Sodium Borohydride Reduction of Aqueous Silver-Iron-Nickel Solutions: a Chemical Route to Synthesis of Low Thermal Expansion–High Conductivity Ag-Invar Alloys
- First Online:
- Cite this article as:
- Sterling, E., Stolk, J., Hafford, L. et al. Metall and Mat Trans A (2009) 40: 1701. doi:10.1007/s11661-009-9859-4
- 1.2k Downloads
Thermal management is a critical concern in the design and performance of electronics systems. If heat extraction and thermal expansion are not properly addressed, the thermal mismatch among dissimilar materials may give rise to high thermal stresses or interfacial shear strains, and ultimately to premature system failure. In this article, we present a chemical synthesis process that yields Ag-Invar (64Fe-36Ni) alloys with a range of attractive properties for thermal management applications. Sodium borohydride reduction of an aqueous Ag-Fe-Ni metal salt solution produces nanocrystalline powders, and conventional powder processing converts this powder to fine-grained alloys. The samples are characterized by X-ray diffraction (XRD), scanning electron microscopy, thermomechanical analysis, and electrical conductivity measurements; thermal conductivity is estimated using the Wiedemann–Franz law. Sintering of Ag-Fe-Ni powders leads to the formation of two-phase silver-Invar alloys with low coefficients of thermal expansion (CTEs) and relatively high electrical conductivities. A sample of 50Ag-50Invar exhibits a CTE of 8.76 μm/(m· °C) and an estimated thermal conductivity of 236 W/(m·K). The Ag-Invar alloys offer thermodynamic stability and tailorable properties, and they may help address the need for improved packaging materials.
Thermal management in electronics components, assemblies, and systems is a critical consideration as circuit board component density continues to rise and as new build-up processes and component integration advance. A major thermal issue facing high-performance circuit designers is the control of thermal energy that integrated circuits (ICs) generate during operation. Effective removal of heat necessitates the use of materials with physical properties dissimilar to those of the semiconductor devices.
Heat generation and coefficient of thermal expansion (CTE) mismatch among semiconductor devices, interconnects, and packaging materials are known to affect electronic assembly performance and reliability. Semiconductors and ceramic packaging materials naturally exhibit low thermal conductivity and low thermal expansion, while the highly conductive alloys that are effective in electrically connecting components and removing heat typically exhibit relatively large CTEs. Further, composite package substrates and underfill materials tend to exhibit low thermal conductivity but relatively high CTEs. If heat extraction is not addressed, the thermally sensitive devices may cease to function. If differential thermal expansion is not addressed, thermal mismatch among dissimilar materials may give rise to high thermal stresses or interfacial shear strains, and ultimately lead to crack formation and premature component failure.
1.1 Substrate-Based Thermal Management
The device or IC substrate represents a key opportunity for improvement of thermal management, and much attention in recent years has centered on the design of substrates that reduce thermal strain, aid heat dissipation, or both. Two approaches employed by circuit designers for substrate-based thermal management include laminate substrates with core constraining layers and substrates with heat sinks or heat spreaders directly joined to devices. Both of these approaches effectively reduce heat or thermal strain, and thereby increase the reliability of modern microelectronic or optoelectronic systems. Successful implementations of low CTE, high conductivity core constraining layers include clad metals such as copper-Invar-copper,[1, 2, 3] carbon fiber-based cores,[4,5] and carbon-SiC laminate composites. A number of composite packaging materials have been explored for use as thermal management substrates in thermally demanding components such as power amplifiers, laser diodes, thermoelectric coolers, and radio frequency (RF) and microwave devices. These include metal matrix composites,[7, 8, 9, 10, 11] metal-metal composites,[8,9,12, 13, 14] and advanced diamond or graphite films and composites.
1.2 Ag-Fe-Ni for Thermal Management
Although many of the composites and advanced alloys used in core constraining and heat sink or heat spreading applications have demonstrated desirable thermal expansion and conductivity properties, some also exhibit significant anisotropy, high material or fabrication costs, high density, poor interfacial characteristics, toxicity, or difficult machinability or processing. This article presents an alternative synthesis approach for the preparation of a range of nanocomposite Ag-Fe-Ni powders, and it describes the properties and phase development of fine-grained, low thermal expansion–high conductivity Ag-Invar alloys formed from the nanoscale powders. The fine-grained Ag-Fe-Ni powder metallurgy alloys described here may offer the desired thermal, electrical, and physical properties necessary for thermal management in high-performance circuits.
