, which is defined by the following claims.

Claims

WHAT IS CLAIMED IS:

1. In a semiconductor manufacturing environment, a method of cleaning chemical-mechanical polishing (CMP) slurry residue from a surface of an item, said method comprising the steps of: providing a solution including hydrogen chloride; applying the solution to the surface of the item; letting the solution remain on the surface of the item for a period of time; and removing the solution from the surface of the item.

2. A method as in claim 1 wherein the solution includes a surfactant.

3. A method as in claim 2 wherein the surfactant comprises an aqueous mixture of propylene glycol and methyl paraben.

4. A method as in claim 1 wherein the solution is substantially free of phosphates.

5. A method as in claim 1 wherein the solution includes a concentration of hydrogen chloride in the range between 5-37% .

6. A method as in claim 1 wherein the solution is removed by rinsing with water.

7. A method as in claim 1 wherein the slurry comprises an iron-containing slurry for use with tungsten polishing.

8. In a semiconductor manufacturing environment, a method of cleaning iron-containing chemical-mechanical polishing (CMP) slurry residue from a surface of an item, said method comprising the steps of: providing a solution including hydrogen chloride and a surfactant, the solution being substantially free of phosphates; applying the solution to the surface of the item; letting the solution remain on the surface of the item for a period of time; and removing the solution from the surface of the item by rinsing with water.

9. A method as in claim 8 wherein the item includes one chosen from the group consisting of CMP tool workpieces, slurry mixing station containers and tubing, cleanroom flooring, cleanroom walls, cleanroom tools, and cleanup stations

10. A method as in claim 8 wherein the surfactant comprises an aqueous mixture of propylene glycol and methyl paraben.

11. A method as in claim 8 wherein the solution includes a concentration of hydrogen chloride in the range between 5-37% .