(734h) Copper-Carbon Nanotube Composites for Lightweight Electronics

As electronics are driven towards progressively smaller scales and higher speeds, bulk metals such as copper are reaching their material limits, and the use of metal composites with ultraconductive nanomaterials, particularly carbon nanotubes (CNTs), is becoming increasingly appealing to make fast, efficient devices with high electrical and thermal performance. However, fully aligned, densely-packed CNT assemblies within metal networks are required to generate optimal wiring and electronic structures. Even small microstructural defects can cause large losses in mechanical strength and electrical/thermal transport, which constrains the range of processing strategies that can be employed. To address this challenge, we have been using experimental techniques to integrate copper (Cu) with CNTs for applications in electronics, and focusing on the homogeneous nucleation of Cu within preformed CNT mats.

Cu/CNT composites have already been demonstrated but their bulk production universally relies on a multi-step process to pre-treat the CNT surface before depositing copper, adding complications and reducing process speed. Pretreatment is required because of the hydrophobic nature of CNT bundles and mats, which restricts deposition of metal directly from aqueous electrolyte solutions. In addition, favored pretreatments often add defects to the CNTs, increasing hydrophilicity at the cost of reducing conductivity. In this work, we introduce a new method to create a uniform Cu-CNT composite from an isotropic CNT mat in single step electrodeposition using an electrolyte solution modified to have a marginally wetting contact angle. We first describe the requirements to transition from nonwetting to wetting systems, considering multiple approaches including chemical additives such as surfactants, organic solvent mixtures (particularly methanol-water mixtures), and electrowetting, and how these influence the copper nucleation, percolation, and final composite conductivity.

We focus on the method of Cu nucleation and percolation by comparing two cases. (1) Nonwetting solutions cause formation of two-dimensional Cu overcoatings on the CNT mat, which percolate at lower volume fractions but ultimately fail to form a true composite. (2) Marginally wetting solutions allow formation of fully three-dimensional Cu networks within the CNT mat. We describe the limits of Cu penetration into dense initial CNT networks using electron dispersion spectrographs of cross-sections of experimental specimens. In addition, we demonstrate what changes are required to obtain good agreement of the Cu deposition with Faraday’s First Law of Electrolysis using conductivity and mass measurements, and dependence of the copper nucleus size on the dielectric constant of the adjusted solvent. We finally discuss the limits to homogeneous Cu concentration and process speed and requirements to turn this into a scalable process, e.g., for scalable Cu-CNT wire manufacturing.

Caption:

(a) The initial substrate consists of a non-woven mat of carbon nanotubes (CNTs). (b) Various treatments to the deposited electrolyte control the apparent contact angle on the textured substrate, leading to marginal wetting behavior. (c) As a direct result, electrodeposition forms Cu nuclei either only on the outer surfaces or throughout the CNT mat, as shown by scanning electron micrographs and electron dispersion spectrographs for each case. (d) An optical micrograph shows copper nuclei at a larger scale. (e) At higher concentrations, bulk conductivity measurements begin to discriminate between core-shell and homogeneous composites prepared via different deposition strategies. These are compared with models for conductivity including Hashin-Shtrikman bounds which delineate thresholds for isotropic composite conductors, a fitted percolation model (with critical density 10% and scaling exponent 1.5), and the standard model for the conductivity of three-dimensional Cu foam, which scales as density ratio cubed and has no fitting parameters.