Abstract Fan Out - Panel Level Packaging (FO-PLP) has redistribution layers (RDLs) which connect IC to a substrate. And each layer in the RDLs is connected through copper micro-vias. Viarelated defects including via separation are very critical because they can escape from electrical test and be found in the field. So many cleaning methods have been developed to keep the target pad surface free of oxides or organic contamination before forming vias. In this paper, we present a via separation case caused by alkaline cleaning introduced before seed metal deposition for electroplating of copper. We investigated the cause by analyzing the microstructure and chemical composition using a focused ion beam (FIB) and a transmission electron microscope (TEM) equipped with an energy dispersive spectrometer (EDS). Via separation, interestingly occurred at the interface between the seed Ti and the seed Cu not the interface between the seed Ti and the target pad..Cu surface which is known to be weak. We suggest a mechanism that structural imperfections at the outer rim of via bottom and galvanic couple of titanium and copper are involved in the separation of vias. Since two dissimilar metals of Ti and Cu are in direct contact, galvanic corrosion can occur in the presence of alkaline solution and discontinuities in the seed Ti layer. We found that galvanic corrosion in the studied system can be further complicated by the existence of copper oxide and titanium oxide as well as Cu and Ti.

Abstract This paper describes the investigation of donut-shaped probe marker discolorations found on Al bondpads. Based on SEM/EDS, TEM/EELS, and Auger analysis, the corrosion product is a combination of aluminum, fluorine, and oxygen, implying that the discolorations are due to the presence of fluorine. Highly accelerated stress tests simulating one year of storage in air resulted in no new or worsening discolorations in the affected chips. In order to identify the exact cause of the fluorine-induced corrosion, the authors developed an automated inspection system that scans an entire wafer, recording and quantifying image contrast and brightness variations associated with discolorations. Dark field TEM images reveal thickness variations of up to 5 nm in the corrosion film, and EELS line scan data show the corresponding compositional distributions. The findings indicate that fluorine-containing gases used in upstream processes leave residues behind that are driven in to the Al bondpads by probe-tip forces and activated by the electric field generated during CP testing. The knowledge acquired has proven helpful in managing the problem.

Abstract At ESREF 2008, the paper “Unusual defects, generated by wafer sawing: Diagnosis, mechanisms and how to distinguish from related failures” had won the Best Paper Award. In the meantime, new experiences were collected, related to new methods as laser sawing and its specific ESD risks and additional failure mechanisms as backside damage, charging of foils, pad corrosion and sawing residue damages. This paper explains in detail these failure sources, including detailed explanations on root causes and physical mechanisms as well as important hints for failure analysts how to distinguish related failure signatures from those, which look similar but are of other origin.

Abstract In wafer fabrication (Fab), Fluorine (F) based gases are used for Al bondpad opening process. Thus, even on a regular Al bondpad, there exists a low level of F contamination. However, the F level has to be controlled at a lower level. If the F level is higher than the control/spec limits, it could cause F-induced corrosion and Al-F defects, resulting in pad discoloration and NSOP problems. In our previous studies [1-5], the theories, characteristics, chemical and physical failure mechanisms and the root causes of the F-induced corrosion and Al-F defects on Al bondpads have been studied. In this paper, we further study F-induced corrosion and propose to establish an Auger monitoring system so as to monitor the F contamination level on Al bondpads in wafer fabrication. Auger monitoring frequency, sample preparation, wafer life, Auger analysis points, control/spec limits and OOC/OOS quality control procedures are also discussed.

Abstract Contamination by particles is one of the major causes of failures in integrated circuits. In some cases, particles may absorb moisture leading to electrochemical migration, dendrite growth, and electrical leakage and short failures. This work presents two case studies of particle induced corrosion of copper wire bond that resulted in an electrical failure. In the first case, adjacent pin resistive short failures were found to fail due to corrosion and electrochemical migration at wires that were in contact with calcium chloride particles. Analysis showed that the highly hygroscopic calcium chloride particles absorbed moisture and resulted in corrosion and electrochemical migration of the copper wires. For the second case, an electrical open failure after temperature cycle reliability test was found to be due to an organophosphorus particle being in contact with the wire.

