Qualitative observations on the diffusion of copper and gold through a nickel barrier

To retard failure of gold plated copper parts by diffusion of copper to the gold surface, a layer of nickel is frequently used between the copper and gold as a diffusion barrier. To evaluate the mechanisms whereby the nickel retards the motion of copper atoms to the gold surface, planar tri-couples of Cu/Ni/Au were prepared by electroplating nickel and gold layers on OFHC copper coupons. Diffusion anneals were carried out at temperatures from 150 to 750°C. A qualitative evaluation of diffusion behavior was provided by an electron microprobe utilizing X-ray wavelength dispersive analysis on polished cross sections. Results demonstrate that the nickel layer retards but does not block the transport of copper to the gold surface. Possible mechanisms for the anomalous buildup of copper at the gold/nickel interface and gold at the copper/nickel interface are discussed.     

Diffusional breakdown of a Ag diffusion barrier in a Cu-Ag-Ni diffusion triple

A diffusion triple with copper and nickel sandwiching a thin diffusion barrier layer of silver was annealed for various times at 760 °C. Silver exhibits a limited solubility (0.135 atom fraction) for copper and a slight (0.01) solubility for nickel at 760 °C. The Ag barrier was breached by successive processes of Cu interdiffusion, interface instability of the Ag-Ni interface, and growth of Cu-Ni protrusions from the Ag-Ni interface to ultimately bridge the Ag barrier. The kinetics of widening of the Ag barrier were determined and interpreted in terms of the diffusion concentration profiles and the plastic deformation of the Ag barrier by the Cu-Ni protrusions. This paper is based on a presentation made at the “Stephen R. Shatynski Memorial Symposium on Surfaces and Interfaces” held at the 114th annual AIME meeting in New York, February 24-28, 1985, under the auspices of the ASM-MSD Thermodynamic Activity Committee.     

Absorption of gaseous oxygen by liquid cobalt,copper, iron and nickel

An experimental investigation of the rates of oxygen solution in molten cobalt, copper, iron and nickel was carried out using pure oxygen and a constant-volume Sieverts’ method. It was found that the volume of gaseous oxygen which initially reacted with the inductively stirred metals was strongly dependent on the physical nature of the oxide film which formed during the first stage of reaction. The initial temperature of the molten iron, cobalt, and nickel was 1600°C, and for copper was 1250°C. For initial oxygen pressures above the melt of about one atmosphere both molten iron and copper, which formed liquid surface oxides, initially absorbed nearly 20 cm³ (STP) O₂/cm² of melt surface area, while molten cobalt and nickel, which formed solid oxides, absorbed about 6 cm³ (STP) O₂/cm² under the same experimental conditions. For approximately 30 s after the initial reaction between these liquid metals and gaseous oxygen, the oxygen absorption rate was proportional to the square root of the oxygen pressure above the melt, and proportional to the melt surface area, but independent of melt volume. The rate-limiting step for oxygen absorption by liquid iron, cobalt and copper can be described by dissociative adsorption of oxygen molecules at the gas/oxide interface. After 30 s of reaction, the rate of oxygen absorption became less dependent on the oxygen pressure above the melt. This indicated that the rate-controlling step was changing from a surface reaction to growth of the oxide layer by cationic diffusion in the bulk oxide. The oxidation rate of liquid nickel appears to be too complex to be described by models for dissociative adsorption of oxygen molecules at the gas/oxide interface and parabolic growth of the oxide layer. The formation of a thin layer of nickel oxide which allows oxygen to migrate through cracks or grain boundaries may be responsible for the relatively high oxygen absorption rate compared to that of liquid cobalt. R. H. RADZILOWSKI, formerly a Graduate Studient at The University of Michigan     

Diffusional breakdown of nickel protective coatings on copper substrate in silver-copper eutectic melts

