Developing metrology for controlling Cu-electroplating additives
10/01/2002
overviewElectroplating baths used for copper processing contain a mixture of additives that are used up during processing, yet tight control of the concentration of each additive is crucial to process success. Metrology is a must for proper control. Engineers have adapted high performance liquid chromatography to provide the precision necessary for rapid control of the concentrations of multiple additives in copper baths.
Plating electrolyte for copper (Cu) processing is usually based on either copper sulfate or sulfonic acid [1]. The use of various additives in a plating bath help to control thickness uniformity and grain structure, ductility, hardness, and surface smoothness.
Depending on the roles in baths, organic additives are categorized as brighteners, suppressors, levelers, or wetters. Inorganic additives such as chloride ions are also used in conjunction with organics. Proper plating additives enable good adhesion on seed layer and prevent problems in subsequent processes, such as Cu CMP [2]. The most widely used additives are brighteners ("accelerators") and suppressors.
The key here is that plating bath additives must be properly controlled for processing success. To accomplish this, we have developed a metrology system using high-performance liquid chromatography (HPLC) for precise and rapid control of multiple additive concentrations in Cu electroplating baths.
 Figure 1. An HPLC system that uses three eluants, a reversed-phase column, and optical detection for simultaneous measurements of brightener and suppressor. |
null
Bath chemistry and analysis
Our work used a sulfuric-acid-based solution containing Cu ions (10–30 g/L), sulfuric acid (50–200 g/L), and chloride ions (20–80 mg/L). Sulfuric acid controls conductivity of the electrolyte. Chloride ions are known to assist the adhesion of suppressorelements on Cu seed layers and to alter diffusion layer thickness.
A brightener — a catalyst with a molecular weight <500 g/mole — has a large diffusion length and enhances charge transfer, contributing to reduction of Cu++ to Cu+. Typical brighteners are saccharine and mercapto-propane sulfonic acid (MPS) and its derivatives.
A suppressor is usually a surfactant having short diffusion length and a molecular weight between 1000 and 10,000 g/mole. The most commonly used suppressors are polyethylene glycol (PEG) and poly-propylene glycol (PPG), and their derivatives.
Void-free interconnects for damascene and dual-damascene structures are obtained by delicate balance of additives. Metrology is important for a plating process because some of the chemical components are consumed during plating and must be replenished based on measurement of constituent chemistry concentrations.
Inorganics such as Cu ions, sulfuric acid, and chloride ions are easily measured with conventional chemical analysis systems such as UV/VIS spectrophotometer, titrator and turbidometer, etc. [3]. For organics, various analytical methods have been developed. One of the most widely used is cyclic voltammetric stripping (CVS) [4]. With CVS, Cu film is alternately deposited on a working electrode and stripped off by anodic dissolution. A potentiostat integrates current over time to quantify the electric charge transferred during plating-dissolution cycles. From the observation of the stripping charge, concentration of individual additive can be determined.
Methods such as the modified linear approximation technique (MLAT) and dilution titration (DT) can measure with reasonable accuracy and precision. These have the advantage of simulating actual plating and demonstrate the capability of detecting all functional materials in terms of the electrochemical activities in plating equipment.
CVS has the drawbacks of low speed and low accuracy, especially with a plating bath using multiple chemicals. It is also highly sensitive to temperature, impurities, and breakdown byproducts. Depending on the number of chemicals and the analysis procedure, it can easily take a few hours to quantify additive concentrations.
HPLC has a simple system configuration that is capable of high speed of operation and raw materials detection. It is the basis of our new analytical method for quantification of organic additives for Cu plating. By utilizing a proper separation column and detector, and a unique change scheme of eluant concentrations, we have developed a metrology system that can simultaneously detect both brightener and suppressor in <10 min without sacrificing accuracy or precision.
Metrology system requirements
A reliable metrology system should have high accuracy and precision. Accuracy is easily determined by comparing the measured value with concentration of a standard specimen
 |
null
A system with accuracy error <5% is a reliable one for wet manufacturing processes, such as electroplating.