Abstract Coating of the Cu bond wire with Pd has been a rather widely accepted method in semiconductor packaging to improve the wire bonding reliability. Based on comparison of a Cu bond wire and a Pd-coated Cu bond wire on AlCu pads that had passed HAST, new insight into the mechanism of the reliability improvement is gained. Our analysis showed the dominant Cu-rich intermetallics (IMC) were Cu3Al2 for the Cu wire, and (CuPdx)Al for the Pd-coated wire. The results have verified the Cu-rich IMC being suppressed by the Pd-coating, which has been extensively reported in literature. Binary phase diagrams of Al, Cu, and Pd indicate that the addition of Pd elevates the melting point and bond strength of (CuPdx)Al compared with CuAl that formed with the bare Cu wire. The improvements are expected to decrease the kinetics of phase transformation toward the more Cu-rich IMC. With the suppression of the Cu-rich IMC, the corrosion resistance of the wire bonding is enhanced and the wire bonding reliability improved. We find that Ni behaves thermodynamically quite similar to Pd in the ternary system of Cu wire bonding, and therefore possesses the potential to improve the corrosion resistance.

Abstract Recently a phenomenon has been found that shows different corrosion rates over Al bond pad regardless of different densities of Cl, F components on Al bond pads in different products. According to the results of analysis, the products showed different corrosion rates for different etch conditions of the bond pad opening. For the cause analysis, we conducted a cross-sectional profile comparison between two products with Al bond pads. Based on the result of comparison, we discovered that the side wall profile of the Al bond pad is affected by the etch conditions of the bond pad opening. In some severe cases, it was observed that a small void was formed between the side wall and Al bond pad. Under moist conditions, this void provided moisture between Al bond pad and TiN barrier metal that the electricity contacted. Through this study, we could conclude that the moisture in the void between Al bond pad and TiN barrier metal may create a galvanic corrosive condition.

Abstract In the authors' previous papers, the failure mechanism and elimination solutions of galvanic corrosion (Al-Cu cell) on microchip Al bondpads in the Al process (0.18un and above) have been studied [1-2]. In this paper, the authors will further study the failure mechanism and root cause of galvanic corrosion (Al-Cu cell) on microchip Al bondpads in the Cu process (0.13um and below) with Ta barrier metal. Based on our results, the root cause of galvanic corrosion (Al-Cu cell) in the Al process is only one way and Al-Cu cell is from Al alloy (Al + 0.5%Cu) on Al bondpads. However, in the Cu process it may be from two ways and Al-Cu cell can be from both Al alloy (Al + 0.5%Cu) on Al bondpads and the Cu metal layer below the barrier metal Ta when Ta has weak points or pinhole. As such, the pinhole defects on Al bondpad caused by galvanic corrosion (Al-Cu cell) in the Cu process might be more serious than that in the Al process. In this paper, TEM is used for root cause identification. Based on the TEM results, galvanic corrosion was due to the weak point/pinhole at the Ta barrier metal layer and Al-Cu diffusion.

Abstract Cu wires were bonded to AlSi (1%) pads, subsequently encapsulated and subjected to uHAST (un-biased Highly Accelerated Stress Test, 130 °C and 85% relative humidity). After the test, a pair of bonding interfaces associated with a failing contact resistance and a passing contact resistance were analyzed and compared, with transmission electron microscopy (TEM), electron diffraction, and energy-dispersive spectroscopy (EDS). The data suggested the corrosion rates were higher for the more Cu-rich Cu-Al intermetallics (IMC) in the failing sample. The corrosion was investigated with factors including electromotive force (EMF), self-passivation of Al, thickness and homogeneity of the Al-oxide on the IMC, ratio of the Cu-to-Al surface areas exposed to the electrolyte for an IMC taken into account. The preferential corrosion observed for the Cu-rich IMC is attributed to the high ratios of the surface areas of the cathode and anode that were exposed to the electrolyte, and the passivation oxide of Al with the lower homogeneity. The corrosion of the Cu-Al IMC is just a manifestation of the well-known phenomenon of dealloying. With the understanding of the corrosion mechanisms, prohibiting the formation of Cu-rich IMCs is expected be an approach to improve the corrosion resistance of the wire bonding.

