In the authors' previous paper, we studied the defects from Fluorine-Induced Corrosion on microchip Al bondpads using SEM, EDX, TEM, AES, IC, XPS and TOF-SIMS techniques. An unknown F-Al compound was found and identified as [AlF6]3-. In this paper, we will further study the chemical mechanisms of Fluorine-Induced Corrosion on microchip Al bondpads and propose a theoretical electrochemical model to reveal the secrets of Fluorine-Induced Corrosion on Al bondpads. To support this new theoretical model, we will provide substantiating data from TOFSIMS analysis and other failure analysis techniques. According to the theoretical model of Fluorine-induced Corrosion proposed, fluorine contamination on Al bondpads will cause two types of corrosions. First, fluorine reacts with Al and forms a complex compound [AlF6]3- on the affected area, which we will refer to as Fluorine Corrosion. Once the compound of [AlF6]3- forms on Al bondpads, it will form an Anode and cause further electrochemical reactions from O2, N2 and H2O (moisture) at the Cathode. The new products of further electrochemical reactions will be [OH]- and [NH4]+ ions. The new product of [OH]- ions will react with Al and cause further Al corrosion on bondpads and form corrosive product consisting of Al(OH)3, which we will refer to as [OH]- Corrosion. The new product of [NH4]+ ions will combine with [AlF6]3- and form a corrosive complex compound (NH4)3(AlF6). This proposed corrosion mechanism results in non-stick bondpad problem.

Fluorine contamination on Al bondpads will result in corrosion, affect quality of bondpads and pose problem such as non-stick on pad (NSOP) during wire bonding at assembly process. In this paper, a fluorine contamination case in wafer fabrication will be studied. Some wafers were reported to have bondpad discoloration and bonding problem at the assembly house. SEM, EDX, TEM, AES and IC techniques were employed to identify the root cause of the failure. Failure analysis results showed that fluorine contamination had caused bondpad corrosion and thicker native aluminium oxide, which had resulted in discolored bondpads and NSOP. It was concluded that fluorine contamination was not due to wafer fab process, but was due to the wafer packaging foam material. XPS/ESCA and TOF-SIMS advanced tools were used to study the chemical and physical failure mechanism of fluorine-induced defects. An unknown Al compound was found using XPS technique and identified it as [AlF6]3- using electrochemical theories and TOF-SIMS technique. This finding was very significance, as it helped developing a theoretical electrochemical model for fluorine-induced corrosion and helped understanding of the mechanism of fluorine-induced corrosion on aluminium bondpads. It was found that fluorine contamination had formed [AlF6]3-on the affected bondpads and it had caused further electrochemical reactions and formed some new products of (NH4)+ and OH-. Then [AlF6]3- and (NH4)+ ions combined and formed a corrosive complex compound, (NH4)3(AlF6), while the OH- reacted with Al and caused further corrosion.

In failure analysis of wafer fabrication, currently, three different types of chemical methods including 6:6:1 (Acetic Acid/HNO3/HF), NaOH and Choline are used in removing polysilicon (poly) layer and exposing the gate/tunnel oxide underneath. However, usage is limited due to their disadvantages. For example, 6:6:1 is a relatively fast etchant, but it is difficult to control the etch time and keep the oxide layer intact. Also, while using NaOH to remove poly and expose the silicon oxide, the solution needs to be heated. It is also difficult to etch a poly layer with a WSi x or a CoSi x silicide using NaOH. In this paper, we will discuss these 3 etchants in terms of their advantages and disadvantages. We will then introduce a new poly etchant, called HB91. HB91 is useful for removing poly to expose the gate/tunnel oxide for identification of related defects. HB91 is actually a mixture of two chemicals namely nitric acid (HNO3) and buffer oxide etchant (BOE) in a 9:1 ratio. The experimental results show that it is highly selective in poly removal with respect to the gate/tunnel oxide and is a suitable poly etchant especially for removing polysilicon with/without WSi x and CoSi x in the large capacitor structure. Application results of this poly etchant (HB91) will be presented.

In this paper, three low yield case studies in wafer fabrication are reviewed. These issues/problems include thicker gate oxide due to contamination from the wafer fab process, QBD failures due to silicon crystalline defects caused by charging during the BN+ implant process and memory failures relating to tunnel oxide defects in EEPROM devices. Chemical deprocessing techniques, 155 Wright Etch, Scanning Electron Microscope, Transmission electron microscopy & Secondary Ion Mass Spectroscopy were used to identify the root causes. Some new chemical deprocessing techniques in exposing the tunnel window & oxide for the memory cell failures were developed. Moreover, some new failure mechanisms relating to the low yield due to thicker gate oxide, silicon crystalline defects and QBD failure were also discussed.

In this paper, some low yield cases in Flat ROM device (0.45 and 0.6 µm) were investigated. To find killer defects and particle contamination, KLA, bitmap and emission microscopy techniques were used in fault isolation. Reactive ion etching (RIE) and chemical delayering, 155 Wright Etch, BN+ Etch and scanning electron microscope (SEM) were used for identification and inspection of defects. In addition, energy-dispersive X-ray microanalysis (EDX) was used to determine the composition of the particle or contamination. During failure analysis, seven kinds of killer defects and three killer particles were found in Flat ROM devices. The possible root causes, mechanisms and elimination solutions of these killer defects/particles were also discussed.

In our previous paper [1], discolored bondpads due to galvanic corrosion were studied. The results showed that the galvanic corrosion occurred in 0.8 ìm wafer fabrication (fab) process with cold Al alloy (Al-Si, 0.8 wt%-Cu, 0.5 wt%) metallization. Galvanic corrosion is also known as a two-metal corrosion and it could be due to either wafer fab process or assembly process. Our initial suspicion was that it was due to a DI water problem during wafer sawing at assembly process. After that, we did further failure analysis and investigation work on galvanic corrosion of bondpads and further found that galvanic corrosion might be due to longer rinsing time of DI water during wafer sawing. The rinsing time of DI water is related to the cutting time of wafer sawing. Therefore, some experiments of wafer sawing process were done by using different sizes of wafer (1/8 of wafer, a quadrant of wafer and whole of wafer) and different sawing speed (feed-rate). The results showed that if the cutting time was longer than 25 minutes, galvanic corrosion occurred on bondpads. However, if the cutting time was shorter than 25 minutes, galvanic corrosion was eliminated. Based on the experimental results, it is concluded that in order to prevent galvanic corrosion of bondpads, it is necessary to select higher feed-rate during wafer sawing process at assembly houses. In this paper, we will report the details of failure analysis and simulation experimental results, including the solution to eliminate galvanic corrosion of bondpads during wafer sawing at assembly houses.