Diffusion couples with electrolessly plated nickel diffusion barriers between copper substrates and silver-copper eutectic alloys were tested at 800 °C and 850 °C, respectively. Growth of (Cu, Ni, Ag) ternary solid solution into the melt was observed at both temperatures. The growth pattern changed from cellular to dendritic as the temperature was increased from 800 °C to 850 °C. The nonplanar growth morphology can be explained in terms of constitutional supercooling in the melt. Kinetics of (Cu, Ni, Ag) solid solution growth were found to be controlled by interdiffusion at the interface of the nickel barrier and the growing solid-state phase. Local breakdown of the nickel diffusion barrier started once the (Cu, Ni, Ag) solid solution reached the copper substrate. Silver diffused from the silver-copper melt, through the ternary solid solution, dissolving copper and forming silver-copper liquid along copper grain boundaries. Ultimately, the nickel barrier was totally converted to the ternary solid solution, broke up, and floated into the liquid. Dissolution of the copper substrate occurred subsequently. A thin layer of chromium undercoating proved to be very effective in extending the protection time of the nickel diffusion barrier, due to the extremely low solubility of both copper and silver in chromium at these test temperatures.     

Absorption of gaseous oxygen by liquid cobalt, copper, iron and nickel

An experimental investigation of the rates of oxygen solution in molten cobalt, copper, iron and nickel was carried out using pure oxygen and a constant-volume Sieverts' method. It was found that the volume of gaseous oxygen which initially reacted with the inductively stirred metals was strongly dependent on the physical nature of the oxide film which formed during the first stage of reaction. The initial temperature of the molten iron, cobalt, and nickel was 1600‡C, and for copper was 1250‡C. For initial oxygen pres-sures above the melt of about one atmosphere both molten iron and copper, which formed liquid surface oxides, initially absorbed nearly 20 cm³ (STP) O₂/cm² of melt surface area, while molten cobalt and nickel, which formed solid oxides, absorbed about 6 cm³ (STP) 0₂/cm² under the same experimental conditions. For approximately 30 s after the initial reaction between these liquid metals and gaseous oxygen, the oxygen absorption rate was proportional to the square root of the oxygen pressure above the melt, and pro-portional to the melt surface area, but independent of melt volume. The rate-limiting step for oxygen absorption by liquid iron, cobalt and copper can be described by dissocia-tive adsorption of oxygen molecules at the gasJoxide interface. After 30 s of reaction, the rate of oxygen absorption became less dependent on the oxygen pressure above the melt. This indicated that the rate-controlling step was changing from a surface reaction to growth of the oxide layer by cationic diffusion in the bulk oxide. The oxidation rate of liquid nickel appears to be too complex to be described by models for dissociative ad-sorption of oxygen molecules at the gasJoxide interface and parabolic growth of the oxide layer. The formation of a thin layer of nickel oxide which allows oxygen to migrate through cracks or grain boundaries may be responsible for the relatively high oxygen ab-sorption rate compared to that of liquid cobalt. Formerly a Graduate Student at The University of Michigan     

Room-temperature diffusion in Cu/Ag thin-film couples caused by anodic dissolution

Thin, 100-nm films of first silver and then copper were deposited consecutively onto inert substrates by magnetron sputter deposition. Constant anodic current densities were applied at room temperature to dissolve the outer copper film to varying depths. The 50Cu/50Ag interface, derived from the auger electron spectroscopic concentration-depth profile, initially moved into the copper toward the outer dissolving surface, indicating enhanced diffusion of copper into silver. After longer times at all anodic current densities, the interface reversed and moved back toward the underlying silver-rich layer, indicating that eventually diffusion of silver into copper predominated. The reversal time was inversely proportional to the anodic current density. These effects are explained by anodic formation of subsurface vacancies which migrate as divacancies to the copper/silver interface where they affect interface movements by the well-known Kirkendall mechanism. Calculated diffusivities up to 10⁻¹² cm²/s at maximum anodic current densities of 900 μA/cm² are dramatically above any that are normally observed at room temperature.     

