Development of the Gel Plating Process for Selective Gold Metallization by Lynne Miriam Svedberg B.S. Materials Science and Engineering University of Michigan, 1994 SUBMITTED TO THE DEPARTMENT OF MATERIALS SCIENCE AND ENGINEERING IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN MATERIALS SCIENCE AND ENGINEERING AT THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY February 1999 © Copyright 1999 Massachusetts Institute of Technology All rights reserved Signature of Author. «0002... 000 eeee eee eee ceee ee . Departindht of Materials Science and Engitéering January 8, 1999 Cenified by......7 ceseees . Michael J. Cima Sumitomo Electric Industries Professor of Ceramics ! ‘Thesis Supervisor Accepted BY... eeesecccccseecretevsseetteenserersescresene ee Linn Hobbs John F. Elliot Professor of Materials Chair, Departmental Committee on Graduate StudentsDevelopment of the Gel Plating Process for Selective Gold Metallization by Lynne Miriam Svedberg Submitted to the Department of Materials Science and Engineering on January 8, 1999 in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Materials Science and Engineering. Abstract ‘A Gel Plating Process for selective gold metallization was developed in this thesis. The Gel Plating Process selectively plates metal on catalytic features without exposing sensitive areag of the substrate surface to a corrosive plating bath. This process utilizes an electroless plating bath and polymeric thickening agent to formulate a gel. The gel is selectively printed onto the areas of the substrate that require plating. Plating occurs autocatalytically at an elevated temperatu.e by the simultaneous anodic oxidation of the reducing agent and catalytic reduction of the metal complex onto the catalytic features under the gel print. The substrate is cleaned, leaving a selectively metallized substrate surface. Aluminum nitride (AIN) is a potential packaging material for high power electronic applications; however, it is severely corroded in the alkaline gold electroless plating solutions often used to selectively plate packaging substrates with electrically isolated lines. An ammonium ion selective electrode was used to study the in situ corrosion of AIN as a function of pH. It was shown that the aluminum trihydroxide corrosion product created on the surface of AIN during exposure to aqueous solutions acts as a barrier layer that decreases the corrosion of AIN in the pH range where this aluminum trihydroxide product is insoluble. An aluminum oxynitride layer of 200 angstroms was found to act as an insoluable barrier layer to corrosion of AIN at a pH of 95. A plating gel was formulated for use in the Gel Plating Process by adding a polymeric thickening agent to an electroless gold plating bath chemistry. The plating chemistry incorporated sodium gold thiosulfate as the gold complex and sodium ascorbate as the reducing agent . The gold concentration in the plating gel was increased to 40 g/L gold from the 4 g/L gold found in typical commercial electroless plating baths. This resulted in a high ion concentration plating solution that was thickened with a hydroxyethyl cellulose (HEC) thickening agent. A stabilizer, 2-mercaptobenzimidazole, was found to increase the room temperature and plating temperature stability of theplating gel. Addition of excess sodium ascorbate caused a significant improvement in plate color. Addition of a surfactant to the plating gel formulation enhanced the printability of the plating gel and reduced extraneous piating of gold onto the non- catalytic areas of the substrate. The rest viscosity of the plating gel was found to be 70,000 cP at 25 °C and 38,000 cP at 60 °C, the highest plating temperature used. The plating mechanism of the Gel Plating Process was found to be reaction rate controlled at short times and diffusion controlled at longer times. A reaction rate model and a diffusion model with a heterogeneous reaction were fit to plating rate data at 50°C and 60°C. The reaction rate model was used to determine the rate constant, k, for the cathodic reaction at the plating surface at short times. A diffusion model with a heterogeneous chemical reaction was used te identify the diffusion coefficient, D, of the gold complex in the plating gel at long times. At 60 °C, it was found that k= 7.5 x 10% ‘om/s and D=4x 107 cm*/s. At 50 °C, it was found that k = 3 x 10° cm/s and D = 2 x 107 cm’/s. The thickening agent was found to decrease both the reaction rate constant for the cathodic reaction at the plating surface and the diffusivity of the gold complex. The gold plate thickness and microstructure was found to be influenced by the gold complex and reducing agent chemistry, the stabilizer concentration, and thickening agent additions. Plating through a gel resulted in a fine grained microstructure. Plating through a solution resulted in a large grained microstructure: SEM cross sections indicated that a dense gold plate resulted from gel plating. The gold film obtained by the Gel Plating Process is free of trapped organics. Thesis Supervisor: Michael J. Cima Title: Sumitomo Electric Industries Professor of CeramicsTable of Contents ACKNOWLEDGEMENTS... ‘TABLE OF CHEMICAL FORMULAS. CHAPTER 1: INTRODUCTION TO SELECTIVE METALLIZATION.....- 1.1 INTRODUCTION. 1.2 SELECTIVE METALLIZATION. 1.3 OVERVIEW OF THE GEL PLATING PROCES! 1.4 ADVANTAGES OF THE GEL PLATING PROCESS... 1.5 ORGANIZATION OF THES... CHAPTER 2: ALUMINUM NITRIDE (ALN) CORROSION KINETICS... 2.1 INTRODUCTION. 2.2 EXPERIMENTAL PROCEDURE. 2.2.1 Preliminary Powder Study.. 27 2.2.3 ‘In situ’ ion selective electrode measurements. 27 2.3 RELAT’ INSHIP BETWEEN SOLUTION PH AND CORROSION KINETICS OF ALN 30 2.4 INSOLUBLE BARRIER LAYER PROTECTION MECHANISM 38 2.5 CONCLUSIONS 43 CHAPTER 3 : PLATING GEL FORMULATION... 3.1 REVIEW OF ELECTROLESS PLATING BATH CHEMISTRY 3.2 INTRODUCTION TO PLATING GEL CHEMISTRY. 3.3. SELECTION OF A THICKENING AGENT FOR THE MODIFIED COMMERCIAL PLATING BATH. 3.3.1 Experimental Procedure. : 3.3.2 Results 3.4. NON-CYANIDE PLATING GEL FORMULATION 3.4.1 Experimental Procedure.. 3.4.2 Effect of Bath Formulation on Thickening Behavior. 3.4.2.1 Low Gold Concentration (4 to 15 g/L Au) 3.42.2 High Gold Concentration Formulations (15 to 40 g/L) I 3.4.3 Effect of lon Concentration on Performance of Plating Gel at Plating Temperature.. 3.44 Stability of Plating Gel as a Function of Chemical Additives. 3.4.5 Gold Plate Appearance as a Function of Gel Chemistry 3.45.1 Discussion 3.5 CONCLUSIONS CHAPTER 4: PLATING GEL PERFORMANCE DURING PROCESSING snes 4.1. SCREEN PRINTING OF PLATING G2L... 41.1 Screen Printing Parameters. 4.1.2 Effect of Surface Tension on the Screen Printing of the Plating Gel 4.1.3 Effect of Humectant Additions on Screen Printing of Plating Gel. 4.2. SELECTIVE BEHAVIOR OF GEL PLATING. 42.1 — Experimental Procedure. 4.2.2 Results and Observations. 4.3 BEHAVIOR OF PLATING GEL AT PLATING TEMPERATURE. 4.3.1 Plating in the Blue M Humidity Oven, 4.4 RHEOLOGY MEASUREMENTS 44.1 Experimental Procedure..462 Result 4.4.3 Discussion. 4.5 CONCLUSIONS .. CHAPTER 5: PLATING RATE KINETICS enn 5.1 INTRODUCTION. —— ‘5.2 LITERATURE REVIEW OF PLATING RATE KINETICS FOR PLATING BATH! 5.3 PLATING RATE STUDIES. 5.3.1. Experimental Procedure. 5.3.2 Observations. ‘$32.1 Plating Rate as a Function of Stabilizer Concentration 5.3.22 Plating Rate as a Function of Surfactant Concentration... 5.32.3 Plating Rate as a Function of Temperature. 5.3.3 Conclusions. — ‘5.4 MECHANISMS OF THE GEL PLATING PROCESS 5.4.1 Diffusion Limited Model. a 5.4.2 Reaction Rate Model.. 5.4.3 Diffusion & Reaction Model. 5.4.4 Conclusions 5.5 INFLUENCE OF THICKENING AGENT ON PLATING KINETICS. 5.6 PLATING GEL STABILITY AT PLATING TEMPERATURE... 5.6.1 Experimental Procedure. 5.6.2 Results. 5.6.3 Conclusions. 5:7 OPTIMIZATION OF PROCESS PARAMETER! 5.8 CONCLUSIONS . CHAPTER 6 : PROPERTIES OF THE GOLD PLATE... 6.1 INTRODUCTION... 6.2 MICROSTRUCTURE .. 6.3 THICKNESS... 64 Purtry 6.5 CONCLUSIONS CHAPTER 7: SUMMARY AND FUTURE WORK..... 7.1 CONCLUSIONS .. 7.2 FUTURE WORK. References. u3List of Figures Figure 1: Commercial Electroless Gold Plating Figure 2: Overview of the Gel Plating Process. Figure 3: Thickness of Gold Plate as a Function of Gel Print Thickness and Gold Concentrator Figure 4: Corrosion of AIN in Aqueous Solutions wrnnm Figure 5: Ammonium Flux and Solubility of Al(OH)s as a Function of pH. Figure 6: Comparison of AIN Powder and AIN Substrate Corrosion. Figure 7: Schematic of AIN Sintering. Figure 8: Corrosion of Oxidized AIN Substrates in # weous Solutions. Figure 9: RBS Spectrum of AION Layer on AIN Suostrates.. Figure 10: Processing-Performance Relationship of Thickening Agents in Commercia! Gold Bath, Based ‘on Viscosity, Dispersion, and Solubility Rankings: 0 = poor; 5 = excellent. Solubility Ranking is Inside the Oval. Figure 11: Visual Differences in the Resulting Gel Print of Two Different Polymer Systems: (a) Hydroxypropyl Methylcellulose (HPMC) based system (b) Polysaccharide based system... Figure 12: Schematic of Syneresis Figure 13: Effect of Propylen&Glycol on the Gelation Temperature Range. Figure 14: Ability of Thickening Agent to Increase Viscosity of Solution as a Function of Gold Concentration. Figure 15: Photos Showing Gel Print Before and After Plating in the Biue M Humidity Oven (Gel number from TableS'12 and 15) 69 Figure 16: Gel Behavior in Humidity Oven (Gel Formulation Number From Tables 12 and 15). Figure 17: Comparison of Plate Color as a Function of Ascorbic Acid Concentratio igure 18: Ascorbic Acid Consumption as a Function of Time . Figure 19: Summary of Plating G&l Formulation... Figure 20: Printing Performance as a Function of Polymer & Surfactant Corcentration. Figure 21: Role of Surfactant in Screen Printing. Figure 22: EPOC substrate Figure 23: Gold Decomposition on Alby Figure 24 Ress om Overplatng Experiment. Figure 25: No Gold on AIN POC... Figure 26: Heating Rate in Blue Figure 27: Humidity Overshoot Behavior in Blue Figure 28: Attempts with PID Control to Minimize Overshoot .. Figure 29: Humidity Profile from Controlled Oven . Figure 30: Identification of Correct Humidity Set-poin Figure 31: Photographs of Syneresis Behavior in Oven Figure 32: Amplitude Sweep for 3 Gel Formulations at (a) 25°C and (b) 60°C. Figure 33: Complex Viscosity as a Function of % Strain at 25 and 60°C Figure 34: Time Sweep for 3 Gel Formulations at 25 °C and 60 °C.. Figure 35: Temperature Sweep for 3 Gel Formulations Figure 36: Frequency Sweep for 3 Gel Formulations at (a) 25 °C ans (b) 60 °C. Figure 37: Initial Plating Rate of the Gold Thiosulfate- Ascorbic Acid Plating System as a Function of mercaptobenzimidazole Concen‘ration Figure 38: Schematics of Accelerating and Decelerating DMTD Species and 2mereapiobeninidazle 135 Figure 39: Plating Rate as a Function of Surfactant Concentration, . Figure 40: Surface Tension as a Function of Surfactant Concentration Figure 41: Plate Thickness vs. Time at 7.5 x 10 M 2-mercaptobenzimidazole for T= 50 °C and 60°C... 142 igure 42: Schematic For Diffusion Model and Boundary Conditions. Figure 43: Effect of Diffusivity on the Plate Thickness as a Function of Time... Figure 44: Plating Rate Data Compared to Reaction Rate Model... Figure 45: Diffusion & Reaction Model Compared to Plating Rate Data at 60 °C...Figure 46: Diffusion and Reaction Model Compared to Plating Rate Data at 50 °C. sense 160) Figure 47: Reacuon Rate & Diffusion Model Comparison to Plating Rate Data from a Plating Solution without Thickening Agent. Figure 48: HRSEM Micrographs of Nu ty Figure 49: Adsorbance Spectrum for Plating Gel Aged at 50 °C for Varying Figure 50: Adsorbance Spectrum for Plating Gel Aged at 60 °C for Varying Times.. Figure 51: Determination of k for Decomposition of Plating Gel Figure 52: Plate Thickness as a Function of Time for Various Gel Thickness at 60 °C. Figure 53: Plate Thickness as a Function of Time for Various Gel Thickness at 50 °C. Figure 54: Comparison of Microstructure from Different Plating Gel Chemistries Figure 55: Morphology at Different Plating Times When Plating Through a Gel (25 KX) Figure 56: Morphology at Different Plating Times When Plating Through @ Gel (50 KX) svn Figure 57: Comparing Initial Nucleation With and Without Thickening Agent Figure 58: Stages of Plating Through a Solution Without a Thickening Agent. Figure 59: Comparison of SEM Cross-sections. 162 164 167 168 169 174 175 178 181 182 183 184 187List of Tables Table 1: Electrolytic versus Electroless Plating. Table 2: Techniques to Evaluate Corrosion of AIN... Table 3: Aqueous Solution Chemistry, pH, and Time of Exposure, Table 4: Buffer Solutions for Adjusting pH Levels. ‘Table 5: Summary of Initial AIN Powder Corrosion Studies.. Table 6: RUMP™ Fitting Parameters for RBS Spectrum ‘Table 7: AIN Thermal Conductivity Results after Oxidation in Air at 750 °C. Table 8: Functions of Constituents in Electroless Gold Baths. Table 9: Types of Thickening Agents Investigated . Table 10: Plating Gel Formulations (Low Gold Concentration) Table 11: Plating Gel Formulations (High Gold Concentration #1)... Table 12: Plating Gel Formulations (High Gold Concentration #2). Table 13: Effect of Ion Concentrations on Thickening Behavior of HPMC.... Table 14: Effect of Ion Concentrations on Thickening Behavior of HEC....... Table 15: Effect of Ton Concentration on Gel Behavior at Plating Temperature. Table 16: Gel Formulations for Stability Study (Repeated from Table 10) Table 17: Plating Gel Formulations for Stability Study (Repeated from Tables 11 and 12).. Table 18: Plating Gel Formulations for Evaluation of Plate Color (repeated from Table 10).. Table 19: Plating Ge! Formulations for Evaluation of Plate Color (repeated from Tables 11 and 12).. Table 20: Stabitizing Agents to Complex with Au(1) Ions in Solution. Table 21: Surfacs Tension of Thickening Agents from Product Literature Table 22: Summary of Printing Trials Table 23: Formulation with AIN decompositio Table 24: Gel Formulations for Overplating Experiments .. Table 25: Plating Gel Formulations for Rheology Studies. Table 26: Parameters for Rheology Tests. Table 27: Plating Gel Formulations for Plating Rate Studies. Table 28: Plating Gel Stability as a Function of 2-mercaptobenzimidazole Concentration Table 29: Plating Gel Formulation for Surfactant Experiment Table 30: Calculated atoms/cm? surfactant. Table 31: Plating Get Formulation for Temperature Study . Table 32: Nucleation Density With and Without Thickening Agent.. Table 33: Plating Gel Formulation for UV/Vis Experiments . Table 34: Parameters Used in Diffusion & Reaction Model From Section 5.5.3. Table 35: Theoretical Plate Thickness. Table 3 Table 37: Plating Gel Formulation for Morphology Stud} Table 38: Nucleation Density from Chapter 5 Table 39: Gel Formulations for SEM Cross-sectioned Samples Table 40: Sample Variations for Blister Testing.Acknowledgements There are so many people that I would like to thank who have given me support, instruction, and have enriched my life here at MIT. I would like to start by thanking my thesis advisor, Professor Michael J. Cima for giving me the opportunity to work on a very interesting project and for advising me over the last 4.5 years. I would like to thank my committee, Professor Latanision and Professor Scott, for their suggestions and time. Professor Sandra Burkett has also helped through scientific discussions as well as moral support. Thanks! I also would like to thank Ken Amdt who started on this project with me. I missed working with you over the last 2.5 years, good times and bad®. You are a true friend, thanks! A big thank you to John Centorino. Icouidn’t have finished without you. Thanks so much for all your help in the lab. Thanks also to Barbara Layne and Lenny Rigione for their constant help and support. Erika Abbas, Larry Panchula, and Jason Grau were my stady group during those ‘lovely’ classes and exams! Thanks a million for putting up with all my questions. Erika, thanks for being patient and explaining things so well. I would also like to thank V.T. Srikar for his help in classes and out of classes, especially the chair! I want to also thank the students in CPRL for making it fun to work! Bill, thanks for your constant friendship and willingness to help. I can’t even begin to list all the ways you have helped me over the last four years, but your friendship means more than all the favors in the world. I want to thank Mani Gopal for all his help with SEM and discussions concerning my thesis. Karina Rigby, Kevin Ressler, Satbir Khanuja, Kamala Crawley, Dai Matusoka and Cesare Monti have all left CPRL but I treasure the fun times we had and the good friendships we formed. Scott Uhland can always bring a smile to my face! Good Luck as ‘senior’ graduate student! I want to thank John Santini for being my buddy in the lab...keep singing! Wendy Katstra has been a great office mate and friend. Thanks for your moral support these last few months. I would like to thank Mindy Tupper for her friendship and ‘group meeting support’. Thanks for listening and sympathizing with me. Thanks also to Mike, Ryan, Rich, and Patrick, good luck during your ‘time’ here. My church family here in Boston supported me with much love and prayer. Thanks especially to Mike, Barbara, Sarah, and Josh Olejarz! Also, the friendship and prayers of Teresita, Quintin, Heather, Valencia, Debbie, Maya, Toyosi, and AJ have enriched my life here, thanks! My parents, Anne, David, Robby, and Maria deserve the biggest thanks that I can give and then some. Thanks for helping me through everything and always saying the right thing. You guys are the best. I bet you will be as glad as I am when this is all over! ‘Most of all I want to acknowledge that God has been my solid rock throughout the good times and bad. He has given me strength, wisdom, and understanding. He deserves all the glory.Table of Chemical Formulas ‘Chemical Name. ‘Chemical Formula and/or (Abbreviation) ‘Aluminum Nitride Hydroxyethyl Cellulose Hydroxypropyl Methyl Ceilulose Dimethylamine Borane Sodium L-Ascorbate 2-mercaptobenzimidazole Octylphenoxypolyethoxy Ethanol (Surfactant) Ethylenediaminetetraacetic acid (AIN) (HEC) (HPMC) (CHs),NH:BH; (DMAB) CeH;NaOg CyHeN2S (CsH7C6H.(OCH2CH2),OH (Triton x-100) CioHiy4NzOs x H,0 (EDTA) ‘Ammonium Citrate (dibasic) H(NH,):C6HsO7 Citric Acid H3CsHsO; Potassium Gold Cyanide KAu(CN)2 Potassium Hydroxide KOH Sodium Hydroxide NaOH Sodium Di-hydrogen Phosphate NaH2PO; Sodium Sulfite Na,SO3 Sodium Thiosulfate Na2S203 Soduim Gold (1) Thiosulfate NasAu(S:03)2 Sodium Citrate NasCcHsO7 Ammonium Acetate NHC3H302 Ammonium Citrate (tribasic) (NH4)3CsHsO7 Ammonium Chioride NH.CI Ammonium Thiosulfate (NH4)2S203 10Chapter 1: INTRODUCTION TO SELECTIVE METALLIZATION 1.1 Introduction Diverse applications utilize thin film technology ranging from decorative coatings for jewelry and automotive parts to functional films in microelectronics. Gold is commonly used in thin film applications because of its unique electrical, chemical, and optical properties. Processes used to deposit thin films of gold include vacuum deposition (evaporation, sputtering, chemical vapor deposition), spin coating, and plating. It is important to offset the high material cost associated with gold by reducing the processing costs as much as possible. Vacuum and spin coating processes require the use of photolithographic techniques and masking to control the area over which the metal is deposited. ‘These processes can be labor intensive and not very economical for a high volume of parts requiring selective metallization. Plating processes are more economical for metallizing large volumes of parts. Plating processes can be divided into two distinct types: electrolytic and electroless plating. Table 1 summarizes the differences between electrolytic and electroless plating. Electrolytic plating is a standard process used to deposit a uniform metal thickness over electrically connected features. Tie pattern to be plated is connected to an external power source by electrical leads. Part specific tooling is usually required to make reliable electrical connections to each part. Excess metallization is used to ensure all features are electrically connected and that uniform potential exists across the part during electrolytic plating. These excess metal features must be removed in a separate process. Terminators are often left that produce undesirable high frequency electrical characteristics. Therefore, electrolytic plating of electrically isolated regions is labor intensive and costly. The electroless plating process deposits a uniform metal thickness over catalyzed features without the application of an extemal power source. This process nNtakes advantage of thermodynamically feasible redox reactions between the catalyzed surface and chemical constituents in the electroless plating bath. Table 1: Electrolytic versus Electroless Plating Electrolytic Gold Plating Electroless Gold Plating © Plating occurs through application of © Plating occurs through external source thermodynamically feasible chemical reactions in bath © Pattern to be plated must be electrically © Pattern to be plated must have a catalytic connected to external power source surface which promotes redox reactions with chemicals in bath Expensive and laborgntensive selective * Most effective method for large scale metallization process selective metallization ‘© Plating solutions available in wide range ~* lating solutions limited to high pH 2 __of pH Gold plating has found use in applications ranging from electric and electronic hardware such as contacts, terminals, connectors, conductors, chasis, and printed circuit boards to components in hybrid microelectronics such as RF conductors and transistors.’ Gold metallization is often the fimal step in microelectronic processes because it provides a layer of metal that has resistance to corrosion, increased conductance, and good properties for wire and die bonding. However, the currently used electrolytic or electroless plating processes can not be applied to the new designs in hybrid microelectronics. High power electronic applications such as new phased array radar require multi- chip module (MCM) technology in order to increase the efficiency of each package. ‘Three-dimensional multichip modules (MCMs) have increased circuit and /C density per ceramic package? The higher power density and multilayer technology requires increased heat dissipation from each layer. Monolithic aluminum nitride (AIN) offers a thermal path in all directions.? It has a thermal conductivity almost 10 times that of alumina and dissipates heat through natural convective cooling. This cooling method is 12advantageous for designing portable equipment and quiet running stationary equipment. * The need for a safe alternative to the toxic beryllium oxide (BeO) used in the wireless industry is another application for AIN. AIN also has a coefficient of thermal expansion close to that of silicon which reduces thermal mismatch and improves reliability. AIN introduces some special manufacturing issues not necessary with alumina processing.” AIN is hydrolytically reactive’, severely corrodes in alkaline aqueous solutions’, and oxidizes in air at high temperatures’, The tape casting, firing, and post- sintering metallization steps of the co-fire alumina manufacturing process must be modified to be compatible with AIN.” One of the post-sintering metallization steps employed in MCM technology is gold plating of electrically isolated features. The most economical and effective selective metallization technique is electroless gold plating. Commercially available electroless gold plating baths have a pH around 14. AIN is severely corroded at this pH. Optional plating methods include coating the ceramic surface with a chemical resistant to the alkaline plating solution, using a lower pH electrolytic plating bath with removable interconnects, or developing a stable, neutral pH electroless gold plating bath.’ These options are expensive and sometimes commercially unavailable. Therefore, a selective metallization technique for depositing gold metal onto specific areas of a substrate without corroding sensitive areas of the substrate is necessary for the current ‘manufacturing demands in the hybrid microelectronics industry. 1.2 Selective Metallization Techniques Substrate protection, cost saving and versatility in manufacturing drive the development of selective gold metallization processes. It is advantageous to plate only a few areas on a substrate where gold metallization is necessary, instead of plating the whole substrate. One substrate may require that gold be deposited on one area for 13corrosion resistance and on another for wire bonding. Corrosion resistance requires approximately 0.2 um of gold; whereas, die bonding may require 2-3 um. The substrate surface will be plated to 2-3 pm over the entire surface to accommodate the wire bonding areas, Therefore, the gold plated to 2-3 um on the areas only requiring 0.2 jim is wasted. Electrolytic selective gold plating techniques are normally very labor intensive and are used only wher the cost savings from the reduction in gold metal usage supersede the additional labor costs. Brush plating, jet plating, laser enhanced jet plating of gold, and selective immersion are a few electrolytic selective metallization techniques.’ Laser-enhanced jet plating of gold® is a good example of the drawbacks to electrolytic selective plating techniques. This technique utilizes the laser- enhanced plating (LEP) method developed for gold plating” but with the improvement of supplying the electrolyte to the substrate surface (which has a constan: current) using a pressurized jet. The laser beam is focused through the jet orifice onto the substrate surface and the gold is reduced upon impact with the substrate surface. The use of a continuous supply of electrolyte via the jet stream helps to limit the problems previously encountered with limited mass transport of reactants to the substrate surface. This technique can be used to selectively plate specific regions of the substrate without using a masking technique because the laser beam can be focused on specific areas of the substrate surface. This technique requires expensive and elaborate equipment. Alternate selective metallization techniques that have been reported in the literature include a technique that incorporates meltable salts into an ink that is printed onto a substrate using ink jet printing’. This technique results in a flash deposit of metal on the substrate surface. Flash deposit occurs because of the galvanic displacement of the less noble metal substrate by the more noble metal in the ink. Therefore, this technique is not an autocatalytic plating process so the resulting metal deposit is limited toa very thin coating. “4The nature of electroless plating requires that a catalytic surface be present to initiate the plating reaction. Therefore, only the areas of a substrate that are rendered catalytic will be plated when the part is immersed in the plating solution, No part specific tooling is required for this process. The cost is reduced because many substrates can be plated at the same time. There are problems associated with commercial applications of some electroless plating solutions. Electroless bath chemistries are thermodynamically unstable and require very specific and precise formulations in order to maintain stability throughout numerous plating runs. Electroless baths also require careful maintenance because very low contamination levels can destabilize the bath. Figure 1 depicts the common problems associated with electroless plating. The chemical composition of the plating bath can change through drag out of the plating solution and water evaporation. The substrates can also contaminate the baths. The costs associated with metal recovery, waste treatment, waste disposal, and maintenance costs of the large plating baths deter the use of electroless plating in many applications. Another issue that arises is the compatibility of the bath chemistry with the material to be plated. For instance, commercially used autocatalytic electroless gold plating baths have a high pH to ensure stability of the reducing agent. These formulations can be corrosive to the material being plated. In addition, manufacturing costs can be reduced if a selective metallization process is used that allows varied metal thickness on the same substrate. Masking (echniques combined with successive runs in a plating bath are often used to achieve variation of metal thickness on the same substrate or to prevent plating on various areas of the substrate. The high pH electroless plating solutions destroy the resists often used in these masking applications. The Gel Plating Process has been developed as a selective metallization technique which takes advantage of the electroless plating bath chemistry without being limited by its drawbacks. The following section will describe in more detail the Gel Plating Process. 15* Bath contaminated by substrates « Bath maintenance is costly ‘Bath chemical composition changes © Large volume of waste generated not through drag out and water evaporation good for environment Plated Substrate + Substrate + Drag In Drag Out (Adapted from Dr. Juan Hajdu’s 37" William Blum Lecture at SUR/FIN °96, Cleveland, OH) Figure 1: Commercial Electroless Gold Plating1.3 Overview of the Gel Plating Process The Gel Plating Process selectively plates metal on catalytic features without exposing sensitive areas of the substrate to a corrosive plating bath. This process utilizes an electroless plating bath and polymeric thickening agent to formulate a gel. This gel is selectively printed onto the areas of the substrate that require plating. The substrate is placed into a heated, humid environment in order to initiate and sustain the plating reaction, Plating occurs autocatalytically at an elevated temperature by the simultaneous anodic oxidation of the reducing agent and catalytic reduction of the metal complex onto the catalytic features under the gel print. The gel is removed from the substrate after the metal has been deposited using a cleaning protocol that is compatible with the substrate surface, The final result is a selectively metallized substzate, metallized directly under the areas where the gel was printed. Figure 2 shows a schematic of the different steps in the Gel Plating Process. In (a), the top view of an AIN substrate with a nickel (light gray areas) metallization pattern is shown. In (b), the gel print is screen printed directly over the nickel pads. Figure 2 (c) is a cross-sectional view of the substrate with nickel pads. In (@), the same cross-sectional view of the substrate is shown as in (c) but with the gel print over the nickel pads. Figure 2 (e) shows the cross-sectional view of the substrate with gold plated directly over the nickel pads. aNickel pad Gel Print v e a a Bi a] (a) Top view of substrate (white) with ___(b) Top view of gel print over nickel pads pickel pads (gray) Nickel pad Pps ——, —_m_a, (©) Cross sectional view of substzate with (4) Cross sectional view of gel print over nickel pads nickel pads Gold Plate ——— (©) Cross sectional view of gold plate over nickel pads Figure 2: Overview of the Gel Plating ProcessThe theoretical thickness of the gold plate after deposition is dependent on the gold concentration in the plating gel and the thickness of the gel print. The theoretical gold plate thickness possible from a gel print can be calculated assuming that all the gold is depleted from the plating gel: Gold Plate Thickeness = Masloc Ceci titnet) Me ay ae where [Au]oer is the concentration of gold in the gel, Mw is the molecular weight of gold, and p,, is the density of gold. ‘The graph in Figure 3 was created from the above equation to show the influence of gold concentration and gel print thickness on the final thickness of gold plate obtained. Commercial electroless gold baths usually have a gold concentration around 4 g/L. An increase in the gold concentration usually decreases the stability of the plating solution. The gold plate will be only approximately 0.1 um thick if a 500m thick gel print with a gold concentration (4 g/L) is used (see Figure 3). Therefore, the gold concentration in the plating gel must be increased to obtain a gold plate with useful commercial applications. Proof of concept experiments for the Gel Plating Process have been performed using two different plating gel formulations. A cyanide-based plating gel was formulated using a commercial thickening agent and had a gold concentration of 8 g/L and pH of 13- 14, This plating gel was screen printed to a thickness of ~S00 ym using a mesh/emulsion screen. The substrate was placed in a reactor at 82°C with controlled humidity and uniform heat transfer to the substrate for about 1 hour. The resulting gold plate was porous and not well developed, but gold did plate. The second plating gel chemistry was formulated using a gold thiosulfate based chemistry with an ascorbic acid 19a Gold Plate Thickness (um) $8 8 0 100 200 © 300-400» 500 600700 Gel Print Thickness (um) Figure 3: Thickness of Gold Plate as a Function of Gel Print Thickness and Gold Concentrationreducing agent. It has a pH between 7.0 to 8.0. The gold concentration was increased from 8 g/L to 40 g/L gold and the gel was screen printed using a metal stencil screen ~250 um thick. The print may be 250 to 700 um depending on the settings of the screen printer. The substrate was placed into a commercial humidity oven at temperatures ranging from 50 to 60 °C and the humidity was set at 97%. The resulting gold plate was dense and uniform. 1.4 Advantages of the Gel Plating Process The Gel Plating Process for selective area electroless gold plating has potential to: (1) allow variation of gold thickness on the same substrate, (2) reduce substrate etching by minimizing the exposure of the substrate surface to the corrosive plating components, (3) eliminate costly bath maintenance and contamination issues, and (4) reduce plating costs partially due to the lower volume of plating medium utilized. A unique feature of the plating gel is that the gold concentration in the plating gel can be raised above the normal concentration in commercial plating baths. A plating gel with a gold concentration of 40g/L can be formulated. This is a 10-fold increase over conventional plating baths. The gold concentration can be increased in a plating gel because the stability criteria is different than that of a plating bath. Electroless gold baths must remain stable at the plating temperature for many months and must be used for numerous plating runs. The plating gel is used only once. Contamination issues are less of a concern than in large plating baths because the plating gel is individually applied to each substrate. ‘The reduced constraints on stability realized by the Gel Plating Process allow for the use of non-cyanide plating formulations that are too unstable to be used as electroless aplating baths. This is very attractive for an industry that is constantly complying to new environmental regulations. A 40 g/L gold plating gel can produce a gold plate of 1 micron when a print thickness of 500 microns is used. Gold plate thickness in excess of 1 micron can be achieved by repeating the gel plating process multiple times on the same substrate. This same technique can be used to vary the metal thickness on the same substrate by utilizing different masks for each successive print. 1.5 Organization of Thesis The goal of this thesis is to develop the Gel Plating Process and investigate the Kinetics governing gold deposition by this process. Chapter 2 of this thesis will examine the corrosion of aluminum nitride to understand the necessity of a selective metallization process for metallizing hybrid microelectronic substrates. Chapter 3 of this thesis will describe in detail the chemistry of the plating gel formulation. The behavior of the plating gel during the screen printing and deposition steps will be discussed in Chapter 4. Chapter 5 investigates the rate limiting step of the Gel Plating Process and the factors that influence deposition through the plating gel. Chapter 6 will describe the properties of the gold plate. Chapter 7 presents a summary of the major conclusions from Chapters 2-6.Chapter 2: ALUMINUM NITRIDE (AIN) CORROSION KINETICS 2.1 Introduction Ceramic packages must be designed to efficiently dissipate the heat generated from high power electronic applications such as cell phone base stations and new phased array radar systems. A potential packaging system for these high power applications is an aluminum nitride (AIN) substrate with a multi-chip module (MCM) design. AIN has a thermal conductivity almost 10 times that of alumina, a coefficient of thermal expansion close to that of silicon, and lower health risks than beryllium oxide (BeO), another high thermal conductivity substrate material. Unfortunately, AIN is hydroscopically reactive and severely corroded in alkaline solutions, Ceramic packages designed for electronic applications encounter numerous aqueous plating and rinsing steps prior to wire bonding or hermetic sealing. Aqueous plating baths vary from acidic to alkaline pH. Electroless gold plating, for example, is an important metallization step for MCM design. Commercial electroless gold baths are aqueous with a pH around 14. AIN is not compatible with this processing step. Furthermore, each plating bath in a cycle usually has a two-stage cascade rinse tank to remove plating residue.'’ Rinsing continues until a “chemically clean” surface is achieved because any contaminants remaining on the surface can render the parts defective in further processing steps. Aqueous cleaning solutions can etch the surface or alter the surface chemistry of AIN. This can have an adverse effect on the substrate properties such as surface roughness. Fundamentals of AIN corrosion must be understood in order to develop plating and cleaning protocols compatible with AIN. AIN corrosion has many similarities to aluminum metal corrosion in aqueous solutions. Various phases of aluminum hydroxides and oxides are detected on aluminum metal when exposed to water:'* 23Al+2H,0 = AIOOH +1.5H, (2.1) Al + 3H,0 — AU(OH),+ 15H, (2.2) 2Al + 3H,O — Al,O,+3H, — (2.3) The morphology of the aluminum hydroxide and oxides formed in the above equations when aluminum metal is exposed to water depends on the pH, temperature, and time of exposure of the aqueous solutions. The initial precipitate is usually an amorphous hydroxide which can be aged to more crystalline corrosion products depending on the aging conditions.'? A typjcal aging sequence may look like the following: AIOOH ~¥Al(OH) sys AU OH 522 PAL Osea (2.4) Exposure of AIN to water results in similar aluminum hydroxides according to the . following reactions:'* AIN + 2H,O > AlOOH gnorpy + NH (2.5) NH, + H,0 © NH,’ + OH™ (2.6) AIOOH guggy + HzO —%—> AU OH) sem (2-7) The majority of the research performed to identify the corrosion mechanisms and Kinetics of AIN in aqueous solutions have used AIN powder because of its compatibility with surface analysis techniques such as Thermal Gravimetric Analysis (TGA), Fourier Transform Infrared Spectroscopy (FTIR), and X-ray Diffraction (XRD). AIN powder is initially converted to an amorphous aluminum oxide hydroxide phase (psuedobochmite phase, AIOOH) when exposed to an aqueous solution (eq. 2.4). ‘The psuedobochmite phase can transform (or age) to the more crystalline aluminum trihydroxide phase (bayerite or gibbsite, Al(OH)sxsta) depending on the time of exposure and aqueous solution conditions: temperature, pH, and ionic composition'’ (equation 2.7). Bowen et 24al'* concluded from their study using AIN powder that the kinetics of AIN consumption are first order and the reaction rate linear. These investigators fit the kinetic data to the unreacted core model with a porous product layer and determined that surface chemical reaction controlled the overall kinetics. AIN substrate corrosion investigations are primarily based on the examination of the corrosion product through SEM, XRD, weight loss, and profilometry.'“""® One study reported that AIN substrates have an etch rate (weight loss/hour) of 3.16-4.10% in sodium hydroxide solutions with a pH of 14; whereas, aluminum oxide does not exhibit etching in the same solution.'? Young et al ' reported that AIN substrates are more quickly corroded in basic solutions than acidic solutions (pH $7). The basic solutions produced a coral-like surface with randomly distributed sharp edges; whereas, the acidic regime demonstrated a jigsaw puzzle surface morphology. Chanchani'® studied the corrosion product formed on AIN substrates in different aqueous environments with varying temperature, time, and pH. It was reported that the corrosion product is an amorphous aluminum hydroxide film which forms on the surface of the AIN after a 1 hour soak at a temperature of 100°C in an aqueous solution with pH between 3.0 and 8.2.18 Sternitzke” investigated oxide formation on AIN samples less than 100 nm thick by using an electron energy-loss spectroscopy (EELS) technique. The oxide layer formation followed a logarithmic rate law, indicating reaction controlled kinetics. TEM micrographs of the AIN samples exposed to water demonstrate that the surface becomes increasingly rough as the time of exposure increases. Sternitzke explains that this is due to the dissolution of the oxide layer. The oxidation rate was determined by evaluating the oxide film remaining on the sample and the dissolution rate of the oxide layer was not 2» included in the rate law interpretation. * The dissolution step must be included for accurate determination of the corrosion kinetics. The above techniques for evaluating AIN powder and substrate corrosion are ex situ, meaning they only look at the corrosion product left on the surface. This paper willdiscuss a new technique developed to measure the in situ corrosion rate of aluminum nitride (AIN) powder and substrates in aqueous solutions. An ammonium ammonia ion selective electrode is used to detect the concentration of ammonium ions or ammonia gas formed as AIN is exposed to an aqueous environment (equation 2.5). This concentration can be used to determine the corrosion rate in terms of both concentration and thickness per unit time. The results obtained from the ion selective electrode technique will be presented along with a preliminary study conducted on AIN powder using TGA, FTIR, and XRD analysis. Table 2 summarizes the above mentioned techniques. Table 2: Techniques to Evaluate Corrosion of AIN Method Sample Pro’s Con’s Type FTIR powder identifies specific phase mainly a qualitative ‘ of aluminum assessment trihydrate on surface not in situ TGA powder quantitative measure of very small samples surface groups required not in situ XRD powder or identifies composition of _ mainly detects substrate surface groups crystalline phases not in situ EELS thin substrate determine oxide thickness data analysis complex (<100 nm) from [NV/{O] signal not in situ Ammonium Ion powder or in situ measurement of _ electrode has limited Selective Electrode substrate surface corrosion pH and quantitative measure of temperature range thickness corroded _electrode sensitive to small interferences 262.2 Experimental Procedure 2.2.1. Preliminary Powder Study ‘An ex situ AIN powder corrosion study was performed using AIN powder from Sumitomo Chemical Co., LTD. 1 gram of AIN powder was placed in 100 ml of the aqueous solutions listed in Table 3 for the time and temperatures indicated. Table 3: Aqueous Solution Chemistry, pH, and Time of Exposure. DI Water Room Temp quick rinse DI Water 85 T hour 0.03 M NaHCO; / DI Water 85 1 hour 0.02 M NaOH / DI Water 35 1 hour 27M NaOH / DI Water 85. hour The solutions were filtered and the powder dried in a vacuum oven for approximately 1 hour at 100 °C. Surface analysis techniques were used to determine the amount and phase of the corrosion products present on the AIN surface. The AIN powdered was heated from 30°C to 1100°C at a heating rate of 10.0 °C/min in a N environment for TGA (Perkin-Elmer TGA 7) analysis. A 2.5 mg of the AIN powder sample was diluted with 38.5 mg of potassium bromide (KBr) for FTIR (Nicolet IR-44) analysis. The wave number range was 4000 to 300 cm. A Kubelka Munk transformation of the spectrum was performed. A @/2-6 XRD (Rigaku powder diffractometer, 18KW rotating anode X-ray generator) spectrum was obtained for each powder sample. 2.2.3 ‘In situ’ ion selective electrode measurements An ammonium (Cole Parmer Instrument Co.) or ammonia (Accumet, Fisher Scientific) ion-selective electrode was used to measure the in situ corrosion rate of AIN powder and substrates in aqueous solutions. Ammoniuin ions or ammonia gas is formed 2as a by product when AIN is exposed to aqueous solutions according to equation 2.5. The ammonium or ammonia ion selective electrode detects the concentration of ammonia/ammonium released while the substrate is in contact with the aqueors solution. This concentration is used to determine the corrosion rate both in terms of concentration and thickness per unit time. The thickness corroded can be calculated by using the following relationships: 1 NH, 1 AIN (2.7) NH,+H,0 — NH,OH (28) ConcentrationNH ,” x Vol.Sol'n x M ,(AIN) ThicknessCorroded= (2.9) PX SurfaceArea where My is the molecular weight of AIN and pam is the density of AIN ‘An ammonium ion selective electrode from Cole-Parmer” is used in solutions with a pH from 4 to 10 and detects ammonium concentrations ranging from 0.1 ppm (5x 10° M NH4* or 9.65 angstroms corroded) to 18,000 ppm (1.0 M NH," or 193 microns corroded). An ammonia ion selective electrode from Accumet™ is used in solutions with a pH greater than or equal to 12 and detects ammonia concentrations ranging from 1.7 x 10° ppm (1 x 107 M NH; or 0.19 angstroms corroded) to 1700 ppm (0.1 MNBs or 19.3 microns). Table 4 outlines the chemistry and pH of the aqueous buffer solutions used in the ion selective electrode experiments. The ionic strength in the buffer solutions must be under the concentration which interferes with ion selective electrode operation. 0.01M NaCI solution was added to the buffer solutions as an ionic strength adjuster (ISA). The electrode was calibrated in 100ml of the ISA/buffer solution. A Labview™ program was used to record the electrode calibration and monitor the ammonium/ ammonia concentration every 5 seconds throughout the experiment. 