Ball Bond Research Articles

Effect of Bromine in Molding Compounds on Gold - Aluminum Bonds

Degradation rates of gold wire ball bonds on aluminum bonding pads were studied in two molding compounds as a function of bromine concentration, temperature, and time at temperature. The measure of degradation was the resistance increase of the aluminum-gold contact. The rate of degradation was observed to increase with increasing bromine content and temperature for both molding compounds. However, the bromine content dependence of the activation free energy of the degradation reaction indicates a different degradation mechanism for each molding compound.

Ball Formation in Aluminum Ball Bonding

The effects of various factors, i.e., shield gas, current density, wire polarity, and wire material, which can influence oxidation of ball surfaces were investigated. Good quality aluminum balls were found to be formed if optimum electric and shield gas conditions were employed; for example, current density <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">geq</tex> 2.5 GA/m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> wire polarity: cathode; shield gas: Ar <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> H <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> . Good quality aluminum balls of 1.7 to 2.5 times the wire diameter could be formed by varying discharge time and/or current density. It was found that they could be obtained independently of the wire material. The relation between morphologies of aluminum balls and the degree Of oxidation was also investigated in order to examine the mechanism for obtaining a good quality bail. A close correlation was found between ball morphologies, i.e., eccentricity, sphericity, and constriction, and the degree of oxidation. The morphologies were significantly improved as the oxide film thickness was reduced.

Gold-Aluminum Intermetallics: Ball Bond Shear Testing and Thin Film Reaction Couples

The gold-aluminum intermetallic phases are known to contribute to the long term degradation of gold wire ball bonds to aluminum metallization on semiconductor devices. The properties of these intermetallics were investigated through two techniques: 1) ball bond shear testing and 2) preparation of known intermetallic compositions by using thin film Au-Al diffusion couples. The influence of thermal aging on gold ball bonds attached to aluminum metallization (on both Si and SiO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> ) was determined through ball shear measurements. The strength of gold wire bonds made to films of Au-Al intermetallics was similarly measured. Failure mode analyses of the sheared bonds for the various systems examined are presented. The effect of a nondestructive ball shear test on the ultimate ball shear strength is given. Ultraviolet (UV)-ozone cleaning of the aluminum prior to gold wire bonding is discussed in light of Auger spectroscopy of the aluminum oxide formed on the surface. The effect of shearing direction relative to the direction of the ultrasonic scrubbing was also examined.

A comparison of the Bond method for sizing wet tumbling ball mills with a size—mass balance simulation model

A comparison is made of the results from a ball mill model simulation with those of the conventional Bond ball mill design method, for a material whose breakage characteristics and Work Index have been determined. In order to perform the simulation, ‘normal’ values were chosen for make-up feed size distribution, mill residence time distribution, ball mix, classifier behaviour, etc. At high flow rates through a mill (low reduction ratio), it is necessary to allow for the reduction in breakage rates caused by over-filling of the mill, using an empirical mass transfer relation: filling ▪. By suitable choice of the constant in this relation, it was found possible to duplicate the variation of mill capacity with feed size and product size (in wet closed-circuit grinding) predicted by the Bond method. The simulation model is then used to predict the variation of performance with design variables not included in the Bond method.

Reliability implications of destructive gold wire bond pull and ball bond shear testing : J. W. Stafford. Semiconductor Int., 83 (May 1982)

Reliability implications of destructive gold wire bond pull and ball bond shear testing : J. W. Stafford. Semiconductor Int., 83 (May 1982)

A New Failure Mechanism: Al-Si Bond Pad Whisker Growth During Life Test

The results are presented of a study conducted to determine reasons for and kinetics of the growth of thin aluminum whiskers from the bond pads of semiconductors. These whiskers were found to grow to lengths of up to 200 µm from thin-film aluminumsilicon on N-type silicon gate metal-oxide semiconductors (NMOS). Whisker growth was found to be due to the compressive stress generated in the metal during wire bonding. The problem is complicated by the presence of silicon grains in the Al-Si film and was found only on parts using gold ball bonds. The growth rate is temperature dependent with a thermal activation energy of 0.7 eV. Copper doping of the metallization as well as the use of aluminum ultrasonically bonded wires were both found to effectively prevent whisker growth.

