Predicting Intermetallic Layer Growth and Interfacial Phase Transformations during Solder Reflow Operations

Melt

When liquid SAC305 solder contacts a copper land pad at 240°C, chemical dissolution begins within milliseconds. Copper atoms dissolve into the tin-rich liquid bath, driven by the concentration gradient between the pad boundary and bulk liquid. Dissolution kinetics follow the Nernst-Brunner model, where concentration changes over time depend on liquid volume, interfacial surface area, and the mass transfer coefficient across the diffusion boundary layer.

In a standard reflow pass, solder remains above the 217°C liquidus temperature for 45 to 75 seconds. During this window, dissolving copper rapidly saturates the local liquid boundary, creating the conditions required for intermetallic precipitation.

Substrate metallization dictates these initial dissolution dynamics. Bare copper lands dissolve at 0.05 to 0.12 micrometers per second at peak reflow temperatures, quickly saturating the adjacent liquid solder. Electroless nickel immersion gold finishes change this sequence entirely.

The gold layer ~ typically 0.05 to 0.10 micrometers thick ~ dissolves into the molten solder almost instantly on contact, leaving the underlying nickel-phosphorus layer exposed to the tin bath. Because nickel dissolves into liquid tin at roughly one-tenth the rate of pure copper, interface erosion slows, forcing nickel-tin intermetallics to nucleate instead of copper-tin compounds.

Thermodynamics of Substrate Dissolution

Liquid alloy composition shifts continuously during peak reflow as pad metal enters the melt. In a SAC305 joint, dissolved copper pushes the local liquid composition past the binary eutectic threshold of 0.7 weight percent copper. Once peak temperature stabilizes, the melt can no longer hold the excess copper in solution, and thermodynamic equilibrium forces solid intermetallic phases to precipitate directly at the liquid-solid boundary.

Free energy of mixing calculations show that binary copper-tin intermetallics have a much lower Gibbs free energy than the surrounding liquid, driving phase formation once solubility limits are crossed.

Flux chemistry modifies dissolution by stripping native surface oxides ahead of the advancing melt. Active organic acid fluxes lower interfacial energy and accelerate wetting across the copper land. Halide-activated fluxes strip copper oxides faster than mildly activated rosin formulas, though high halide concentrations speed up copper erosion during extended time-above-liquidus profiles.

Removing the oxide barrier allows direct metallic contact, driving up the kinetic rate constant for copper transport into the liquid bath.

Phase Equilibrium in Lead-Free Binary Systems

Binary phase diagrams for copper and tin show two primary intermetallic phases stable at reflow temperatures: eta-phase Cu6Sn5 and epsilon-phase Cu3Sn. The Cu6Sn5 phase forms first at the interface during reflow. It grows quickly because its stoichiometry requires less copper transport per unit cell than the copper-rich Cu3Sn phase.

Peak temperatures near 245°C raise copper solubility in liquid tin to roughly 1.5 weight percent, driving rapid Cu6Sn5 crystal growth.

Electroless nickel finishes introduce complex ternary interactions. Without copper in the substrate, Ni3Sn4 intermetallic layers form along the nickel-phosphorus boundary. When SAC305 is used on an electroless nickel pad, copper dissolved from the solder bulk migrates to the interface and substitutes for nickel, forming a ternary (Ni,Cu)6Sn5 phase.

Thin nickel barriers require strict thermal budget control. Excessive time above liquidus consumes the barrier layer completely, bringing liquid solder into direct contact with underlying copper and triggering severe interfacial dewetting.

Dewetting is frequently blamed on plant humidity or pad oxidation occurring between unbagging and reflow. Cross-sectional EDX spectroscopy shows, however, that incomplete coverage usually stems from weak flux activation or peak temperatures that fail to drive copper dissolution past saturation.

Nucleation

Crystalline phase selection during the initial seconds of reflow sets the chemical gradient across the joint boundary. Heterogeneous nucleation happens directly on the substrate surface because the activation energy barrier is far lower than for homogeneous nucleation in the bulk liquid. Substrate roughness, grain orientation, and local chemical impurities serve as preferential nucleation sites.

Eta-phase Cu6Sn5 nucleates as discrete hexagonal crystallites that soon coalesce into a continuous polycrystalline layer across the pad.

