Jun Peng1 Ian Witting1 Nicholas Geisendorfer1 Christoph Kenel1 David Dunand1 Ramille Shah1 G. Snyder1 Matthew Grayson1

1, Northwestern University, Evanston, Illinois, United States

Thermoelectric materials have broad applications in energy harvesting and Peltier cooling, though the typically rigid configuration limits potential device applications, especially where flexible platforms are required. Most explorations of flexible composite-based thermoelectric materials use a conducting polymer thermoelectric to improve the conductivity while sacrificing the thermoelectric performance. Thin films or layers deposited on soft substrates can also provide a flexible solution, though such processing requires high cost deposition systems.

Here we propose a 3D printing method to make thermoelectric threads out of standard high-performance thermoelectric materials as the initial step towards weaving thermoelectric fabrics. Two methods are demonstrated: one employing a 3D printed polymer matrix surrounding the thermoelectric powder, and another whereby the polymer is sintered away, leaving behind a pure thermoelectric polycrystalline thread. Here, we 3D-printed fine n-type and p-type thermoelectric fibers with a diameter less than d = 50 µm by extruding composite thread using nonconductive poly (lactic-co-glycolic acid) (PLGA) as a polymer binder, followed H2 sintering to remove the binder, and liquid-phase compression to reduce porosity with an excess tellurium at T = 440 C and P = 45 MPa. The thermoelectric powders were produced by high energy ball milling, and the diameter of grains or clusters was c = 0.1~10 μm. The composite fibers were the most flexible, and even the sintered fibers acquired a bend radius of order cm-scale.

Sintering was observed to significantly improve the electric conductivity. For an example, in comparison with the conductivity of the initial 3D-printed p-type composite thread, σ = 1 * 10-6 S/cm, that of sintered counterpart was almost 8 orders of magnitude greater at σ = 44 S/cm, which is comparable to that of bulk pellets of the same Bi0.5Sb1.5Te3 powder, σ = 120 S/cm. The Seebeck coefficient of sintered samples was preserved, S = 160 µV/K, also comparable to that of bulk materials, S = 200 µV/K. Therefore, the calculated power factor (PF) of p-type fiber was PF = 112.6 µW/mK2. With the assumption that the sintered fiber manifests a thermal conductivity similar to the identically processed bulk p-type pellets, κ = 1.1 W/mK, the material figure-of merit of sintered fiber is approximately, zT = 0.03 which is only four times lower than that of bulk pellets, zT= 0.13, at room temperature, T = 300K. With optimally doped p-type material with zT = 0.8, one can therefore project an eventual goal of at least zT = 0.19 for these thermoelectric threads. The relatively lower zT of the threads could be attributed a porous structure after sintering. Preliminary results of a thermoelectric module fabricated by 3D-printing of n-type and p-type fibers will be demonstrated.