Taking a New Approach to Copper ECD Bath Management

Olivier Blachier, Jim Clark BOC Edwards, Chemical Management Division Chaska, Minn. -- Semiconductor International, 7/1/2000

Traditional feed-forward approaches have tried to correlate the Cu deposition tool throughput and its related current charge density to breakdown rates of bath organics (typically measured in mg/amp-hr). These methods have been complemented by analysis-based management techniques using voltammetric stripping (CVS, PCVS, AC voltammetry) to continuously adjust the plating tool organics replenishment algorithm. This, however, does not focus on the process engineer's need to better assess the impact of organic byproducts buildup on the feature-fill capability. Moreover, the matrix effects between "active" and "inactive" organics during the bath analysis often lead to misinterpretation of the real organic concentrations and their corresponding replenishment needs.

We have developed a system called CuBIS (Copper Bath Integrated System) that is based on a "remove-and-reconstitute" approach; a bleed stream is recycled by complete routine organic stripping and replacement. This method replenishes the bath constituents at a rate that automatically varies with the plating conditions and byproduct formation, with the chemistry-specific parameter incorporated into the control logic. The system also incorporates modules to control potential contaminants present in the chemistry, avoiding buildup of heavy metals or other compounds. We also have developed a proprietary copper abatement system to remove copper from the plating rinsewater stream and treat any waste generated by the organics regeneration process itself.

1. “Batch dump” control vs. “Bleed” control (0.3 amp-hr/wafer used to plate 1µm thickness).

Additive chemistry and characterization

Organic additives are essential components for successful copper electrochemical deposition in any application. Generally, the additives are categorized as either suppressors or accelerators, depending on their role in the electroplating process. Chemically, each category consists of compounds that may have various organic functional groups. During normal electroplating, these organic additives may be consumed in stoichiometric reactions or may break down in a variety of oxidative side reactions, some of which are non-stoichiometric. In any event, the original concentrations must be maintained, and the organic breakdown products must be removed for consistent performance of the plating bath.^2-4

As their names imply, suppressors act to impede the deposition of metallic copper on the cathodic surface, while accelerators enhance the deposition. Suppressors can be further characterized as either carriers or levelers, while accelerators are considered to be brighteners. The suppressors generally are polymeric surfactants. In the case of carriers, they form a monolayer at the cathode, which offers a diffusion barrier to cupric ions and enhances cathodic polarization needed for fine grain structure (carriers require the presence of chloride ions to perform their function properly; hence the inclusion of HCl, the only inorganic additive, in plating solutions). Levelers typically are multiply charged and adhere preferentially to highly charged areas such as corners and edges, and thus prevent overhanging at trench mouths. On the other hand, their large size impedes their migration into trenches, which in turn impedes conformal filling and allows for better bottom-up filling.⁵

2. TOC concentration after addition of additives and oxidation/removal through CuBIS stripping system.

Accelerators usually are unsaturated compounds containing a polar sulfur, oxygen or nitrogen functional group. They adsorb strongly and uniformly on seed surfaces, promoting dense nucleation and, consequently, growth of fine grains. This leads to a uniformly smooth, well-textured (i.e., bright) finish. It has been noted that accelerators tend to break down more rapidly than suppressors.⁵ In a simplified approach, it has been estimated that a commercially available chemistry has accelerator agents with a stoichiometric breakdown rate estimated at 2 mg/amp-hr while its suppressor agents break down at a rate of 10 mg/amp-hr (Fig. 1).

The simplest approach to control organics is to run the process until plating results are unacceptable, then dump the entire bath and begin with a fresh one. This generates large quantities of waste that must be treated due to the copper and acid content. There is essentially no control, in either the general or statistical sense, over the bath chemistry. A somewhat more sophisticated approach is to continuously remove a sidestream of the bath solution and replace it with fresh solution. This is known as the "bleed and feed" approach. Typically, both methods lead to the same amount of waste generated over time, ranging from 10 cc/wafer to 25 cc/wafer at high throughputs. In addition, while it does remove some of the contaminants from breakdown of the organic additives, it does not completely remove them, and only dilutes them somewhat to a steady-state concentration.