Abstract A land-grid array connector, electrically connecting an array of plated contact pads on a ceramic substrate chip carrier to plated contact pads on a printed circuit board (PCB), failed in a year after assembly due to time-delayed fracture of multiple C-shaped spring connectors. The land-grid-array connectors analyzed had arrays of connectors consisting of gold on nickel plated Be-Cu C-shaped springs in compression that made electrical connections between the pads on the ceramic substrates and the PCBs. Metallography, fractography and surface analyses revealed the root cause of the C-spring connector fracture to be plating solutions trapped in deep grain boundary grooves etched into the C-spring connectors during the pre-plating cleaning operation. The stress necessary for the stress corrosion cracking mechanism was provided by the C-spring connectors, in the land-grid array, being compressed between the ceramic substrate and the printed circuit board.

Abstract In this study, a comprehensive investigation of the Ag-Al bond degradation mechanism in an electrically failed module using the argon ion milling, scanning electron microscopy (SEM), dual beam focused ion beam-SEM, scanning transmission electron microscopy energy dispersive x-ray spectroscopy, and time-of-flight secondary ion mass spectrometry is reported. It is found that the bond degradation is due to the galvanic corrosion in the Ag-Al bonding area. Specific attention is given to the information of microstructures, elements, and corrosive ions in the degraded bond. In this study, it is believed that the Ag-Al bond degradation is highly related to the packaging designs.

Abstract It is well-known that underetch material, contamination, particle, pinholes and corrosion-induced defects on microchip Al bondpads will cause non-stick on pads (NSOP) issues. In this paper, the authors will further study NSOP problem and introduce one more NSOP failure mechanism due to Cu diffusion caused by poor Ta barrier metal. Based on our failure analysis results, the NSOP issue was not due to the assembly process, but due to the wafer fabrication. The failure mechanism might be that the barrier metal Ta was with pinholes, which caused Cu diffused out to the top Al layer, and then formed the “Bump-like” Cu defects and resulted in NSOP on Al bondpads during assembly process.

Abstract In wafer fabrication, Fluorine (F) contamination may cause fluorine-induced corrosion and defects on microchip Aluminum (Al) bondpads, resulting in bondpad discoloration or non-stick on pads (NSOP). Auger Electron Spectroscopy (AES) is employed for measurements of the fluorine level on the Al bondpads. From a Process control limit and a specification limit perspective, it is necessary to establish a control limit to enable process monitor reasons. Control limits are typically lower than the specification limits which are related to bondpad quality. The bondpad quality affects the die bondability. This paper proposes a simulation method to determine the specification limit of Fluorine and a Shelf Lifetime Accelerated Test (SLAT) for process monitoring. Wafers with different F levels were selected to perform SLAT with high temperature and high relative humidity tests for a fixed duration to simulate a one year wafer storage condition. The results of these simulation results agree with published values. If the F level on bondpad surfaces was less than 6.0 atomic percent (at%), then no F induced corrosion on the bond pads was observed by AES. Similarly, if the F level on bond pad surfaces was higher than 6.0 atomic per cent (at%) then AES measured F induced corrosion was observed.

Abstract This paper presents a quick, reliable, and fully quantitative method of measuring the intermetallic coverage of copper to aluminium bonding at time zero and post reliability stressing. This method is currently used in select manufacturing quality control processes, as well as during product release procedures. By applying this measurement method after various life-tests, it has been possible to collect information on degradation in the copper aluminium system which is currently being used to make a model of the corrosion mechanism in the copper aluminium system.

Abstract Lead-free manufacturing regulations, reduction in circuit board feature sizes and the miniaturization of components to improve hardware performance have combined to make data center IT equipment more prone to attack by corrosive contaminants. Manufacturers are under pressure to control contamination in the data center environment and maintaining acceptable limits is now critical to the continued reliable operation of datacom and IT equipment. This paper will discuss ongoing reliability issues with electronic equipment in data centers and will present updates on ongoing contamination concerns, standards activities, and case studies from several different locations illustrating the successful application of contamination assessment, control, and monitoring programs to eliminate electronic equipment failures.