Composites and alloys prepared by chemical synthesis and processing of nanoscale powders offer some distinct advantages over bulk materials, including better mechanical properties, increased densification rates, and higher purity in the bulk alloys. In addition, the intimately mixed nanoscale powders produced by chemical synthesis exhibit phase evolution, diffusion, and interfacial energy properties and behaviors that more closely match those expected in thin films–scale metallic materials. The Ag-Fe-Ni alloys investigated in this study may also help researchers address manufacturing issues of increasing importance. The Ag-Fe-Ni alloys are ductile, and they may be easily formed into other shapes by punching, shearing, bending, drawing, and extruding. Disadvantages of Ag-Fe-Ni include high material costs and relatively high density compared to some of the metal matrix composites and carbon-based thermal management materials. In addition, the low Young’s modulus of Ag-Fe-Ni alloys may be advantageous in applications involving direct mounting of devices to heat extracting substrates, but it could also serve as a concern for core constraining applications that rely on high laminate composite stiffness to reduce thermal strains in the electronic assembly.
Like many of the core constraining composites described previously, the alloys developed in this study leverage the thermal properties of Invar, a face-centered-cubic alloy of 64 wt pct Fe and 36 wt pct nickel. Invar and near-Invar alloys exhibit zero, negative, or very low thermal expansion near ambient temperatures as a result of a balance between thermal expansion and volume magnetostriction. Above room temperature, the CTE of Invar gradually increases, with more marked changes in CTE occurring above temperatures of approximately 225 °C, where the thermal expansion is no longer balanced by the magnetostriction.
This investigation of Ag-Fe-Ni alloys builds on prior work in powder metallurgy Cu-Invar thermal management materials. Although copper appears to be a logical choice for the high-conductivity phase of an Invar-based composite, investigations of powder Cu-Fe-Ni alloys have demonstrated that production of metallurgically stable, two-phase, low thermal expansion–high conductivity alloys in this system is difficult. The solid-state solubility limits in the ternary Cu-Fe-Ni system present a significant challenge to powder processing of these materials, as diffusion of Fe and Ni into the Cu has deleterious effects on both phases. Iron and nickel impurities dramatically decrease the electrical and thermal conductivity of the Cu phase, and changes in the Fe-Ni ratio increase the thermal expansion of the Invar phase.
Silver offers a promising alternative to copper for use as a high conductivity phase. Although little is published on the phase development and ambient-temperature phase stability in the Ag-Fe-Ni ternary system, the Ag-Ni, Ag-Fe, and Fe-Ni binary phase diagrams indicate that the formation of a range of thermodynamically stable, two-phase Ag-Invar materials using chemical synthesis and powder processing techniques is possible. The feasibility of Ag-Fe-Ni alloy use requires that the Ag and Invar phases remain metallurgically unmixed to avoid the harmful effects of impurities on Ag conductivity. Silver-Invar composites in a limited compositional range have been produced by Ag liquid infiltration of an Invar powder compact;[9,18] and these alloys have been used in some GaAs IC thermal management applications. The phase stability, microstructural development, and properties of ternary Ag-Fe-Ni alloys produced from nanocrystalline metal powders, however, have not been thoroughly investigated.
The present study aims at the production of nanocrystalline Ag-Fe-Ni powders using solution-based chemical synthesis techniques; formation of fine-grained, low CTE–high conductivity Ag-Invar alloys by conventional powder processing and heat treatment; and characterization of the resulting products. This exploration of a wide compositional range of fine-grained Ag-Invar alloys has the potential to provide insights into the behavior and performance of these promising ternary alloys and to reveal Ag-Invar compositions that are useful in a variety of thermal management applications.
2 Experimental Procedures
2.1 Chemical Synthesis
The resulting nanoscale powders were filtered using a fritted glass filtration funnel and vacuum assist; rinsed with warm (~90 °C) deionized water to assist in the removal of NH3, NaOH, and B(OH)3; and dried with acetone. The powders were heat treated in a H2 atmosphere at 520 °C to reduce any oxides present in the sample and to dissociate any metal borates that remained after initial washing into B2O3 and metal particles. After heat treatment, the powders were washed with warm water to remove B2O3, filtered, and dried with acetone.
