abrasive technology TECHVIEW Impact of Diamond CMP Conditioning Disk Characteristics on Removal Rates of Polyurethane Polishing Pads Mark Bubnick, Sohail Qamar, Thomas Namola, and Dave McClew Abrasive Technology, 8400 Green Meadows Dr., Lewis Center, OH 43035 www.abrasive-tech.com
Diamond CMP Conditioning Disk Characteristics Impact of Diamond CMP Conditioning Disk Characteristics on Removal Rates of Polyurethane Polishing Pads Mark Bubnick, Sohail Qamar, Thomas Namola, and Dave McClew Abrasive Technology, 8400 Green Meadows Dr., Lewis Center, OH 43035 Tel: 1.740.548.4100, Fax: 1.740.548.7617, mbubnick@abrasive-tech.com ABSTRACT Common perception is that a diamond-conditioning disk with one set of characteristics will perform equally and optimally for all known CMP processes. This concept is outdated and costing the wafer fabs significant time, money, yields and quality. Just as slurries, pads, and process parameters differ for the various CMP processes (Ref 1); a process optimized diamond-conditioning disk is warranted. The technology now exists to design and manufacture diamond-conditioning disks for specific CMP processes on any equipment platform. The ability to optimize the polishing pad removal rate through diamond conditioning disks for various CMP processes is desired. Surface characteristics and removal rates of polyurethane polishing pads can be controlled and accurately manipulated through the selection of diamond crystal type. This relationship holds true for both randomly placed and structured grid diamond crystal arrangements. A relationship between diamond type and its impact on polishing pad removal rate and RAS (relative abrasive sharpness) (Ref 2) of the conditioner disk was established. Diamond concentration was held constant for this study. Polishing pad removal rates generated by the diamond-conditioning disk were measured using a special test apparatus that simulated standard CMP processes and setup. RAS values were measured using Abrasive Technology proprietary equipment. This paper presents details of diamond disk characteristics, test methodologies, test results, and field relevance. INTRODUCTION The Preston equation (Ref 3) is used to explain the material removal rates on silicon wafers. It states that film removal rate is a function of both the pressure and relative speeds of the platen on the wafer s surface: Film Removal Rate = K*P*S where S is the table speed relative to the wafer and P is the applied pressure. The Preston Coefficient, K, is a function of the system s conditions including the pad, slurry, and material being removed. The Preston equation and other research (Ref 4-8) recognize that the type of polyurethane conditioning pad, the type of slurry, and the interaction between the wafer material and the polyurethane polishing pad depend specifically on the CMP process.
A similar interaction exists between the polyurethane conditioning pad and the diamond-conditioning disk. The removal rate of a glazed conditioning pad whose pores are plugged with material removed from the wafer can be expressed by a similar equation: Polishing Pad Removal Rate = K *P *S where S is the diamond disk speed relative to the polishing pad and P is the applied pressure. The Preston Coefficient for the conditioning disk, K, is a function of the loaded pad properties, the type of slurry, and the characteristics of the diamond disk: K = f(loaded pad, slurry, diamond disk) In current CMP processes, the choice of pad and slurry are process specific, i.e., a tungsten slurry isn t used for copper CMP process and visa-versa. This theory also holds true for the diamond-conditioning disks. In the past, the same diamond disk was forced on all of the different CMP processes. It is now known that the disk characteristics, such as diamond type, diamond configuration, and diamond concentration, impact pad life and pad removal rate. Diamond disk characteristics have been known to indirectly influence film polishing rate, film non-uniformity, etc. In past publications, Abrasive Technology has established that disk characteristics can be optimized and consistently controlled during manufacturing. Therefore, it has become possible to optimize diamond conditioning disks toward specific CMP processes. In this study, two different diamond disk characteristics were created by: 1. Using three types of diamond (Type I, II, and III). 