28Table 4: Buffer Solutions for Adjusting pH Levels” “pH Buffer Solution (100 ml) 5.5 DI water/ no buffer 7 50 mlof 0.1 M tris(hydroxymethyl)amino methane + 46.6 ml 0.1 M HC! + balance H20 8.5 50 mlof0.1 M tris(hydroxymethyl)amino methane + 14.7 ml 0.1 M HCI + balance H20 9.5 50 ml of 0.025 M Borax + 8.8 ml of 0.1 M NaOH + balance H2O 12__1 ml 10M NaOH + 100 ml H2O 12__1 ml 10MNaOH + 100ml H,O 0 The sample solution contained 20 ml of the ISA/buffer solutions listed in Table 3 plus 0.4 ppm ammonium standard. The background ammonium concentration ensures electrode stability at low ppm concentrations. The AIN sample was placed into the sample solution after collecting 20 minutes of background ammonium concentration. Both AIN substrates and powder were evaluated in the ISA/buffer solutions using the ion selective electrode technique. AIN powder (Dow Chemical Company®, Midland, MD) was evaluated at pH values of 5.5 and 9.5. AIN substrates (Carborundum Microelectronics, Phoenix, AZ) 2.53x 2.53 cm” and 0.025 cm thick were evaluated at pH values of 5.5, 7, 8.5, 9.5, and 12 after cleaning by a simple rinse with isopropyl alcohol and methanol. The AIN powder or substrate remained in the ISA/buffer solution for approximately 2.5 hours while the electrode continuously monitored the corrosion. Another experiment measured the corrosion of AIN substrates that were first oxidized in air before being placed in the pH=9.5 ISA/buffer solution. A tube fumace was used to perform the oxidation of the AIN substrates in an air environment at a temperature of 750°C. The AIN substrates were placed in the tube furnace at room temperature and ramped to 750°C at a rate of 10°C /min. ‘The substrates were held at 750°C for ‘0’, ‘5’, and ‘10’ minutes. The ‘0’ minute sample was removed from the furnace as soon as the temperature reached 750°C and air quenched. The resulting oxidation layer was not detectable with weight gain measurements sensitive to 100 micrograms. The thermal conductivity of the oxidized AIN substrates was measured at 29Carborundum Microelectronics. They used the laser flash method that utilizes a CO) laser. The thickness of the “oxide” layer was semi-quantitatively determined using Rutherford Backscattering Spectroscopy (RBS). AIN substrates from Carborundum Microelectronics® were polished to a 0.25 micron finish. The polished AIN substrates were then oxidized according to the procedure described in the previous paragraph. The RBS analysis was performed at the Harvard University’s Gordon M‘ Kay Laboratory by John Chervinsky. The AIN samples were bombarded with a 2.0 MeV He* ions. The RBS spectrums collected from the as-received and 10 minute oxidized polished AIN substrates wete evaluated by using the RUMP™ software. 2.3 Relationship Between Solution pH and Corrosion Kinetics of AIN Initial AIN corrosion studies were performed on AIN powder to determine the corrosion product present after exposure to aqueous solutions for various pH, times, and temperatures. The corrosion products formed on the AIN powder were various phases of aluminum hydroxide: psuedobochmite (AIOOH) and bayerite/gibbsite (Al(OH) sum). Table 5 summarizes the results of this study. The %AI(OH); was determined from the weight loss detection during TGA experiments. The type of surface hydroxyls present on the AIN powder were identified from the FTIR spectrum and associated with the correct aluminum hydroxide phase. ‘The intensity of the peaks recorded from both the FTIR and XRD measurements gave a qualitative idea of the relative amount of each phase present. The notation, s= strongly detected, w= weakly detected, or o= not detected, used in Table 5, represents these qualitative results. The study shows that 91% of the AIN powder transforms to bayerite with very little residual psuedoboehmite present when the powder is held in a pH=14 aqueous solution for 1 hour at a temperature of 85°C. Whereas, 83% of the AIN powder transforms to both psuedoboehmite and bayerite when the aqueous solution has a pH=5, although the time and temperature are the same. These results in 30Table 5 illustrate the aging effect that increasing pH has on the phase and morphology of the corrosion product. Temperature also contributes to the aging of the corrosion product. For example, a quick rinse with DI water (pH~5) at room temperature shows a 249% conversion of the AIN powder to an amorphous psuedoboehmite phase compared to the one hour treatment at 85°C at a pH-5 mentioned above. These trends are similar to the aging scheme discussed in section 2.1, equation 2.4 for the corrosion products formed ‘on aluminum metal. Table 5: Summary of Initial AIN Powder Corrosion Studies Quick Rinse Thr. 85°C pH=5 1 hr, 85°C pH=8 1 hr. 85°C pH=11 1 hr. 85°C pH=14 pB= psuedoboehmite B/G= Bayerite/ Gibbsite trongly detected w= weakly detected = not detected The relationship between the corrosion products formed on the surface of AIN and the corrosion rate of AIN at different pH values was studied using the ammonium / ammonia ion selective electrode technique. In situ measurement of the ammonium concentration evolved from the AIN substrate surface during corrosion as a function of time from the ion selective electrode reading is shown in Figure 4. The initial concentration (first 10 minutes) was converted into an ammonium flux. This ammonium flux is plotted as a function of pH in Figure 5 (a). Figure 5 illustrates that the initial corrosion rate (ammonium flux) of AIN over the given pH range shows a similar trend as 31the solubility curve for aluminum hydroxide over the same pH range. The preliminary powder study (Table 5) and the literature cited in section 2.1 define the corrosion products which are found on the AIN surface when exposed to aqueous solutions as different phases of the aluminum hydroxides. Therefore, the corrosion products found on the surface of AIN follow the same solubility as Figure 5 (b). Comparing Figure 5 (a) and Figure 5 (b) suggests that the corrosion raté of AIN is at a minimum when the corrosion product is not very soluble (pH=6 to 8). The corrosion rate of AIN increases at pH=9.5 and 12, where the solubility of the corrosion product increases. This indicates that the corrosion product, an aluminum hydroxide, remains on the surface of the AIN when it is not soluble and acts as a barrier layer. More of the AIN corrodes when the aluminum trihydroxide is soluble in the aqueous buffer solution. ‘ The pH range of 6 to 8 does show decreased corrosion of AIN. However, the AIN surface is transformed to the aluminum tihydroxide product. Aluminum hydroxides reversibly absorb water, anions, hydrated cations, and a number of organic compounds.* It also has high specific surface area and porosity.* Aqueous solutions within this pH_ range may appear safe to use because of minimized AIN corrosion; however, care should be taken to ensure that aluminum hydroxide surface does not contain contaminants that will interfere with post processing steps and degrade device reliability. 32.o m0 wm mw 10 ‘Tine(nin) Figure 4: Corrosion of AIN in Aqueous Solutions 33“Amnon Fie (mon it) PERLE LL EE 5 6 7 8 8 © mH @ . s* =. 2 8s. ¢ Qe. z 2 7 é 7 ny 7 PH 0) (a) Ammonium flux as a function of pH for AIN substrates (b) Solubility of aluminum trihydroxide as a function of pH (adapted from K.Wefers et al, “Oxides and Hydroxides of Aluminum”, Alcoa Technical Paper No. 19, Revised, p. 3) Figure 5: Ammonium Flux and Solubility of Al(OH); as a Function of pH. 34AIN powder and substrate corrosion rates obtained using the ion selective electrode in ISA/buffer solutions of pH=5.5 and 9.5 were compared and the results are shown in Figure 6, The AIN substrate corrodes at a higher rate than the powder (concentration normalized according to surface area). AIN powders are commonly stabilized against hydrolysis by reaction with oxygen at high temperature. The overall ‘oxygen content is only changed a small amount since the oxygen contamination is limited to the surface by adjusting the reaction temperature. Dissolved oxygen is removed during high temperature processing for substrate manufacture. Pure AIN single crystals have a thermal conductivity on the order of 320 W/(m K)*; however, polycrystalline AIN substrates formed using low oxygen concentration powders have thermal conductivity values less than 100 W/(m K).™ Polycrystalline AIN substrates have dissolved oxygen in their lattice which creates aluminum vacancies. These vacancies scatter phonons resulting in low thermal conductivity values. Alkaline-earth additives, such as CaO and ‘Y203, react with the Al,O3 on the surface of the AIN powder to form a liquid which promotes sintering and removes dissolved oxygen from the AIN grains. The removal of the dissolved oxygen enhances the phonon mean free path, resulting in an increased thermal conductivity.* Figure 7 shows this schematically. ‘As a result, the AIN substrate is composed of relatively pure AIN grains; whereas, AIN powder surfaces have a protective layer purposely created before shipping to prevent further degradation of the powder. Therefore, the powder shows more of a passivation layer than the sintered AIN substrates. 35<< — Substrate bo <— Powder be pop oo om mmm Teen) (@ pH=55 + Simoes 1 Poe ps (b) pH=9.5 Figure 6: Comparison of AIN Powder and AIN Substrate Corrosion 36Figure 7: Schematic of AIN Sintering 37‘The findings presented above indicate that AIN has similar barrier layer protection and passivation mechanisms to aluminum metal. Aluminum metal forms a protective oxide layer when exposed to air. This oxide layer is 1-3 nm thick when formed in air under 300 °C and has a relatively high corrosion resistance.”” Chemical and anodic oxidation coatings can be formed to enhance the corrosion resistance of aluminum metal in severe environments. Anodizing is an electrochemical process which passes a current through the electrolyte in which an aluminum anode is suspended. Oxygen, a negatively charged anion, migrates to the anode, losses its electrons, and unites chemically with aluminum to form an oxide layer.”” A variety of oxide thickness can be grown by varying the current and electrolyte composition. Normal anodic coatings from sulphuric acid anodizing have film thickness ranging from 5-30 pm and can be sealed to obtain maximum corrosion resistance.” ‘ 2.4 Insoluble Barrier Layer Protection Mechanism The present research adapted the idea of an insoluble barrier layer protection mechanism by creating a protective oxide layer on the AIN substrates. AIN substrates were oxidized in air for short times (0 to 10 minutes) at a temperature of 750°C. Bellosi™ cites that AIN reacts with oxygen to form a “more-or-less protective Al,O;” at temperatures above 600°C. Oxidation of AIN starts at 600°C, but crystalline Al;Os is net observed until above 100°C. The layer formed below 1000°C may be a thin oxynitride layer.” The oxidized AIN samples were placed into the pH=9.5 buffer solution and monitored with an ammonivm ion selective electrode. The results from samples oxidized for different times are shown in Figure 8. The corrosion rate decreases as the time of oxidation for the AIN substrate increases. 38‘Sample 20-23 ss recived Sarple 025: 0rrin. @750C Sarple20-28 Strin, @7S0C sree Sarple 20-21: 10min, @750C = -x+ <— Onin ‘Thickness Corroded (Angstroms) 8 s 8 8B ° <— 10min, & 0 7D 0 © © 10 1 40 160 16 ‘Tine(nin) Figure 8: Corrosion of Oxidized AIN Substrates in Aqueous Solutions 39The as-received substrate is most severely corroded; whereas, the sample oxidized for 10 minutes before it is exposed to the aqueous solution shows no corrosion over the 2.5 hour period. Therefore, the oxidation step forms an insoluble barrier layer on the AIN surface that protects the sample from corrosion in the aqueous buffer solution. The RBS spectrum collected from the 10 minute oxidized AIN substrate showed that the insoluble barrier layer was roughly 200 angstroms. Figure 9 compares the RBS spectrum collected from both the as-received and 10 minute oxidized AIN substrates. These spectrums were fitted using the RUMP™ program. The fitting parameters used are illustrated in Table 6. Table 6: RUMP™ Fitting Parameters for RBS Spectrum Thickness (angstroms) ‘Composition Layer 1 9 ‘Au Layer 2 200 AION Layer 3 99999 AIN +3 wt% Y203 Layer 1 in Table 6 is a sputtered gold layer necessary to minimize charging problems during the collection of the RBS spectrum. Layer 2 represents the oxynitride layer formed during the oxidation step described in the experimental procedure sections. Layer 3 represents the bulk AIN substrate which contains yittrium oxide as a sintering aid. In Figure 9, the line from the fitting program is the solid line, The hollow circles represent the actual spectrum collected. Figure 9(a) shows the spectrum for the as-received AIN substrate compared to the fit line from the above fitting parameters. Figure 9(b) shows the actual spectrum collected for the AIN sampled oxidized for 10 minutes compared to the fit line generated from the parameters in Table 6. The fit line has an oxygen peak which is representative of Layer 2, the AION layer created from the oxidation treatment. The RBS spectrum for the as-received layer does not show this oxygen peak. However, the RBS spectrum from the 10 minute oxidized sample does demonstrate this peak. Therefore, it can be concluded that the insoluble barrier layerformed on the AIN substrate from the 10 minute oxidation treatment at 750 °C is a aluminum oxynitride layer that is roughly 200 angstroms thick. ‘The 200 angstrom oxynitride layer on the AIN surface is not expected to influence the thermal conductivity value of AIN significantly. However, the thermal conductivity results from the oxidized AIN samples do show a decrease in thermal conductivity with increasing oxidation time as shown in Table 7. The 10 minute sample shows a 6.6% decrease in thermal conductivity from the as-received sample. Table 7: AIN Thermal Conductivity Results after Oxidation in Air at 750 °C ‘Time of Oxidation (min) ‘Thermal Conductivity (W/mK) As-received 197.7 0 189.9 5 183.6 10 184.7 4[simulation + As received AIN og mo om ooo » i Channel (@) As-received sample [Simulation 2 _AIN oxidized 10 min @ 7506 (b) 10 min oxidized sample Figure 9: RBS Spectrum of AION Layer on AIN Substrates 422.5 Conclusions In situ measuring of AIN corrosion rates for both powder and substrates was demonstrated with the ion selective electrode method. It was shown that the corrosion of AIN substrates over the pH range of 6 to 12 correfates with the solubility data of aluminum trihydroxide over the same pH range. It was shown from the preliminary ex situ AIN powder studies that aluminum hydroxide phases form on the AIN surface when exposed to an aqueous environment. Therefore, the aluminum trihydroxide product, when insoluble in the aqucous solution, acts as a barrier layer which decreases the corrosion of AIN. It was shown that the corrosion rate of the AIN substrate is approximately four times faster than the corrosion rate of the AIN powder when evaluated with the ammonia/ ammonium ion selective electrode. Corrosion protection of AIN substrates was achieved by creating a 200 angstrom oxynitride layer on the surface of the AIN before exposing it to the ISA/buffer solution. However, there is a 6.6% decrease in thermal conductivity associated with this oxynitride layer. Ex situ study of AIN substrate and powder corrosion was shown to be inaccurate method for explaining AIN substrate corrosion. The ex situ methods used to evaluate AIN substrate corrosion do not take into account the dissolution of the corrosion product layer. The solubility of the corrosion product is a critical factor when evaluating the extent of AIN corrosion. Therefore, in situ techniques that evaluate the aqueous corrosion of AIN are necessary to accurately understand the fundamentals of AIN substrate corrosion. 43Chapter 3 : PLATING GEL FORMULATION 3.1 Review of electroless plating bath chemistry Electroless gold plating is an economical technique for selective plating and can be used to plate electrically isolated features common to MCM designs. Electroless plating is an autocatalytic process which plates metal without the use of an extemal current and deposits metal selectively on areas of the substrate surface where a catalyst is present.” An important feature of a truly autocatalytic electroless gold plating process is that gold can be plated on a gold substrate, Electroless baths operaté through a chemical reduction of a gold complex to gold metal. Two simultanesus electrochemical reactions occur at the catalytic surface: 1) catalytic oxidation of the reducing agent produces electrons and 2) reduction of the gold complex to gold metal uses electrons. Therefore, the electroless plating bath is designed to maintain a balance between stability of the gold complex in solution and the thermodynamically favorable reduction of the gold complex to gold metal. The thermodynamic instability of the electroless plating bath may cause problems such as homogenous precipitation of metal in the solution or “plate out” onto the walls of the plating tank. The gold electroless plating baths that are stable over long periods of time and used in industry usually contain a strong gold complex. This strong gold complex requires high plating temperatures before the gold complex is reduced to gold metal on the catalytic surface.The most stable and commercially used electroless gold bath contains a gold- cyanide complex (KAu(CN):) that is reduced by a borane reducing agent (DMAB, (CH3)NH:BH;). This bath has a pH around 14, gold concentration of 4 g/L, and a plating temperature of 82°C. It can be used to plate alumina substrates, but corrodes AIN substrates (see Chapter 2). ‘Thermodynamic considerations control the cathodic gold deposition (equation 3.1):°" 3Au(CN),"+3e"93Aut+6CN (3.1) The reducing agent, DMAB, forms a hydrolysis intermediate which is used for further reaction (equation 3.2): (CH,),NH : BH,+OH~— (CH),NH+BH,OH™ (3.2) ‘The anodic oxidation of this hydrolysis intermediate (equation 3.3) supplies the electrons at the catalytic surface, for example nickel, that are necessary for the reaction in equation 3.1. BH,OH~+30H~— BO, +: 4+2H,0+3e" (3.3) The overall plating reaction for gold onto the nickel surface is (equation 3.4): BH,OH~+3Au(CN),+30H”—> BO,” +3/, H,+2H,0+3Au+6CN" (3.4) Electroless gold plating baths contain many constituents besides the gold complex and reducing agent. Table 8 describes some of the important bath constituents and their functions.2? 45Table 8: Functions of Constituents in Electroless Gold Baths ‘Component Function ‘Some Examples Gold complex ions To supply metallic gold K[Au(CN)], NasAu(SOs)2 Reducing Agents To reduce gold ions to gold DMAB, ascorbic acid ‘metal Organic chelating To act as a buffer and prevent EDTA, citric acid, tartaric acid rapid decomposition Stabilizers To inhibit the solution Alkali metal cyanide, thiourea decomposition by masking active nuclei Exaltant To increase the rate of Succinic acid deposition and to counteract the slowing effect of the chelating agent Buffers To maintain pH Alkali metal salts Weiting Agents To promote wetting of the Sulphonates of fatty acids, soh to the parts to be plated _sulphated alcohols Sullivan et al’? developed a non-cyanide, non-alkaline electroless gold plating bath with a citric acid buffer, sodium gold(]) thiosulfate (Na;Au(S2Oz)s) as the gold complex and sodium L-ascorbic acid (CeHeNaO¢) as the reducing agent. (see equation 3.5) C,H 0q+2Au(S,0,),° A CoH 0gt2Aut45,0,°+2Ht (3.5) This bath has a pH of 6.4, deposition rate of 1 pm/h , and a plating temperature of 30°C. The non-toxic, nonalkaline nature of this bath makes it an attractive alternative to the presently used cyanide, alkaline baths, especially for AIN substrates, However, these baths are not as stable and reliable as the high pH cyanide baths currently used in manufacturing. Kato et al®? formulated an electroless plating bath with an extended lifetime by using a sodium chloraurate as the gold complex and sodium L-ascorbate as the reducing 46agent. They set the pH using a phosphate buffer and added sodium thiosulfate and sodium sulfite to extend the life of the bath. They also used 2-mercaptobenzothiazole and 2-mercaptobenzimidazole as stabilizers in their plating solution to prevent homogenous precipitation of gold in the plating solution. 3.2 Introduction to Plating Gel Chemistry ‘The Gel Plating Process requires that an electroless plating solution be thickened to a viscosity that is compatible with the desired method of placing the gel onto the substrate. The starting point for formulating a plating gel is: 1) an electroless plating bath chemistry that will autocatalytically plate gold and 2) a thickening agent that is compatible with the plating solution and will thicken the solution to the appropriate printable rheology. Two separate approaches can be used to formulate the plating gel. The first is to modify an existing commercially available plating bath. ‘The second is to formulate a unique plating solution specifically designed for the gel plating process. The unique plating solution can be formulated with a non-cyanide chemistry because of the advantages offered by the Gel Plating Process described in section 1.5. ‘The selection of a thickening agent that is compatible with the plating solution and thickens the solution to the appropriate rheology is a critical step in the development of the plating gel. Section 3.3 will describe the procedure necessary for selecting a thickening agent for the modified commercial electroless plating solution. Section 3.4 will describe the formulation of the non-cyanide plating gel specifically designed for the Gel Plating Process. 473.3 Selection of a Thickening Agent for the Modified Commercial Plating Bath 3.3.1. Experimental Procedure ‘A commercial plating bath was used called Sel-Rex Lectroless Au 2000™ (Ethone Omi, Hartford, CT). This plating solution came in two parts: 1) the gold solution which contains a gold cyanide complex and 2) the reducing agent solution which contains a dimethylamine borane reducing agent. The standard make-up procedure results in a plating solution with 4 g/L gold and pH=14. This make-up procedure was’ modified™to increase the gold concentration to 8 g/L and reduce the potassium hydroxide additions. The commercial thickening agents listed in Table 9 were mixed into ihe modified commercial plating bath and evaluated according to their dispersion, solubility, and viscosity behavior in the plating bath. Next, the successfully thickened plating gels were evaluated by screen printing performance using a DeHaart™ SPSA-10 standard screen printer. The specifics of screen printing parameters were described by Amdt“. The substrate with the printed plating gel was placed into a heated, humid environment to initiate and sustain the plating reaction. The plating gel performance during deposition was also evaluated.Table 9: Types of Thickening Agents Investigated Chemical Type Properties ‘Trade Name ‘Supplier Hydroxypropyl water soluable,non _ MethocelK4M Dow Chemical Methylcellulose ionic, psuedoplastic ‘Company (HPMC) . Hydroxyethyl Water soluable, Cellosize QP-100M Union Carbide Cellulose (HEC) non-ionic, thixotropic Poly (Methyl Vinyl Gantrez AN International Ether / Maleic Specialty Products Anhydride Poly(ethylene oxide) _ water soluble, non Polyox Union Carbide (PEO) ionic, psuedoplastic Polysaccharide water soluble, _Tota (Carrageenan) FMC thixotropic Polyurethane non ionic, near- Duramax T-5000, Rohm and Haas Ethylene Oxide Newtonian to shear —_T-5003, B-1051 copolymer thinning 3.3.2 Results Figure 10 illustrates the performance of the thickening agents in a high pH gold bath that contains a gold metal complex and reducing agent. The bold numbers inside the oval regions in Figure 10 identify the solubility ranking. Dispersion rankings are on the x-axis and viscosity rankings are on the y-axis. Dispersion rankings were assigned from 0 to 5 based on how well the polymer powder mixed into the solution. A ranking of 0 was given if the powder agglomerated in the solution and did not mix uniformly. A ranking of 5 was given if the powder mixed uniformly into the solution. The viscosity rankings were assigned from 0 to 5. A value of 0 was assigned if the solution viscosity was similar to water. A value of 5 was assigned if the solution was ‘very’ viscous. A ‘very’ viscous solution is described as a gel in a beaker that barely flows when the beaker 49is inverted. ‘The solubility rankings were assigned from 0 to 5. A ranking of 0 was given if the dispersed polymer powder did not hydrate in the solution. A value of 5 was given if the dispersed polymer powder hydrated into a homogenous gel structure. Figure 10 illustrates that the hydroxypropyl methylcellulose (HPMC), polysaccharides, polyethylene oxides (PEO), and hydroxyethyl cellulose (HEC) obtained good rankings (4 or 5) for viscosity and solubility. ‘The polysaccharides were given a low dispersion ranking (-0.5) compared to the other thickeners mentioned above. The polyurethane/ethylene oxide co-polymers were eliminated because of their low viscosity (ranking = ~1). The poly(methyl vinyl ether/maleic anhydride) was eliminated due to its undesired reaction with the plating bath components. In other plating baths, particularly those of lower pH, these polymer thickening agents may be appropriate. HPMC, HEC, PEO, and polysaccharides were the polymer systems that performed best in the high pH plating solution. HPMC, HEC, PEO, and polysaccharide thickening systems were further evaluated by performance during screen printing. Figure 11 shows the resulting print from the polysaccharide and HPMC systems. The polysaccharide appears to pull off the substrate and adhere to the screen; whereas, the HPMC adheres to the substrate and results in a good gel print. The PEO and HEC systems exhibited a stringing behavior during snap-off. It was determined through these experiments that the plating gel viscosity and structure, printer set-up, and print thickness influence the quality of the gel print, HPMC produced the gel viscosity and structure most suitable for screen printing. 