Due to the increasing need for lowering the variability related to byproducts buildup in the bath and decreasing the cost of ownership of the electochemical deposition tool, we have employed a "closed-loop" method that regenerates the chemistry by continuously stripping all organics (active and byproducts) from the recirculating bath, verifying on-line the performance of this removal method, and then adding the original components back in. This also is complemented by an inorganics control module that automatically measures and adjusts any inorganic components (CuSO₄, H₂SO₄, HCl, DI water) that might be needed. The organics are removed by a proprietary method that uses a combination of oxidation techniques combined with activated charcoal filtration.

Typically, bath solution is removed continuously from the tool baths and fed into the CuBIS at rates ranging from 5 ml/min to 25 ml/min per tool. This can, however, be extended to higher regeneration flow rates in the future if byproducts stabilization is needed at a lower concentration. This flow rate is adjusted automatically based on input from the process tool: amp-hours, wafer throughput, tool status, etc. The solution is continuously stripped of any organics present. The stripped solution is sent to a holding tank, from which batches are withdrawn periodically for regeneration, then sent to a daytank.

3. CVS parameters of treated and non-treated chemistry (measured by ECI Technology Quali-Lab system).

Various organics removal techniques

Electroplating of metals is common in the printed circuit board manufacturing industry, and maintaining consistent chemistry for high-quality plating using copper-based electrolytes has been fairly well understood for the past 20 years. Circuit board manufacturers and plating shops have developed the "carbon" treatment to regenerate their plating bath from time to time (Shipley ELECTROPOSIT 1000 data sheet, 1991, Enthone-OMI SEL-REX MICROFAB Cu 200 technical data sheet, 1995). As the standard of electroplated copper required by the semiconductor industry is much higher, a more stringent chemistry control is necessary. The easiest and commercially available way to remove dissolved organic species (contaminants) from copper plating baths in PCB applications is to treat them with H₂O₂ addition and activated carbon followed by an in-line filter cartridge.

H₂O₂ oxidizes and deactivates most of the organic additives, and carbon filtration removes up to 95% of the TOC. This traditional approach, if used in the semiconductor industry, leads to a dilute plating chemical containing trace deactivated additives. Due to the amount of H₂O₂ required to achieve a high removal efficiency, the inorganic chemistry cannot be regenerated.

Bulk carbon filtration also has been investigated and shows a repeatable >90% TOC removal efficency. However, activated charcoal is sensitive to organic polymer chain sizes and allows trace active and depleted organics to pass through its media. Therefore, using carbon filtration as the main method to remove organic byproducts may not be adequate. Moreover, the filter volume required to continuously treat the bleed chemical from multiple plating tools leads to a high disposal and maintenance cost.

Considerable research has been performed by a leading filtration company to study the efficiency of destroying organic contaminants by using oxidizing agents combined with heat and UV light exposure. The major advantage of this method is the selectivity that can be obtained by varying process parameters. Also, organic compounds can be photochemically decomposed to carbon dioxide by exposure to certain wavelengths of UV radiation. This is considered to be a clean process with no direct contact with the process fluid, and it also can kill any microorganism growth in the solution.

Combined with powerful oxidizing agents that can be used to oxidize organic content into carbon dioxide, the results in Figure 2 and Figure 3 show the process can be controlled very repeatably over time. Initial CVS data and HPLC/MS organic species characterization demonstrate that the plating parameters remain the same over time.

Bath impurity control

4. Inorganic concentration (Cu2+, H2SO4, Cl-) over multiple replenishment cycles.

Because of this closed-loop regeneration approach, impurity control and management becomes critical. A detailed study has been performed, and new techniques have been incorporated into the system design.

