Ind. Eng. Chem. Res. 2000, 39, 3249-3254
3249
MATERIALS AND INTERFACES Stabilization of Alumina Polishing Slurries Using Phosphonate Dispersants Qiuliang Luo Rodel Inc., 451 Bellevue Road, Newark, Delaware 19713
Stability of alumina slurry in high ionic concentrations was evaluated using the settling rate of alumina abrasives in various slurries. The ζ-potential was used to interpret the effect of electrostatic interaction. The presence of high ionic concentrations reduces the ζ-potential significantly by compressing the electric double layer. The addition of potassium phthalate does not change the ζ-potential at low concentrations (2.5 wt %) due to the adsorption of phthalate on the alumina surface. Citric acid decreases the ζ-potential and changes the sign at lower concentrations than that of potassium phthalate. Various phosphonate dispersants were evaluated in terms of the settling rate of alumina abrasives in the slurry. Their stabilization behavior is not well correlated to the surface charge on the alumina particle surface and may be related to the strength of adsorption and followed by the steric hindrance from the adsorbed dispersants. Introduction Chemical-mechanical polishing (CMP) has become an important process technology in semiconductor manufacturing to planarize the surface topography on the wafer and facilitate further processing.1-3 CMP can be used to planarize both insulating and conducting materials used in device wiring. Insulating materials include various silicon dioxides and low dielectric constant materials (low K materials), typically polymeric materials. The conducting materials include tungsten, aluminum, and copper. The CMP of insulators uses silica or ceria as the abrasives while the CMP of conductors uses alumina under most cases. The stability of the slurry is an important issue in slurry development for CMP because the rapidly settled abrasives may clog the slurry distribution lines and cause unpredictable problems during semiconductor manufacturing. In addition, the coagulation of the abrasive particles is extremely detrimental to wafer polishing because it increases the defects on the polished wafers and reduces the yield. Polishing chemistry is required to achieve a reasonable removal rate for all the polishing slurries (slurries for polishing insulating materials and conducting materials) and the chemistry is more complicated for a metal polishing slurry than that for a dielectric polishing slurry. The alumina slurry used to polish metals needs oxidizers, etching (complexing) agents, and passivation agents.3-5 The oxidizers are used to convert the metal to metallic oxides and the etching (complexing) agents are used to enhance the metal polish rate. The passivation agents reduce the chemical etch rate during polishing to reduce metal dishing during polishing. The addition of these chemicals decreases the slurry stability significantly by compressing the electric double layer and thus lowering the ζ-potential.6,7 Therefore, stabili-
zation of these polishing slurries and keeping these alumina abrasives well dispersed is required to improve the slurry polishing performance. Alumina slurries can be stabilized by adding polymeric surfactants.8 Those surfactants could be nonionic, cationic, or anionic, depending on the media. For example, DAPRAL (a maleic anhydride R-olefin copolymer), an anionic polymeric surfactant, can stabilize alumina suspensions in aqueous and nonaqueous solutions at low pHs.9,10 It is believed that the electrostatic attraction between positively charged alumina particles and negatively charged surfactant molecules is responsible for the adsorption of surfactant on the alumina particle surface, resulting in the negatively charged alumina particle surfaces. Because the alumina particles are negatively charged due to the adsorption of polymeric surfactants, both steric (due to the sterically hindered polymeric surfactant) and electrostatic repulsion between alumina particles play important roles in the stabilization of alumina suspensions. Another agent used for the stabilization of alumina suspensions is polyacylic acid. The stabilizing behavior of poly(acrylic acid) is dependent on its concentration and solution pH.11 Relatively high concentrations, e.g., 100 ppm or higher, and higher pH such as pH > 10 lead to stabilization of alumina suspensions while low concentrations, e.g., 5 ppm or lower, and lower pH such as pH < 4 lead to flocculation. This is due to the conformation effect of poly(acrylic acid) at different pHs and concentrations. At low polymer concentrations, because the alumina surface coverage by poly(acrylic acid) is low, the increase in pH causes the coiled poly(acrylic acid) on the surface to stretch out. This stretching out favors a better bridging of the alumina particles by the polymer and thus enhances the alumina flocculation (destabilization). However, at higher poly(acrylic acid) concentra-