50Hyéroxypropyl methylcellulose olyethylene oxide, Hydroxyethyl cellulose (@)~Potysaccharides oly (methyl vinyl ether/maleic anhydride 2 Polyurethane Ethylene 15 Oxide copolymer : Os 1 ts 2 25 5 a5 4 45 8 ‘Disperion ‘* Number represents solubility ranking Figure 10: Processing-Performance Relationship of Thickening Agents in Commercial Gold Bath, Based on Viscosity, Dispersion, and Solubility Rankings: 0 = poor; 5 = excellent. Solubility Ranking is Inside the Oval. 51(a) Hydroxyp IPMC) based gel (b) Polysaccaride based gel Figure 11: Visual Differences in the Resulting Gel Print of Two Different Polymer Systems: (a) Hydroxypropyl Methylcellulose (HPMC) based system (b) Polysaccharide based system.Understanding the performance of the plating gel at the plating temperature is also critical for formulation of the plating gel. The plating temperature for the cyanide-based plating gel is ~85 °C. The HPMC precipitated out of solution at this temperature. The gelation temperature of the polymer is below 85 °C. The polymer will hydrate and form the gel below the gelation temperature. The gel undergoes syneresis and the polymer precipitates out of solution expelling the solvent above the gelation temperature. Figure 12 is a schematic of the gel before and after syneresis occurs. Propylene glycol can be added to the solution to increase the gelation temperature. Figure 13 shows the effect of propylene glycol on the gelation range. The gelation range (GR) can be defined as the temperature range where the polymer is precipitating out of solution. The solution is clear below the gelation range indicating that all the polymer is hydrated into a gel. The solution is cloudy above the gelation range because the polymer powder has precipitated out of solution. The lower limit of the gelation range depicted in Figure 13 is the temperature at which the first signs of polymer precipitation are observed. The upper limit of the gelation range in Figure 13 is the temperature at which the majority of the polymer has precipitated out of solution. The gelation temperature also depends on the concentration of polymer added. For example, Figure 13 shows that the gelation range for a 1 wt% polymer solution is higher than that for a 4 wt% polymer solution. The gelation range lower limit increases above 90 °C when propylene glycol is added to the mixture. It follows that the polymer will not precipitate out of solution below 90 °C. Therefore, plating can be performed at 85 °C using a plating gel made with HPMC and propylene glycol. 33Homogeneous printed circle (a) before syneresis (homogeneaps gel print) clear liquid residue polymer agglomeration (b) after syneresis occurs Figure 12: Schematic of Syneresis6 w% polymer + 359% clear propylene glycol 4wt% polymer + 35%! propylene glycol : 6wt% polymer} clear “GR 4wt% polymer clear a | 1 wt% polymer clear 0 10 20 30 40 50 60 70 80 90 “Temperature (C) *GR = Gelation Temperature Range Figure 13: Effect of Propylene Glycol on the Gelation Temperature Range 553.4 Non-cyanide Plating Gel Formulation Gold thiosulfate and ascorbic acid were selected as the metal complex and reducing agent for the formulation of a non-cyanide plating gel chemistry. The nature of the Gel Plating Process establishes criteria which guide the development of the plating gel. First, the plating solution must not contain high concentration of ions that interfere with the thickening capabilities of the polymeric thickening agent. Secondly, the plating gel must not decompose during the plating process. Thirdly, the gold concentration in the plating gel must be increased to a value that will produce a commercially useful gold plate. The correlation between gold concentration and gold plate thickness can be seen in Figure 3. Finally, the appearance and properties of the gold plate must meet industry standards. Development of the plating gel using the above criteria is an iterative process. A thorough experimental procedure section describing the Gel Plating Process from start to finish will be presented. Next, observations made during gel formulation will be presented. These observation will be organized in the following way: 1) effect of ion concentration on thickening behavior of plating solution, 2) effect of ion concentration on performance behavior of plating gel at the plating temperature, 3) effect of chemical additives on stability of plating gel, and 3) effect of chemical additives on gold plate appearance. Finally, the resulting gel formulation arrived at by consideration of the above criteria will be presented. 4 Experimental Procedure Plating gels were formulated according to Tables 10-12. A table of chemical formulas can be found at the beginning of this thesis. First, the butfiers, stabilizers, chelating agents, surfactant, gold complex, and reducing agent (see Table 8) are mixed with DI water to form a plating solution. Next, the thickening agent is added to formulate 56the plating gel. The thickening agent can be mixed with the plating solution through three different methods. ‘Method 1 is the ‘hot’ method. The plating solution was heated to a temperature of 70 °C and stirred. The thickening agent was added to the hot plating solution gradually to ensure that the polymer powder dispersed well into the hot plating solution. The solution gelled as it cooled to room temperature. Method 2 is the ‘cold’ method. The thickening agent was mixed with propylene glycol at room temperature. The thickening agent dispersed well because it is insoluble in propylene glycol. The plating solution was then poured into this propylene glycol / thickening agent mixture while stirring. The solution gelled very quickly after being mixed with the propylene glycol and thickening agent. Method 3 is the ‘pH sensitive’ method. The thickening agent was poured directly into the stirred plating solution at room temperature. The time for the solution to thicken will depend on the pH of the solution. Method 1 end 2 were used in this study with the HPMC thickening agent. The row in Table 10 labeled ‘Bath Make-up: hot or cold’ identifies which method was used for each gel formulation. Method 3 was used with the HEC thickening agent (Table 11 and Table 12). The solution thickened in approximately 10 minutes for a pH above 7. The gel formulations listed in able 10 had a pH ranging from 5.5 to 7.5. The gel formulations listed in Table 11 and Table 12 had a pH ranging from 7.0 to 8.0. 37Pio 20 704, yoy | 104 wa mo p09 | pio2 | pros | pico | proo| pros } rq | soy | 104 [104] _ | :dn-oxew 12D 9ltolos}|o9lolo}loltstolelo| oo {ol - |e) oWdh (py) 109715 se se | Se} se se se se se SE Se | SE] SE se - - auaidorg - feool-|-1-1-1-1/-|-|-1f-|-]-]|-]|- | ‘ose -[-1-T-T-7-7T-[-T-T- 7-1 Trool - | - | osten =| - |= [soo [eso] z50-0] sooo lecoo] ro [eco] - | - | - | - | - vied ~|-{-|-|-|-|- | - | - |] - | - [esol zo] - | - HOw so | - | - |svo| zo |9zso|szv0| spo | 60 |6av0} - | 90] v0] eo] - HON f= |-}-}-}-]-]-]- 1] - [vo] co] zo ]ro| - | ‘0'HPo%N zo = j€o} VO | LTO | ero solo] 10 | ZO |soro} - | yO | vO ZO} - ‘O°HO'H soo|soo|-| - | - |szoolzzio| soo} - |uzrvo| - | soo| soo| - | soo] *OPN‘H’O goo] eoo| - | - | - |1s00|9200| coo| - |ozo0| - | coo] coo] - | evo| COsnveN altajalojolse{s|e.iolsirietzi{rio (Grure}oyy Uy sy) sxoquiNN YoREMUTIO [2D (uopesquau0y ploy M07) suopEInUOg fH Bue 201 148, 58ele} stri[ele|sltele}| ec letv] ¢ |v | Moan (ey) - fe]: fete fe te te fe ft | b+ | freottg auaytdorg zo|zo| - | -|-| - }|-|-|]-{ro}]-|-! - | - FOFS*(HN) - | - |isoo} - [zo] ro | - | - | - : - | vo | - SOrS"EN ro | ro}ero| - |sro| ro | - | - | - | zo -| - | - ‘OSeN coleo| - | -[eo| - | - | - [veo] vo |-|-| - | - | veornn ~ | - |izoo} ro] - | zo |sro| - | - | - [stro] zo] ro | zo vida sezo| €0 | suo | coo | 6zo| sevo| set] - | vo | sto | zt |981| ace | 981 HO | -f-[-][-T-[-[-1-]-[- [- | oaten go |co| - | - |so| co} - | - | - | co] - ("HN en ee ee eee “OWEN -{-1- 1-1-1 [- lecoleol - | “O*H°O"CHN) - | +] - |izo] - | - |evo| - | - | - |seolzo| vo |zvo| ‘O*HO'H sco |reo| ceo |svo| co] seo |ccolreoleco| vo | - liso] eco isof PORNO zo |zo| vo |srol|zo| zo | zo|zo| zo| zo | - | - | zo | co} Co'smveen wxlzlaxi[x[xleleleleloalalal os [a (Axe oN UT SHUN) stequINN UOHEINUNIET [25 (1# wopesyuasu0d ploy YIH) SuoNEINULIO [9H BuNEIA ‘TT AGEL, 59.ve| se | se | se [ce | ce | se [ce sel (%94) OHH - |Lzoo0| - o - - - | oo} - 2 OW ,.0T] uej9ej1ng |: : s = a : ee a oc | sz | scot se | se | se | se | st | sit | - OW 01] S'N°H‘O zo = = G : zo | zo | zo| zo} - (W] £0"S*HN) -T- 1-1-1 -.--[-[- [el mirosen svo| sro | sro | sro | sro | erro | epro| sro| sto | ro GAl fosten £0 | 8000 | 9100 | 9400 | 800° | r600 | s000| €0 | co OW] VLGa"HN 6co| - 2 2 2 e =| svo| szo| - C OW] HOeN so|_- -{- | -.| - | -|-]- | - [w]1"AN - | so so | so | so | so| sol] - OW] ‘O°H“O"HN Lo | Lo zo | vo | co | co | zo | eco UW) °0eN‘H?O zo| zo | zo} zo | co | zo | zo | zo} zo} co GW) *C0%s)nvFeN se | uc | vc | se | ve | ce | ce | ve | oc | S19qUINN WOHEINULIOT PD (24 wonexjua.u03 Plog yaIH) suoHEINALIOY [PH BUNA *ZT AILThe sample to be plated was activated with acid before the plating gel was screen printed over the catalytic surface. The type of acid activation will depend on the catalytic surface. A gold surface was activated by immersing the substrate into a 50 vol. % hydrochloric acid solution at a temperature of 35 °C for 30 seconds. A nickel surface was activated by immersing the substrate into a 10 vol.% citric acid solution at 70 °C for 2 minutes. The sample was then rinsed with DI water and dried with air. A screen printing machine (DeHaart SPSA-10) was used to screen the plating gel onto the substrate surface. Two different screens were used with this machine: 1) a polymeric emulsion and metal mesh screen and 2) a metal stencil screen. The screen was visually aligned with the substrate pattern, The plating gel was printed onto the screen. The printing can be performed by using the automated squeegee apparatus or by using a manual printing method. The manual printing method was preferred in this study because it uses less gel per print. The substrate with gel print was then placed into the deposition chamber. Two deposition chambers were used in this study: 1) a custom built reactor that has a saturated nitrogen atmosphere (described in detail in reference 34) and 2) a commercial humidity oven (Blue M) that can be set to 97% relative humidity. The humidity was set as close to 100 % relative humidity as possible for this study. The plating time varied from sample to sample. The plating gel was washed off the surface after deposition by rinsing with DI water. 3.4.2 Effect of Bath Formulation on Thickening Behavior Plating bath formulations contain more chemical constituents than just the metal complex and the reducing agent as can be seen in Table 8. A typical plating bath can have a high concentration of ions in an aqueous solution. This is not a problem for 61typical plating bath operation; however, this high ion concentration does become a problem when attempting to thicken a plating solution with a polymeric thickening agent. 3.4.2.1 Low Gold Concentration ( 4 to 15 g/L Au) Plating gels in Table 10 were formulated using HPMC as the thickening agent. HPMC is non-ionic and will not complex with metallic salts or other ionic species in solution. However, HPMC can be salted out of solution when the concentration of electrolytes or other dissolved materials exceeds certain limits.* This occurs because there is competition for water between the electrolytes and the hydration of the cellulose tive tolerated in different solutions of ether. The product literature lists grams of addit HPMC; however, this: information was not directly applicable to the plating gel chemistry. Therefore, HPMC was tested in various chemical formulations to determine the concentration and species of ions that can be tolerated. Table 13 summarizes some of the ion concentrations in the different baths evaluated using the HPMC thickening agent. The actual formulations containing all the species in the bath are listed in Table 10. Sodium ions are the most detrimental to the thickening capabilities of HPMC solutions. All formulations containing greater than 0.86 M sodium ions (baths 1-6 in Table 13) did not thicken to a gel when mixed with HPMC. Other ions also may influence the gelling behavior of HPMC; however, sodium ions appear to have the greatest effect throughout this study.Table 13: Effect of Ion Concentrations on Thickening Behavior of HPMC Bath [Au] [Na‘] [S03] [EDTA] [C,Hs07*] _ Gel/No Gel (M)_™) ™) (™). (™) 1 0 15 0 0 06 No Gel 2 0.03 1.208 0.06 0 06 No Gel 3 0.03 1.2 0.06 0 06 No Gel 4 0 1.2 04 0 0 No Gel 5 0.076 1.144 0.152 0.093 0.195 No Gel 6 0 09 0 Ol 0.2 No Gel 7 0.03 0.86 0.06 0.033 0.17 Gel 8 0.076 078 0.152 0.05 0.1 Gel 9 0.051 0.764 0.102 0.062 0.13 Gel 10 oO 0.72 0 0.033 0.17 Gel WW 0 045 0 0.05 O1 Gel 12 Oo 03 0 o 0 Gel One of the goals in the development of the Gel Plating Process is to increase the gold concentration to a value such as 40 g/L which will result in a useful gold plate thickness. However, sodium gold thiosulfate (NasAu(S203)2) and sodium L-ascorbate (CsHyNaOg) which are the gold complex and reducing agent used in the non-cyanide plating gel formulation both contain sodium ions. Every 1 [MJ increase in gold ions is connected with a 3 [M] increase in sodium ions. Therefore, HPMC cannot be used to thicken plating gel formulations with a gold concentration in excess of 10 to 15 g/L (baths 5&8 in Table 13) gold because of its sensitivity to sodium ion concentration. A new thickening agent must be found if the gold concentration is to be in excess of 15 g/L. 3.4.2.2 High Gold Concentration Formulations (15 to 40 g/L) HEC and PEO are non-ionic, water soluble polymers which are more tolerant to high electrolyte solutions (according to their product literature). These thickening agents were compared to HPMC in order to determine which could provide a better viscosity at 6an increased gold concentration. Figure 14 illustrates the viscosity of the polymer solution as a function of gold concentration. A bar originating from a specific gold concentration extends to the viscosity attainable at that gold concentration with a specific thickening system. The viscosity ranking is a qualitative number that identifies 0 as a viscosity similar to water and 5 as a viscosity suitable for screen printing. The HPMC thickener is only useful for solutions with gold concentrations less than 15 g/L and the PEO thickening agent does not give the desired viscosity for screen printing. Therefore, the HEC thickener is preferred for the high gold concentration plating gels. There are some constraints associated with utilizing the HEC thickening agent. ‘The particular type of HEC used in these experiments was Cellosize™ QP-100MH. This polymer is modified so that is disperses easily in an aqueous solution. However, it requires time and the proper pH to thicken. The solution pH must be above 7.0 or it will take approximately one hour to reach the desired viscosity. Also, HEC can be salted out of solution similar to HPMC. Therefore, experiments were performed to determine the compatibility of HEC with plating bath formulations. Table 14 summarizes the ion concentrations of cerain species for most of the plating gels listed in Tables 11 and 12. It is obvious from looking at the results in Table 14 that the HEC behavior as a function of ion concentration is different than that of the HPMC. The highest Na* concentration tolerated by HPMC was 0.86 M; whereas, a Na* concentration of 2.281 M is tolerated by HEC as seen in Gel 22 in Table 14.© J Hydroxyethy! cellulose HEE Hydroxypropy! methylcellulose iE Polyethylene oxide 8 I Gold Cone. (g/L) 8 Viscosity Rank Figure 14: Ability of Thickening Agent to Increase Viscosity of Solution as a Function of Gold ConcentrationTable 14: Effect of Ion Concentrations on Thickening Behavior of HEC Bath [Aw] [Na°] _[S,0x"]_[CsH,0,*]_[NH."] Gel No Gel 3.27 0.6 0.42 0 15 03 ‘No Gel 16 02 2992 05 0.4 0 No Gel 17 0 2.36 0 0.42 0 No Gel 18 0 2.169 0s 0.35 0 No Gel 19 02 «1856 (0.7 0 17 No Gel 20 02 1.178 04 0.5 2.176 No Gel 21 02 0.938 «04 0879 2.537. NoGel 22 02 2.281 «04 0.29 0 Gel 23 02 2013 © 05 - 03 Gel 24 0.2 1.891 0.6 0 15 Gel 25 0.15 1.@29 0.3 0.21 0 Gel 26 02 1.556 0.481 0 0 Gel 27 0.2 1.438 0.6 Ou. 13 Gel 28 0.2 1.413 05 0 13 Gel 29 G2 1.388 0.5 0 0 Gel Gels 15-21 did not thicken when mixed with HEC. The negative result from gels 15-18 indicated that some sodium salts were interfering with the thickening capabilities of the hydroxyethyl cellulose. The product literature** for HEC lists inorganic salts that under saturated conditions lead to the precipitation of the polymer from solution. Most of the ammonium complexes do not precipitate the polymer under saturated conditions. Therefore, ammonium complexes were used to replace many of the sodium complexes used in the plating baths. Gels 19-21 did not gel even after these substitutions, indicating that the sodium ions are not the only species that hinder the thickening capabilities of HEC. Comparing gel 21 to gels 22-29 leads to the conclusion that either the citrate or ammonium ion concentration is interfering with the thickening capabilities of the polymer. Further comparison of gel 19 (which did not thicken) to gel 24 (which did thicken) strengthens the argument that the sodium ion concentration is not the determining factor but instead a combination of sodium, thiosulfate, and perhaps ammonium ions. Bath 15 was formulated with 60 g/L gold and did not thicken.Therefore, 40 g/L (0.2 M) gold concentration is a good upper limit to use with this thickening agent. 3.4.3 Effect of lon Concentration on Performance of Plating Gel at Plating Temperature Thickening is not the only property of interest for the polymer solution. The performance of the thickened polymer solution at the plating temperature is also important, Plating is performed in a humid environment to prevent the gel from drying. This will be discussed in more detail in Chapter 4. However, it is important that the gel print retains its shape at the plating temperature. The following section wili describe how the ion concentration of the plating gel influences its behavior at the plating temperature. Plating gels 30-37 in Table 12 were examined during plating to determine the effect that ion concentration has on the behavior of the plating gel during plating. Each plating gel was printed onto a substrate and placed into the commercial humidity oven. The gel print was weighed before and after deposition. The change in weight of the gel print was calculated to determine if the gel print dried during deposition. Photographs of the gel print before and after deposition for 3 different plating gels are shown Figure 15. Pictures (b) and (4) show a pool of liquid surrounding the gel print after deposition. These plating gels are formulated according to gels 30 and 31 in Table 12. A gel print that exhibited good behavior during plating is shown in Picture (e). The evaporation rate as a function of gel formulations can be seen in Figure 16. The legend in Figure 16 correlates the gel chemistry with the gel number in Table 12. The ion concentration in each plating gel is summarized in Table 15. The plating gel formulation used in Figure 15 (b) gains weight during plating; whereas, the plating gel formulation in pictures (d) and (e) dry during plating. Gels 32-37 exhibit good behavior orduring plating. Therefore, the ion concentration does effect the behavior of the plating gel at the plating temperature. The plating gels that demonstrated a pool of liquid around the gel print after deposition can be explained in two ways. First, water was adsorbed from the humidity oven during plating as seen in Figure 15 (b). Second, the gel print undergoes syneresis as, the high ion concentration plating gel dries during deposition (Figure 15 (d)). Syneresis, was discussed is section 3.3 and a schematic is shown in Figure 12. It is hypothesized that syneresis occurs when the highly concentrated plating gel experiences drying at the plating temperature. The loss of water forces the ions and polymer in solution to fight for the water remaining, The polymer begins to expel solvent and will eventually return to its pre-hydrated powdeg form. A gel print of reduced volume surrounded by a pool of expelled solvent is shown in Figure 15 (d). Syneresis of the plating gel at the plating temperature can be avoided if the ion concentration of the plating gel is reduced as demonstrated by plating gels 30-35. A low ion concentration plating gel that performed well at the plating temperature is shown in Figure 15 (f). EDTA appears to be the main ‘component that influenced the behavior of the plating gel at the plating temperature. Table 15: Effect of lon Concentration on Gel Behavior at Plating Temperature Gel_[Na*] [S:0;°] [EDTA] [OH] % Propylene Surfactant Result {M]__[M]. (M] [M] Glycol [M1 30 1.891 0.603 03 0.291 5 0 Liquid 311.891 0.603 0.3 0.291 5 6x10" Liquid 32 1.586 0.603 0.00828 0 0 0 No Lig. 33° 1.586 0.603 0.09390 ° 0 No Lig. 34 «16 = 0.4 0.008280 0 0 No Lig. 35° «1904 0.046 0 5 0 No Lig. 36 «1904 0.046 0 0 0 No Lig. 37:16 0.4 0.008280 0 2x107___NoLig. _ 68(©) Gel 36 before oven ( Gel 36 after oven Figure 15: Photos Showing Gel Print Before and After Plating in the Blue M Humidity Oven (Gel number from Tables 12 and 15) 0°© SurfactantsHumeciant, High Elecroyte (Gel 31) tent © Humectant, High Electoiyte (Gel 30) . #Humectant Lower Electolyte (Gel 35) onto" ‘A Nother, Lower Electrolyte (Gels 22-04,26) Surfactant, Low Electolye (Gel stmt oe t peice) 5 6.0x10*] a a aca 2 sono] © i . 