Sodium and phosphorus buildup, for example, are stabilized in the regeneration loop below a set concentration, which can be adjusted by the process tool operator. Metals are continuouly being leached out from the bath chemistry using a proprietary membrane technology. This allows the chemistry purity to increase over time on-line and provides a direct control on Fe or Zn. However, the bath chemistry impurity specification still represents an unknown for the semiconductor industry: How does the concentration of impurities in the bath correlate to the deposited film impurity concentration? Pourbaix diagrams show electrochemical deposition of P or Na is unlikely. This was confirmed by various studies.⁶For the long term, new chemistry additives are being developed with Na-free brighteners, and non-phosphorus based anodes are under development.

5. Consumables cost per wafer at various regeneration rates.

Process and CoO benefits

Process stabilization of both byproducts and active organics, combined with a tighter control of key inorganic chemicals such as hydrogen chloride, are evident advantages provided to the copper electrochemical deposition process control.

Optimized feature filling and better process repeatability at the wafer level, over time and between multiple baths in a production environment, should result from such stabilization. This new approach is now undergoing extensive wafer marathon-tests to demonstrate these benefits.

Figure 4 shows the results of an extensive test of the autotitrator capabilities. The analyser used achieved repeatable performance with a relative standard deviation of less than 3% for the chloride concentration determination. The use of a dedicated titration vessel, kept separate from the Cu and H₂SO₄ titration vessel, made this result possible. This allows back-to-back measurements at a higher frequency and leads to a more precise adjustment of the Cl- in the bath.

Finally, this new approach brings tremendous cost benefits to the end user by lowering the cost of ownership (CoO) of the plating tool. Detailed cost estimates per wafer calculations using a commercially available chemistry are shown in Figure 5.

6. Total cost savings with regeneration approach.

The cost of consumables also is greatly reduced in this approach (Fig. 6). Only the organic additives (and organic byproducts) are being removed, so only these organic additives need to be replaced. This is in contrast to the previous approaches, where the complete solution is discarded and then replaced, either by the bleed stream or by dumping the entire bath.

Conclusion

The "remove-and-reconstitute" ap-proach has several advantages over previous modes of plating bath management. By selectively stripping only the organic components and then replacing them, cleaner and more repeatable bath chemistry is maintained. By establishing a closed loop for solution regeneration, costly treatment of waste copper and acid is eliminated. Finally, the control of inorganic chemicals within a tight range and the continuous purification of the regenerated chemistry ultimately will result in an optimized process control for the plating tool manufacturer and the end user.

By combining these different modules, the integrated system addresses key challenges in the area of copper plating process repeatability and control, optimized feature fill, and copper waste elimination. Additionally, the cost of ownership of the plating tools is reduced significantly as compared with the tools using traditional peripheral systems found in advanced semiconductor fabs. •

---

REFERENCES

1. "International Technology Roadmap for Semiconductors: 1999", Table K, Environment, Safety, and Health Difficult Challenges — Summary of Issues, Semiconductor Industry Association.
   
2. R. Hurtubise, E. Too, C.C. Cheng, Enthone-OMI, "Copper Electrochemical Deposition: Is It Capable of Meeting the Stringent Reliability Criteria of the Copper Damascene Process?" Future Fab, Issue 5.
   
3. Internal BOC reports RE-99-064 and RE-2000-014.
   
4. Peter Bratin, Gene Chalyt, Michael Pavlov, "Control of Damascene Copper Processes by Cyclic Voltammetric Stripping"; Plating & Surface Finishing, March 2000, pp. 14-15.
   
5. R.D. Mikkola, Q.-T. Jiang, B. Carpenter, "Copper Electroplating for Advanced Interconnect Technology," Plating & Surface Finishing, March 2000, pp. 81-85.
   
6. Peter Bratin, "New Developments in the Use of Cyclic Voltammetric Stripping for Analysis of Plating Solutions," UPA Technology Internal Report, p.6.
   

Talkback

We would love your feedback!

Post a comment

» VIEW ALL TALKBACK THREADS

Related Content

• Topics
• Author

Related Content

 

By This Author

There are no other articles written by this author.

SPONSORED LINKS