10.1021/ie0000717 CCC: $19.00 © 2000 American Chemical Society Published on Web 08/06/2000
3250
Ind. Eng. Chem. Res., Vol. 39, No. 9, 2000
tions, the surface coverage of alumina particles is high and the adsorbed polyacylic acid exhibits less stretching (coiled structure), which reduces the effect of bridging alumina particles in the suspension leading to stabilization. Citric acid can be used as an effective dispersant for the high alumina-loading supspensions, even though these suspensions contain much lower ionic concentrations than a metal CMP slurry.12 High molecular poly(ethylene glycol) can be used to stabilize an alumina slurry at high ionic concentrations for copper CMP.5 In addition, the addition of surfactant or dispersant to the polishing slurry reduces the surface defects significantly by lowering the adhesion of abrasive particles on the wafer surface.13 Obviously, the stabilization of the alumina CMP slurry is dependent on the polishing chemistry used. In this paper, the alumina slurry was used to polish tungsten in manufacturing semiconductor interconnects. Because tungsten can be easily passivated in acidic media by forming tungsten oxides, no extra passivation agent is required4,5 for these slurries studied. The polishing chemistry includes 2% potassium iodate as the oxidizer and 2% citric acid as the complexing agent and 2% potassium phthalate as the oxide suppressant to reduce the silicon dioxide polish rate. All kinds of phosphonate dispersants were evaluated to stabilize these polishing slurries in terms of slurry stability and particle size distribution. The ζ-potential was used to determine the electrostatic effect, facilitating the understanding of slurry stability under various conditions. Experimental Section 1. Preparation of Test Slurries. A 250-mL Nylgene plastic bottle was used as the container for the test slurry. At the beginning, the amount of predetermined chemicals was added to the container and dissolved with a minimal amount of deionized (DI) water. Then, the premilled 30% alumina slurry (mixture of R-alumina and γ-alumina) was added to the container and diluted with DI water. This concentrated alumina was a mixture of 75% R-alumina and of 25% γ-alumina. This mixture with the necessary chemistry was sonicated to be sure to break all the possible agglomerates. The concentration of abrasives was 5 wt % for all the test slurries and all the test slurries were prepared at room temperature; the suspension viscosity was about 1.12 cp and varied slightly with the chemistry, including the addition of phosphonate dispersants. The slurry pH was adjusted using diluted KOH solution or diluted nitric acid. Phosphonate dispersants were supplied by Albright-Wilson Americas Corporation. The rest of the chemicals used in this paper were purchased from Aldrich Chemicals and were used without further purification. 2. Evaluation of Slurry Stability. The slurry stability was evaluated in terms of the settling rate of the interface between the supernatant and the abrasive suspension. The settling rate was determined using a QuickScan supplied by Coulter Corporation. This instrument measures the light transmittance and backscattering through the test tube with test suspensions. It scans the entire test tube from the bottom of the test tube to the top of the test tube to find the position of the interface between the clear supernatant and the abrasive suspension. The dramatic transmittance difference around the suspension-supernatant interface
Figure 1. Effect of KIO3 concentration on the slurry stability and ζ-potential. The slurry contains 5 wt % alumina abrasives at pH 4.
indicates the position of the interface. It scans automatically at a given time interval as programmed until the predetermined settling time is reached, for example, letting the slurry settle for 20 h without any perturbation. The predetermined settling time is dependent on the slurry stability; the higher the slurry stability, the longer the predetermined settling time. Backscattering intensity reveals the solids concentration variation at different positions in the test tube. In this paper, the transmittance was used to determine the position of the suspension-supernatant at various times because it has been found that the clear interface has always been observed for all alumina test slurries used in this paper. Then, the position of the interface was plotted with respect to the settling time. The initial slope of the plot is referred to as the settling rate of the interface and thus the stability of the slurry. Because the settling rate is time-averaged and the experiment was time-consuming, the experiment was not repeated. 