5 i 3 # [Lceing < zoxoe . Water & oo 4 é ° Fj @ 2.0x10°7 Adding Water sont{ 9 5 oe ~6.0x10° 7} [I 0 15 90 45 60 75 90 105 120 135 150 165 180 Time (rin) All solid symbols are for high electrolyte gels. All open symbols are for low electrolyte gels. Figure 16: Gel Behavior in Humidity Oven (Gel Formulation Number From Tables 12and 15) 703.4.4 Stability of Plating Gel as a Function of Chemical Additives Electroless plating solutions contain buffers, chelating agents, and stabilizing agents to prevent the homogeneous precipitation of metal in the plating solution. Table 8 reviews these different chemical constituents and the roles that they play. The stability concems for a plating gel are different than those for a plating bath. An electroless plating bath must be stable at the plating temperature for long periods of time. The electroless plating bath requires careful maintenance to prevent decomposition of this bath over time, The plating gel on the other hand is a one shot process. The plating gel must only remain stable during the gel plating cycle. No bath maintenance or chemistry adjustments are needed. However, the increased gold concentration in the plating gel has a large influence on its stability. ‘The plating bath usually contains only 4 g/L gold; whereas, the commercially useful plating gel should have 40 g/L gold. This increase in gold concentration increases the driving force for the gold to precipitate out of solution. ‘Therefore, stabilizing agents must be identified that prevent homogenous precipitation of gold in the plating gel without hindering the heterogenous plating of gold on the substrate surface. Some of plating gels formulated according to Tables 10-12 have been used to study the stability of the plating gel and are repeated in Table 16 and Table 17. Observations were made conceming the room temperature stability of the plating gel before it was screened onto the substrate. In addition, observations were made concerning the stability of the plating gel during deposition on the substrate. These observations are summarized in Table 16 and Table 17. The following section will go into greater detail highlighting the important chemical additives necessary to prevent decomposition of the plating gel at room temperature and at the plating temperature. Decomposition of the plating gel can be described as any obvious discoloration or particle formation in the plating solution or gel. The following section is not a ncomprehensive study of the kinetics of decomposition during plating; but, it is a qualitative description how different chemical additives influenced the stability of the plating gel. Table 16: Gel Formulations for Stability Study (Repeated from Table 10) Gel Formulation Numbers (Units in Molarity) oy72)73)6,)7)]8]9j]i3] 1 NasAuGSi0y, | 003 [0.03 [0.03 | - [0.03 [0.076]0.051| 0.03 | 0.03 CeHyNad, | 0-05 | 0.05 0.05 | - | 0.05 [0.127]0.075] 0.05 | 0.05 H3C fos2jos2}-])-]-)-)-1- EDTA = [i= [01 [0.033 | 6.05 ]0.062.- | - Na,SO3 - |0.004) - = - + - = 5 NaS203 . . . - - - - [0.008] - Propylene Glycol! - | 35 | 35 | 35 | 35 | 35 | 35 | 35 | 35 (wt%) HPMC(wi%) | - | 6 | 6,6 ]6|6)6!| 6 | 6 Gel Make-up: | - | hot | hot | cold | cold | cold | cold | hot | hot hot or cold Stableat RT for | No | No | No | No | Yes | No | Yes | Yes | Yes 3+ brs Stable at Plating | No | No | No | No | Yes | No | Yes | No | Yes ‘Temp. 2Table 17: Plating Gel Formulations for Stability Study (Repeated from Tables 11 and 12) Bath Formulation Numbers 16 | 22] 24] 26 | 38 NasAu(S203)2 [M1] 02 [02] 02] 02, 02 (C6H;NaOg [M] 0.33 | 0.33 | 0.7 | 0.33 0.7 H3CHs07 [MI 04 0291 - 5 7 NH.CI[M] = |- |os] - | 05 NaH,PO, [MJ - a ~ [oo] - NaOH [M] 1.858 | 1.35 | 0.29 | 0.075 | 0.29 EDTA {M] 0.1 | 0.15 - 0.021 ~ NH,EDTA [M] - - 03 - 0.3 'NaSO3 [M] = = | 0.15 | 0.143 | 0.15 Na;S;0; [M] 0.1 - 0.2 | 0.081 - (NH4)2S205 [M] : 7 - - 0.2 CoHAN,S [10% M] - - - - 30 Propylene Glycol (wt%) | ~ a 5 - - HEC (wt%) 3 }s{3]5 | 37 Stable at RT for 3+hrs | Yes | No | Yes | Yes | Yes Stable at Plating Temp. | No | No | No | Yes | Yes The main components necessary for electroless plating to proceed are the metal complex and the reducing agent. Gel ‘0’ (Table 16) was formulated with only these components. This bath did not plate on a nickel surface at room temperature and it decomposed (turned yellow with streaks of black) in one hour and thirty minutes. The bath did plate when the temperature was increased from 50 to 60 °C; however, it began to immediately decompose at the higher plating temperature. It is obvious from the above Btrials that a functional plating bath requires additives in addition to the metal complex and reducing agent. Sodium sulfite and sodium thiosulfate have been suggested”? as complexing agents to prevent bath instability. ‘Therefore, Gel 13 ( Table 16) was formulated using sodium thiosulfate to increase the stability of the plating bath. Gel 13 (without the thickening agent) was heated to 55 °C and visible plating was observed on the nickel surface in 5 minutes. The bath remained stable for 20 minutes at 55 °C. A gel was also formulated from this composition using ~6 wt% HPMC. The gel began to decompose at room temperature in less than 16 hours. Sullivan et al®? used a citric acid buffering system with potassium hydroxide to establish the pH at 6.5. Gel 3 (Table 16) was formulated with the citric acid buffer and alkaline hydroxides. Gel 2 (Table 16) added sodium sulfite to the bath composition of Gel 3, Gel 3 (without the thickening agent) started to plate immediately on the nickel foil; however, after 20 minutes at 55-60°C , the bath began to decompose. Gel 2 and Gel 3 started to decompose during gel formulation using both the ‘hot’ and ‘cold’ methods of gel formulation. Gel 14 (Table 16) had a similar formulation to Gel 3 except that the concentration of Na* and K* where minimized by adding smaller amounts of NaOH and eliminating all KOH and sodium citrate from the bath formulation. Gel 14 did not decompose during gel formulation. However, Gel 14 did decomposed within 44 hours at room temperature. Therefore, a low gold concentration plating gel can be formulated which is stable over short plating times and storage times according to the formulation of Gel 14. Ethylenediaminetetraacetic acid (EDTA) is a well-known chelating agent used in plating bath formulations. It is used as a brightener and stabilizer in some bath formulations. Gel 6 (Table 16) was formulated with EDTA additions. This bath was not formulated into a gel because of the high ion concentration. Gel 7 (Table 16) was 4formulated with approximately half the concentration of EDTA, citric acid, and sodium hydroxide. Gel 7 decomposed in less than 26 hours at room temperature, as opposed to 44 hours with Gel 14, It is assumed that the concentration of citric acid plays a significant role in the gel stability at room temperature. Therefore, at low gold concentrations, citric acid, sodium sulfite and sodium thiosulfate increase the stability of the plating gel. The gold concentration used in the previously discussed bath formulations was at 0.03 M (5.91 g/L). The gold concentration in the plating gel must be increased to make this a viable process for industry. The gold concentration was increased to 0.041 M (8 g/L). Gel 9 (Table 16) lists the formulation for this increased gold plating gel. This plating gel was stable throughout the plating process. Gel 8 (Table 16) was formulated with 0.08 M (15 g/L) gold concentration. Only the gold complex and reducing agent were increased in this formulation. All the other components were similar in concentration as Gel 9. This plating gel decomposed during plating at 50 °C within 5 minutes. Gel 22 (Table 17) was formulated with a gold concentration of 0.20 M (40 g/L). This plating gel turned black during plating and decomposed in the beaker at room temperature in less than one hour. It is evident that gels with increased gold concentration gel require more stabilizers than the gels with low gold concentration. The EDTA concentration was increased from 0.15 M to 0.2 M but did not stabilize the gel formulation. Gel 26 (Table 17) was formulated at 40 g/L gold concentration using phosphate buffer. This plating gel was stable for 2 days at room temperature. The increased 15stability of this plating gel could be from either the phosphate buffer or the additions of sodium thiosulfate and sodium sulfite to the plating gel. Therefore, Gel 16 (Table 17) was formulated using a citric acid buffer and sodium thiosulfate additions. The addition of sodium thiosulfate appears to have increased the room temperature stability of the plating gel. Gel 24 (Table 17) was formulated with increased reducing agent concentration and a new buffer (ammonium chloride), and stabilizers (sodium sulfite and ammonium thiosulfate). The ratio between the reducing agent and the gold complex previously was 1.67. This ratio is now 3.5 with the added sodium L-ascorbate. This gel still shows decomposition at the plating temperature; however, the gel is stable at roorn temperature for ~3 hours. Therefore, citric acid is not the only buffer that can be used without sacrificing room temperature stability. The phosphate and the ammonium chloride buffers in combination with the stabilizers result in room temperature stability for short periods of time. ‘The chemical additives discussed above did not keep the high gold concentration gel from decomposing during plating. Gel 38 (Table 17) was formulated using another stabilizer, 2-mercaptobenzimidazole. Kato et al”? used this stabilizer to extend the bath lifetime of their gold sodium sulfite plating bath. Gel 38 did not turn black after 3 hours of plating at 50°C. In conclusion, 2-mercaptobenzimidazole and sodium sulfite or thiosulfate complexes are stabilizers required to maintain the stability of the increased gold concentration plating gel at room temperature and during plating. The concentration of EDTA appears to have a minor influence on the stability of the high gold concentration gels. 163.4.5 Gold Plate Appearance as a Function of Gel Chemistry Some of the plating gel formulations listed in Tables 10-12 were used to evaluate the gold plate appearance. Table 18 and Table 19 summarize the plating gels from Tables 10-12 that were used to evaluate gold plate appearance. The gold plate appearance was evaluated after the deposition and rinsing steps. The plate color observations are summarized in Table 18 and Table 19. Morimoto et al”” state that there is an empirical knowledge which prevails that gold plating films having good function have an aesthetic Jemon yellow color tone. Gold plating films of poor tone are unacceptable as commercial products even if they are functional. The goal of this study was to identify the plating chemistry that provides a Jemon yellow color gold plate. Therefore, this section will not g0 into details about the microstructure and functionality of the gold plate (see Chapter 6). Table 18: Plating Gel Formulations for Evaluation of Plate Color (repeated from Table 10) Gel Formulation Nevibers (Units in Molarity) 718 [| 9 | 13 7 14 NajAu(S,03), 0.03_| 0.076 [ 0.051 | 0.03 [0.03 CoHNaO, 0.05 [0.127 [0.075 | 0.05 | 0.05 H3CoHs07 0.1 [0.105 [013 | - 0.2 NaOH 0.45 [0.425] 0526} - | 05 EDTA [0.033 [0.05 [0.062 | - : Na,S;03 = = = [0008 | - Propylene Glycol @wt%) | 35 | 35 | 35 | 35 | 35 HPMC (wt%) 6é[6]/6fe6l| 6 Gel Make-up: hot or cold | cold | cold | Cold | hot _| hot Bright | Dark Dark Plate Color ang o | ¥ | Blo where Y=Yellow, O=Orange, B=Brown The formulation of plating gels with increased gold concentration is a necessary goal for successful implementation of the Gel Plating Process. Extensive time was not spent in perfecting the color of the low gold concentration plates. However, the information leamed using the low gold concentration plating gels aided in the 1formulation of high gold concentration plating gels with a good resulting gold plate appearance. Table 19: Plating Gel Formulations for Evaluation of Plate Color (repeated from Tables 11 and 12) Gel Formulation Numbers 19 | 2] 2 | 28) 30 | 35 | 38 Na3Au(S,0;)2 [M] 02 [02 [02 102] 02 | 02 | 02 CgHNaO, [M] 07 | 033 | 033 [038] 07 | 10 | 07 ncaonm | = [ot = ft = | = | NH,CH30; [M] - - - a 05 | 05 - NH,CI[M] 03 | - ee 05 NaHiPO, [M] - [> joosp- fT - [- | - NaOH [M] 0.156 | 1.35 [0.075 [0.225] 028 |~- | 029 EDTA [M] k= | 01s [oor |= 5 NHEDTA [M] o4 | = > [03 | 03 [004] 03 Na,SO3[M] 02 | - [0143] 01 | 015 [0.15 | 0.15 NazS,03 (M] - - |0.081) - a 5 - (NHL)28:03 [M] or] - [ - [02] 02 | - | 02 CrH.N;S [104 M] - - = S - 75 | 30 Propylene Glycol (wt%) | ~ a --- | 5 5 = HEC (wt%) 3 5 | 5 | 3 {35 | 35 | 37 Bright | Dark | Dark ‘Bright | Bright | Bright Cee yo| a | «6 |°| vol} yo] o where Y=Yellow, O=Orange, B=Brown, G=Green Low gold concentration gels were formulated according to Table 18. Gel 13 was formulated with just the gold complex, reducing agent, and sodium thiosulfate as the stabilizing agent. It had a gold concentration of 4 g/L. This sample was then placed into Bthe custom built reactor at 60 °C for 20 minutes. Visible plating was observed in 8 minutes. The plate was very dark (almost brown) in color. Gel 14 was formulated using a citric acid buffering system without the sodium thiosulfate stabilizing agent. It had a gold concentration of 5.9 g/L. This gel was placed into the custom built reactor at 40 °C. This plate was dark orange in color. Gel 7 was formulated using the citric acid buffering system and EDTA additions, This gel had a gold concentration of 5.9 g/L. The resulting gold plate was a bright yellow-orange color that was brighter than the plate from Gel 14. Therefore, the EDTA additions did improve the appearance of the gold plate. Gel 9 was formulated using EDTA additions and the citric acid buffer with 8 g/L gold. Plating was obvious after 2 minutes in the reactor at 50 °C. The resulting gold plate was a yellow color. Therefore, this plating gel formulation is a good guide to aid in preparation of the increased gold concentration plating gels. Gel 8 was formulated using the same ratio’s of EDTA and citric acid as in Gel 9 but the gold concentration was increased to 15 g/L with a subsequent increase in the reducing agent concentration. The resulting plate was dark in color. High gold concentration plating gels were formulated according to Table 19. Gel 22 was formulated at 40 g/L gold concentration, EDTA and citric acid additives were used in this plating gel. The resulting gold plate was not bright in color. Gel 26 was formulated at 40 g/L gold concentration but with a phosphate buffer and sodium thiosulfate and sulfite additions. ‘The resulting gold plate was dark and greenish in color. The sodium complexes of EDTA and thiosulfate were replaced with ammonium complexes in Gel 28. The plate is more orange than green, but it is still darker in color than desired. 9The plate color improved drastically when the amount of ascorbic acid added to the plating gel was increased. The ratio between the reducing agent and the gold complex in Gel 19 is 3.5, previously it was 1.67. The resulting gold plate from Gel 19 is a bright yellow-orange color. The plate color was acceptable for the samples formulated with Gel 19; however, under the optical microscope there appeared to be clusters of gold deposited on the surface. Gel 38 was formulated using 2-mercaptobenzimidazole as a stabilizer. The resulting gold plate was shiny and bright orange in color. Ammonium acetate was substituted for ammonium chloride in Gel 30. This substitution was necessary because of the detrimental affects that chloride ions have on the corrosion of metal in humid environments. Gel 30 was the best gel for stability and plate color; however, the high ion concentration of this plating gel effects it behavior at the plating temperature. Specifically, this high ion concentration gel undergoes syneresis and expels liquid from the polymer network as discussed in section 3.4.3. Gel 35 summarizes the typical formulation for a plating gel that does not undergo syneresis at the plating temperature but still has good stability and plate color. Figure 17 illustrates the improvement in plate color when the ascorbic acid ~ gold thiosulfate ratio is increased. 80Ascorbic Acid : Gold Thiosulfate Ratio = 5 Gold Concentration = 40 g/L 1.5 x 10° M 2-mercaptobenzimidazole Ascorbic Acid : Gold Thiosulfate Ratio 17 Gold Concentration = 40 g/L ‘No 2-mercaptobenzimidazole Figure 17: Comparison of Plate Color as a Function of Ascorbic Acid Concentration 813.4.5.1 Discussion The ascorbic acid and the 2-mercaptobenzimidazole concentration was found to have the most significant influence on resulting plate color. Ascorbic acid concentration in the plating gel can decrease over time. The plating reaction will not occur at the desired efficiency if there is not enough ascorbic acid. Ascorbic acid can be oxidized through two different routes in electroless plating.® Homogenous decomposition of ascorbic acid can take place in the electroless plating bath from dissolved oxygen. Heterogenous decomposition of ascorbic acid can occur in the presence of a catalyst. Ideal operation of an electroless plating bath requires that heterogenous oxidation of the ascorbic acid takes place on the catalytic substrate. Sullivan et al ® studied the decomposition rate of ascorbic acid with various catalysts. They report that ascorbic acid in the presence of dissolved oxygen gave an 11 +0.5 % decrease in ascorbic acid concentration in two hours. The addition of a gold substrate does not effect consumption rate significantly. The presence of gold thiosulfate did increase the consumption rate of ascorbic acid to 18 +0.5 % in two hours. Experiments studying the homogenous oxidation of ascorbic acid confirm that the rate is psuedo first order with dissolved oxygen concentration assumed constant.’ The rate constant is 1.75 x 10° +3.184 x 10% (+20) min" for pH = 6.4 and 30 °C. Rate constants at pH 6 and 9 have been reported to be 1.4.x 10? min’ and 1.8 x 10? min’ respectively. This indicates that 5 % of ascorbic acid is consumed in 30 min. At higher temperature, the rate constant will increase but the amount of dissolved oxygen should decrease.** A plot of ascorbic acid consumed by dissolved oxygen as a function of time for different initial starting concentrations of ascorbic acid has been calculated using the first order reaction of ascorbic acid and the rate constant mentioned above (see Figure 18) This is assuming that the dissolved oxygen concentration is constant over time. It can be concluded from Figure 18 that the homogeneous oxidation of ascorbic acid continues in 22the presence of dissolved oxygen. Increased initial ascorbic acid concentration ensured that there was enough ascorbic acid available for heterogeneous oxidation at the catalytic substrate. Therefore, the plate color improved with the increase of initial ascorbic acid concentration because the plating reaction was operating efficiently. This homogeneous oxidation of ascorbic acid in the plating bath due to dissolved oxygen increases the chance that gold will homogeneously precipitate in the plating gel. Honma et al ® believe that the cause of decomposition of gold plating systems (especially the sulfite gold complex in acidic to neutral regimes) is due to the disproportionation reaction that takes place between the free gold (I) ions that accumulate at the reaction interface during deposition: 3Au*—>2Au+ Au? (3.6) Metallic gold and gold (III) ions are formed in the solution. The bath becomes turbid from metallic gold precipitation and finally decomposes. They tried different amines to complex with the Au(D ions in solution. They found that EDTA reduced the decomposition of the bath during make-up and operation but the plate appearance was poor. The stability only lasted for a few hours. The concentration of Au was only 0.006 M. Other stabilizing agents that they tried are listed in Table 20 with a summary of their results: 8Table 20: Stabilizing Agents to Complex with Au({) Ions in Solution ‘ppm ‘Stabilizer 0.1 1.0 10 700 [ KafFe(CN)e] A A B B Ni(CN); « 2KCN A A A A 2,2'-bipyridine B B A A Cupferron B A A A 2-mercaptobenzothiazole _| A A Cc Cc Neocuproin B B A A [ Cytosine B B A A Phenanthroline B B A A ‘A: Plating proceeds without decomposition B: Decomposition after plating potty plating or deplating Honma et al *° concluded that these stabilizers improve the bath stability because they form complex ions with gold (1) ions and supress the disproportionation reaction of free gold (1) ions. The addition of 2-mercaptobenzimidazole to the plating gel formulation (Gel 35) did result in a bright yellow-orange gold plate. In addition, a plating gel formulation with this additive did not decompose during plating (section 3.4.4, Gel 38). The 2-mercaptobenzimidazole must complex with the gold (1) ions in solution and prevent rapid decomposition of the plating gel. As a result, the plate color is improved because the plating reaction is operating in a controlled manner as opposed to uncontrolled plating. 84ono Figure 18: Ascorbic Acid Consumption as a Function of Time 85 rr3.5 Conclusions ‘A summary of the main components evaluated when developing the plating gel is presented in Figure 19. The three main issues to consider when formulating the plating gel were thickening the increased gold concentration plating solution, stabilizing the plating gel, and improving the plate color to a yellow-gold color. Figure 19 illustrated that the HEC thickening agent was able to thicken the 40 g/L gold plating solution. The stabilizer, 2-mercaptobenzimidazole, increased the stability of the plating gel at the plating temperature. An increase in ascorbic acid concentration caused a significant improvement in the plate color. A final formulation was developed from the results of the studies presented in sections 3.4.2 to 3.4.5. This formulation is Gel 35 (see Table 12). This plating gel formulation thickens properly, contains stabilizers to prevent decomposition of the plating gel over a specific time period, eliminates syneresis of the plating gel at the plating temperature, and produces a good quality, good color gold plate. 86Gel Formulation Stability at Room Temp. and Plating Increase Gold Yellow-Gold Concentration Plate Color Temp. | ¥ I . Added Increase Use HEC TRE RET 2-mercaptobenzimidazole Ascorbic Acid 40 g/L Gold Concentration Ratio of Ascorbic Acid : Gold Thiosulfate = 3.5 RT Stability > 12hrs 60 °C stability for 2.5 hours Figure 19: Summary of Plating Gel Formulation 87Chapter 4: PLATING GEL PERFORMANCE DURING PROCESSING 4.1 Screen Printing of Plating Gel ‘The “printability” of the plating gel can be varied by using different types and concentrations of polymeric thickening agents, humectants and surfactants. Printability of the plating gel was evaluated by performance during silk screen printing. Silk screen printing of the plating gel requires that it have a low enough yield stress to allow the gel to flow during printing but high enough to retain the shape of the gel print at the plating temperature. It also must wet the surface to allow adhesion to the substrate but not to the screen during snap-off. ‘A hydroxypropyl methylcellulose (HPMC) polymer (see section 3.3) was determined to be an effective thickening agent for the high pH, low gold concentration plating bath. Amdt™ provides a detailed explanation of the factors attributed to successful screen printing with this thickening agent. Hydroxyethyl cellulose (HEC) was determined to be a good thickening agent for formulating the high gold concentration, non-cyanide plating gels; however, its performance during screen printing is not as good as that of the HPMC. The following sections will explain in detail the screen printing behavior of the HEC thickening agent. 4.1.1 Screen Printing Parameters The plating gel can be printed onto the substrate by a conventional screen printer in a pattern defined by a stenciled emulsion or a metal stencil. The squeegee pressure, loading speed, and printing speeds can be adjusted according to known procedures to 88optimize the print thickness and quality. The wet deposit thickness can vary from less than 50 microns to greater than 500 microns. The stenciled emulsion printing screen is generally made of two materials that have been laminated together. The first material is a metal mesh that is stretched onto a frame. The second material is a polymeric emulsion that defines the pattern and the thickness of the wet print. The metal mesh is defined by a mesh size, open area (%), type of material, and tension. The mesh size and the open area are responsible for the largest variations in the print quality. For the particular gels tested, a finer mesh (200+) with an open area of only 46% resulted in an unacceptable print; whereas, an 80 mesh (coarser) screen with an open area of 71% resulted in a good print. The metal stencil can be made by either wet etching the correct thickness of metal or by laser cutting the metal into the desired pattern. This type of metal stencil screen limits the intricacy of the patterns that can be replicated. However, the metal stencil screen has 100% open area for the open features. The plating gel can easily flow through the metal stencil screen. The thickness of the printed gel is a function of the metal stencil thickness, squeegee pressure, and snap off distance (distance between the substrate and the bottom of the screen). The squeegee durometer also plays a significant role during screen printing. A lower durometer squeegee allows more material to be pushed through the mesh emulsion screen. A higher durometer squeegee works well with the metal stencil screen. Both the mesh/emulsion and metal stencil screen were used to evaluate the printing of the HEC plating gel. The metal stencil screen was found to be simpler to use for the particular applications described in this thesis. 