The crystallographic structure of the metal substrate directly influences initial grain orientation. Single-crystal or coarse-grained copper substrates induce epitaxial growth patterns in the Cu6Sn5 layer. Certain copper planes show low lattice mismatch with hexagonal Cu6Sn5, lowering interfacial strain energy.

This alignment accelerates early growth along specific orientations, creating irregular grain heights across the initial intermetallic boundary.

Interfacial Free Energy and Phase Selection

Gibbs free energy calculations determine which chemical phase precipitates first when molten solder wets the pad. Although Cu3Sn has a more negative enthalpy of formation, Cu6Sn5 nucleates first because tin atoms are far more abundant at the liquid interface. Copper diffusion from the substrate cannot supply the 75 atomic percent copper needed for immediate Cu3Sn formation.

Consequently, Cu6Sn5 ~ which requires only 54.5 atomic percent copper ~ nucleates exclusively during initial reflow.

Ternary alloy additions in modern solder pastes shift these free energy landscapes. Adding 0.05 weight percent nickel to SAC solders changes the interfacial energy between liquid solder and the growing intermetallic crystal. Nickel atoms replace copper sites in the lattice, forming (Cu,Ni)6Sn5.

This substitution lowers the nucleation barrier, yielding a finer grain structure with reduced scallop heights. Micro-alloying with bismuth or antimony similarly modifies local surface tensions, altering growth kinetics.

Polymorphic Phase Transformations in Copper Tin Compounds

Hexagonal eta-Cu6Sn5 undergoes a phase transition to monoclinic eta’-Cu6Sn5 as the joint cools below 186°C. This polymorphic transformation involves a lattice reconfiguration that generates a 2.15 percent volumetric contraction. Under rapid cooling, this volume change induces micro-strain at the interface between the intermetallic layer and the surrounding tin matrix. Stress concentrations along these boundaries create localized micro-cracks, compromising joint integrity before post-assembly testing.

A polymorphic volume change during solid-state cooling generates internal micro-strain that initiates micro-cracking along the intermetallic boundary.

Micro-alloying strategies target this structural shift directly. Adding trace amounts of nickel or gold into the Cu6Sn5 lattice stabilizes the hexagonal structure down to room temperature, preventing the shift to monoclinic geometry and eliminating the internal strain from volume contraction. Cross-sectional analysis confirms that stabilized hexagonal layers maintain coherent, crack-free boundaries with the copper substrate after cooling.

Substrate surface finish selection further alters the activation energy required for secondary phase precipitation during subsequent thermal exposures.

Morphologies

Intermetallic growth during reflow proceeds through distinct stages. Initial nucleation forms an array of hemispherical scallops projecting directly into the liquid solder. Scallop geometry depends on rapid liquid-state diffusion through channels between adjacent intermetallic grains.

Tin atoms move inward along these channels to react with dissolved copper, while copper atoms migrate outward into the melt. Scallop growth follows a kinetic power law where thickness increases with the cube root of time above liquidus, typical of ripening-controlled growth.

Extended time above liquidus or higher peak temperatures flattens these rounded scallops into a consolidated layer. As adjacent scallops expand, liquid channels between them narrow and close. Once liquid access to the substrate is cut off, copper mass transport relies entirely on solid-state diffusion through the intermetallic layer.

Because solid-state transport is orders of magnitude slower than liquid diffusion, the growth exponent shifts from 0.33 to 0.50. This transition marks the boundary between reflow-driven growth and post-reflow solid-state aging.

Scallop Coarsening Driven by Ripening Mechanisms

Ostwald ripening governs structural coarsening of Cu6Sn5 scallops while the solder is molten. Smaller grains have higher surface curvature and chemical potential than larger adjacent scallops. Copper atoms dissolve from smaller scallops, diffuse through the liquid gap, and redeposit onto larger ones.

This process grows average grain size while reducing total grain boundary surface area across the interface.

Peak reflow temperature heavily dictates the rate of Ostwald ripening. Raising peak temperature from 235°C to 255°C increases copper solubility in liquid SAC305 from 1.1 weight percent to 1.6 weight percent. This higher solubility speeds mass transport between scallops, doubling average scallop diameter during a standard 60-second time-above-liquidus window.

Coarse scallops leave deep troughs at grain boundaries, concentrating mechanical stresses during thermal shock testing.