3. Measurement of ζ-Potential and Particle Size Distribution. The ζ-potential of alumina particles in various suspensions was measured using a Coulter Delsa 440 also provided by Coulter Corp. The slurries were centrifuged to separate the abrasives from the suspension. The supernatant was used to dilute the slurry for the measurement of the ζ-potential to keep the same chemistry while the particle concentration was being reduced. Each ζ-potential was measured three times and the average of the three values was reported in this paper. The maximum error observed in the experiments was Briquest 301-50A > Briquest 462D-AO = Briquest 462-23K. ζ-Potential data at various Briquest ADAP-60AW concentrations reveal that the ζ-potential remains at about -10 mV, demonstrating again the minimal effect of electrostatic repulsion on the stability of the alumnia slurry. It seems that the effect of stabilization of the alumina slurry by adding a dispersant is dependent on the adsorption strength of dispersant on the alumina surface with respect to citric acid. This is because citric acid can also be adsorbed on the alumina surface and the dispersant has to compete with citric acid for the active sites on the alumina surface. When the dispersant is adsorbed on the alumina particle surface, it seems that the steric hindrance may be responsible for the stabilization of the alumina slurry because the measured ζ-potentials are all very small (less than -10 mV). On the basis of the molecular structure of those dispersants, Briquest 543-45AS is more bulky than Briquest 221-50A, and the slurry stability follows the same trend, so is Briquest 221-50A relative to Briquest ADAP 60AW. Briquest 301-50A, Briquest 462D-AO, and Briquest 462-23K may have less adsorption strength on the alumina surface compared with that of citric acid, which may be responsible for their stabilization behavior for this polishing slurry. Further investigation is needed to explain the behavior of these phosphonate dispersants. Conclusions Stability of the alumina slurry at high ionic concentrations was evaluated using the settling rate of alumina abrasives in various slurries. The ζ-potential was used to interpret the effect of electrostatic repulsion. The presence of high ionic concentrations reduces the ζ-potential significantly by compressing the electric double
3254
Ind. Eng. Chem. Res., Vol. 39, No. 9, 2000
layer. The addition of potassium phthalate does not change the ζ-potential at low concentrations. However, the ζ-potential changes from positive to negative at high concentrations due to the adsorption of phthalate on the alumina particle surface. Citric acid decreases the ζ-potential and changes the sign at lower concentrations than that of potassium phthalate, indicating a stronger adsorption of citric acid on the alumina particle surface than that of potassium phthalate. Various phosphonate dispersants were evaluated in terms of the settling rate. The stabilization behavior of those dispersants is not well correlated to the ζ-potential because the ζ-potentials are all small (less than -10 mV). It may be related to the adsorption strength on the alumina surface relative to citric acid on the alumina particle surface. And then the slurry stability follows the order of steric hindrance from the dispersant when it is adsorbed on the alumina particle surface. Literature Cited (1) Fury, M. A. Solid State Technol. 1995, April, 47. (2) Sivaram, S.; Bath, H.; Legget, R.; Maury, A.; Monnig, K.; Tolles, R. Solid State Technol. 1992, May, 87. (3) Landis, H.; Burke, P.; Cote, W.; Hill, W.; Hoffman, C.; Kaanta, C.; Koburger, C.; Lange, W.; Leach, M.; Luce, S. Thin Solid Films 1992, 220, 1. (4) Stein, D. J.; Hethering, D. L.; Guilinger, T.; Cecchi, J. L. J. Electrocehm. Soc. 1998, 145 (9), 3190.
(5) Luo, Q.; Campbell, D. R.; Babu, S. V. Langmuir 1996, 12, 3563-3566. (6) Stein, D. J.; Hethering, D. L.; Cecchi, J. L. J. Electrocehm. Soc. 1999, 146 (5), 1934. (7) Lyklema, J. In Colloid Dispersion; Goodwin, J. W., Ed.; The Royal Society of Chemistry: London, 1982. (8) Napper, D. H. Polymeric Stabilization of Colloid Dispersion; Academic Press: London, 1983. (9) Yu, X.; Somasundaran, P. Colloids Surf. A 1994, 89, 227. (10) Li, C.; Yu, X.; Somasundaran, P. Colloids Surf. A 1992, 69, 155. (11) Tjipangandjara, K. F.; Somasundaran, P. Colloid Surf. A 1991, 55, 245. (12) Hidber, P. C.; Graule, T. J.; Gauckler, L. J. J. Am. Ceram. Soc. 1996, 79 (7), 1857-1867. (13) Free, M. L. Micro 1998, May, 29-37. (14) Pourbaix, M. Atlas of Electrochemical Equilibria in Aqueous Solution; NACE: Houston, TX, 1974. (15) Hunter, R. J. Fundamentals of Colloid Science; Oxford University Press: Oxford, 1987; Vol. 1, pp 329-332. (16) Koz’mina, Z. P.; Belova, M. P.; Sannikov, V. Kolloid. Zh. 1963, 25, 169. (17) Lide, D. R. Handbook of Chemistry and Physics, 80th ed.; CRC Press: Boca Raton, FL, 1999. (18) Sposito, G. The Environmental Chemistry of Aluminum, 2nd ed.; CRC Press: Boca Raton, FL, 1996; pp 58-59. (19) Product Literature, BRIQUEST Phosphonates, Albright & Wilson Americas.
Received for review January 19, 2000 Revised manuscript received June 12, 2000 Accepted June 20, 2000 IE0000717