894.1.2 Effect of Surface Tension on the Screen Printing of the Plating Gel The plating gel formulated with HPMC screen printed better than the plating gel formulated with HEC. The HEC plating gel exhibited a stringing behavior and appeared to stick to the screen more than the substrate surface. The surface tensions of the two thickening agents are compared in Table 21. The surface tension of HPMC is lower than that of HEC. Table 21: Surface Tension of Thickening Agents from Product Literature Trade Name ‘Chemical Name ‘Surface Tension in 1.0% sol’n Cellosize™ QP- _ Hydroxyethyl Cellulose (HEC) 66.3 mN/m 100MH Methocel™ K4M —— Hydroxypropyl Methylcellulose 43-55 mN/m (HPMC) A surfactant was added to the plating gel formulated with HEC to determine if the surface tension difference plays a major role in the different printing behaviors of the two thickening agents. The surfactant used was Triton™ x-100 (Union Carbide). It is an octylphenoxypolyethoxy ethanol non-ionic surfactant from the chemical family of alkyl aryl polyether alcohol (CsHi7CsH.(OCH,CH2)xOH) with a molecular weight of 628. Table 22 summarizes the printing trials that were used to evaluate the effect that surfactant and thickening agent concentration have on the printing behavior. The influence of propylene glycol will be discussed in the next section. Table 22: Summary of Printing Trials Trial #1 ‘Trial #2 Trial #3 ‘Trial #4 Trial #5 [Surfactant] M6 x 10" 6x10" 6x10 6x10" 6x10" % Propylene Glycol 0 35 0 10 5 % Cellosize™ QP- 100MH 2 2 3 3 3 Print Performance fluid = More brittle += Good More brittle. Great thanelastic Print ___than elastic __PrintThe effect of polymer and surfactant concentration on the screen printing behavior of the plating gel is described by Figure 20. Good screen printing behavior is obtained in a specific window of surfactant and thickening agent concentration. The schematic shown in Figure 21 illustrates the role that the surfactant plays during screen printing. The gel wets the substrate surface better when the surfactant is used to reduce the solid-liquid interfacial energy (ys). 4.1.3 Effect of Humectant Additions on Screen Printing of Plating Gel ‘Arndt found that humectant additions of propylene glycol improved the printing performance of the HPMC plating gel. However. the HEC plating gel behaves differently. For example, Table 22 shows the results of printing trials conducted with the HEC plating gel containing propylene glycol additions. The plating gel behaved in a brittle manner and did not flow in the viscoelastic manner observed from the plating gels without propylene glycol additions. The HEC product literature states that glycols can act as a plasticizer for HEC products when added in concentrations of 5 to 30%." This may explain why plating gels with a concentrations of propylene glycol grester than 5 % exhibited this brittle behavior during screen printing. The humectant was initially added to reduce the vapor pressure of the plating gel to prevent drying during deposition. However, the results in Figure 16 shows that there is not much of a difference in evaporation rate between the low electrolyte gel with or without propylene glycol. Therefore, for screen printing purposes, it is advantageous to eliminate propylene glycol from the HEC plating ge! formulation. 910.020 Viscosity ors: o.ot0: % Surfactant 0.005: Gel does not go through screen 0.000 20 25 30 35 40 % Thickening Agent ----> Printing Performance as a Function of Polymer & Surfactant Concentration 92Liquid / Gel “ eS Old: Young’s equation for liquid-vapor-solid junction (8 > 0°) : Ysv ~YSL YLV ‘Yev> Yet» YLv = interfacial tensions © = contact angle (wetting = 6 < 90°) cosO= Figure 21: Role of Surfactant in Screen Printing 934.2 Selective Behavior of Gel Plating The selective properties of the Gel Plating Process make this an attractive process for plating substrates that have electrically isolated features, especially if these features require a thicker gold metallization than other areas of the substrate, A gel print can be placed directly over catalytic features of the substrate or overprinted to include both catalytic and non-catalytic areas of the substrate. The resolution needed to print exactly over certain metallization patterns cannot be obtained when screen printing down a 500 micron gel print. Specific pattern designs that have closely spaced features require that gel be printed over the catalytic features and the non-catalytic areas that separate the features. An example of closely spaced, electrically isolated lines is shown in Figure 22. The bulk substrate is AIN. The surface is metallized in a pattern that consists of co-planar transmission lines. These lines are basically three metallized features that are separated by thin lines of the AIN substrate. The gel print is shown Figure 22(b). The gel is printed over the three metallized lines by one continuous rectangle of gel. The co-p'anar transmisson lines require a thick gold plate (2 microns) to operate correctly; whereas, the rest of the metallization pattern requires a only a thin layer of gold for corrosion resistance. ‘The gel must be printed over all three of the co-planar transmission lines by one rectangular gel print. The plating gel will be printed on the non-catalytic areas separating these metal lines. No plating can occur between these metal lines or the feature will be shorted and the function of the lines ruined. Electroless baths are thermodynamically unstable and overplating between closely spaced lines can be a problem with these baths. In traditional electroless gold plating baths, the reactants can be thought of as infinite for one substrate; however, in the gel plating process the reactants are limited by the volume of the gel print. This will limit the severity of overplating that can occur for each feature. 94In theory, electroless plating should only occur on a catalytic surface (that is one in which the reducing agent can be oxidized to give off the electrons needed for plating). However, extraneous plating can result on non-catalytic features of a substrate because of bath instability or substrate contamination. K. Morimoto et al *” state that neutral baths tend to deposit metal at non-intended places causing stains or blur. This phenomena is more noticeable when plating a metal feature on a ceramic base. The resulting metal film extends from the metal portion to the surrounding ceramic base surface. Morimoto et al >” used a non-ionic surfactant and/ or a non-ionic polymer as a stabilizer to reduce the plating spread. They also state that the above mentioned stabilizer may aid in preventing the plating bath from decomposing. Their explanations are as follows: 1) The stabilizer is adsorbed to a gold plating film or gold colloid particles resulting from solution decomposition. The plating proceeds on the plating film, but not on the gold colloid particles, thereby preventing solution decomposition and plating spread and 2) The stabilizer is adsorbed to a ceramic portion of the substrate to prevent gold decomposition on the ceramic but allow plating on the plating film. Extraneous gold was deposited on the AIN surface when attempting to selectively plate gold using the Gel Plating Process (see Figure 23). The formulation of the plating gel used for plating the part in Figure 23 is listed in Table 23, along with the resulting formulation that did not have gold on the AIN surface. The plating gel used for the part in Figure 23 did not contain a surfactant and contained a low concentration of 2-mercaptobenzimidazole. Previously plated samples that contained a surfactant in the plating gel formulation did not exhibit this overplating behavior. ‘The following sections describe the experiments performed to minimize the overplating (or extraneous plating) of gold on AIN. 95Co planar Transmission lines AIN Substrate Figure 22: EPOC substrate 96Table 23: Formulation with AIN decomposition ‘Chemicals Gold on AIN [M] No Gold on AIN [M] NaAu(S,05) 02 02 CeHpNaOg 1.0 07 ‘Ammonium Acetate 0.5 05 Di-ammonium EDTA’ 0.046 0.046 NaSO3 0.15 0.15 2-mercaptobenzimidazole 75x10" 75x 107 ‘Surfactant - 2.7 x 107 Hydroxyethyl Cellulose 35 wi% 3.5 wi% . 4.2.1. Experimental Procedure Plating’gels were formulated with similar composition to Table 23 except that the 2-mercaptobenzimidazole, EDTA, propylene glycol, and surfactant concentrations were varied according to Table 24. The surfactant used is a non-ionic surfactant from the family of alkyl aryl polyether alcohol. Specifically, its trade name is Triton x-100™ (Union Carbide) and the chemical name is octylphenoxypolyethoxy ethanol. It is assumed since it is non-ionic in nature that the surfactant does not interfere with the ions in solution. Experiments were performed to determine the chemical additives in the plating gel that reduced the extraneous gold plating that can occur during electroless deposition. A substrate similar to the one used in Figure 23 was used to test the plating gel formulations. Small droplets of varying formulation (see Table 24) were placed on the substrate so that part of the droplet covered the metal pattern and part covered the AIN surface. The sample was placed into the Blue M humidity oven at 50-60 °C for 1 hour at 97% humidity. The droplets of gel were removed by rinsing with DI water. Pictures of each step were taken to compare the amount of gold decomposed on the surface of the AIN as a function of bath composition. 7rrneoeend oo a (a) before gel plati Extraneous Auon AIN ~ (b) after gel plating 1 layer Figure 23: Gold Decomposition on AIN 98Table 24: Gel Formulations for Overplating Experiments Gel [2mercapto- [EDTA] [Surfactant] %Propylene Observed benzimidazole} Glycol B 15x 107 0.046 0 0 Gold c 75x 10% 0.046 0 9 Gold E 75x 10° 0.046 0 0 Gold G 15x 10° 03 6x10? 0 No gold H 75x 10% 03 0 0 Gold I 75x 10% 0.046 6x 10° 5 No Gold J 75x 10° 0.046 6x 10? 0 No Gold K 7.5.x 104 0.046 2.7.x 107 0 No Gold 4.2.2 Results and Observations The results from the overplating experiments described above are shown in Figure 24, Alll gel formulations without the surfactant demonstrate extraneous plating of gold on the AIN surface. It also appears that as the stabilizer decreases the amount of plated gold increases, This finding may be attributed to an increase in the plating rate with a decrease in stabilizer concentration (see Chapter 5). Gels ‘G’, ‘I’, and ‘J all contain a high concentration of surfactant solution. This high concentration of surfactant solution does decrease the plating rate (see Chapter 5). Therefore, the ‘no gold’ observation could be attributed to a low plating rate, Gel ‘K’ has a low concentration of surfactant that does not decrease the plating rate (see Chapter 5). No gold is plated on the AIN when this plating gel is used. The resulting plate from Gel ‘K’ can be compared to the plates from Gel ‘C’ or Gel ‘H’ which have similar stabilizer concentrations. Gold plating on the AIN is definitely observed with Gel ‘C’ and ‘H’; whereas, no gold is plated on the AIN when Gel ‘K’ is used. 99Therefore, the surfactant helps to prevent extraneous gold plating on the AIN surface while still allowing selective plating on the desired catalytic surfaces. A new substrate was plated using the formulation listed in Table 23. This substrate is shown in Figure 25. Some gold was plated onto the AIN after a 3 hour plating time. However, this gold was removed from the AIN surface by ultrasonicating the substrate in DI water for 5 minutes or less. Therefore, it appears that the surfactant adsorbs onto the ceramic substrate and reduces the adhesion of the extraneously plated gold to the ceramic surface. 100Figure 24: Results from Overplating Experiment 101Figure 25: No Gold on AIN EPOC 1024.3 Behavior of Plating Gel at Plating Temperature The environment surrounding the plating gel during the deposition step is critical for proper plating through the gel medium. The gel is an aqueous based system that can experience significant drying at the plating temperature. The electroless plating bath chemistry is very sensitive to slight changes in composition. Significant loss of water through evaporation will alter the gel chemistry and change the quality of the resulting plate In addition, the gel formulation contains a high electrolyte concentration and a polymeric thickening agent. The loss of water from the plating gel can cause syneresis of the gel. The polymeric Thickening agent agglomerates into the center and a liquid is expelled (see Figure 12). This will cause a non-uniform gold plate as well as a loss of resolution at the plating temperature. Therefore, the deposition environment must be at a high humidity in order to keep the gel from drying during plating. The plating gel has an evaporation rate of approximately 6.05 x 10° g/(cm* sec) at room temperature. Water has an evaporation rate of 1 x 10° g/(cm? sec). The high electrolyte concentration and the polymeric thickening agent additions in the plating gel lower the vapor pressure of the aqueous solution, which lowers the evaporation rate of the plating gel. 4.3.1 Plating in the Blue M Humidity Oven ‘The plating step was performed in a commercial humidity chamber, Blue M. This ‘oven is 6 ft’ in the inner chamber. The heating rate of the sample, the humidity profile, and the humidity set point during plating were determined. ‘The heating rate of the samples were determined by attaching thermocouples to various substrates used during plating experiments and monitoring the temperatur~ change. The samples were placed onto a setter inside the humidity oven. The setter was equilibrated to the temperature of the oven. Figure 26 illustrates the heating rate of a glass slide, a thick AIN substrate (EPOC), and a thin AIN substrate. 103‘The humidity profile was recorded after the sample was placed into the oven and the door was shut. The humidity was first set to the desired point. Then the door was opened to place the sample into the oven. The humidity readings from the dry bulb sensor in the oven were recorded as a function of time. The humidity profile is shown in Figure 27. The humidity drops to a low reading when the oven door is opened. After the door is shut, the humidity overshoots the set point of 85% relative humidity. The oval inserts in Figure 27 illustrate how the shape of the gel print changes as the humidity profile changes. Ovals #1 through #4 are cross-sectional schematics of the gel print. PID controls were used to try and compensate for this behavior, but the attempt was not successful. Figure 28 shows the attempts made to correct the humidity profile with PID control. ‘The sample loading procedure into the humidity oven was modified to resolve the humidity overshoot problem. First, a cover was placed over the oven door opening that had a small opening for the sample to be inserted. This prevented the humidity from escaping when the oven door was opened. In addition, the sample was loaded while the oven was in manual mode. The above modifications helped to minimize the humidity overshoot problem as can be seen in Figure 29. Figure 29 shows four different humidity profiles for four different humidity set points. The humidity only decrease slightly when the oven door is opened and the sample inserted The humidity set point required for plating was determined. A low ion concentration plating gel (see chapter 3) was screen printed onto a substrate and placed into the oven at varying humidity set points for 10 minutes. The weight change of the plating gel during the plating process was determined. The results of this study can be seen in Figure 30. The evaporation rate of the plating gel after 10 minutes in the Blue M humidity oven at a temperature of 60 °C is plotted in Figure 30 as a function of the humidity set point. The evaporation rate of the gel print is very close to zero when the 104relative humidity is set to 95%. Photographs of each gel print before and after the humidity oven are shown in Figure 31 for these humidity set points. Picture (b) illustrates the syneresis of the gel print at a relative humidity of 75% for 10 minutes. Syneresis is still observed at a set point of 85% relative humidity; however, picture (4) shows that less water has been expelled from the gel print. Picture (f) illustrates that 95% relative humidity prevents the gel from drying at the plating temperature and the plating gel does not undergo syneresis. Syneresis behavior was discussed in detail in Chapter 3. Therefore, a humidity set-point of 95% or higher is necessary to prevent the plating gel from drying during deposition. 10555 50: Gea @ @ @ = e . le © Thin AIN substrate © yg Lee + EPOC Tile e je & * Glass slide 5 roa 40 } | 2 0 ya0,” 240” 980” 480 600 720 eH 860 Time (sec) Figure 26: Heating Rate in Blue M 106% Humid. © T=55C,Hum=85% a © 800 600 8001200” 1500 1800 2100 2400 2700 3000 Time(sec) Overshoot Behavior in Blue M 107100: = Initial Conditions © Decrease Gain ° No Gain = Increase Gain 80 Qea aadbaa 70: % Humid 60 50 40 ° 300 600 900 1200» 1500-1800 Time(sec) ‘Temperature = 50 °C, Humidity = 75 % Figure 28: Attempts with PID Control to Minimize Overshoot 10875% 75% 85% 90% 95% 95% ‘%Humidity (0 60 120. 180 240 300 360 420 480 40 600 650 720 Time (sec) Humidity Profiles for Four Different Humidity Set-points: 75%, 85%, 90%, 95% Figure 29: Humidity Profile from Controlled Oven 109Evaporation Rate % Humidity Set Point Figure 30: Identification of Correct Humidity Set-point 0(a) Gel Print Before n aaa (@ Gel Print (e) Gel Print Before (f) Gel Print After, 95% RH Figure 31: Photographs of Syneresis Behavior in Oven Mt4.4 Rheology measurements 4.4.1. Experimental Procedure The theology of various plating gels was studied using 2 DSR4000 rheometer from Paar Physica (Eatontown, NJ). Tests were performed in rotation mode to obtain the viscoelastic properties of the plating gel. The plating gel was characterized at 25 °C and 60 °C (the highest plating temperature used). ‘Three plating gels were formulated according to Table 25: control gel, gold gel, and ascorbic gel. A temperature sweep, amplitude sweep, time sweep, and frequency sweep were performed on each plating gel at 25 °C and 60 °C. The parameters for each test are listed in Table 26, A 50 mm diameter plate spindle was used. A 1 mm gap between the base plate and top plate spindle was set for each test. A low viscosity mineral oil was placed around the edges of the gap to prevent the sample from drying. Table 25: Plating Gel Formulations for Rheology Studies Control Gel Gold Gel Ascorbic Gel ‘Na3Au(S20s)2 [M] - - 0.2 CHNaQg [M] - 07 - NHC;H30, [M] 0s 0.5 05 NHEDTA [M] - 0.00828 0.00828 NazSO3{M] - 0.15 0.15 CyHeN2S [M] - 75x10" 7.5.x 104 HEC wt% 3.5 wt% 3.5 wt% 3.5 wt% 12Table 26: Parameters for Rheology Tests Time Sweep Amplitude Frequency Temp. Sweep Sweep Sweep % strain 3 0.1 to 150 5 3 Frequency (Hz) 1 i 0.03 to 100 1 Temperature (°C) __25 & 60 25 & 60 25 & 60 25 t0 60 4.4.2 Results The amplitude sweep was performed on each gel formulation at 25 °C and 60 The storage modulus (G’) and loss modulus (G”) are plotted as a function of % strain in Figure 32, The purpose of this test is to determine the %strain range that has a linear behavior for each gel. The value of strain used in subsequent tests will be chosen from this plot. At 25°C, all gel formulations demonstrate a fairly large linear region from about 2 to 20% strain. The storage modulus (G’) begins to decrease around 20 to 30 % strain. At 60°C, the gold and ascorbic gels demonstrate the same type of behavior and similar linear region as seen at 25°C. However, the storage modulus (G") for the control gel at 60°C starts to decrease around a strain of 10 %. It is also interesting to note that the loss modulus (G"} is higher than the storage modulus (G’) for the control gel at 60°C; whereas, the opposite is observed with the other gel formulations. More evidence of this behavior will be seen in later tests. A strain value of 5 % was chosen from these plots for use in later tests. The rest viscosity of the plating gel can be determined from the linear region of the amplitude sweep plot. The complex viscosity I7*l is plotted as a function of %strain in Figure 33 for both 25 and 60°C. All gel formulations have a viscosity of approximately 70,000 cP at 25 °C. The viscosity decreases to approximately 38,000 cP for the gold and ascorbic gels at 60 °C. The viscosity of the control gel at 60 °C is an even lower value of ~ 23,000 cP. 13A time sweep was performed for each gel formulation at 25°C and 60°C. This test was used to determine the amount of time the gel must settle after sample loading and before running the test. G’ and G” are plotted as a function of time in Figure 34 for (a) 25°C and (b) 60°C. All three gel formulations show very linear behavior with time at 25°C. There appears to be no need for waiting after sample loading and before running the test. The gold and ascorbic gels are both linear over time at 60 °C. However, the control gel does show a decrease in both the G’ and G” modulii with time. G’ and G” cross after about 20 minutes at 60 °C. This is another indication that the control gel changes structure at 60 °C. Therefore, the control gel was left at 60 °C for over 20 minutes before testing. A temperature sweep from 25 °C and 60 °C was performed for each gel formulation. The complex modulus (G*) is plotted as a function of temperature for the three gel formulations in Figure 35. There is a linear decrease in the value of G* over this temperature range. This is the expected behavior. A frequency sweep was performed for each gel formulation at 25 °C and 60 °C. G’ and G” are plotted as a function of angular frequency in Figure 36. All three gel formulations behave similarly at 25 °C. However, the control gel behaves in a different manner than the gold and ascorbic gel at 60 °C. The next section will discuss the conclusions of the results presented above. 14100) pEEERSELESUEEEESEE EY ease — (Gott Gen +6 (Gold Ge) s-6 (Ascabie Ged s-*+6 Ascorbic Get (Contra Ge (Contra Ge) No + *e 10 ‘strain > (@) Amplitude Sweep at 25 °C , PPerteaeon Woo ae teastoen 6 (aso ot (sexe Gob “w-G (Conti 6" (Conteh a = ao (St, 33 Oe ree TN i “ Pr 52 ge er eanain (b) Amplitude Sweep at 60°C Figure 32: Amplitude Sweep for 3 Gel Formulations at (a) 25°C and (b) 60°C 151000 1000000, ——mn'l Gaid Gel T=25 C = in" Ascorbic Gel T=25 C g i Contol Gel T=2s © a -4--In'l Gold Gel T=60 C 9 2 In" Ascorbic Gel T=60 C 3 8 --¥--In*l Control Gel T=60 C g S 100 100000 < & 2 8 ERIE 3 = oe OD Ey 10 10000 % strain Figure 33: Complex Viscosity as a Function of % Strain at 25 and 60°C 6tg eseeeeee eeeencns 33 3 soo a3 fre a —eteesen gs ome ont ae ‘Shece ot ° ~G' (Contra! Gat) Geom ob ws s 10 18 20 25 Time (nin « (a) Time Sweep at 25 °C 1000 60 [AS raawcen Fne-fasee) pose sh 33 Vesna) “© (Conve et) Fo consa en Babin aeeeenepeperepn gc SE Storage Modulus (Pa) G* Loss Modulus (Pa) 8 10 . 