Abstract A case study of Fluorine (F)-outgassing is presented in this paper that caused the corrosion of Aluminum bond pad. It will be shown that the source of F-contamination is not the typical residue left behind after the passivation etch with Fluorine-based gas chemistry and the subsequent removal of the etch polymer generated with solvent (chemical) clean. Rather, it is introduced as a result of F-outgas over a period of time from the intermetallic dielectric (IMD) film, fluorosilicate glass (FSG), during the post-fab wafer storage. The methodology used in our failure analysis (FA) lab to identify and characterize this type of failure mode is presented in the paper.

Abstract Cu needs a higher level of ultrasound combined with bonding force to be bonded to the Al pad properly, not just because Cu is harder than Au, but it is also harder to initiate intermetallic compounds (IMC) formation during bonding. This increases the chances of damaging the metal/low k stack under the bondpad. This paper presents a fundamental study of IMC as well as one example of a failure mode of Cu/Al bonded devices, all based on detailed analysis using scanning electron microscopy, scanning transmission electron microscopy, energy dispersive spectrometers, and transmission electron microscopy. It presents a case study showing a corrosion mechanism of Cu/Al ballbond after 168hr UHAST stress. It is observed that all Cu9Al4 was consumed, while very little copper aluminide remained after 168 hours of UHAST stressing.

Abstract The authors use electron spectroscopy for chemical analysis and Auger electron analysis to study the interaction of Cl and F with Al thin-films deposited as thin-films on Si wafers and as Al bondpads. The motivation behind the study is F contamination being the putative source of poor throughput at wafer probe. F species stemming from NH4F and XeF2 exposure behave quite differently from HF on the Al surface. Whereas HF tends to attack the Al metal and leave an extended oxygenated-fluorinated surface, NH4F and XeF2 promote the formation of a stable, non-deliquescent fluoride salt of aluminum. HCl is far less corrosive to Al than HF, leaving a thin chlorinated-oxygenated surface. Immersion of Al thin-films in tetra-methyl-ammonium hydroxide (TMAH) and NH4OH provided non-halogenated surfaces for comparison. With exposure to air, the surface coated with the fluorinated Al salt (NH4F) adsorbs oxygen from the air to form a segregated AlF3/Al2O3 bilayer that remains stable with a total thickness on the order of 5 nm to 10 nm. Furthermore, wafers treated with NH4F display stellar throughput performance at wafer test despite having surface F contamination. A mechanical rather than chemical model is proposed to explain the improved performance at wafer probe with the immersion of wafers in a bath containing fluoride salts before wafer probe.

Abstract The use of in-situ lift-out combined with focused ion beam milling has become a favorable choice as it offers several indispensable advantages compared to the conventional mechanical and ex-situ lift-out sample preparation techniques. This paper discusses the procedures of the multiple-post in-situ lift-out grids preparation using a dicing saw. In addition, a real case is described to show that the multiple-post in-situ lift-out grids have been successfully applied to failure analysis. The multiple-post in-situ lift-out grids provide more positions and flatter surfaces for TEM sample mounting. The flat surface greatly increases the mounting efficiency and success rate. For the real case application, a thick Al fluoride oxide layer and Al corrosion were found above the Al bond pads, which had NOSP problem, and their neighbor area, respectively.

Abstract Fluorine-induced corrosion is one of the well-known failure modes of Al bondpad leading to non-stick on pad (NSOP) issues. Exposure to moisture (H2O) and atmosphere (O2) play an important role in determining the shape and size of the Al-F corrosion defects on the Al bondpad surface. In this paper, we will propose a laboratory simulation methodology that can reproduce F-induced defects observed either at the wafer fab or the assembly house. The methodology, known as SLAT ( S helf L ifetime A cceleration T est), is used to study the relationship between the F-corrosion defects and the relative temperature (T) and humidity (RH %). It is observed that Al-F corrosion defects simulated are similar to the real defects found in wafer fab and assembly house in our previous studies. A relatively higher T and lower RH % results in the formation of the “crystal-like” defects, but if a relatively lower T and higher RH % condition is used, the “oxide-like” defects were formed. In this paper, we will compare the simulation results to the real defects found in the previous cases and discuss the failure mechanism. From the present study, the importance of controlling T and RH % during wafer storage to eliminate F-induced defects will be highlighted.