Evaluation Of Wire Bonding Performance, Process Conditions .
EVALUATION OF WIRE BONDING PERFORMANCE, PROCESSCONDITIONS, AND METALLURGICAL INTEGRITY OF CHIP ONBOARD WIRE BONDSDaniel T. Rooney, Ph.D., DeePak Nager, David Geiger, and Dongkai Shanguan, Ph.D.Flextronics Inc.130 Mosswood Blvd.Youngsville, NC 27596919-570-1486dan.rooney@flextronics.comdata. Wedge bond shear testing will become more popularas the metrology and test standards are developed.2However, wire bond integrity should not be characterized byonly performing wire pull and bond shear tests.ABSTRACTChip on board wire bonding presents challenges to modernwire bonding technology which include smaller, closelyspaced wire bond pads; bonding to soft substrates withoutspecial processing and pad construction; and diverse firstbond and second bond metallurgies. These challenges areaddressed by extensive bonding accuracy tests, a design ofexperiments approach for optimizing wire bond processparameters, reliability testing, and detailed materialscharacterization of the metallurgical integrity of the wirebonds. The thermo-mechanical integrity of the wire bondinterconnects was evaluated by wire pull and hot storagetests.The methodology of materials analyses of themetallurgy of the wire bond interconnects is described. Thepaper illustrates a wire bond lift technique that is used toinspect for cratering damage and the “area-uniformity” ofgold aluminum intermetallics. An improved understandingof the wire bonding process was achieved by showing thedependence of the visual appearance of the wire bonds onwire bond process parameters.This paper describes several analytical techniques and thereliability testing used to evaluate the wire bonds.3 Thesetechniques included: traditional wire pull tests; a visualcharacterization of the size and shape of the wire bonds; agrain structure evaluation of the bonding wire; and a“wedge-lift” method for evaluating intermetallic metallurgyand area uniformity of the wire bonds. Characterization ofvisual appearance of the wire bonds and bond widthmeasurements are important, but often overlooked,indicators of wire bond quality. The reliability testingincluded two chamber thermal cycling and high temperaturestorage.Wire Bond Process ParametersOptimizing a wire bond process begins with a clearunderstanding of the machine set-up, the response variablesinvolved, and their relationship to one another.Experimenting with these parameters is time well spent, andis an important step toward developing a robust wire bondprocess. An overview of thermosonic bonding and themachine set-up parameters is helpful for interpreting theresults of wire bond DOE data.4INTRODUCTIONOur chip on board wire bond applications require a finepitch gold wedge bond process to interconnect a silicon diewith aluminum bond pad metallization to an FR4 epoxyglass printed circuit board, (PCB) with electrolytic goldplating. Wire bonding on soft, easily deformed PCBsubstrates is a complex process which requires anunderstanding of multiple disciplines including: (1) themetallurgy and properties of the materials of construction;(2) the evaluation of mechanical integrity by wire pull andbond shear test methods; and (3) the methodology ofreliability testing. Wire bond processes are traditionallyoptimized by conducting designed experiments (DOE’s),with wire bond machine set-up parameters, includingultrasonic power and time duration, bonding tool force, andstage temperature as control factors. The response factors inthese DOE’s are typically wire pull and bond shear tests.1Thermosonic wedge bonding utilizes a combination of heat,pressure, and ultrasonic vibration to form a metallurgicalbond between two materials. The four key machine settingsare: bonding work stage temperature, ultrasonic transducerpower, bonding tool force, and bonding time. Heat softensthe gold bonding wire and gold board metallization. Powerand time relate to the ultrasonic generator settings used to“ultrasonically soften,” the bonding wire. Tool force is theamount of weight applied to the bonding wire tomechanically couple the bonding wire to the bonding padsurface. During the bonding process, the bond tooloscillates back and forth to soften a short piece of bondingwire that is wedged between the tool and bonding padWire bond shear testing is not commonly used to test thestrength of wedge bonds, but does provide complimentary90
