Chemical Sensors for Monitoring the Manufacturing Process
in Chip-Scale Packages

Omowunmi A. Sadik, Assistant Professor, Chemistry

    One of the requirements of modern electronics manufacturing technology is the need to deliver high yield product packages at low costs. During "chip-scale" package manufacturing processes, analytical information is required to understand the mechanism of the plating process. This will enable the modification and control of packaging conditions needed to achieve high reproducibility in packaging performance. Process analysis can often be accomplished in two different ways, including: (i) Classical, off-line, manual sampling method of determining parameters, (ii) Batch technique or continuous feed batch processes where the control of various parameters is essential.

    The continuous technique is used in combination with automated analytical devices such as chemical sensors. Chemical sensors are devices consisting of a suitable transducer (such as electrochemical, optical, or mass) with chemically selective layers and a signal processor. They are capable of continuously recognizing concentrations of chemical constituents in liquids or gases and converting this information in real-time to an electrical or optical signal. The focus of this project is to design and test suitable chemical sensors that will help in the processing of chemical signals obtained from circuit card manufacturing. These sensors must be capable of being integrated into other devices that can exercise real-time feedback required in the electroless plating bath technologies used in the fabrication of circuit boards and cards found in electronic devices.

    In the fabrication of most circuit cards and boards, electrolytic or electroless methodologies are used for both the bulk circuitization and surface finishing. Electrolytic processes utilize an external current source for metal ion reduction. On the other hand, electroless methods involve the use of chemical reducing agents to drive the heterogeneous reduction of metal ions or complexes. Compared to electrolytic plating technique, electroless-plating principle is highly attractive because of its simplicity of operation, which requires no external source of current or elaborate equipment. The use of electroless deposition takes care of the problems associated with dendrite formation and uneven deposits associated with electrolytic deposition in which non-uniform current densities occur.

    The increasing use of wirebond in assembling single or multichip organic modules (SCM-L or MCM-L) and chip-on-board (COB) schemes require surface finishing that utilize electroless gold (Au) plating. IBM, a leading computer manufacturer, currently employs a proprietary electroless Au-plating process for these applications and for ceramic modules. Electroless gold deposition process requires frequent and accurate monitoring of numerous chemical parameters, including gold, reducer and stabilizer concentrations. Therefore, this project seeks to identify a means of increasing the quality of plating process by:

1. Providing further insights into how chemical transformation in electroless systems influence plating performance, and

2. Developing the prototype of a chemical monitoring device or system for use in electroless plating processes.

    The long-term objective is to extend the use of this on-line prototype system to other plating systems beyond electroless gold. A typical electroless gold bath formulation contains KAuCN₂ as the source of gold ions. Other compositions include potassium cyanide (KCN) as stabilizer, aqueous potassium hydroxide (KOH) as the support medium, and either potassium borohydride (KBH₄) or dimethyl aminoborane (DMAB) as the reducing agent. In addition, small amounts of organic chelating agents such as ethylene diamine tetra amino acid (EDTA), citric acid and inorganic catalysts like Pd²⁺ and Ti²⁺ are added to increase the plating rate. Organic stabilizers can complex undesirable metallic contaminants or may otherwise allow the operation of the bath at higher temperatures. We are focusing on the use of a gold bath that employs DMAB as reducing agent because DMAB can be used to reduce gold on a variety of noble metals. These include lead, platinum, gold and other active metals such as copper, nickel, and iron. While DMAB is reported to yield significantly lower plating rates than potassium borohydride, it is sufficiently stable in alkaline medium and does not hydrolyze to the same extent as borohydride. Our approach is the use of impedance spectroscopy for the characterization and prediction of bath performance as well as the development of chemical sensors for monitoring bath components.

