How Does Copper Content in Tungsten–Copper Affect Thermal Conductivity and CTE? A Grade Comparison from W90Cu10 to Higher-Copper Grades
Published:
How Does Copper Content in Tungsten–Copper Affect Thermal Conductivity and CTE? A Grade Comparison from W90Cu10 to Higher-Copper Grades
Tungsten–copper (W–Cu) is a pseudo-alloy, not a solid solution. Tungsten and copper are essentially insoluble in each other, so a part is a two-phase composite: a tungsten skeleton infiltrated with copper, or a blended powder compact consolidated near net shape. Properties follow copper fraction and microstructure. They do not follow a heat-treatment temper, and they do not follow a trade name.
Newlife - MIM runs a MIM/PM chain from compounding and feedstock through injection molding, debinding, sintering and finishing, and produces tungsten–copper and copper parts, including near-net micro-fins, micro-channels and cavities. Copper fraction is still a drawing variable. The same “W–Cu” label can sit at very different expansion and conductivity, and a grade inherited from a previous design should be checked against the counterface before tooling is cut.
Copper content is the variable that moves both thermal conductivity and CTE, and it does not move them at the same rate
Copper weight fraction is the grade attribute that shifts both thermal conductivity and thermal expansion enough for a designer to act on. W90Cu10, W80Cu20 or W70Cu30 is usable only after it is tied to that weight fraction and to measured properties on the supplier datasheet. A proprietary trade name is not a specification until the same mapping is written down.
The two properties do not scale together. Handbook values for the pure metals are about 4.5 ppm/K and about 170 W/m·K for tungsten, against about 17 ppm/K and about 400 W/m·K for copper (CRC Handbook of Chemistry and Physics). The ratios are what selection uses: copper expands between about three and a half and four times as much as tungsten, and conducts only a little more than twice as well. Adding copper therefore buys expansion faster than it buys conductivity.
On most die-attach and ceramic-package programs, CTE match to the counterface is the binding constraint. Thermal conductivity is then whatever the matched grade measures. Extra copper specified “for thermal margin” can spend the expansion budget for a modest conductivity gain, which is the mismatch that drives joint fatigue. If the thermal budget is short and the counterface cannot tolerate more expansion, change the cross-section, add a copper cladding on the already matched grade, or — where the geometry justifies tooling — mold micro-channels. Change the grade only after those options are compared.
Use mean CTE over the operating interval. A single room-temperature point is not the expansion the joint sees. Free expansion of the substrate is not itself a life metric: bonded-joint strain follows the CTE difference and the temperature swing; length enters as edge displacement and warpage; joint thickness sets the shear. A grade step can matter on a short part, and it can be irrelevant if the counterface moves with it. Compare mean CTEs, then judge the joint.
The manufacturing route, not the grade name, decides which shapes are practical
Four routes reach a W–Cu part. The route sets the geometry that is affordable. It is not implied by the grade name.
| Route | How the composite is made | Shape | Volume where it usually fits | Constraint to design around |
|---|---|---|---|---|
| Press and sinter a W skeleton, then infiltrate with Cu | Liquid copper is drawn into open porosity in a tungsten skeleton | Simple to moderate: discs, plates, blocks | Medium to high | Incomplete fill in thick or blind sections |
| MIM, then sinter, with infiltration only if that route requires it | Binder feedstock molded near net, debound and sintered; copper may be in the feedstock or infiltrated afterward | High: micro-channels, cavities, fins, stepped and thin walls | High once tooling is amortised | Debinding, section uniformity, shrinkage; any infiltration step still has to reach every section |
| Die pressing or extrusion of blended powder | Copper is in the blend; a separate infiltration step may or may not be used | Low: rods, tubes, sheets, simple profiles | High for simple forms | Segregation and density gradients; do not assume the blend reaches infiltrated density |
| CNC from sintered or wrought stock | Copper is already in the purchased stock | Unrestricted, but subtractive | Low volume, and prototypes | W–Cu is abrasive; most of the blank can become scrap |
A flat, simple, thick plate with loose tolerances is usually cheaper to press or to machine from stock than to mold. MIM earns its tooling when the part carries features that would otherwise be milled away. That is the class of geometry in Newlife - MIM’s tungsten–copper and copper scope: near-net micro-fins, micro-channels and cavities, as an alternative to expensive copper machining.
A grade is comparable only after copper fraction and measured CTE and conductivity are stated
The table is a selection map, not a datasheet. It does not list CTE or thermal conductivity. Those values depend on density, infiltration quality and test direction, and they have to come from the supplier for the grade actually made. Because copper’s expansion advantage over tungsten is much larger than its conductivity advantage, a step from a 10 wt% copper grade to a 30 wt% copper grade moves CTE by a larger fraction than it moves thermal conductivity. The size of either move is a measured number, not a mixture-rule output.
| Designation | Role in selection | Do not assume |
|---|---|---|
| W90Cu10 | A high-tungsten grade: lower expansion and lower conductivity than W80Cu20 | A match to any named ceramic or semiconductor until both mean CTEs are compared over the operating range |
| W80Cu20 | A common mid grade. The step from W90Cu10 is a joint decision, not a rounding difference | That conductivity has improved in proportion to expansion |
| W70Cu30 | Higher expansion and higher conductivity. Copper–tungsten in this higher-copper region is also used for electrical duty, including EDM electrodes and resistance-welding electrodes; contact grades may be called out under ASTM B702 rather than by a packaging name | That the grade is still the right composite if CTE is unconstrained |
| Proprietary trade names | Not a specification | Any comparison until copper wt%, density, CTE and thermal conductivity are stated |
Electrical-contact specifications and packaging heat-sink datasheets are not interchangeable. Confirm which document the drawing is actually calling.