Precision can be defined in various ways. One of the most widely used terms in the semiconductor industry is P/T (i.e., the ratio of precision to process tolerance). It is defined as
 |
null
Precision (P) is 6σ metrology and is taken from repeated measurements of a reference sample. Process tolerance (T) is the process window or the difference between the upper control limit (UCL) and the lower control limit (LCL). A precision metrology tool should have a small P/T, which indicates that measurement error is a small fraction of the operating range of the product parameter represented by the reference specimen. Overall error (σObserved) includes errors from both process tool and measurement systems or
 |
null
Therefore, if σMetrology = (P/T)σProcess, then
 |
null
The goal of our development work was to establish a metrology system with P/T <0.3 and an observed process error <1.044σProcess (i.e., the error caused by the metrology toolaffected overall measurement error <4.4%).
 Figure 2. Typical HPLC chromatograms from three different concentrations: target, upper, and lower limits. |
null
Configuring an HPLC system
HPLC is a well-known system in analytical chemistry. Since it detects a specific chemical entity after separation from a matrix, it does not involve a "convolution" (side) effect [5]. Depending on the materials property, various detectors are used, including optical, electrochemical, and refractive index (RI) detectors. In our work, we used reversed-phase column and UV detectors (Fig. 1) and three eluants: acetonitrile, sulfuric acid, and de-ionized water. The wavelengths of our UV detector were determined by using a separate UV/VIS spectrophotometer.
We configured the HPLC to establish a measurement system of high precision, manufacturable accuracy, short run time, and simultaneous detection of both organic additives. In this method, composition of eluants was changed in a step-wise manner so that measurement time was reduced to <10 min (i.e., brightener and suppressor were detected at 3.6 min and 6.5 min, respectively).
In a typical chromatogram, the first peaks are brighteners, the second suppressors (Fig. 2). In the graph, the top line is from a solution containing both chemicals at UCL concentrations, the center line is target concentrations, and the bottom line is LCL concentrations of the process window.
System accuracy and precision
 |
We determined our target and upper and lower process control limits by optimizing our plating process and Cu thin film quality. Then, we prepared calibration standards for both brightener and suppressor at these concentrations using de-ionized water. Thus, we prevented hydrolysis of additives by acid, which can reduce the lifetime of the mixture. We used the average values from five runs with each of the standards to generate our calibration curves.
We prepared three sets of test specimens in copper sulfate electrolyte containing the same amounts of inorganic chemicals as an actual bath, to simulate a plating bath in production. The first had brightener and suppressor at their lower control limit concentrations (i.e., 1.5 mL/L and 2.0 mL/L); the second had both concentrations at the target concentrations (i.e., 2.0 mL/L and 4.0 mL/L); and the third had both concentrations at upper control limit (i.e., 3.0 mL/L and 8.0 mL/L). Our result of 10 repetitive measurements showed a P/T for all concentration of 0.3 (see table). The P/T values from 10 repetitions were between 0.01 and 0.16, well below 0.3. In addition, system linearity was excellent at R2 ≈1.0 for both brightener and suppressor measurements (Fig. 3).
 |
null
Acknowledgments
We thank Mike Bowen at Intel's Portland technology development fab for his help on metrology capability requirements, and Prof. B.H. Han at the Chung-Nam National U. for his helpful comments.
References
- F.A. Lowenheim, Modern Electroplating, John Wiley & Sons Inc., New York, 1975.
- V.M. Dubin, et al., "Electroplating Bath Composition and Method Using," US patent pending.
- G.H. Jeffrey, et. al., Vogel's Textbook of Quantitative Chemical Analysis, 5th ed., Longman Scientific and Technical, New York, 1991.
- Qualilab operation manual, ECI technology, East Rutherford, NJ, 1999.
- G.D. Christian, Analytical Chemistry, John Wiley & Sons, New York, 1980.
For more information, contact: Hok-Kin Choi, Intel Corp., 2880 Northwestern Pkwy, M/S: SC3-06, Santa Clara, CA 95052; ph 408/765-6093, fax 408/765-4881, [email protected].
Kimin Hong, Chung-Nam National University, Daejeon, Korean
Hok-Kin Choi, Intel Corp., Santa Clara, California