maintenance due to contamination build-up and wear, anddamaged or burnt appearing wire bonds in areas where thetool contacts the bond.surface. As heat and ultrasonic energy soften the wire andpad metallization the bonding tool deforms the bonding wireagainst the bond pad and forms small microwelds betweenthe two materials. A more detailed description on how thesebonding parameters affect the quality of the wire bonds is asfollows.Bonding TemperatureThe test vehicle consisted of FR4 printed circuit boardsubstrates, which were bonded on a work holder withvarious temperature settings of 120 C, 130 C, and 140 C.The actual PCB temperatures were 105 C, 115 C, and 125 C at these respective settings. Within this temperaturerange, the board materials lose rigidity, which is known toresult in bond lifts from “ultrasonic cupping”. Ultrasoniccupping occurs when the bonding wire slips out from underthe contact area between the smooth (slippery), boardplating surface, the bonding wire and tool during wirebonding. The use of a cross-bonding tool can help reducethe extent of ultrasonic cupping problems.5Ultrasonic powerUltrasonic power has the strongest influence on bondquality and visual appearance, because it controls the extentof softening of the bonding wire. Insufficient power canresult in narrow, under formed bonds and tail lifts.Excessive power results in wire bonds with a “squashed’appearance, heel cracks, cratering damage to thesemiconductor die, undesirable build-up of residual bondpad metallization on the bonding tool, and poor mechanicalintegrity of the wire bonds. Uneven bond ears, excessivesplash, and high deformation (bond width to wire diameterratio), are visual indicators of excessive bond power.Excessive bond deformation can occur if either the devicebeing bonded is not properly secured or if the bond forceapplied to the bonding tool is too light. These can result ineither mechanical chattering of the device or insufficientmechanical contact between the bonding wire, bonding tooland bond pad surfaces. Both of these setup problems canresult in mechanical overworking of the bonding wire orcratering damage.Wire Bond Process OptimizationA wire bond process optimization study is usuallyconducted using a design of experiments approach when anew product is launched or when new equipment ispurchased. The control factors for the DOE are machinesettings including power, force, time, and temperature, andthe response factors are typically wire pull strength andshear strength. One objective of a DOE study is to vary thebonder settings in a systematic way to maximize the wirepull test values. A common strategy is to determine theoptimum settings for the machine and then vary theparameters until no sticks occur. This preliminary workprovides ranges for the machine set up parameters that areuseful for designing the DOE test matrix. A typical DOEstudy provides process windows where the bond integrity isacceptable as well as machine settings that cannot be used.A more ambitious DOE study determines process windows,or usable ranges for the machine settings, while making theprocess immune to noise factors such as bonding tooldegradation, variations in materials of construction, anddaily fluctuations of the process equipment.Bond ForceThe bond power and bond force parameters areinterdependent. Increases in bond power often necessitateincreases in bond force to allow for proper coupling ofultrasonic energy from the bonding tool to the bond wireand substrate materials. Conversely, excessively high bondforces can stall the scrubbing action of the bonding tool.Excessive bond force inhibits the transfer of energy from thebonding tool to the bonding surfaces, thereby inhibiting orpreventing the formation of a metallurgical bond. Excessivebond force can also result in the so - called unbonded centralregion. This occurs when welds form around the peripheryof the wedge but not in the middle, because contaminantsand oxides on the bonding surfaces were swept into thecenter of the bond during bonding. Good area uniformityof the metallurgical bonds occurs when the bondingparameters allow for breaking up and consumption ofcontaminants and oxides uniformly across the weld areasurfaces, rather than just at the periphery. The bond lifttechnique to be described later allows for inspection of thearea uniformity of the intermetallics that from on the firstbond to the die.Wire pull test results are not always linked to acceptablewire bond performance for a given product. One of theproblems with DOE studies for a new