Heat-treated powders were compacted in stainless steel pellet dies. A 5-mm pellet die with 0.3 g powder was used to produce small cylindrical pellets for thermal expansion measurements, and a 25-mm pellet die with 3 g powder was used to create coin-shaped pellets for use in all other analyses. The pellets were sintered in a H2 atmosphere at 900 °C for 1 hour to yield fine-grained Ag-Invar alloys.
2.2 Characterization Techniques
The crystalline phases of the as-synthesized powders, heat-treated powders, and sintered pellets were characterized using a Shimadzu MAXima XRD-7000 X-ray powder diffractometer (Shimadzu Corporation, Kyoto, Japan). A copper Cu Kα radiation source at 40 kV and 30 mA was used for all X-ray diffraction (XRD) measurements, and XRD data processing included background subtraction, smoothing, and Cu Kα2 reflection subtraction. Compositions of the sintered pellets were measured using a Shimadzu AA-6650 atomic absorption spectrophotometer (AAS, Shimadzu Corporation, Kyoto, Japan) in flame mode. For the AAS analyses, approximately 0.5 g of the solid samples was dissolved in nitric acid and serially diluted until the metal content of the dilution was nominally between 1 and 5 ppm.
Particle sizes of the as-prepared and heat-treated powders were characterized using a JEOL1 200CX transmission electron microscope (TEM). All TEM samples were prepared by ultrasonically dispersing powders in acetone and dropping the dispersed powder onto holey carbon Cu grids. Sintered pellets were ground with SiC papers and polished to 0.05-μm Al2O3, and a JEOL 6060LV scanning electron microscope was used to characterize the microstructures of the sintered pellets. Phase compositions were investigated by compositional dot mapping using a Thermo Scientific Noran System SIX energy dispersive spectrometry (EDS) microanalysis system (Thermo Scientific, Waltham, MA).
3 Results and Discussion
3.1 Sample Composition and Phase Identification
Sample Identification, Nominal Compositions, and Actual Compositions as Determined by AAS and Fe:Ni Ratio in the Sintered Ag-Fe-Ni Pellets
Nominal Composition, Wt Pct
Actual Composition, Wt Pct
Sample Identification and Densities of the Sintered Ag-Fe-Ni Pellets
Percent of Theoretical Density
3.2 Thermal Expansion Behavior
Rule-of-Mixtures CTE, μm·m−1·°C−1
Turner Model CTE, μm·m−1·°C−1
Experimental CTE, μm·m−1·°C−1
100 Ag (ref)
100 Invar (ref)
The TMA data show a reduction in the average CTE with increasing Invar content. All of the average CTEs fall within the lower bound of the Turner model and the upper bound of the rule-of-mixtures model, indicating that the Fe:Ni ratios in the sintered alloys closely match the desired 64Fe-36Ni ratio for the Invar phase. The good fit between the measured and predicted values also indicates that the Ag and Invar phases in the sintered alloys are metallurgically distinct, i.e., that little Fe and Ni diffuses into the (Ag) phase and that little Ag diffuses into the γ-(Fe,Ni) Invar phase. This desirable phase separation was verified by EDS compositional dot mapping of several sintered samples.
3.3 Conductivity of Ag-Invar Alloys
Measured Experimental Electrical Conductivities of Sintered Ag-Invar Pellets, and Estimated Thermal Conductivities for the Ag-Invar Alloys Based on the Wiedemann–Franz Law; Parallel and Series Model Predicted Values, as well as Typical Property Values for Silver and Invar, are Included for Comparison
Electrical Conductivity, Ohm−1·cm−1
Thermal Conductivity, W·m−1·K−1
5.1 × 105
7.8 × 104
4.3 × 105
4.4 × 105
4.3 × 104
3.6 × 105
3.8 × 105
3.0 × 104
3.2 × 105
3.2 × 105
2.4 × 104
1.9 × 105
2.6 × 105
2.0 × 104
1.2 × 105
2.0 × 105
1.7 × 104
6.9 × 104
1.5 × 105
1.5 × 104
3.3 × 104
1.0 × 105
1.3 × 104
3.0 × 104
5.6 × 104
1.2 × 104
1.4 × 104
100 Ag (ref)
5.8 × 105
100 Invar (ref)
1.2 × 104
The experimental electrical conductivity values and calculated thermal conductivity values for Ag-Invar alloys match closely with the son Frey model for high Ag and low Ag content. The trends in conductivity may be explained through an examination of phase continuity through the microstructure. In samples of high Ag content, the (Ag) phase forms a continuous, high-conductivity matrix that enables excellent electronic conduction through the sample; and the Invar phase in high Ag samples forms isolated, low-conductivity grains that do not significantly disrupt the flow of electrons. Similarly, at low Ag contents, the low-conductivity Invar phase is continuous, and the high-conductivity (Ag) exists only in isolated grains. In samples of low Ag composition, the conducting electrons are forced to move through the high resistivity Invar regions. Since the conductivity difference between pure Ag (σ = 5.81 × 105 Ohm−1·cm−1) and Invar (σ = 1.20 × 104 Ohm−1·cm−1) is so great, the loss of continuity in the (Ag) phase quickly drives the alloy conductivity to lower values. Conductivities for samples of intermediate Ag content are expected to transition along an S-shaped curve between the predicted values for continuous (Ag) and continuous Invar.