    How chemical transformations in electroless gold plating bath influence performance: A detailed chemical and physicochemical study of electroless gold plating baths using spectroscopic, electrochemical and magnetic resonance measurements has been undertaken. This study is to enable us to understand the factors that govern the properties of the DMAB/gold electroless systems. Electroless deposition often consists of two chemical reactions occurring in an electrochemical cell, where the chemical oxidation reaction of DMAB liberates electrons, and the reduction of gold consumes the electrons. The mechanism of electroless deposition in the presence of reducing agents and metal ions is believed to proceed according to mixed potential theory. The use of mixed potential theory leads to the assumption that electroless Au plating can be considered as the superposition of anodic and cathodic reactions at some mixed potential (or deposition potential), E_m. We have tested the validity of mixed potential theory for the electroless gold plating using DMAB as a reducing agent. At steady-state equilibrium potential, the rate of deposition is equal to (i) the rate of oxidation of DMAB (anodic current, i_Red), and to (ii) the rate of reduction of gold ions (cathodic current). Therefore:

i_deposition = i_Red = i_M     (Equation1)

where i_M is the rate of metal ion reduction (cathodic reaction) and i_Red is the rate of DMAB oxidation (anodic reaction). From Faraday's law, the rate of gold deposition can be expressed as:

Rate (mg/cm/hr) = 1.09 x i_deposition (mA/cm²)      (Equation2)

    The experimental acquisition of E_M was accomplished by investigating the electrode polarization behaviors after subjecting the metallic substrate to a slow potential scan such that steady-state currents are obtained. Polarization experiments were performed to assess the effects of plating rates that fully describe the electroless bath. A complementary gravimetric determination of the electroless gold deposition rate was conducted in a circulating water bath using water-jacketed thermostated glass cells under identical conditions. This was used to assess the correlation of standard techniques with the plating rates obtained. Deposition was obtained under different bath compositions, temperature and pH, and the rate was determined as mg/cm2/hr. Figure 1 shows the polarization curve obtained for the electroless gold bath using DMAB. The point of intersection of the anodic and cathodic curves yielded E_M and i_dep from which the deposition rate was calculated using Equation 2. This resulted in E_M of _615 mV and a deposition current of 3.873 mA/cm2. This result corresponds to a theoretical rate of 4.222 mg/cm/hr and a gravimetrically determined rate under similar conditions produced the rate of 10mm/hr.

    The driving force (E_D) for the overall reaction in the bath is the difference in potential between the metal reduction and oxidation of DMAB. E_D is on the order of several hundred mVs, depending on the pH and cyanide concentration. Using the plating bath, other electrochemical experiments performed confirmed that increasing hydroxide concentration has two effects on DMAB concentration. First, it decreases oxidation potential with no net current flow, and secondly, it increases the rate of oxidation at any given potential. In the presence of 0.035M KCN, the DMAB polarization curve shifts approximately by nearly150 mV more positive. This results in a factor of two decrease in plating rate. Currently, we are investigating the influence of varying the concentration of these parameters on the overall plating efficiency.

    How chemical sensors improve the efficiency of the Manufacturing Process: In order for the bath to function effectively, certain parameters such as gold concentration, the reducing agent, pH and other chemical byproducts must be frequently monitored and controlled. If the results fall outside of the set levels, the resultant coatings may be off-color, too thin, brittle, and uneven. This may result in spotty coverage. In addition, the plating rates may be inconveniently long if the components become depleted. We are developing automated chemical sensors for analyzing the gold/reducer concentrations and for providing early warning signals for determining plating efficiency (Figure 2). Currently, this analysis is usually carried out in remote analytical laboratories where an operator manually samples the bath at specific intervals in order to test for the chemical concentration levels. Since the operator's decision on defects is made manually, any tools to aid this decision making will reduce the chance of a false call. Many times, the operator may want to electronically control the plating rates and efficiency, and this may require a considerable amount of time and effort.

    Chemical sensor techniques provide an unparalleled monitoring capability of the chip carrier making process, by helping to define a time-integrated response to a fluctuating concentration level that can aid in detecting defects. With the placing on-line of an automated chemical system, the operator and the chip carrier making process have different roles. The system may perform the fluctuating component levels or defect handling duties automatically using the cyanide, pH, and oxygensensors placed on-line (Figure 2), while the operator may be involved in placing the inspection samples into an on-line robotic system for feeding through the manufacturing process. More importantly, the system will make the defect call without relying on the operator's manual sampling or off-line analysis and interpretation. The off-line system provides more analytical examination of the concentration levels at the expense of throughput. The addition of operator variability provides an automated sensing system with high throughput and repeatability at the expense of the initial setup plus the measurement details that need to be defined.

Copyright © 2002 [IEEC]. All rights reserved. Revised: Wednesday, 20 November 2002 .