1 me ww ww ‘Time (min) (b) Time Sweep at 60°C Figure 34: Time Sweep for 3 Gel Formulations at 25 °C and 60 71000 |r (Gold Gel) --G* (Ascorbic Gel) Gt (Control Gel) G* Complex Modulus (Pa) 3 2 30 40 30 6 Temperature (C) Figure 35: Temperature Sweep for 3 Gel Formulations 118G' Storage Modulus (Pa) G* Loss Modulus (Pa) 8 1000. Storage Modulus (Pa) G* Loss Modulus (Pa) a seat frease [8 (ao1d cop [6 (Geta cen 5-6! (Ascobie Ge} 10-6" (Aseotle Gad (Convo! Ga) mG" (Ascoie Ge) 7 By 7 ‘Angular Frequency (a) Frequency Sweep at 25 °C ‘6'(Contel Ged +6"(Contol Gop 7 0 7 ‘Angular Frequency (b) Frequency Sweep at 60 °C Figure 36: Frequency Sweep for 3 Gel Formulations at (a) 25 °C and (b) 60°C 94.4.3 Discussion The sinusoidal deformation that the sample experiences in the rotation experiments discussed above produces a response than can be described by two dynamic modulii: the storage modulus (G") and the loss modulus (G’). G" is the in-phase response or elastic modulus. G” is the out-of-phase response or viscous modulus.” The behavior of G’ and G” as a function of angular frequency can be used to describe the structure of the sample. A highly cross-linked material will have a flat slope for G’ over the angular frequency range. G” is much lower than G’ for a highly cross-linked material. The slope of G’ increases as the amount of cross-linking decreases. The results in Figure 36 indicate that at 25 °C the structure of all three gel formulations (control, gold, and ascorbic) are similar. At 60 °C, the slope of G’ for the control gel is higher than the slope of G’ for the ascorbic and gold gels. Therefore, the ascorbic and gold gels have a higher degree of cross-linking at 60 °C than the control gel. The control gel was formulated with only the buffer (ammonium acetate) and the thickening agent. The ascorbic and gold gels were highly concentrated with the components listed in Table 25. Therefore, it appears that the gels formulated with more components maintain their structure better at 60 °C than the gel with only the buffer. ‘There is an interaction between the components in the plating gel and the thickening agent that increases the cross-linked nature of the gel at 60 °C. 4.5 Conclusions Surfactant additions aid with printing and selective deposition from the plating gel. The surfactant reduces the surface tension of the plating gel which enables it to wet the substrate surface. The surfactant also adsorbs onto the non-catalytic surface that the plating gel has been over printed onto and reduces the adhesion of gold overplating. 120Understanding the processing parameters during the deposition step in the humid environment is critical for maintaining the gel structure during deposition. The gel structure must be maintained to produce a uniform plate thickness from the Gel Plating Process. The correct humidity set point and sample loading procedures have been identified to minimize gel drying or swelling during the deposition step. Rheology measurements were used to determine the viscoelastic properties of the plating gel with and without the high concentration of ions. It was determined that the structure of the plating gel at 60 °C demonstrated more cross-linking with the high ion concentration formulatiag (gold and ascorbic gels) as opposed to the low ion concentration formulation (control gel). The gold and ascorbic acid plating gel formulations were stable over the temperature range of interest (25 to 60 °C). The rest viscosity of the plating gel is ~70,000 cP at 25 °C and ~38,000 cP at 60 °C. 121Chapter 5: PLATING RATE KINETICS 5.1 Introduction The plating kinetics for the Gel Plating Process will be evaluated in this chapter. First, a literature review of the methods used to evaluate the plating kinetics of electroless plating baths will be presented. Most of the baths evaluated in the literature are low metal concentration baths with a low viscosity. The plating gel is a high gold concentration, high viscosity system. In addition, the plating gel is a complex chemistry that contains many additives. In literature, the plating systems evaluated are usually simple systems that do not contain many additives. ‘Therefore, the techniques used in literature to evaluate the plating kinetics of plating baths are not directly applicable to evaluation of the plating gel. Tite mechanisms of plating through the plating gel will be evaluated by measuring the plating thickness as a function of time for various experimental conditions. This information will be compared to models for diffusion control, reaction control, and diffusion with a chemical reaction to determine the rate controlling steps involved in the Gel Plating Process. 5.2 Literature Review of Plating Rate Kinetics for Plating Baths Electroanalytical techniques are used to evaluate the plating kinetics of electroless plating baths throughout the majority of the literature surveyed. The techniques used are based upon the mixed potential theory. Paunovic* states that the mixed potential theory is based on the fact that the partial anodic and cathodic reactions take place simultaneously at the catalytic surface. Each reaction is trying to establish its own equilibrium with its own equilibrium potential. A steady state is reached which has a compromised equilibrium potential called the steady state mixed potential (Emp). A general description of the reactions taking place at the catalytic surface is shown in 122equation (5.1). The individual anodic and cathodic plating reactions are shown in equations (5.2) and (5.3), respectively. The reducing agent acts as the electron donor and the metal as the eletron acceptor if the equilibrium potential (E.,) of the reducing agent or reduction reaction is more negative than the cathodic or metal reaction.’. The characteristics of the steady state mixed potential are described below:*! 1) Both redox systems (partial anodic and cathodic reactions) are removed from their equilibrium potentials for a specific overpotential (11) (see eq. 5.4) 2) A net electrochemical reaction occurs in each redox system since both the partial cathodic (eq. 5.2) and partial anodic (eq. 5.3) reactions are removed from their equilibrium by establishment of the mixed potential. 3) The condition for steady state is that the rate of reduction for M**, the cathodic current density iy, 1s equal to the rate of oxidation of the reducing agent Red, the anodic current density izes. (see eq. 5.5) 4) A system at the steady state mixed potential is not at equilibrium since a net overall reaction occurs. Therefore, the change of the free energy is not zero, which is the requirement for the thermodynamic equilibrium. M Evie t+ REA setui¢n EL M panic + OXsetuion (5-1) lution M Sion + 0” Sete Mf (cathodic partial reaction) (5.2) hon enlace Red oiyjon AE OXsetuion +E” (anodic partial reaction) (5.3) Hy = Emp — Era (5.4) Hy = Ene ~ Eeyxes ‘cote dpontin = (la eng = Creat ene (8.5) 123,Therefore, according to the mixed potential theory the plating rate and mixed potential can be determined by evaluating the catholyte and anolyte individually. Bindra et al” used the mixed potential theory to evaluate the kinetics in a copper electroless plating bath. They tested the applicability of the mixed potential theory to the study of electroless plating solutions by comparing the results obtained when using the complete electroless plating bath to the results obtained when the catholyte and the anolyte were studied separately. They concluded that studying the catholyte and anolyte separately and using the mixed potential theory to determine the electroless plating rate is not valid because the two partial reactions are not independent of each other. Electrons released from the anodic partial reaction are consumed by the cathodic partial reaction, which also happens to be the metal deposition reaction. The overall rate of the electroless plating process is therefore governed by the slower of the two partial reactions. Bindra et af” used a technique that allows determination of the polarization curve for one of the partial reactions during electroless plating while the other partial reaction is occurring simultaneously. This technique is based on the classification of electroless plating processes according to their overall mechanisms. They determined that the cathodic partial reaction is entirely diffusion controlled and the anodic partial reaction is, entirely reaction rate controlled. The overall mechanism of plating is determined by the slower of the two partial reactions. They state that the cathodic partial reaction is the rate determining step for any electroless plating of metal. Therefore, diffusion control is the ‘overall mechanism of the electroless plating reaction. Sullivan et al‘? limit the use of studying the catholyte and anolyte separately to determining the mixed potential and predicating deposition rate of the electroless bath. They also state that the oxidation of the reducing agent and the reduction of the gold ‘complex must not interfere with each other when occurring simultaneously. Therefore, 124electroanaytical techniques which study the anodic and cathodic reaction separately should be used within the confines of the limitations stated above. Pauxiovic’’ describes in detail the partial cathodic reaction occuring in the electroless plating bath. The formation of the electroactive species (M"*) by dissocation of the metal complex /ML,}*”” (where p is the charge of the ligand L, z the charge of the noncomplexed metal ion, and (z+xp) the charge of the complexed metal ion) is shown in equation (5.6). Equation (5.7) illustrates the charge transfer from the catalytic surface to the electroactive species M'*. The first charge transfer is the rate determining step (RDS) as shown in equation 5.8. Paunovic’ shows that there is a definite correlation between the rate of dissociation of the complex and the rate of metal deposition. The concentration of the metal ions and ligands, pH, and the type and concentration of additives effect the kinetics of the partial cathodic reaction. [ML aM“ +x1? (5.6) M* +2 3M 6D M*+ e& 8 yyMey (5.8) The following 2-step general mechanism for the partial anodic reaction was proposed by Van den Meerakker.*® Equation 5.9 illustrates the formation of the electroactive species, which consists of dehyrogenation or breaking the R-H bond from the reducing agent. The charge transfer from the hydroxyl ion is shown in equation 5.10. The catalytic properties of the substrate, pH, and plating bath additives influence the oxidation rate of the reducing agent. The effect of pH is direct and significant for the reducing agent requires OH ions in the charge transfer step.*" RH See Rr easesecerttee) +H (5.9) 125R+OH~—>ROH+e — (5.10) Haruyama et al“ defined the the electrochemical conditions for electroless deposition to take place as: 1) the oxidation potential of the reductant should be less noble than the reversible potential of the metal to be plated, and 2) the metal should have enough catalytic activity for the anodic oxidation to take place at a reasonable rate. The first condition can be determined through thermodynamic consideration. The second condition can be harder to predict. The catalytic activity of the substrate plays a significant role in electroless plating. ‘There are many characteristics of electroless plating which suggest a catalytic effect:* 1) Electroless deposition occurs selectively on an active surface. 2) A catalyzing procedure is necessary for electroless deposition taking place on a non- active surface such as plastics and ceramucs. 3) There is an induction period before deposition occurs. 4) A limited variety of hydrogen-containing reductants are used in electroless process, e.g., hypophosite, formaldehyde, borohydride, dialkylamine borane, and hydrazine. 5) Specific combination of metal and reductant is necessary for electroless plating to take place. 6) Electroless deposition usually accompanies hydrogen evolution, the rate of which is not directly related to that of metal deposition. The hydrogen evolved orginates mainly from the reductant molecule. 7) The poison for hydrogenation catalysts such as thiourea and mercaptobenzothiazole functions as stabilizer in an electroless process. Ohno et af” determined that the order of catalytic activities with each metal are different for each reductant. There is no simple correlation that can be used to predict the 126metal with the best catalytic activity for a specific reducing agent. Each substrate should be evaluated to determine the best reducing agent for that metal. A review article by Molenaar“* suggests that autocatalytic metallization reactions can be divided into the following classes by studying the mechanism of the oxidation reaction: 1) Oxidation determined by a dehydrogenation reaction. Examples are hypophosphite on nickel, cobalt, and palladium, formaldehyde on copper, and borohydride on gold. Dehydrogenation reagtions occur when atomic hydrogen is formed and supplies electrons in a subsequent reaction, 2) Inorganic reducing agents which are oxidized by direct catalytic electron transfer at the metal’ surface. Examples are iron(ll) ions, which are used to develop photographic nuclei with silver and copper, vanadium(II) ions for selective metallization of copper and tin and chromium (II) ions which can deposit autocatalytically pure nickel and cobalt. 3) Inorganic redox systems in which the reducing agent and the metal ion to be reduced are identical. This catalytic disproportionation reaction occurs for example with tin(ID) ions in alkaline solutions. The only example of this class that formulates an autocatalytic bath is a special tin (II) compound which supplies electrons for the reduction of a second tin(II) ion. As with formaldehyde and hypophosphite, the adsorption of the active tin (11) compound is of importance in the catalytic reaction. 4) Enol compounds such as tartrate and ascorbic acid. A catalytic oxidation of an adsorbed tartrate species takes place, but stabilizers must be used in the solution. Therefore, it is demonstrated that the catalytic activity for the metal being plating must be taken into account when choosing a reducing agent. A reducing agent with a high catalytic activity on the metal will have an increased plating rate. However, there are two issues to consider when choosing a reducing agent system for an autocatalytic 127electroless plating bath: 1) what metal will be deposited and 2) what is the initial substrate material. Iacovangelo” separates plating into three stages: 1) initial stage of nucleation and growth where the catalytic activity of the substrate dominates, 2) intermediate stage where enough metal has been deposited for the reaction to occur autocatalytically; however, the plate is porous enough that the reducing agent can reach the underlying substrate and proceed along the mechanism used in stage (1), and 3) final stage is when the initial substrate surface is completely covered by the plated metal and the plating reaction continues according to the kinetics of autocatalytic plating. Okinaka et al” explain that electroless gold deposition can take place on noble metals such as Pd, Pt, Rh, and Au as well as active metals such as Cu, Ni, Co, Fe. The mechanism is different on the two classes of metals. The reaction is catalytic from the very beginning on noble metals; whereas, on active metals gold deposition in initiated by a galvanic displacement reaction. Therefore, the initial mechanism of plating may be very different from the kinetics of autocatalytic plating. For example, the catalytic activity of the initial substrate for oxidation of the reducing agent may be very different than the cataytic activity of the metal that is being plated. Also, the initial kinetics will be that of the galvanic displacement reaction when plating on an active metal. In addition, the later stage kinetics may be affected by the dissolved ions in the solution from the galvanic displacement. A substrate material can be chosen with the same catalytic activity as the metal being plated so that the confusion associated with the different initial mechanisms discussed above is minimized. Sullivan et al’ used gold substrates to study the autocatalytic deposition of gold. They evaluated the oxidation of hydrazine, hydroxylamine, hypophosite, and ascorbic acid on gold. They determined that ascorbic acid exhibited the highest catalytic activity for the deposition of gold onto the gold substrate, They also used a rotating gold disk electrode at different rotation rates and a buffered ascorbic acid solution to determine the kinetics of the electro-oxidation of 128ascorbic acid (anodic reaction). They found the anodic reaction to be mass transfer controlled at low rotation rates. The experimental diffusion coefficient of ascorbic acid was found to be 6.5 x 10° cm/sec compared to 1 to 2 x 10° cm/sec which is the typical diffusion coefficient of an electroactive species in an aqueous solution. The bulky structure and subsequent lower mobility of ascorbic acid may explain its low diffusion coefficient."* Sullivan et al * studied the electrochemical reduction of gold thiosulfate (cathodic reaction) in buffer solutions using a gold rotating disk electrode. They reported the experimental diffusion coefficient of gold thiosulfate in an aqueous solution to be 7.0 x 10° cm/sec. The diffusion coefficient of the gold thiosulfate electroactive species and the ascorbic acid electroactive species are very similar according to this study. The following section will describe the experiments performed to determine the plating kinetics of the Gel Plating Process. A gold substrate and plating gel are used to obtain the kinetics for the autocatalytic deposition of gold onto a gold surface. The kinetics of the Gel Plating Process would be difficult to evaluate using electroanalytical techniques because of the number of chemicals added to the plating gel formulations. Therefore, the plating thickness as a function of time was determined for different gel formulations and temperatures. These results will be correlated to diffusion and reaction rate models to determine the rate controlling mechanisms of the Gel Plating Process. 5.3 Plating Rate Studies The literature presented above helps to understand the fundamental kinetics of electroless plating from a dilute solution, However, the Gel Plating Process has two features which distinguishes it from the plating baths analyzed in the above review: 1) the high concentrations of ions, specifically the gold complex and reducing agent and 2) the viscous nature of the plating solution. Mechanisms that govern the deposition of gold through a low viscosity, low ion concentration plating bath are likely to be different from the mechanisms governing the kinetics of the Gel Plating Process. There are, however, fundamental parameters that will have an influence on both plating systems such as the 129fundamental parameters that will have an influence on both plating systems such as the substrate surface, plating temperature, plating bath formulation, plating bath stability, additive concentrations, and pH. Factors such as pH and substrate surface were not varied in the following section. However, plating temperature, gel formulation, and additive concentration were investigated. 5.3.1 Experimental Procedure Plating, gel formulations were made according to Table 27. Each plating gel had a 40 g/L (0.2M) gold concentration and a pH between 7.0 -7.5. The effect of EDTA, di- ammonium thiosulfate, ascorbic acid, 2-mercaptobenzimidazole, and surfactant on the plating rate was determined by varying the concentrations of these components added to the plating gel formulation and looking at the resulting plate thickness. ‘The plating gel was screen printed onto an immersion gold plated rectangular sample, 2.54 cm x 1.27 cm x 0.04 cm. The initial substrate is phosphor bronze electroplated with nickel. The immersion gold layer was applied to the nickel surface by using a commercial immersion plating bath, Auruna 511 (Uyemura International Corporation; Sheffield, CT). The immersion plating bath was maintained at a pH of 4.7 at a temperature of 90 °C. The final immersion gold Is yer thickness ranged from 0.05 to 0.2 microns. The rectangular sample was mounted onto a glass slide and held on the base plate of the DeHaart SPSA-10 standard screen printer by vacuum. A metal stencil (purchased from SMT Inc.; Merrimack, NH) was attached to the screen printer and lowered over the sample. The metal stencil was 254 wm (10 mils) thick with a 2.3 x 1.0 cm? rectangular open area. The plating gel was applied to the metal stencil and hand squeegeed onto the substrate using a squeegee of durometer 90 shore. The substrate was weighed before and after gel printing. The substrate with the gel print was placed into either the Blue M Humidity oven with the humidity set at 97% relative humidity or a custom made reactor with a saturated nitrogen atmosphere at 130~100 % relative humidity. The plating temperature was either 50 or 60 °C. The substrate was weighed after removing from the deposition chamber and rinsed with DI water to remove the plating gel. Gold Plate Thickness measurements were performed by AMP Inc. using X-ray Fluorescence (XRF) analysis. Table 27: Plating Gel Formulations for Piating Rate Studies Bath Components [M] Range for Studies NasAw(S:05)2 0.2 CeHyNaOg 0.7 t0 1.0 Ammonium Acetate 05 Di-ammonium EDTA 0.00828 to 0.046 NazSO3 0.1 00.15 (NH,)28203 0 t0 0.203 2-mercaptobenzimidazole 75x 10% to 1.5x 10° Surfactant 0to6x 10° Hydroxyethyl Cellulose (HEC) 3.5 wi% 5.3.2 Observations The gold plate thickness for each gel formulation and temperature tested was measured by XRF measurements. Table 27 lists the variation in ammonium thiosulfate, EDTA, sodium sulfite, ascorbic acid, 2-mercaptobenzimidazole, and surfactant concentrations used throughout these plating rate studies. Variations of EDTA, di- ammonium thiosulfate, sodium sulfite and ascorbic acid did not change the plating rate for the molarity ranges studied in Table 27. The stabilizer, 2-mercaptobenzimidazole, and surfactant did have an effect on the plating rate. 5.3.2.1 Plating Rate as a Function of Stabilizer Concentration Figure 37 illustrates the plating rate at 60 °C as a function of 2-mercaptobenzimidazole _concentration. Low concentrations of 2-mercaptobenzimidazole result in a higher plating rate than high concentrations of 1312-mercaptobenzimidazole. However, the plating gel must have a minimum concentration of 2-mercaptobenzimidazole to remain stable at room temperature and during plating, as discussed in section 3.4.2. Table 28 summarizes observations made about the plating gel stability during this plating experiment. A 2-mercaptobenzimidazole concentration greater than 7.5 x 10° M is required to maintain a stable plating gel at room temperature for longer than one hour. However, Figure 37 illustrates that the plating rate decreases as the concentration of 2-mercaptobenzimidazole is increased. Therefore, by comparing the observations in Table 28 with plating rate data in Figure 37, a 2-mercaptobenzimidazole concentration of 7.5 x 10% M was determined to be optimal for plating at 60 °C. Table 28: Plating Gel Stability as a Function of 2-mercaptobenzimidazole Concentration [2-mercaptobenzimidazole] _Room Temp. Stability Stability at 60°C 1.5 107.5% 10° Thour Gel tums black in 15 mins. Discoloration in resulting 7.5.x 10% ~ 1 day plate ‘Some orange powder in 75x 10* >I day plating gel after 2.5 hours Ogden and Tench* studied the effect that dimercaptothiadiazole (DMTD) had on plating from a copper pyrophosphate electroplating bath. They found that two different species of DMTD were formed in the plating bath. The structure of DMTD and the different species formed in the plating bath are illustrated in Figure 38. They concluded that one species acted to accelerate the plating rate while another acted to slow down or decelerate the plating rate. Both accelerating and decelerating species could be formed in the plating bath depending on the concentration of DMTD added to the bath. In addition, they found that additives that contained only one -SH or -NH2 group complex with the Cu (II) ion to produce a deceleration in the plating bath. 132Plating Rate (um/min) 18 Te {og [2-mercaptobenzimidazole} Figure 37: Initial Plating Rate of the Gold Thiosulfate- Ascorbic Acid Plating ‘System as a Function of 2-mercaptobenzimidazole Concentration. 