3.4 Solubility and Phase Stability of Ag-Invar
As noted in Section I–B, little published information on the phase development and ambient-temperature phase stability in the Ag-Fe-Ni ternary system is available. As such, the phase stability, microstructural development, and properties of ternary Ag-Fe-Ni alloys are still in need of thorough research.
This investigation has demonstrated that thermodynamically stable, two-phase Ag-Invar alloys are formed along a constant 64Fe-36Ni compositional line in the ternary system. The formation of two-phase Ag-Invar alloys is not entirely surprising, given the solubility information available in the Ag-Fe, Ag-Ni, and Fe-Ni binary phase diagrams. Solubility limits of Fe and Ni in the (Ag) solid solution at silver’s melting temperature (961.93 °C) are 0.00337 and 0.7 wt pct, respectively. These low solubility limits, even at high temperatures, indicate that Fe and Ni are unlikely to diffuse into (Ag) and thus unlikely to negatively affect electrical conductivity. Similarly, silver has low miscibility with iron and nickel, with limits of 0.0004 and 1.8 wt pct for Ag in (Fe) and (Ni), respectively. The low solubility of Ag in iron and nickel, and vice versa, indicates that the critical ratio of Fe to Ni in the γ-(Fe,Ni) Invar phase should be maintained in the Ag-Invar alloys.
Since the Ag-Ni solubility limits are higher than those in Ag-Fe, the solid-state diffusion of Ni in the ternary Ag-Invar alloys represents a potential problem. Although EDS compositional dot mapping indicates that the desired phase separation is occurring, it is possible that some Ni may be dissolved in the (Ag) phase. Dissolved Ni atoms in the silver phase would simultaneously decrease the (Ag) conductivity and increase the Invar thermal expansion. Since solubility limits of Ni in (Ag) and Ag in (Ni) decrease with temperature, however, further atomic segregation and improved conductivity and CTEs may be achieved through aging heat treatments. It is expected that most dissolved Ni may be driven out of the (Ag) phase by heating the Ag-Invar alloys to temperatures around 0.5 TM and holding for some time.
Nanoscale Ag-Fe-Ni ternary metal powders were prepared using a simple, ambient-temperature, solution-based chemical technique involving sodium borohydride reduction of metal nitrates. The Ag and Invar phases form upon annealing of the nanocrystalline Ag-Fe-Ni powder at 520 °C in H2, and sintering of the annealed powder in H2 at 900 °C leads to the formation of fine-grained, two-phase Ag-Invar alloys. Compositional dot mapping indicates nearly complete compositional separation between the Ag and γ-(Fe,Ni) Invar phases, resulting in high conductivity in the Ag and low thermal expansion in the Invar. The CTEs and electrical conductivities of the Ag-Invar alloys decrease with increasing Invar content, and specific values of CTE and conductivity are easily obtainable by this synthesis and processing approach. The CTE values closely match those predicted by common models, and the electrical conductivity values are described by simple models for two-phase mixtures. The thermal and electrical properties of the Ag-Invar alloys prepared by this method are comparable to some of the existing low CTE–high conductivity materials, indicating that a range of Ag-Invar alloys may be viable for use in thermal management applications. The simple synthesis, relatively low processing costs, small grain size, and isotropic properties of the ternary alloys may offer distinct advantages over some of the materials currently in use.
JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.
The authors thank Clara Cho and Alex Epstein for their contributions to this project and the Franklin W. Olin College of Engineering for partial support of the project through the Olin College Faculty Research Program.