Out-diffusion of the noble component during the initial stage of the oxidation of γ′-Ni3Al

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) studies of the oxidation behaviour of single crystalline γ′-Ni₃Al revealed that Ni diffuses through the oxide layer to the surface of the material during oxidation at low oxygen partial pressure. Small pure nickel particles (diameter ∼ 50 nm) form on top of the oxide scale. In this paper, this observation is explained on the basis of a micromechanical model. The model assumes that at intermediate temperatures the prevailing transport mechanism is the diffusion along the oxide/metal interface caused by compressive stresses generated during oxide growth. Relaxation of internal stresses as well as the formation of the small Ni-particles on top of the oxide surface are controlled by interface diffusivity. Other possible stress relief mechanisms can be excluded by theoretical estimates.     

Room-temperature diffusion in Cu/Ag thin-film couples caused by anodic dissolution

Thin, 100-nm films of first silver and then copper were deposited consecutively onto inert substrates by magnetron sputter deposition. Constant anodic current densities were applied at room temperature to dissolve the outer copper film to varying depths. The 50Cu/50Ag interface, derived from the auger electron spectroscopic concentration-depth profile, initially moved into the copper toward the outer dissolving surface, indicating enhanced diffusion of copper into silver. After longer times at all anodic current densities, the interface reversed and moved back toward the underlying silver-rich layer, indicating that eventually diffusion of silver into copper predominated. The reversal time was inversely proportional to the anodic current density. These effects are explained by anodic formation of subsurface vacancies which migrate as divacancies to the copper/silver interface where they affect interface movements by the well-known Kirkendall mechanism. Calculated diffusivities up to 10⁻¹² cm²/s at maximum anodic current densities of 900 μA/cm² are dramatically above any that are normally observed at room temperature.     

Mass diffusion in polycrystalline copper/electroplated gold planar couples

The Au/Cu system is common to electrical connectors and recent trends toward higher ambient temperatures and thinner gold electroplates focuses attention on a new failure mode. This mode is degradation of thin gold electroplates by mass diffusion of the base metal (copper) to the surface of the electroplate at low temperatures (less than 250°C). Therefore, diffusion experiments were conducted for polycrystalline copper/electroplated gold planar couples over the temperature range of 50° to 750°C. The chemical interdiffusion coefficients, ~D, were calculated using the Boltzmann-Matano solution on the concentration-distance profiles which were determined using an electron microprobe. Results of this study show that ~D is a small function of concentration, generally with a variation of less than a factor of 3 to 5. The correlation of the temperature dependence of ~D between 250° and 750°C with existing data is excellent. The data conform to an Arrhenius equation: ~D= 1.5×10⁻⁵ e^−23,600/RT At temperatures below 250°C the values of ~D deviate from this equation and below 150°C are significantly greater than would be predicted by extrapolating the high temperature Arrhenius equation. No correlation was found between electroplate thickness and diffusion rate.     

Reaction layer formation at the graphite/copper-chromium alloy interface

Sessile drop tests were used to obtain information about copper-chromium alloys that suitably wet graphite. Characterization of graphite/copper-chromium alloy interfaces subjected to elevated temperatures were conducted using scanning electron microscopy (SEM), energy-dispersive spectroscopy (EDS), Auger electron spectroscopy (AES), and X-ray diffraction analyses. These analyses indicate that during sessile drop tests conducted at 1130 °C for 1 hour, copper alloys containing greater than 0.98 at. pct chromium form continuous reaction layers of approximately 10-μm thickness. The reaction layers adhere to the graphite surface. The copper wets the reaction layer to form a contact angle of 60 deg or less. X-ray diffraction results indicate that the reaction layer is chromium carbide. The kinetics of reaction layer formation were modeled in terms of bulk diffusion mechanisms. Reaction layer thickness is controlled initially by the diffusion of Cr out of the Cu alloy and later by the diffusion of C through chromium carbide. This article is based on a presentation made in the symposium “High Performance Copper-Base Materials” as part of the 1991 TMS Annual Meeting, February 17–21, 1991, New Orleans, LA, under the auspices of the TMS Structural Materials Committee.     