2. Controlling the diamond surface configuration (random vs. structured). EXPERIMENTAL Experimental diamond disks were manufactured to compare the effects of abrasive type and abrasive configuration. The three abrasive types were: 1.) Type I - Nominal abrasive 2.) Type II - Less aggressive abrasive 3.) Type III - More aggressive abrasive Pad Removal Rate: The pad removal rates of four different disks were compared: Type I and Type III diamonds were used on both the random and structured configurations. Tests were run using a pad removal rate testing machine customized to Abrasive Technology s specifications with Rodel IC-1000 pads and DI water. RAS Testing: RAS testing was performed on disks containing the Type I and Type III abrasive in both the random and structured configurations; and on a random configuration using Type II abrasive. The mean RAS values presented are from production parts. RESULTS Type I and Type III Pad Removal Rate Tests: The test parameters for the polishing pad removal rate tests are given in Table I. The results of the polishing pad removal rate tests are shown in Table 2 (Fig. 1). Table 1: Test parameters for polishing pad removal rate tests Loading force Conditioner disk rotation Pad rotation Disk translation speed Disk translation stroke Slurry 27 N 90 rpm 100 rpm 5 mm/s 15 mm DI water @ 60 ml/min
Table 2: Polishing pad removal rate test results for Type I and Type III diamond Diamond Type Abrasive Configuration Polishing Pad removal rate (mm/hour) Type I Random 0.11 Type I Structured 0.12 Type III Random 0.32 Type III Structured 0.28 By controlling diamond type, the pad removal rate can be varied by a factor of 2.6. Additionally, changing the diamond configuration from random to structured, while keeping diamond type constant, does not affect the polishing pad removal rate. The key factor in controlling the polishing pad removal rate is the diamond type. RAS Tests Values: RAS test results for three diamond types are given in Table 3. Table 3: Mean RAS values Type I, Type II and Type III diamond Abrasive Type Abrasive Configuration RAS value Type I Random 5.6 Type I Structured 5.7 Type II Random 5.4 Type III Random 6.2 Type III Structured 6.1 The mean RAS values and pad removal rate data have a strong correlation; higher RAS values producing higher removal rates. The RAS data and the pad removal rate data show that the diamond type and not the diamond configuration is the controlling factor in pad removal rate. CONCLUSIONS: 1. The ability to accurately manipulate polishing pad cut-rate has been proven to depend on the diamond type. 2. The removal rate of the polishing pad can be varied by a factor of 2.6 by choice of diamond type. 3. The diamond configuration (random vs. structured) was shown to not have an impact on the polishing pad removal rate. 4. Increasing RAS values were shown to correlate to higher pad removal rates. RECOMMENDATIONS: The characteristics of diamond disks used to condition polyurethane polishing pads should be optimized and controlled to suit the many different CMP processes.
Figure 1: Polishing pad removal rate for Type I and Type III diamond in random and structured configurations Figure 2: SEM picture of a random disk Figure 2A: SEM picture of a structured disk References: 1. J. M. Steigerwald, S. P. Murarka, R. J. Gutmann, Chemical Mechanical Planarization of Microelectronic Materials, P 40-46, 65-84, John Wiley & Sons, Inc. New York, 1997 2. G. Prabhu, S. Kumaraswamy, D. Flynn, S. Qamar, T. Namola, CMP-MIC Proceedings, P.293 (2000) 3. F.W. Preston. The Theory and Design of Plate Glass Polishing Machines. J Soc Glass Tech 11:214-254, 1927 4. Ross E Barker, Glenn C. Mandingo, Craig D. Lack, CMP-MIC Proceedings, P 144 5. Amy L. Moy, Joseph L. Cecchi, Dale L. Hetherington, and David J. Stein, CMP-MIC Proceedings, P 189 6. W.C. Chiou, T. Shih, S.M. Jang, C.H. Yu, M.S. Liang, CMP-MIC Proceedings, P 358 7. Michael R. Oliver, Robert E. Schmidt, and Maria Robinson, ECS Fall Conference, October 2000 8. Dr. David B. James, CAMP 5th International CMP Symposium, August 2000
World headquarters - Lewis Center, Ohio USA abrasive technology WORLD HEADQUARTERS 8400 Green Meadows Dr. P.O. Box 545 Lewis Center, Ohio 43035 USA Ph: 1.740.548.4100 Fax: 1.740.548.7617 EUROPEAN HEADQUARTERS Roxby Place, Fulham London SW6 1RT United Kingdom Ph: 44.20.7471.0200 Fax: 44.20.7471.0202 ASIAN HEADQUARTERS Blk 1093 Lower Delta Rd. 05-11/13 Tiong Bahru Industrial Estate Singapore 169204 Ph: 65.6270.6878 Fax: 65.6270.6877 www.abrasive-tech.com TAIWAN OFFICE 27 Alley 46, Lane 43 San Ming Rd, Yang Mei Taoyuan, Taiwan Ph: 886.3.482.8151 Fax: 886.3.482.8150 2002 Abrasive Technology 1/06