How Does Reflow Cooling Rate Dictate Interphase Structure?

Cooling dynamics dictate the final structural state of the interfacial zone and bulk matrix. Cooling rates below 1.0°C per second allow extensive secondary precipitation of coarse Ag3Sn plates and large Cu6Sn5 rods in the bulk matrix while thickening the interfacial layer. Faster cooling at 3.0°C to 5.0°C per second freezes the liquid structure, refining eutectic phase spacing and keeping the interfacial layer compact.

The table below outlines intermetallic layer growth and phase development across multiple reflow cycles.

Double-sided SMT assembly subjects primary-side joints to a second reflow profile. The first pass forms an initial Cu6Sn5 scallop layer averaging 1.45 micrometers. During the second pass, exposure to liquid solder at 240°C for another 60 seconds remelts the bulk matrix and reactivates Ostwald ripening at the interface.

Scallop thickness increases to 2.20 micrometers, while solid-state reaction at the copper boundary produces the first measurable sublayer of epsilon-phase Cu3Sn.

Calculating intermetallic growth across multi-pass reflow requires combining liquid-state ripening equations with solid-state growth exponents. Total Cu6Sn5 thickness follows the expression:

x(t) = x_0 + K_L (t_L)^0.33 + K_S (t_S)^0.50

where x_0 represents native oxide-affected initial thickness, K_L is the liquid-state ripening rate constant at peak temperature, t_L is total time above liquidus, K_S is the solid-state diffusion coefficient, and t_S is thermal soak time spent between 180°C and liquidus. For a double-sided process running peak temperatures of 245°C with 70 seconds time-above-liquidus per side, calculated total Cu6Sn5 thickness reaches 2.42 micrometers, matching cross-section measurements closely.

A Cu3Sn sublayer forms between Cu6Sn5 and the copper substrate once local tin concentrations drop below stoichiometric thresholds. Solid-state diffusion drives copper atoms from the pad into the base of the Cu6Sn5 layer. Cu3Sn layer thickness grows strictly through planar solid-state diffusion governed by parabolic kinetics.

Excessive Cu3Sn growth signals high thermal exposure during rework, leaving a brittle sublayer prone to vacancy coalescence.

1. Substrate Land Preparation requires removing organic contaminants and native oxides via chemical cleaning to establish uniform surface energy across the land.
2. Thermal Profile Calibration aligns conveyor speed and heating zone setpoints to restrict total time above liquidus to under 70 seconds per pass.
3. Cooling Zone Control regulates nitrogen flow in zones 9 and 10 to hold cooling rates between 2.5°C and 4.0°C per second across all thermal masses.
4. Metallographic Sample Preparation requires vacuum epoxy encapsulation and diamond suspension polishing to prevent edge-rounding during cross-sectional analysis.

Scallop heights over three micrometers elevate mechanical failure risks during drop testing regardless of bulk solder ductility.

Profiling

Thermal profiling verifies that actual reflow profiles match theoretical growth kinetics models. Sensors attached directly to component leads and board pads measure peak temperatures, heating slopes, and liquidus dwell times. Small-mass components like 0402 passives heat rapidly and spend extended time above liquidus.

High-mass components like shielded inductors or heavy BGAs lag behind, reaching lower peak temperatures and shorter dwell times. Profiling ensures that all joints across the assembly stay within the thermal window needed for controlled intermetallic growth.

Overheating low-mass components to ensure wetting on adjacent BGAs creates localized areas of excessive intermetallic growth. A joint exposed to 255°C peak temperatures for 100 seconds forms an overly thick, coarse intermetallic layer. Excess thermal energy accelerates substrate erosion, consuming thin protective surface coatings and triggering early Cu3Sn sublayer growth.

Multi-channel profiling highlights these thermal differentials across the layout, allowing engineers to balance zone setpoints and blower speeds.

Thermocouple Attachment Protocols for High-Density Assemblies

Sensors with poor mechanical attachment give inaccurate temperature readings that corrupt kinetic growth predictions. High-temperature metal-filled epoxy or direct micro-spot welding provides solid thermal contact between sensor tip and land pad. Adhesive tape or loosely clamped sensors introduce air gaps that misreport peak reflow temperatures by up to 12°C. An uncalibrated sensor reporting 235°C when pad temperature actually reaches 247°C causes undetected structural degradation across production runs.

