Generation and Prevention of Water Marks in IPA Vapor Drying
Technical Report 1102-21

Abstract:
Wafer drying after clean and rinse steps can result in the formation of water marks that adversely effect wafer processing. These water marks are the result of both residual contamination from evaporation and oxidation of the silicon surface from oxidizing species in the DI water before drying. Elimination of this effect can only be done in part by improving DI water quality. Complete resolution of this problem must include use of a IPA Vapor Dryer. For cases where the thermal load on the dryer is high (dual 150 mm cassettes or larger), the dryer design plays a critical role in the rate of recovery of the IPA vapor zone and thus, elimination of water marks.

Introduction and Objectives:
The use of IPA Vapor Drying has been well established for complete removal of water on silicon wafer surfaces following aqueous based clean steps. This technique has relieved users of the problems associated with standard rinse-drying such as poor adhesion and out gassing due to residual moisture in addition to reducing particulate generated by, and caused to be electro-statically held by, the spin cycle.

DI water inherently contains significant amounts of particulate, bacteria, inorganic (silica) and organic material. Removal of the DI in a ‘drying’ environment that allows evaporation of the water results in these materials being left as residuals on the wafer surface. Since the extent of this problem is dependent on the purity of the water used in the final rinse step, a partial solution includes improving DI water quality.

However, this work has determined evidence that even highly pure DI water will result in the formation of water marks due to oxidation and hydrolysis of the wafer surface if the water is allowed to remain on the wafer for even short time periods.

Water droplets along the topographical edges will result in residual contaminants on the wafer surface if they are not quickly displaced from the wafer after the aqueous clean cycle. These droplets, if allowed to evaporate or to remain on the wafer surface for extended times, will adversely effect subsequent processing. The key to resolving this problem lies in the ability to quickly remove these water droplets, without evaporation, before the oxidizing species can diffuse to the wafer surface.

Proper IPA Vapor Drying can solve this problem by displacing water on the surface of the wafer with IPA without allowing the water to evaporate. Since water evaporates very quickly in this environment, this can only be accomplished by rapid replacement of the ambient with IPA. Thus, the rate of thermal recovery of the IPA vapor zone plays a critical role in the ‘water mark’ formation problem of IPA vapor drying.

This paper reviews some of the more common process problems identified to be caused by the ‘water mark’ problem, proposes a possible mechanism of formation independent of water purity, and discusses an existing solution to the problem.

Experimental Method and Results:

DI Water Purity:
Experimentation was conducted on 150 mm patterned wafers to determine the effect of DI water purity on the formation of water marks using IPA Vapor Drying techniques. Full loads were used in all studies.

DI water was used from two different sources that varied in resistively and in Total Organic (TOC). All wafers were cleaned, rinsed, and dried under identical conditions with the exception of the DI source. The Vapor Drying technique used was of conventional system design.

Two important results were determined. First, significant reduction of the water spot formation was observed by using DI water from the cleaner source. This was to be expected since the conventional vapor dryer design is known to allow evaporation of water droplets prior to IPA replacement for full loads where thermal loading is highest. Secondly, water spot formation could not be totally eliminated under these conditions even when using water that meets current industry standards for purity. This result not only suggests another mechanism for water spot formation, it clearly indicates that complete solution of the problem lies in the vapor drying technique.

Table 1 details the water purity levels that were used in the high purity portion of this study. DI water contamination above these levels resulted in significant formation of water marks. This data is in agreement with published water purity guidelines by Balazs Laboratories for VLSI manufacturing of £ 1.0 m m minimum geometries.

Effects of Water Marks on Subsequent Processing:
Innumerable processing problems can result from water mark formation in the drying sequence depending on the placement of the sequence in the process flow. Two common problems are discussed here to typify the effects. Both of these effects have direct results on the physical properties of the wafer; the resulting electrical performance can only be inferred. However, it is clear that degradation of key indicators like threshold voltage, capacitance, and yield would be affected by shifts in means and increased variability. Increased variability across the wafer would be the pronounced affect with the magnitude being dependent on the extent of the water mark formation problem.

The two process problems identified here are abnormal oxidation growth rate of a Gate Oxide (Vapor Drying being used as the final step of the Gate Oxidation preclean) and WSi2 peeling (Vapor Drying being used as the final step of the WSi2 deposition preclean).

Effects on Gate Oxide:
Auger depth profile analysis was completed on wafers generated in the DI comparison study to determine the chemical nature of the water mark. See Figure 1a. Results indicated that these marks were actually locations where the Gate Oxidation proceeded at accelerated rates giving final films thicknesses 30% higher that the 150 A average. These defective areas were highly concentrated along LOCOS edges on the gate regions; intuitively the locations where water droplets would remain prior to vapor drying. Figure 1b shows the effect.