Interdiffusion in thin conductor films — chromium/gold,nickel/gold and chromium silicide/gold

The diffusion of chromium, nickel and chromium silicide into gold has been studied using thin film techniques. Electrical measurements of film diffusion couples were made at several temperatures and these data are related to the interdiffusion of the metals. It was found that the activation energy for resistance changes in the films is somewhat greater than one-half of that for bulk diffusion indicating the predominance of a grain boundary diffusion mechanism; however, a drastically lowered temperature coefficient of resistance also indicates some interdiffusion by lattice diffusion. Under the conditions of these experiments, the chromium, nickel, and chromium silicide diffuse from the substrate through the gold to the outer surface of the gold where they are oxidized.     

Effect of thermal spray on the microstructure and adhesive strength of high-velocity oxy-fuel-sprayed Ni-Cr coatings on 9Cr-1Mo steel

Coatings of 80Ni-20Cr and 50Ni-50Cr on a 9Cr-1Mo steel substrate were produced by high-velocity oxy-fuel (HVOF) spraying to protect the steel against steam oxidation in ultrasupercritical (USC) boilers. The oxidation studies on the coated specimens showed good protection against the scale growth on the steel substrate. Both the 80Ni-20Cr and 50Ni-50Cr coatings formed a thin protective oxide film on the coating surface. The 80Ni-20Cr coating showed Fe diffusion from the substrate to the coating and nickel diffusion from the coating to the substrate during the oxidation process. In the case of 50Ni-50Cr coatings, the diffusion process was reduced, but a continuous layer of chromium carbide was observed at the coating/substrate interface during the oxidation. The adhesive/cohesive strength of these coatings was evaluated on aged specimens at 750 °C by using a simple tensile test. The results of the as-coated 80Ni-20Cr specimens showed an adhesive-strength value of 68 MPa. On extended aging, the strength of the coating increased beyond the detection limit of the resin. The nickel diffusion from the coating to the substrate and the iron diffusion from the substrate to the coating caused the increased adhesive strength. In the case of 50Ni-50Cr, the as-coated specimens showed an average adhesive strength of 76 MPa and showed a decreasing trend on the aged specimens. The formation of chromium carbide at the interface caused inferior values in the adhesive/cohesive strength of the 50Ni-50Cr coatings. The chromium carbide formed on the coating/substrate interface was identified as M₂₃C₆-type carbide.     

A kinetic study that compares the leaching of gold in the cyanide, thiosulfate, and chloride systems

Due to the increasing environmental and public concerns over cyanidation, there has been a large amount of research into viable alternative lixiviants. This article presents a detailed kinetic study of gold leaching in cyanide, ammonia/thiosulfate, and chloride/hypochlorite solutions. The gold leach rates were measured using a rotating electrochemical quartz crystal microbalance (REQCM). This instrument allows the mass of a gold sample to be measured in situ, with a sensitivity of less than 10 ng. It will be shown that for the cyanide system, the leach rate of a gold/silver alloy is substantially higher than that of pure gold; for the gold/silver alloy, the reaction is diffusion controlled. For the thiosulfate system, reaction rates which are substantially higher than those for cyanide can be achieved with freshly prepared solutions containing copper(II), although the leach rate decreases as the copper(II) reacts with thiosulfate. A steady-state copper(II) concentration is obtained once the rate of copper(II) reduction by thiosulfate matches the rate of copper(I) oxidation by oxygen; at these steady-state concentrations of copper(II), the leach rates are slower than those obtained for cyanide. It will also be shown that reaction rates similar to the cyanide system can be readily achieved in chloride media at pH 3 with 2.5 mM hypochlorous acid; under these conditions, the reaction is limited by the diffusion of HClO.     