process is that it isdifficult to know whether the response factors such as wirepull test values are actually indicative of whether the wirebonds will meet the requirements of the product in it’s reallife application. A common approach is to choose a wirepull requirement based on similar processes for otherproducts, a military standard, or a commercial productstandard. Then a range of machine settings is determinedbased on the pull test values without considering otherresponse factors such as visual appearance and metallurgicalintegrity.Bonding TimeThe time in which the ultrasonic energy is turned on istypically on the order of tens of milliseconds; however, oncethe microwelds form they prevent further scrubbing of thewire into the substrate, effectively limiting the time in whichthe bonding process can actually occur. Time has the widestprocess window; however, excessive bond time can result inslow throughput in manufacturing lines, increased toolFor example an over compressed wire bond can have a goodwire pull strength in an as-bonded condition, but can bemechanically overworked and fail during reliability stresstesting or in the field. Another example would be an underformed first bond to the aluminum die metallization that91
meets the minimum wire pull test requirement, but hasspotty, non – uniform mixtures of gold-aluminumintermetallic phases with poor metallurgical characteristics.If non-uniform mixtures of intermetallic phases and unreacted aluminum are formed, these can grow into mixturesof more mature intermetallic phases that have differentvolume expansivities.Wire bonds with mixtures ofintermetallics have high levels of internal mechanicalstresses, and could exhibit premature wear out failures. Anideal wire bond process consumes all of the aluminum bondpad metallization forming a single intermetallic phase withgood area coverage that will increase the uniformity ofintermetallics, which continue to form during exposure ofthe wire bond to heat during post wire bonding packagingprocesses and field service.6surface roughness on wire bond ability for aluminum wire togold plated substrates in COB applications.8 The resultsindicated that smoother surfaces generally wire bond betterthan rougher surfaces.Samples with higher surfaceroughness could still be wire bonded, but exhibited a higherpotential for having bad bonds. The rougher the surface, theless ideal the bonding is due to a reduced contact areabetween the bonding wire and substrate pad. Also, therougher the surface is, the more likely the possibility thatcontaminants could become entrapped. Factors that affectsurface roughness include the grain refiners and currentdensity used in the gold and underlying nickel-platingprocesses. Surface roughness is also affected by brushing orpumice cleaning processes of the copper, which occurduring PCB processing.TEST VEHICLE CONSTRUCTION AND TESTMETHODSTest Vehicle Construction and PreparationThe test matrix consisted of three boards with eight panelswith six test dice on each panel. The PCB substrates werespecified as a high Tg FR4 (170 C), with ½ ounce copper,electroplated with 100 to 200 microinches of nickel,followed by 20 to 32 microinches of gold. The dimensionsof the test die are 4.3 x 4.3 x 0.57 millimeters, and thealuminum pad sizes and pitches are listed in Table 1. Dieattach was performed using Ablebond 84MILRS4 epoxy,and die mounting was performed using an ASM AD809machine. Bond placement accuracy, wire pull strength, andelectrical resistance of the wire bonds were measured toevaluate the process capability.Figure 1.a: SEM micrograph of plating, which exhibitedorange – brown color and poor bond abilityRow numberPad Size (mil)Pad Pitch (mil)(1)3.7 x 3.74.0(2)2.9 x 2.93.2(3)2.7 x 2.73.0(4)2.5 x 2.52.8Table 1: Pad size and pad pitch of the four largest rows ofthe ASM 96 test die.Figure 1.b: SEM micrograph of plating, which exhibitedgold color and good bond abilityMaterials Characterization of PCB PlatingIn the chip on board wire bond process, the quality of theplating has a significant impact on the integrity of the wirebonds.The thickness of the plating layers, surfacecontaminants, surface roughness and metallurgical hardnessare important characteristics, which affect the bond abilityof the PCB substrate.7 We encountered several problems inprocuring test vehicles that could be wire bonded andperform acceptably in reliability tests.Surface roughness, color, reflectivity, and hardness areinter-related. Smooth