Bondability Problems Associated with the Ti-Pt-Au Metallization of Hybrid Microwave Thin Film Circuits

The cause for unacceptable bonding behavior of a new high-resolution, close-tolerance hybrid thin film circuit metals system was investigated. The conductor/resistor technology was developed at Hewlett-Packard for microwave applications. It consisted of Ti-PtAu/Ta <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> N where conductors were defined by a sputter etching process with the aid of an auxiliary Ti mask at the Au surface. The cause of the bondability problem was determined by Auger electron spectroscopy (AES) to be residual Ti-O compound at the conductor surface. An evaluation scheme which specifically addressed bond acceptability and integrity with respect to mechanical, visual, and reliability testing was developed. Types of bonds and the methods of testing associated with each bond type were investigated in terms of the apparent bond failure mechanism. Various wire and ribbon bonding processes were examined. A series of bond evaluation techniques which include both qualitative and quantitative ball shear and wire pull tests were performed. A qualitative shear test of the thermocompression Au ball bond was shown to be the most reliable indicator of bond integrity and acceptability. Bond stick rates and wire pull test results were found to be, in themselves, unreliable yardsticks for determining bond quality. Additional information was obtained from AES samples with excellent (close to 100 percent) stick rates but poor shear test characteristics. Various surface treatments to enhance bondability were investigated. Three different concentrations of H <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> O:HF:HNO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</inf> solutions as well as a Au etch were examined. A Ti etch consisting of H <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> O, HF, and HNO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</inf> , (10:1:1, respectively) was selected and easily implemented as a final processing step before bonding. Results comparing as-processed with Ti etched (10:1:1) circuits revealed a marked increase in stick rate (from ~50 to ~100 percent), ball bond shear strength and wedge bond pull strength, without altering existing bonding processes currently employed in production. Furthermore, environmental testing of exposed bonds on Ti etched (10:1:1) conductor surfaces confirmed bond integrity.

Resolution of the Gold Wire Grain Growth Failure Mechanism in Plastic Encapsulated Microelectronic Devices

The relative statistical influence of bonding parameters and assembly materials on the failure frequency rate of microelectronic devices in plastic encapsulated packages due to grain growth, failure are resolved. Wire creep is a stressinduced break in a thermocompression gold wire ball bond which most frequently occurs within several wire diameters above the ball in plastic encapsulated microelectronic devices. Thermal Shock, Condition C, was used as the standard test by which to measure performance in this investigative program. A complete analysis was made of all materials and processing factors which would potentially contribute to the wire creep, grain growth failure mechanism. Over 75 test matrices were defined and executed with the demonstrated result that only four variables had apparent major influence on this failure mode: 1) the gold wire impurity constituents, 2) the type of wire bonding capillary, 3) the gold wire ball formation technique, and 4) the lead frame plating impurities. A full factorial experiment was conducted to delineate the relative influence of these four contributors. The results of this investigation show that: 1) lead frame plating impurities can significantly accelerate the wire creep, grain growth failure frequency; 2) hydrogen torch flame-off ball formation always results in wire creep, grain growth failure; 3) the type of capillary contributes to this failure to a lesser but still significant extent; and 4) the type of wire is the least significant contributor, but within this variable the level of grain modifying impurities is significant. Hypotheses, supported by additional experimental investigations, with foundations in metallurgy have been formulated. Corrective actions implemented as a result of this investigation have reduced the average cumulative failure frequency rate of 30 percent at 250 thermal shock cycles to zero cumulative failures at greater than 3000 thermal shock cycles.

Contamination Control in Lead Bonding to Thick and Thin Films

Degraded wax residues from processing steps have been found to be a major source of contamination on beam lead devices for hybrid microcircuits. Two avenues of approach were used to solve bonding problems traced to this contamination: 1) a search was initiated for a more thermoxidatively stable and more easily removed wax, and 2) an effort was made to determine the level of cleanliness necessary for successful bonding and the most efficient method of attaining this level. A series of halocarbon waxes (polytrifluorochlorocthylenes) was found to be stable both chemically and thermally. The carbon contamination levels were decreased after improved solvent cleaning techniques and measured ~2 rim. This level compares with values as high as 9 nm after typical solvent cleaning. Plasma and ultraviolet cleaning both reduced carbon residues to 0.2-0.8 nm. Ball bonds made with weld forces of 1.0-1.5 N were successful on ultraviolet (UV) and plasma cleaned surfaces near 200°C and near 327°C for surfaces cleaned by improved solvent techniques. The beeswaxes presently used had adequate and better bondability than the more stable halocarbon waxes when only solvent cleaning was used and temperatures did not exceed 125°C to avoid charring.