Figure 1

a) Auger Peak Height
Sputtering Time (min)

1: Abnormal Oxide, 2: Normal Oxide
Water Drop

b) Unusual Oxidation

The second effect of water marking is demonstrated by poor adhesion of WSi₂ films to the substrate. This problem is illustrated in Figure 2. Poor drying of the Polysilicon surface prior to WSi₂ deposition results in a water spot; oxidation of the wafer surface. WSi2 is deposited over the unwanted oxide and patterned. Subsequent oxidation of the WSi2 to form LDD spacers results in a W rich area in the WSi2. During the oxidation, silicon in the WSi2 film is oxidized on the surface to form SiO2. Silicon in the WSi2 film is replenished from the underlying Polysilicon. The presence of the unwanted oxide layer prevents silicon from diffusing upward to recombine with tungsten. The result is a tungsten rich area within the WSi2 layer. Subsequent oxidations in the process flow rapidly oxidizes the tungsten which, in turn, forms a complex with silicon dioxide; SiO2 . WO3. Thermal expansion differences between this compound and WSi2 results in the WSi2 peeling from the underlying silicon.

Figure 2

Mechanism of Water Mark Formation in Pure DI:
Water mark formation was still present even when using high quality DI and conventional IPA Vapor Drying techniques with large thermal loads. This suggested another mechanism was present other than the simple residual after evaporation. Infrared spectroscopy was completed on the small water drop-lets found on the wafers prior to vapor drying.

The results, shown in Figure 3, correspond with IR spectra of standard samples of hydrated silicon dioxide (SiO2.H2O). Presence of the SiO absorption indicate that not only is the silicon surface being oxidized in the aqueous environment, but the resulting oxides are being hydrolyzed by the water droplet.

Conventional vs. Improved IPA Vapor Drying:
Since the extent of water mark formation proved to be dependent on both DI quality and the time of contact with the wafer surface, additional investigations were conducted to determine if the IPA Vapor Dryer design played a significant role in the formation of water marks.

Conventional dryer design has proved adequate for smaller thermal loads, but as wafer size increased (along with product sensitivity) redesigns of the system have been made to allow for more rapid recovery of the vapor zone. The rate of recovery is known to be dependent on the rate of heat transfer to the IPA. More importantly however, uniformity of heat transfer to the liquid IPA, redirection of condensate to heated surfaces, and controlled vaporizing of the IPA, play critical roles in minimizing the recovery time for large thermal loads.

As the cassettes of wafers are removed from the rinse bath and inserted into the IPA vapor zone, the vapor zone collapses due to the thermal capacitance of the cassettes, holder, and wafers. As the system thermally recovers, the vapor zone rises from the bottom of the wafers upward. During this time droplets of water near the top of the wafer are exposed to a heated and dry environment that gives rise to rapid evaporation. If the vapor zone does not rise rapidly enough, water marks form either from the residual of evaporation or from the oxidation of the surface from the oxidizing species present in the water.

See Figure 4. As the cassette of wafers is lowered into the IPA Dryer, a), a water film covers the wafers if they are hydrophilic, or water droplets sit adjacent to topographical edges if the wafers are hydrophobic. The thermal load on the system causes the IPA vapor zone to collapse, b), exposing the top portions of the wafer to a rapid-drying environment. The vapor zone recovers, c), progressing from the bottom of the wafers upward with a rate determined by system design and the size of the thermal load. If the vapor zone does not rise rapidly enough, water spots are formed on the top of the wafers, d).

Studies were conducted using patterned wafers sent through identical cleaning and rinsing steps then split for Vapor Drying using either the conventional or improved dryer design. A large thermal load was used; two 150 mm cassettes.

The recovery time for each dryer could be determined visually simply by watching the vapor zone rise from the bottom of the wafers to the top. EPI’s improved design allowed the vapor zone to recover much more quickly than the conventional design. See Figure 5. Water mark formation was eliminated in the improved design but was present in significant amounts on the conventional design.

Figure 4

Discussion:
Water mark formation has been shown to be caused both by evaporation of water droplets on the wafer surface and by oxidation of the wafer surface by oxidizing species in the water itself. The first effect can be reduced, but not eliminated, with the use of highly pure DI water. Both effects are time dependent with the most critical time being immediately after insertion into the dryer, and subsequent collapse of the IPA vapor zone, when water evaporation rates are the highest. The size of the thermal load (the cassettes, holder, and wafers), along with system design, determine the rate that the IPA vapor zone will recover. For large thermal loads, improved dryer designs are required to eliminate the water mark problem.

Figure 5

Conclusions:
Water mark formation can only be completely eliminated on large thermal loads by the use of a IPA Vapor Dryer that allows vapor zone recovery quick enough to prevent evaporation of water or oxidation of the wafer surface. The EPI Vapor Dryer with the improved design (Patent Pending) allows vapor zone recovery significantly faster than conventional models.

Post Script:
Information and data contained in this report has been provided to EPI by it’s customers under various nondisclosure agreements. As always, EPI upholds the utmost respect for customer confidentiality and has received permission from the specific customer to reveal this information in the format shown. In some cases the data has been normalized or similarly altered to prevent disclosure of proprietary information. This report was compiled and edited by the EPI Engineering Group. If you have any questions regarding this report contact Kevin Schumacher of Essex Products International, Inc. At the time this article was published Kevin Schumacher was the Vice President of Operation and Engineering at S&K Products International Inc. He is now President and owner of Essex Products International.