银铜侧向复合材料界面结合机理分析

本文采用“包覆锭坯+扩散烧结+冷轧复合”联合工艺制备了银铜侧向复合带材,利用金相显微镜（OM）、扫描电镜（SEM）和能谱仪（EDS）观察分析银铜复合界面结构和元素分布,并分析其银铜复合界面的结合机理。结果表明,银铜复合界面形成过程为：1）银铜接触界面处凹凸不平的表面在轧制力的作用下相互咬合,形成机械结合界面;2）接触面在轧制力的作用下,银铜表面氧化膜破裂,新鲜表面质点间在轧制变形热的作用下产生原子结合;3）在扩散烧结过程中,银铜界面处的原子在高温作用下被激活,银铜原子相互扩散,在界面处发生银铜共晶反应形成液相金属层,随着烧结时间的延长,其共晶反应液相层厚度逐渐增加,随后冷凝结晶,使银铜实现侧向冶金结合。4）在后续中间退火过程中,共晶层与两侧的铜、银基体相互扩散,铜、银原子向更深的方向逐渐扩散,在靠近共晶层铜侧和银侧逐步形成固溶体层,使银与铜的结合强度进一步提高。银铜侧向复合界面结合机理包含机械咬合结合、接触共晶反应自钎焊结合和原子扩散结合3种,复合界面结合强度较好,剪切强度达220 MPa。     

Effect of Scale Formation on Copper Enrichment Behavior in Continuously Cast Slab

Copper enrichment behavior in continuously cast slab induced by scale formation during continuous cooling was experimentally investigated,and the effects of initial slab surface temperature and oxygen potential in atmos-phere were discussed.The results showed that a loose scale adhered to the substrate was formed in H2 O-N2 atmos-phere at higher slab surface temperature compared to a gap formed between the scale and the steel substrate after continuous cooling in H2 O-O2-N2 atmosphere.Under the condition of continuous cooling in H2 O-N2 atmosphere, the copper enrichment occurred both within the loose scale and at the scale/steel interface with simultaneous Ni en-richment near the interface at higher slab surface temperature.The combined effects of the loose scale and nickel en-richment were thought to promote the back-migration of Cu-rich phase from the interface and occlusion within the scale layer.While in H2 O-O2-N2 atmosphere,the Cu enrichment was found on the steel side and the formed gap pre-vented the migration of Cu to the scale.     

Oxidation of nickel sulfide

The oxidation of nickel sulfide whose atomic fraction of sulfur,x _s, is 0.40 to 0.44 was studied in a mixed O₂-N₂ gas stream at 923, 973, and 1023 K. The oxygen partial pressure was maintained at 2.0 x 10⁴ Pa. In the oxidation of nickel sulfide ofx _s = 0.40 and 0.41, a dense NiO layer was formed on the sulfide surface without the evolution of SO₂ gas, because of the low sulfur activity. Diffusion of nickel within the inner sulfide core toward the surface controlled the oxidation rate during the first one minute of oxidation. Subsequently, the oxidation rate was controlled by the diffusion of nickel through the formed NiO layer. In the oxidation of nickel sulfide ofx _s = 0.44 at 973 and 1023 K, the reaction proceeded irregularly to the interior of the sulfide core with the evolution of SO₂ gas, and a porous oxide layer was formed, due to the high sulfur activity of nickel sulfide. For the same reason, this oxidation was also accompanied by the dissociation of nickel sulfide. Under the experimental conditions ofx _s = 0.42, 1023 K and x_s = 0.44,923 K, the oxidation started with weight increase and without the evolution of SO₂ gas, and in the subsequent stage the weight decreased and SO₂ gas was evolved. K. HAJIKA, former Graduate Student     

Maximum Copper and Tin Contents in LC‐steel Thin Strip ‐ Hot Shortness Model Calculations