Electrical bias level influence on THB results of plastic encapsulated NMOS 4K RAM's

Samples of plastic encapsulated NMOS/LSI dynamic 4K RAM's were placed on 85°C/85-percent relative humidity (RH) tests at four different bias levels. No level exceeded specification maxima. Tests were continued as long as practical to allow observation of the entire failure distribution of most of the test groups. Over 394 000 actual device-hours were accumulated. Detailed failure analysis was performed on all failures to allow for proper censoring of life test data. Censored failure mechanisms included mobile ionic contamination, package lead corrosion and plating, electrical overstress (mishandling), tester error, and in several cases, electroplating of gold from the ball bonds. Failure data representing only failures due to moisture-related electrochemical corrosion of the IC aluminum metallization were analyzed. This failure mechanism accounted for 78 percent of the failures. Failure distributions of each test group were of log-normal nature. Three test groups were studied (Chip A/Package 1, Chip B/Package 1, and Chip B/Package 2). Test results indicated an exponential acceleration of the failure rate with increase in electrical bias level. Bias versus median life curves for the three test groups examined as a composite allow for statements as to the moisture-resistance characteristics of the packages and of the IC chips.

High-Reliability Thermocompression Bonding for Hybrid Applications

The problem of producing reliable thennocomprassion gold ball bonds to aluminum metallization has been discussed by many authors. In most of these discussions, little information is provided to relate bond reliability to process parameters-cleanliness and bonder settings. The influence of bond and capillary temperatures upon bond integrity is discussed. The use of beryllia capillaries to achieve the high capillary temperatures achievable with pulse tip bonding without sacrificing production throughput is discussed. An accelerated life test for ascertaining bond reliability is discussed and shown to be an effective tool for process prove-in. Finally, a bonding technique is discussed that not only meets the requirements of MIL STD 883A, but also will withstand temperature stresses equivalent to 300°C storage for 4 h, 2000 h of accelerated operating life, and 1300 temperature cycles.

Nondestructive Bond Pull in High-Reliability Applications

The problem of producing reliable thermocompression gold ball bonds to aluminum metallization has been discussed by many authors. In most of these discussions, little information is provided to relate bond reliability to process parameters-cleanliness and bonder settings. The influence of bond and capillary temperatures upon bond integrity is discussed. The use of beryllia capillaries to achieve the high capillary temperatures achievable with pulse tip bonding without sacrificing production throughput is discussed. An accelerated life test for ascertaining bond reliability is discussed and shown to be an effective tool for process prove-in. Finally, a bonding technique is discussed that not only meets the requirements of MIL STD 883A, but also will withstand temperature stresses equivalent to 300°C storage for 4 h, 2000 h of accelerated operating life, and 1300 temperature cycles.

Application of scanning electron microscopy to integrated circuit failure : J. R. Devaney, Solid St. Technol., March (1970), p. 73

Integrated circuit failure analysis with scanning electron microscopy taking into account ball bond contamination and open metallization at contact windows

Failure Modes in Gold-Aluminum Thermocompression Bonds

Electrical opens and mechanical fracture have been reported as two distinct failure modes in gold-aluminum thermocompression ball bonds. It is found that the thickness of the aluminum film to which the gold wires are bonded and the temperature at which the bonds are held during aging determine which failure mode will be observed. The electrical failure mode predominates when bonds made to aluminum films 5000A or more in thickness are aged at temperatures above 200°C but at 200°C both failure modes are observed. Mechanical fracture is expected to be the dominant failure mode when ball bonds to these 5000-10000-A films are aged at temperatures below 200°C. Bonds to thinner aluminum films (500-3000 A) exhibited only the electrical failure mode.

An improved form of thermocompression bond

The mechanical reliability of both the ball and the chisel thermocompression bond, which are employed at the present time in many semiconductor devices, has been assessed by testing large numbers of sample bonds in a high speed centrifuge and in a microtensile testing machine. The rapid formation of detrimental compounds at a gold-aluminium interface prohibits the use of the gold wire-aluminium film ball bond in the fabrication of long life transistors and the all-aluminium chisel bond is considered to be relatively unreliable. A study of the plastic flow of metal which occurs during the formation of a thermocompression bond has led to the development of a new bond structure, the eyelet bond, which enables aluminium wires to be bonded successfully to a number of substrates, including heavily oxidized aluminium films and both clean and lightly oxidized Kovar surfaces with high and very reproducible bond strengths. Transistors in which the new structure has been used for bonding aluminium wires to the aluminium contact areas and to the plain Kovar posts have now been stored at a temperature of 300°c for 1000 hours. The initial characteristics have remained virtually unchanged and no deterioration in mechanical reliability has been observed.