Wall uniformity, feature size and the following joint set the design window
Infiltration and MIM are aimed at the commercial middle of the composition range, and they become the wrong tool at the edges for different reasons. At low copper, the usual industrial route is a densified tungsten skeleton that is then infiltrated. If that skeleton is left porous, or if infiltration misses a thick or blind section, conductivity falls below the value used in the thermal design. At high copper, the composite is copper-rich; if CTE is not constrained, a copper or copper-alloy route is often simpler and more conductive. Exact composition limits are process-specific. Take them from the process owner, not from a generic cutoff.
Inside that middle, the drawing is constrained by what the process actually feels. MIM feedstocks prefer reasonably uniform sections. Micro-channels, fins, pockets and stepped pedestals are where near-net shaping pays for tooling. A later braze or sinter-bond will demand flatness that an as-sintered percentage tolerance does not automatically provide. Heat flow may also be directional: a pressed skeleton can be mildly anisotropic, so a datasheet value is safe only in the direction in which it was measured.
Newlife - MIM’s published near-net-shape control is relative density ≥ 98% and dimensional tolerance of ±0.3% to ±0.5%. The published sintered-density minimum for the pure-copper, copper-alloy and tungsten–copper MIM family is ≥ 8.4 g/cm³. That 8.4 g/cm³ figure is a family minimum, not the density to write on a high-tungsten drawing. Theoretical density rises as the tungsten fraction rises, and the grade density has to be agreed explicitly. The ±0.3% to ±0.5% band is an as-sintered capability. On a die-attach or braze face it often still needs a finishing allowance, because flatness and roughness are tighter than a percentage of overall size. Agree the finished interface separately from the as-sintered band.
Debinding, shrinkage and incomplete infiltration are the reasons a W–Cu MIM part is rejected
The dominant process risks are incomplete debinding, non-uniform shrinkage and incomplete infiltration. Residual binder carbon changes how the skeleton sinters and shows up as lot-to-lot density scatter, so the debinding profile and the furnace atmosphere are process parameters worth auditing. Uneven wall thickness produces differential shrinkage at thin-to-thick transitions; the distortion appears after sintering and is expensive to correct. Thick or blind sections can remain locally uninfiltrated, which removes the conductivity the thermal model assumed. Shrinkage compensation on an infiltrated skeleton is not the same problem as single-material MIM, because copper occupies porosity the skeleton leaves behind. The compensation model has to be the one qualified for that route.
Do not force W–Cu MIM onto a plain, thick, low-complexity plate; pressing and machining is usually the shorter path. Do not use it when the only requirement is maximum conductivity and CTE is free; copper is the better conductor. Do not use it at a copper fraction where a copper alloy meets the expansion budget at lower cost. Do not use it for a single piece that exceeds the practical mold and furnace envelope, or for a wide-tolerance prototype where machining from stock wins on lead time.
Measured properties and lot traceability qualify a supplier; the grade name does not
Newlife - MIM develops its own ultrafine copper, iron and nickel powders at D50 ≤ 3 µm and supplies customized MIM feedstock. That particle-size statement covers copper, iron and nickel. It is not a tungsten powder specification, and it should not be read as the reason a tungsten skeleton shrinks predictably. Near-net tungsten–copper and copper MIM is offered against the density and tolerance figures above, inside IATF 16949, ISO 9001 and ISO 14001 systems. Those certificates indicate a documented quality system. They do not replace a capability check on the dimensions the assembly uses, or a datasheet line for CTE, thermal conductivity and density of the grade actually molded.
Before tooling is released, ask for powder-lot, feedstock-lot and furnace-lot traceability, and for measured CTE, thermal conductivity, density and the two or three dimensions the assembly actually cares about, in the test direction heat will flow.
Send the drawing and the CTE target, and confirm the grade before tooling is cut
Grade names do not match thermal expansion. Copper weight fraction and measured CTE do. Bring the counterface material, the operating temperature range, the flatness and finish the braze or sinter-bond needs, and the annual volume. Copper content and process route can be narrowed before a tool is cut. If the copper content on the current drawing was inherited rather than derived from the counterface, that is the first item to check.
Visit http://www.newlife.cn to discuss a tungsten–copper or copper MIM project with the Newlife - MIM application engineering team, request a free consultation, and agree the grade, density and tolerance against the drawing before tooling is released.
References
1. ASM Handbook, Volume 7: Powder Metal Technologies and Applications. ASM International. 2. German, R. M. Powder Metallurgy and Particulate Materials Processing. Metal Powder Industries Federation, 2005. 3. ASTM B702/B702M. Standard Specification for Copper-Tungsten Electrical Contact Material. 4. ASTM B328. Standard Test Method for Density, Oil Content, and Interconnected Porosity of Sintered Metal Structural Parts and Oil-Impregnated Bearings. 5. CRC Handbook of Chemistry and Physics. CRC Press. Thermal and physical properties of pure tungsten and copper. 6. Newlife - MIM: http://www.newlife.cn