For conventional casting processes low copper and tin contents have to be ensured in LC‐steel to avoid hot shortness. It is expected that higher cooling rates, e.g. in thin strip casting, permit higher copper and tin limits. Hot shortness occurs because of selective oxidation of the iron whereby the more noble copper is enriched at the steel‐oxide interface. A liquid metallic copper phase which wets the grain boundaries supports cracking during hot deformation. The enrichment of the liquid copper phase depends on the oxidation temperature: At low temperatures copper is solid, cannot wet the steel surface and is incorporated into the growing oxide layer. At mid temperatures (1083‐1177 °C) the copper phase is liquid, wets the grain boundaries of the steel surface and causes hot shortness. At high temperatures a liquid fayalitic slag is formed in the oxide layer if the steel contains silicon. The fayalitic phase occludes parts of the steel surface and removes copper from the steel surface; then hot shortness is reduced or even avoided. Other mechanisms to remove copper from the steel surface need the presence of Fe₃O₄ and Fe₂O₃ in the oxide layer. These iron oxides are not formed for short oxidation times where linear oxidation takes place. Diffusion of copper into the steel is too slow to reduce hot shortness if copper has an elevated concentration in the steel, e.g. 0.5 wt.‐%. Therefore, only the occlusion mechanism is of importance during linear oxidation. A model is established on the basis of these observations in order to predict an upper copper limit in dependence of the steel strip thickness (cooling behaviour) and the oxygen content in the cooling atmosphere (nitrogen‐oxygen mixture). The model is compared to experimental results from KIMAB which are presented in this issue. It is demonstrated that a copper layer thickness of 0.098 μm at the steel‐oxide interface is sufficient to cause cracks of a depth of more than 0.2 mm. For strip thicknesses below 5 mm a simple approximation can be used to predict the maximum copper content in LC‐steel to avoid hot shortness. For example, thin strip of a thickness of 2 mm will have no cracks (above 0.2 mm) even if 0.7 wt.‐% of copper is contained in the LC‐steel. For atmospheres with a reduced oxygen partial pressure even higher copper contents are possible. Tin is with short oxidation times not a problem concerning hot shortness, as shown by the KIMAB results. This may be explained by the much higher diffusivity of tin in iron compared to copper.     

Rate of dissolution of solid nickel in liquid tin under static conditions

The dissolution kinetics of solid nickel in liquid tin have been investigated under static conditions. The cylindrical nickel specimens were immersed in liquid tin over the temperature range of 551 to 803 K in the reaction time range of 0.9 to 6.0 ks. A natural convection model for mass transfer and the dissolution rate equation derived by considering intermetallic compound layer formation were used to interpret the experimental dissolution data. A larger dissolution at the upper part of specimen causing natural convection and an intermetallic layer formation with a linear relationship at solid/liquid interface occurred. Below 628 K, the dissolution rate appears to be controlled by chemical reaction of an intermetallic compound layer. At mid-range temperatures (of 681 K), the dissolution process was governed by a mixed control mechanism involving diffusion in liquid tin and chemical reaction of the intermetallic compound layer. At temperatures above 735 K, the rate seems to be controlled by diffusion across a concentration boundary layer in liquid tin. The formation of an intermetallic compound layer did not interfere with the dissolution.     

Observations on interdiffusion in planar copper/tin-nickel/gold tricouples

Many of the basic properties of the electrodeposited tin-nickel (Sn-Ni) alloy have been investigated, but two previously unreported aspects are interdiffusion behavior in the Cu/Sn-Ni/Au system and the thermal stability of Sn-Ni when associated with gold. Planar couples were aged at temperatures between 100 and 500°C for times to one year and evaluated by scanning electron microscopy, electron probe microanalysis, and Auger electron spectroscopy. Results show that above 250°C the layer disintegrates into fine particles and the gold and copper interdiffuse as if the Sn-Ni layer were never present. At lower temperatures, the disintegration of the layer is not detectable by the techniques used or within the time period studied. However, tin is still released from the layer and diffuses very rapidly, even through thick gold, to the surface where it is oxidized. The thicker the tin-nickel layer, the greater the amount of surface oxide which is produced for a fixed aging condition. Thus some quantities of tin-oxide can be expected to form on the surface of gold in times of just a few years for application temperatures between 100°C and room temperature.