The performance of a thermoelectric material is determined cooperatively by the Seebeck coefficient (S), thermal conductivity

(κ), and the electrical conductivity (σ) of the material [4]. Unfortunately, these three parameters have some intercorrelations in bulk, #ITF2357 randurls[1|1|,|CHEM1|]# limiting the thermoelectric performance of a bulk material [5]. In this regard, one-dimensional (1D) nanowires have been highlighted, where a combination of quantum confinement effect and phonon boundary scattering drastically enhances the thermoelectric performance [6–8]. However, the controlled growth of thermoelectric nanowires and the reproducible fabrication of energy conversion modules based on them should be further demonstrated. Two-dimensional (2D) thin films have the superiority in terms of the ease GDC-0449 in vitro of material and module fabrication

and the reproducibility of the thermoelectric performance. The best thermoelectric materials reported to date include Bi2Te3 [9], AgPbmSbTe2+m [10], and In4Se3−δ [11]. These materials, however, contain chalcogens (Se, Te), heavy metals (Pb, Sb), and rare metals (Bi, In), all of which are expected to restrict the widespread use of these materials. Recently, it has been demonstrated that even a conventional semiconductor, silicon (Si), can exhibit thermoelectric performance by adopting nanostructures such as nanowires [12], nanomeshes [13], and holey thin films [14]. Although Si has a high S of 440 μV/K, its electrical conductivity is poor (0.01 ~ 0.1 S/cm) [15]. Thus, alloying Si with a good metal could lead to the improved

thermoelectric performance. Aluminum (Al) is a typical good metal that has Celecoxib the advantages of high electrical conductivity (approximately 3.5 × 105 S/cm) [16], light weight, and low cost. Despite the expected high electrical conductivity, the thermal conductivity of Si-Al alloys may be still high due to the large thermal conductivities of the constituents: κ Al = 210 ~ 250 W/m K and κ Si = 149 W/m K at room temperature [17]. The thermal conductivity of the alloy can be reduced by introducing nano- or microstructures on the alloy film. For this reason, embodying nano- or microstructures on Al-Si alloy films is a critical prerequisite for the study of thermoelectric performance of heterostructures made of Al-Si alloys. In this work, aluminum silicide microparticles were formed from Al thin films on Si substrates through self-granulation. This process resulted from solid-state interdiffusion of Al and Si at hypoeutectic temperatures, which was activated by compressive stress stored in the films. This stress-induced granulation technique is a facile route to the composition-controlled microparticle formation with no need of lithography, template, and chemical precursor.