1. Select high-temperature solder alloy with a melting point above 280°C for attaching profiling wires to critical test joints.
2. Route sensor leads along low-profile board channels to prevent thermal shading of adjacent components inside convection zones.
3. Perform three consecutive profiling runs on production assemblies to confirm temperature repeatability within a 2.0°C band.
4. Verify sensor calibration against a certified reference thermal block every 50 profiling passes to eliminate measurement drift.

Reflow profiles maintaining peak pad temperatures between 238°C and 244°C for 55 to 68 seconds maintain intermetallic layer thickness within the optimal range of 1.2 to 2.2 micrometers.

Calibrating Nitrogen Atmospheres to Limit Surface Oxidation

Nitrogen inerting brings residual oxygen levels inside the oven chamber below 50 parts per million. Eliminating oxygen prevents solder and pad re-oxidation during ramp and soak stages. Lower oxidation reduces liquid solder surface tension, enhancing flux action and speeding up wetting across land pads.

Faster wetting shortens the required time-above-liquidus, allowing shorter thermal profiles without compromising joint quality.

Nitrogen atmospheres also alter intermetallic scallop geometry. By preventing oxide films at the triple-point wetting line, nitrogen enables smooth, uniform spreading of molten solder. This uniform contact ensures homogeneous copper dissolution across the pad, eliminating localized pits and thin spots in the intermetallic layer.

Reducing oxygen concentrations from 1,000 ppm to 30 ppm improves intermetallic thickness uniformity by 35 percent across fine-pitch BGA pads.

IPC-J-STD-001 Section 4.5 mandates documented thermal profiles for every high-reliability surface mount assembly class, prohibiting deviations that exceed intermetallic growth limits.

Shear

Mechanical verification of solder joint integrity tests structural strength at the interface. Standard low-speed ball shear testing pushes component solder balls at 0.1 to 1.0 millimeters per second. Under these low strain rates, plastic deformation occurs predominantly within the bulk solder matrix.

Ductile shear fractures pass through the soft tin matrix, yielding high force values that mask underlying interfacial vulnerabilities.

High-speed ball shear testing per JEDEC JESD22-B117A exposes brittle interfacial defects by increasing impact velocities to 1,000 ~ 4,000 millimeters per second. High strain rates prevent plastic deformation in the bulk solder, transferring applied energy directly to the intermetallic layer. Under impact, brittle layers fracture along the interface between Cu6Sn5 scallops and the substrate, or along the internal Cu6Sn5/Cu3Sn boundary.

Fracture mode distribution shows whether reflow profiles yielded robust metallurgical bonds or brittle interphases.

High-Strain Speed Testing and Interfacial Micro-Cracking

Brittle fractures during high-speed shear testing signal degraded interfacial structure. Mode 4 separation occurs when fracture propagates entirely within the intermetallic layer, showing low impact energy absorption. Layers thicker than 4.0 micrometers consistently fail via Mode 4 brittle fracture.

High peak temperatures and prolonged reflow passes increase layer thickness while growing interfacial grain size, creating cleavage planes that split under mechanical shock.

Cold bump pull testing provides complementary failure mode analysis by applying tensile force directly to component solder spheres. Solder balls are clamped and pulled vertically off the pad at speeds up to 500 millimeters per second. Micro-cracks formed during the polymorphic transition of Cu6Sn5 act as stress concentration points during pull testing, resulting in pad cratering or intermetallic detachment at forces well below design specifications.

Mapping failure modes across high-speed impact tests validates reflow profile limits.

Kirkendall Void Agglomeration under Thermal Exposure

Solid-state thermal aging following reflow activates vacancy diffusion that forms Kirkendall voids. Copper atoms diffuse outward from the substrate through the Cu3Sn layer toward the Cu6Sn5 phase much faster than tin atoms diffuse inward toward the copper. This unbalanced atomic flux creates an excess of vacant lattice sites in the copper-rich boundary layer.

Over time at elevated temperatures, these vacancies coalesce into micro-voids along the interface between base copper and the Cu3Sn sublayer.