133,‘The accelerating and decelerating species formed from the DMTD are compared to the molecular structure of 2-mercaptobenzimidazole in Figure 38. 2-mercaptobenzimidazole has one -SH group which according to the above theory should form a decelerating species with the gold ion in solution. An increase in the 2- mercaptobenzimidazole concentration does have a decelerating effect on the bath, as shown in Figure 37. This behavior agrees with Ogden and Tench’s prediction that additives with one -SH group produce a deceleration in the bath®'. They hypothesize that the DMTD dimer decelerates the electrodeposition rate by blocking growth sites. An altemative explanation could be that the gold complex formed with the 2- mercaptobenzimidazole ligand requires a higher activation energy before the gold ion will separate from the ligand and be available for reduction at the catalytic surface. Organic compounds containing sulfur are well-known ligands for transition-metal ions.* Neither of the above explanations is disputed by the plating rate data. However, an increase in 2-mercaptobenzimidazole concentration in a plating gel definitely increases the room temperature stability of the plating gel while it is not in contact with a catalytic surface. ‘The greater the concentration of 2-mercaptobenzimidazole in the plating gel, the longer the plating gel is stable at room temperature without homogeneously precipitating gold into the plating gel. Therefore, an increase in the 2-mercaptobenzimidazole concentration appears to increase the activation energy needed to homogenously precipitate gold into the plating gel. The activation energy for the homogenous precipitation of gold may be increased because the gold ions complex with the 2-mercaptobenzimidazole in solution. 134DOL Dimercaptothiadiazole (DMTD) 2-mercaptobenzimidazole H H TN | TA TN cus ‘s ‘sCu* AY AK Copper = Na 2 Complexes S—— ‘l c) I AA) RK Accelerating Species from DMTD Decelerating Species from DMTD Figure 38: Schematics of Accelerating and Decelerating DMTD Species and 2-mercaptobenzimidazole 135‘The mechanism by which 2-mercaptobenzimidazole lowers the plating rate can be explained two ways: 1) the 2-mercaptobenzimidazole acts as a poison to the catalytic surface and/or 2) the gold ion - mercaptobenzimidazole complex increases the activation energy necessary to separate the gold ion from the ligand. The 2-mercaptobenzimidazole is definitely a necessary component for preventing homogeneous decomposition of gold in the plating gel at room temperature. Therefore, this component is a necessary ingredient in the plating formulation even though it decreases the plating rate. 5.3.2.2 Plating Raté as a Function of Surfactant Concentration Surfactant additions to the plating gel were found to increase the printability of the plating gel (see section 4.1) and decrease the overplating problem (see section 4.2). The concentration necessary to prevent extraneous plating of gold yet not interfere with the plating mechanism had to be identified. Plating gels were formulated with the concentration of surfactant varied according the concentrations listed in Table 29 (all the other components held constant). The initial plating rate for each surfactant concentration was determined by plotting the plate thickness (determined by XRF) as a function of time and calculating the slope from 0 to 30 minutes. All surfactant concentrations exhibited a linear region from 0 to 30 minutes on the plate thickness versus time graphs. This plating rate was then plotted as a function of surfactant concentration (see Figure 39). Table 29: Plating Gel Formulation for Surfactant Experiment Plating Gel ‘Surfactant Concentration (M) High 6x 107 High Medium 1x10? ‘Medium 5x 10° Low 2x 107 Control 0 136Plating Rate (unvsec) oooo16 ] noo. o.00012 iii oe oe i ° 1.006-07 1.00608 1.006-05 1.006-04 1.00809 1.006-021.00-011.006+00 {og [Surfactant] Figure 39: Plating Rate as a Function of Surfactant Concentration 137Below a critical concentration (5 x 10° M) of surfactant there is no change in plating rate and above that concentration the plating rate is lowered (see Figure 39). The plating rate for surfactant concentration below 5 x 10° M is very similar to the plating rate with no surfactant added to the plating gel formulation. The adsorption of the surfactant on the catalytic surface and the subsequent decrease in the number of sites available for nucleation and growth of the gold plate can explain the decrease in plating rate with the increase in surfactant concentration. A calculation of the surface sites occupied by the surfactant will give a rough estimation of whether the above theory is plausible. The concentration of surfactant in the plating gel can be used to calculate the theoretical number of gold atoms occupied by adsorbed surfactant, It can be assumed that one molecule of surfactant will occupy one gold atom. The concentration of surfactant in the plating gel can be used to calculate the moles of surfactant in the gel print. The moles of surfactant in the gel print is related to atoms using Avogadro's number (N4). The gold atoms per surface area occupied by surfactant can be obtained by multiplying by the gel print thickness. The gel print thickness used in this calculation is 700 microns. A value of 10'* atoms/cm? can be assumed as a good estimate for the number of atoms on the plating surface. Table 30 summarizes the results of this calculation. 4 x 10'° atoms/cm? are occupied with surfactant when the plating gel contains 6 x 10° M surfactant. This is in excess of the 10" atoms/em? approximation of the gold atoms on the plating surface. Therefore, a significant decrease in plating rate would be expected since the nucleation and growth sites for the gold are severely limited. 3 x 10" atoms/cm? are occupied by surfactant when the plating gel contains 5.0 x 10° M surfactant. This is one order of magnitude below the estimation for surface sites on the plating surface, so a decrease in plating rate is not expected. These findings correlate well with Figure 39. 138Table 30: Calculated atoms/cm” surfactant Surfactant Concentration (M) ‘Atoms/em™™ 6x 10-3 4x10 5x 10-5 3x 10% 2x 10-7 ix 10 0 0 Many physical properties of surfactant solutions show a sharp change at the concentration regime which is termed the critical micelle concentration (CMC). The CMC is determined by titrating the surfactant into water and measuring the surface tension. The surface tension at the CMC essentially "flattens" and there is no further decrease in surface tension with addition of surfactant.* The surface tension at the CMC that is reported for Triton x-100™ (the surfactant used in these experiments) is 31 dynes/cm. Figure 40 shows the surface tension of the plating solution as a function of the surfactant concentrations listed in Table 29. The surface tension starts to level off at a value of 31 dynes/em when the surfactant concentration is 1x 10° M. This indicates the that critical concentration of surfactant for micelle formation is 1 x 10° M. The plating rate shows a sharp decrease between 5 x 10° and 1 x 10° M surfactant. Therefore, the plating rate decrease is not a function of exceeding the CMC of the surfactant. 1392 a Surface Tension (dyuies/em) 5 8 7 ee ESE TES OT log[Surfactant} Figure 40: Surface Tension as a Function of Surfactant Concentration 1405.3.2.3 Plating Rate as a Function of Temperature ‘Two plating temperatures were investigated in this study: 50 °C and 60°C. A plating gel with the composition in Table 31 was printed onto the substrate with a thickness of ~ 700 microns. The substrate was then placed into the custom made reactor with a humid, nitrogen atmosphere at 50 °C and 60°C. Table 31: Plating Gel Formulation for Temperature Study Bath Components {MJ Range for Studies Na3Au(S205)2 0.2 CeHyNaQg 07 Ammonium Acetate 0.5 Di-ammonium EDTA 0.00828 NaSO3 0.15 2-mercaptobenzimidazole 7.5 x 104 Surfactant 2.7% 107 Hydroxyethyl Cellulose (HEC) 3.5 wt% Figure 41 compares the plate thickness as a function of time for plating temperatures of 50°C and 60 °C. The plate thickness was determined by XRF measurements. The initial slope (0 to 30 minutes) of the plate thickness data was evaluated to compare the change in initial plating rate with temperature. The initial plating rate at 60 °C is 1.2 x 10 m/sec compared to 4.9 x 10° m/min at 50°C. The activation energy was approximately 75 + 18 KJ/mol. The theoretical gold plate expected from a 700 micron thick plating gel that has a gold concentration of 40 g/L (0.2 M) is 1.4 microns. The plate thickness is only 0.67 + 0.14 microns after 3 hours at 60°C. Therefore, 53% of the gold is stil] left in the plating gel. Section 5.4 will compare models to the above plating data to identify the rate limiting steps of this plating system. M4Plate Thickness (um) 0 2000 4000» 6000" 8000-10000 12000 * Time (sec) Figure 41: Plate Thickness vs. Time at 7.5 x 10% M 2-mercaptobenzimidazole for T= 50°C and 60°C. 1425.3.3 Conclusions The above studies indicate that the concentration of 2-mercaptobenzimidazole and surfactant and the plating temperature have the largest influence on the plating rate. Increasing the concentration of the stabilizer, 2-mercaptobenzimidazole, in the plating gel formulation decreases the plating rate. The stabilizer is necessary to prevent homogeneous precipitation of the gold at room temperature. Therefore, the optimum concentration of stabilizer for plating and gel stability was chosen to be 7.5 x 10%M. A concentration in excess of 5 x 10° M surfactant was found to decrease the plating rate. However, below this concentration, the plating rate was similar to that of a system without surfactant. The plating rate does increase with increasing temperature. The initial plating reaction has an activation energy of 75 +18 KJ/mol. Further investigation of the mechanisms involved in this plating reaction will be discussed in section 5.4. 143,5.4 Mechanisms of the Gel Plating Process Bindra et af” state that the overall reaction in the electroless plating of metals is diffusion limited. This statement is valid for plating solutions that have a low gold concentration. Transport of the Aut ions to the plating surface is increased by the use of agitation in these plating baths. The gel plating process operates at high gold concentrations, specifically 40 g/L, which is a 10-fold increase in concentration from the typical commercial electroless plating bath. In addition, no agitation is supplied to increase the transport of species to the surface. Plating through a gel with a high gold concentration may show a different behavior than plating through a dilute solution. The following sections will compare the plating rate data presented in Section 5.3.2.3 with different models of diffusion control, reaction rate control, and diffusion and reaction rate control to determine the overall mechanism of electroless gold plating through the gel. Possible rate determining steps for the Gel Plating Process may include 1) diffusion of the electroactive gold species to the plating surface, 2) diffusion of the electroactive reducing agent species to the plating surface, 3) diffusion of the by products from the cathodic reaction at the plating surface away from the plating surface, 4) diffusion of the by products from the anodic reaction at the plating surface away from the plating surface, 5) the cathodic reaction at the plating surface, or 6) the anodic reaction at the plating surface. The diffusion away from the plating surface of the by products from the anodic and cathodic reactions at the plating surface are not considered likely rate limiting steps in this analysis because the by products from the reaction will have a lower mass than the initial diffusing species. Therefore, the flux of the by product species away from the plating surface should be faster than the flux of the initial diffusion species to the plating surface. This leaves four possible rate determining processes to consider: 1) diffusion of gold complex, 2) cathodic reaction at the surface, 3) diffusion of reducing agent, 4) anodic reaction at the surface. The gold complex is assumed to be a gold 144thiosulfate complex; however, the actual diffusing species has not been experimentally determined. The overall plating reaction can theoretically be controlled by any of the above mechanisms or combination of these mechanisms. The operating parameters of the Gel Plating Process can be used to rule out some of the mechanisms stated above. For example, the reducing agent concentration is over 3 times the concentration of the gold complex, The anodic reaction at the surface can be eliminated as a possible rate controlling mechanism since this species is in excess. The diffusivites that Sullivan et al*®? determined for the gold thiosulfate and the ascorbic acid electroactive species are very close in value, 6.5 x 10° cm*/sec for ascorbic acid and 7 x 10° cm*/sec for gold thiosulfate. However, the concentration of ascorbic acid in the plating gel is more than 3 times the concentration of the gold complex. Since the diffusivities are the same for the gold complex and reducing agent, the 3 to 1 ratio of reducing agent to gold complex will be maintained throughout the plating process. Therefore, it is likely that the diffusion of the gold complex to the plating surface will limit the reaction more than diffusion of the reducing agent to this surface. From this argument, diffusion of the reducing agent can be eliminated as a rate determining process for the overall plating reaction. The possible rate determining steps are now: 1) diffusion of the gold complex to the plating surface and 2) cathodic reaction at the plating surface. Electroless plating of metal is usually considered to be diffusion controlled. Therefore, a diffusion model, based on the diffusion of the gold complex to the plating surface, will be presented in the first section. This model does not correlate well with the experimental data, The Gel Plating Process does not exhibit the same rate controlling mechanism as the typical electroless plating solution with the dilute metal concentration. Therefore, other models are necessary to determine the rate determining step of the Gel Plating Process. A reaction rate model, based on the cathodic reaction at the plating surface, will be presented in section 5.5.2. Finally, section 5.5.3 will present a model that incorporates both diffusion and reaction to show that both mechanisms must be involved to explain the plating rate data. 1455.4.1 Diffusion Limited Model The series solution to Fick’s second law was solved using the boundary conditions shown in Figure 42. A schematic illustrating the diffusion parameters for modeling of the diffusion through a gel print to the catalytic surface is shown in Figure 42. The boundary conditions describing this situation are: =e, for O0 (6.12) af.=0 for x=L, at t>0 (5.13) Manipulation of the series solution to address these boundary conditions can be simplified by looking at the solution for ‘diffusion out of both sides of a rectangular slab’, The boundary conditions for this problem are: for O0 (5.15) L, at t>0 (5.16) boundary condition at x = L (eq. 5.13) for the gel problem. Therefore, the series solution can be solved using the following boundary conditions and considering one half of the ‘diffusion out of a rectangular slab problem’: for O0 (5.18) c=0 for x=2L, at t>0 (5.19) The final solution obtained using the above boundary conditions is: De of (BE or) (6.20) 2L 2L 146Xel Gel Thickness x=0 Catalytic Surface, c= 0 (a) Schematic for Gel Problem 0 L Concentration Profile for Gel Problem Concentration Profile for ‘Diffusion out of a Rectangular Slab’ Problem ) © Figure 42: Schematic For Diffusion Model and Boundary Conditions 147‘The concentration of the diffusing species is determined as a function of time and position according to equation (5.20). The gold complex is assumed to be the diffusing species in this reaction (see argument at beginning of section 5.5). The accumulation of gold at the catalytic surface over a specific time interval can be calculated by determining the flux of gold ions at x = 0 and integrating the flux over the specific time interval of interest. The flux at x = O can be calculated from the partial derivative of equation (5.20) with respect to x times the diffusivity: Qj+I)| 2b Integrating th flux over a specific time interval gives the following equation: ee 5S eh SSE |e a (CGE) Jorn pial [a] cA P ) 5.22) where L = gel thickness, C, = initial gold concentration, ime, D = diffusivity of gold ions, j= summation index, {{ J dt = moles of gold deposited over time interval (mol,,, ) The plate thickness at that specific time interval can then be calculated using equation (5.23). mol plate thickness = (5.23) where M, = molecular weight of gold and pay = density of gold 148Therefore, this model can be used to predict the plate thickness expected for various values of D. The parameters L = gel thickness, C, = initial gold concentration, and t = time were set equal to values used in the plating rate study (see section 5.3.1). Plots of plate thickness as a function of diffusivity were compared to the actual plating rate data. The best fit of the diffusion model to the plating rate data is shown Figure 43 for both 50 and 60 °C. The diffusion model did not fit to the plating rate data over all times. The shape of the diffusion model does not fit the initial plating rate data; however, at longer times, the shape of the diffusion model does appear to correspond with the experimental data. The diffusivites chosen for the graph in Figure 43 are not accurate numbers for the diffusivity of the gold complex because the initial plating rate data was not fit by the model. The next section will compare the reaction rate model to the plating rate data. M92 Te pron Model OnBe8 on ee a T2906 Diffusion Model for 60 °C Figure 43: Effect of Diffusivity on the Plate Thickness as a Function of Time 1505.4.2 Reaction Rate Model A reaction rate model has been developed to compare to the experimental plating rate data. The reaction was assumed to be first order with the gold concentration as the only species influencing the reaction (see the discussion at the beginning of section 5.5) ‘The equations of interest for a first order chemical reaction are:®© [Au] =[Au], expCkr) (5.24) There is assumed to be no mass transfer resistance. The rate of disappearance of gold out of the gel will be called the flux (J) of gold out of the gel and will be assumed to be equal to: -kjAu]J=R=J (5.25) The [Au] is equal to the concentration of gold (C) in the plating gel so that equation 5.25 can be re-written as: J=-kC (5.26) The thickness of the gel (L) determines the overall amount of gold for the plating reaction. Therefore the following definitions apply: Amount Au_ unit area CL (5.27) Change inamount Au unit area xtime (5.28) 151Mass balance dictates that the flux of gold out of the system must equal the change in the amount of gold per unit area per unit time: ~ee=L2 (6.29) or This equation can be rearranged and solved by integrating dC over the limits of initial concentration to concentration in the gel at time =r and integrating dt with the limits of time = 0 to time ES--£ fa (5.30) The solution to this equation gives the concentration C expressed as a function of initial concentration, gel thickness, and time as shown below: k = ot (5.31 c Coen i) 631) This expression for C can be substituted into equation (5.29) and (5.26) to obtain the expressions: ac k Lean k, on(- 5s) (5.32) J=-KkC, exp (- #1) (5.33) 152The flux (J) is defined as the amount of gold leaving the system; therefore, this ‘can be converted to the plating rate by the following equation: 6.34) where PR = plating rate, h = gold plate thickness, J = flux of gold ‘The plate thickness as a function of time can be determined by rearranging and integrating equation (5.34). fans [ya 635) a Substituting the absolute value of equation 5.33 for J and integrating over the above limits, the expression for plate thickness is: M,KC,(LY,_ (ky Mekeeh on i) (5.36) The plate thickness as a function of time for various values of k, the rate constant can be plotted using equation (5.36). The plating rate data at temperatures of 50 °C and 60 °C is compared to the reaction rate model in Figure 44, The reaction model fits the initial plating rate data (0 to 30 minutes) for the 60 °C plating temperature when k=7.5x 10° cm/s. The reaction rate model fits the initial plating rate data for the plating temperature of 50 °C when k = 3 x 10° cm/s. The following equation can be used to determine the activation energy for the initial plating time: kek, 2 9. a ) (5.37) 153,‘The activation energy for the initial plating time calculated from the k values listed above is 82 KI/mol. The activation energy calculated from the initial plating rates was 75 + 18 KJ/mol. These two values are similar which indicates that the initial plating reaction is reaction rate controlled. However, plating times et 60 °C which are longer than 30 minutes do not line up with the reaction rate model (see Figure 44). The next section will discuss a model that takes into account diffusion with a reaction at the plating surface. 1542 a Tae erase restin Model for 60°C Pate Tete om) ° ‘0 os ‘ooo woo Figure 44: Plating Rate Data Compared to Reaction Rate Model 1555.4.3 Diffusion & Reaction Model The reaction rate model indicates that the initial plating time is governed by reaction rate contro]. Longer times are assumed to be governed by diffusion. However, the diffusion mode] presented in section. 5.4.1 is not a valid model to determine the diffusion coefficient at long times because it does not take into account the reaction rate at the surface for short times. A third model has been developed which does take into account the cathodic reaction at the plating surface and the diffusion of gold ions through the gel. The series solution to Fick's second law was solved to satisfy boundary conditions (eq. 5.38 to 5.40) that take into account the chemical reaction at the surface: C=C, for O0 (5.39) ve sicn 0 for x=2L, at t>0 (5.40) where C= concentration of gold, C, = initial concentration of gold, D = diffusion coefficient of gold ions, k= reaction rate constant, and L = gel print thickness. To model the geometry of the gel plating process, one half of the above geometry is considered (see Figure 42 and the discussion in section 5.4.1). Carslaw and Jaeger’” solve a similar problem but their limits are 0 < x < L and they are dealing with the heat analog. The method they used to obtain their solution was modified for the above boundary conditions. The following solution was found to satisfy Fick’s second law and the above boundary conditions. c=BA, (cole 7 Da,?t) (5.41) x sola where A, is an arbitrary constant and @,, is any root other than zero of the equation: 1562aDk tanQLa)= gr (5.42) This equation was determined from the above listed boundary conditions. The coefficient, A,, which satisfies the boundary conditions was determined to be: ff c, (] IMPRINT: (COPYRIGHT) » ADD: DEGREE: > DEPT: SUPERVISORS: | Notes: cat, date: pvenr_COQ?™_roecree:_ Fy DD NAME: E