surfaces with fine grain structures arehighly reflective and have a yellow color, but can bemetallurgically hard. Hard gold can be wire bonded, but thebonding process parameters must be adjusted accordingly.Soft gold has large grains that reduce its reflectivity and canresult in a brown color, which is not cosmetically appealing.However, soft gold can be easier to bond to, since it morereadily deformed during the wire bonding process. On theother hand, brown colored gold plating can be indicative ofexcessive levels of grain refiners, surface contaminants, or arough surface, which can negatively impact wire bonding.Unfortunately, the visual appearance of the gold is not asimple indicator of the bond ability. Maintaining lot-to-lotconsistency in the critical plating characteristics can helpminimize the need for adjustments to the wire bond processparameters, and is a key step toward achieving a robust wirebonding process.Significant differences were observed in the color,reflectivity, and surface texture of PCB plating that waswire bondable and plating that was not wire bondable,Figures 1.a – 1.b. We found that samples with very thingold plating were wire bondable; however, these samplesperformed poorly in hot storage reliability tests. The wirebondable gold exhibited a shiny, yellow color and a smoothappearance, yet the non-bondable gold was brown andrough appearing. DiGirolamo evaluated the effect of92
Plating thickness is also an important factor in the bondability of PCB substrates. The gold and nickel requirementsspecified by ANSI/IPC-SM-784 for gold thermosonicbonding are 1 to 1.3 microns (40 to 50 microinches), of goldminimum over 1.3 to 5 microns (50 to 200 microinches), ofnickel. The gold should be “soft” gold with 99.9 % purity,and a hardness of Grade A-Knoop 90 maximum. Thehistorical “rule of thumb” for gold plating thickness hasbeen a minimum of 30 microinches. In practice, newerproducts and processes use thinner gold. In our DOE , agold thickness of 25 microinches was the target for the goldplating with a minimum allowed thickness of 20microinches. Plasma cleaning helps remove inter-diffusednickel oxides from the surface of gold allowing for the useof thinner gold plating in some applications. Cross-sectionsof bondable and non-bondable boards exhibited excessivethickness (400 - 500 microinches), in the nickel plating onnon-bondable boards, Figures 2.a – 2.b.The excessivenickel thickness and the rough surface finish of theoverlying gold suggested that there was a problem in henickel plating process, which affected the surface textureand bond ability of the bond pads.Nickelmedium deformation of 1.8 is typical for 60 kHz bondwidth, whereas a medium deformation of 1.4 is typical for100 kHz. Lower deformation is a major benefit of usinghigh frequency ultrasonic energy; however, the tooldisplacement and resultant scrubbing action is greater forthe 60 kHz machine, which makes a 60 kHz machine lesssensitive to problems with surface contaminants than a 100kHz machine. The bonding wire was American Fine Wire(P/N AW 14), 1 mil gold wire with a tensile strength of 12– 17 gf and elongation of 0.5 – 3.0%. The bonding tool wasa Small Precision Tool titanium carbide wedge tool (modelFP30B-TI-2020-L-CGM), with a cross-groove design and2.8 mil foot width. The extra mechanical gripping action ofthe cross-groove tool allows for more efficient transfer ofenergy from the tool wire interface to the bonding surfaces.The bonding parameters for the bonding accuracy, the wirepull, and resistance tests are listed in Tables 2 and 3. Theinput factors for the DOE study were power, force, time andwork stage temperature. The output factors for the DOEwere wire pull strength and bond width. 112 wires werebonded on each test vehicle for the bonding accuracy test.30 wires were bonded in series on PCB pads for theelectrical resistance test.Rough 2324copperFigure 2.a: SEM photo of cross section of plating with poorbond ability. Over-plating reduces gold smearing.Over-platingSmooth goldFigure 2.b: SEM photo of cross section of plating withgood bond ability, Note the gold layer is whiteWire BondingAn ASM model AB559A-06 rotary bond head automaticwedge bonder equipped with accessories for gold wirebonding was used to perform the evaluation. The modelAB559A-06 uses a 60 kHz ultrasonic generator, and is fullyautomated. Use of a 60 kHz transducer requires greaterwire deformation (ratio of bond width to bonding wirediameter), to achieve optimum bond strength than theincreasingly popular 100 kHz high frequency transducer. AStageTemp.( C)120 C120 C120 C120 C120 C120 C120 C120 C130 C130 C130 C130 C130 C130 C130 C130 C140 C140 C140 C140 C140 C140 C140 C140 4070/12070/14070/12070/14070/12070/140Table 2: Bonding parameters for bonding accuracy andwire pull tests. Actual PCB temperatures were 105 C, 115C, and 125 C at 120 C, 130C, and 140 C settings93