• Kirkendall Void Agglomeration reduces net load-bearing pad area, causing sudden brittle joint detachment under low mechanical stress.
• Gold Embrittlement Phase Precipitation forms thin, brittle AuSn4 plates that migrate to the intermetallic interface during extended thermal exposure.
• Black Pad Hyper-Corrosion leaves a phosphorus-rich Ni3P layer underneath ENIG finishes, inducing catastrophic micro-voiding along the intermetallic boundary.
• Scallop Trough Stress Concentration focuses impact shear forces into deep intermetallic grain boundaries, initiating micro-cracks that propagate through the joint.

Substrate copper purity directly affects Kirkendall void density. Impurities like sulfur, phosphorus, and carbon trapped in electroplated copper land pads act as nucleation sites for vacancy condensation. Electroplated copper containing sulfur concentrations above 10 parts per million exhibits severe Kirkendall voiding after 500 hours of thermal storage at 125°C. Clean, high-purity copper finishes resist vacancy aggregation, maintaining interface shear strength throughout product operating lifespans.

Per IPC-TM-650 Method 2.4.43, solder joint intermetallic interfaces subjected to high-speed impact testing must demonstrate over 80 percent ductile bulk failure modes across sampled test pads.

Ignoring Kirkendall void growth and intermetallic coarsening leads to catastrophic field failures where BGA components detach under normal operating vibration, triggering line-stop warranty recalls that dwarf initial manufacturing margins.

Indemnity

Contractual agreements between original equipment manufacturers and assembly facilities must explicitly state intermetallic layer growth parameters and failure criteria. Supply contracts frequently specify electrical continuity testing upon assembly completion, but continuity testing fails to detect brittle intermetallic layers or early-stage Kirkendall voiding. Defining mandatory cross-sectional microstructural analysis in the Quality Service Agreement establishes enforceable quality standards before assemblies ship to customers.

Responsibility for microstructural defects depends on establishing clear root-cause boundaries between material supply and reflow profile execution. Board pad plating defects, such as hyper-corroded electroless nickel or contaminated electroplated copper, originate from substrate vendors. Incorrect peak reflow profiles, excessive multi-pass thermal exposures, and improper cooling rates belong entirely to the assembly contractor.

Standardized quality dossiers containing thermal profiles, solder paste lot certificates, and SEM/EDX cross-sectional records resolve commercial disputes over scrap allocation and rework costs.

Commercial Allocation of Board Rework and Warranty Claims

Reworking defective surface mount components subjects adjacent solder joints to additional thermal cycles. Applying localized hot-air desoldering nozzles heats nearby component pads above liquidus temperatures, initiating secondary intermetallic scallop growth and accelerating Cu3Sn sublayer thickening. Quality contracts limit maximum component rework passes to two operations per pad location.

Exceeding two rework cycles requires complete assembly scrap, with costs assigned to the party responsible for the initial component placement failure.

Warranty claims arising from field failures trigger extensive failure analysis costs. Third-party metallographic lab testing, including cross-sectional SEM imaging and EDX elemental mapping, costs between $1,500 and $3,500 per sample location. Contracting agreements assign these analytical fees to the assembly contractor if intermetallic layer thickness exceeds the maximum agreed threshold of 3.5 micrometers due to uncalibrated oven profiles.

Inspection Sign-Off and Metallographic Dossier Criteria

Cross-sectional inspection sign-off protocols require random lot sampling during volume production runs. Production lots undergo sampling per ANSI/ASQ Z1.4 Level II standards. Sampled boards undergo sectioning, mounting, polishing, and optical micro-analysis to verify intermetallic layer thickness, phase distribution, and wetting angles.

Lots exhibiting average intermetallic thickness outside the 1.0 to 3.0 micrometer range trigger immediate process holds and thermal profile re-verification.

Establishing presence on the assembly floor eliminates disputes over thermal profile compliance and sample authenticity. Quality auditors verify profile datalogger calibrations directly at the reflow oven, audit solder paste storage temperatures, and sign off on cross-sectional microstructural dossiers prior to board lot release. Direct operational oversight ensures that theoretical intermetallic kinetic models match workshop reality, protecting field reliability and defending client capital.

Direct floor audits of thermal profiling records and cross-sectional SEM dossiers prevent uncalibrated reflow runs from turning into expensive field recalls.

Audit dossiers must contain raw datalogger files for every production shift alongside certified chemical assay reports for all solder paste lots. Retaining these records for seven years fulfills automotive and aerospace quality traceability mandates while providing defensive documentation during component warranty litigation.