Fail CodeDescription1Wire break at midspan2Wire break at die3Wire break at PCB4Lifted wire at die5Lifted wire at PCB6Lifted die pad metallization7Wire touching adjacent wire8Bond Placement EccentricTable 5: Failure modes for bond accuracy and wire pull c)(grams)( C)1 - 242020120130 CTable 3: Bonding parameters for wire resistance testBonding accuracy testThe bonding accuracy test was performed by visualinspection of die sets 1 – 4. Table 4 lists the number ofoccurrences of the bonding failures, and Table 5 lists theclasses of failures that could have occurred. No seriousanomalies were found, with the exception some of the wirebonds were not centered with respect to the bond pads.These failures resulted from misaligned teach points for thelocations of the wire bonds, when the wire bond machinewas programmed.TestPCBStageTemp.( C)12345678120 C120 C120 C120 C120 C120 C120 C120 C91011121314151617130 C130 C130 C130 C130 C130 C130 C130 C140 C18140 C19140 C20140 C21140 C22140 C23140 C24140 CDie 1Defectcode(pad)8(113)Die 2Defectcode(pad)Die 3DefectcodeTemp. C120130140Table 6:Die 4Defectcode(pad)#1234567891
the gold bonding wire and gold board metallization. Power and time relate to the ultrasonic generator settings used to “ultrasonically soften,” the bonding wire. Tool force is the amount of weight applied to the bonding wire to mechanically couple the bonding wire to the bonding pad surface. D
the Cu wire bonding technology presents a promising alternative. 2.2. Cu wire bonding In general, to judge the wire bonding ability, we control the bonding ball shear and the bonding wire pull (Fig. 2.3). Value set-tings differ according to the type of assembly package. Moreover, wire diameters also differ depending on whether Au or Cu was used.
comparing with Au wire bonding. Bonding force for 1st bond is the same range, but approx. 30% higher at 2nd bonding for both Bare Cu and Cu/Pd wire bonding but slightly lower force for Bare Cu wire. Bonding capillary is PECO granular type and it has changed every time when new cell is used for bonding
black black white white white white red green blue black Wire 3 Wire 3 Wire 4 Wire 4 Wire 5 Wire 6 Wire 8 Wire 7 Wire 2 e 2 Wire-number Wire color Wire-number e color For multicolor configurations For all other configurations * internal pull-up resistor to vcc 3.3 V Dimension 15.15
Wire bonding was conducted on a gold ball bonder at a bonding temperature of about 155-160 C on the PCB surface using 25um 4N gold wire. The bonding tool used for the wire bonding process was for off-the-shelf 80um pitch applications with capillary tip dimensions of about 100um. Wire pull tests were
Wire Size Wire Size 20.75mm 1.50mm2 E1010RD EN2502 (10mm) (12mm) Wire Size Wire Size 21.00mm2 2.50mm E1510BK EN4012 (10mm) (12mm) Wire Size 21.50mm 4.00mm2 E2512GY EN6012 (12mm) (12mm) Wire Size Wire Size 2.50mm2 26.00mm E4012OR EN10012 (12mm) (12mm) Wire Size Wire Size 4.00mm2 10.00mm E6012GR EN61018 (12mm) Wire Size Wire Size 26.00mm2 16.00mm .
Modern Chemistry 1 Chemical Bonding CHAPTER 6 Chemical Bonding SECTION 1 Introduction to Chemical Bonding OBJECTIVES 1. Define Chemical bond. 2. Explain why most atoms form chemical bonds. 3. Describe ionic and covalent bonding. 4. Explain why most chemical bonding is neither purely ionic or purley 5. Classify bonding type according to .
from electric shock. Bonding and earthing are often confused as the same thing. Sometimes the term Zearth bonding is used and this complicates things further as the earthing and bonding are two separate connections. Bonding is a connection of metallic parts with a Zprotective bonding conductor. Heres an example shown below.
lations of physical systems, using the Python programming language. The goals of the course are as follows: Learn enough of the Python language and the VPython and matplotlib graph-ics packages to write programs that do numerical calculations with graphical output; Learn some step-by-step procedures for doing mathematical calculations (such as solving di erential equations) on a computer; Gain .
