Nanoporous Ca3Co4O9 Thin Films for Transferable Thermoelectrics

The development of high-performance and transferable thin-film thermoelectric materials is important for low-power applications, e.g., to power wearable electronics, and for on-chip cooling. Nanoporous films offer an opportunity to improve thermoelectric performance by selectively scattering phonons without affecting electronic transport. Here, we report the growth of nanoporous Ca3Co4O9 thin films by a sequential sputtering-annealing method. Ca3Co4O9 is promising for its high Seebeck coefficient and good electrical conductivity and important for its nontoxicity, low cost, and abundance of its constituent raw materials. To grow nanoporous films, multilayered CaO/CoO films were deposited on sapphire and mica substrates by rf-magnetron reactive sputtering from elemental Ca and Co targets, followed by annealing at 700 °C to form the final phase of Ca3Co4O9. This phase transformation is accompanied by a volume contraction causing formation of nanopores in the film. The thermoelectric propoperties of the nanoporous Ca3Co4O9 films can be altered by controlling the porosity. The lowest electrical resistivity is ∼7 mΩ cm, yielding a power factor of 2.32 × 10–4 Wm–1K–2 near room temperature. Furthermore, the films are transferable from the primary mica substrates to other arbitrary polymer platforms by simple dry transfer, which opens an opportunity of low-temperature use these materials.


INTRODUCTION
Nanoporous materials are promising in the area of thermoelectricity, as they can enable simultaneous tailoring of electronic and phononic properties in a single material system, leading to multifold enhancement of thermoelectric efficiency. 1−3 The thermoelectric efficiency of any material system is related to dimensionless thermoelectric figure of merit ZT (=S 2 T/ρκ), where S, ρ, κ, and T are the Seebeck coefficient, electrical resistivity, thermal conductivity, and absolute temperature, respectively. High thermoelectric efficiency requires high Seebeck coefficient simultaneously with low electrical resistivity and thermal conductivity. However, design of such materials is quite challenging because these parameters are interdependent with electrically conducting materials having low Seebeck coefficient and high thermal conductivity, and vice versa.
Bulk nanostructured thermoelectric materials can be used to achieve low phonon thermal conductivity, while retaining good electronic properties. 4−12 Nanoscale features with dimension comparable to the phonon mean free path have been incorporated to preferentially scatter the phonons to reduce thermal conductivity and thus enhance ZT. An alternative approach for selective scattering of phonons can be the incorporation of nanopores with controlled size and periodicity. 2,3 The average mean free path of electrons in most materials is typically 1 order of magnitude lower than phonon mean free path. For example, the mean free path of electrons in silicon (Si) is in the range 1−10 nm for heavily doped Si with carrier concentration of the order of 1 × 10 19 cm −3 , while the phonon mean free path is 300 nm at 300 K. 13 Thus, the reduction in thermal conductivity of nanoporous materials is possible without adversely affecting electronic properties, by controlling the characteristic length scale of the porous structure in the range in-between electronic and phonon mean free path. For example, the thermal conductivity of thin holey silicon with 55 nm pitch (periodicity of pores) can be reduced by almost 2 orders of magnitude as compared to the pristine bulk value, while retaining a high power factor, resulting in enhanced ZT ≈ 0.4 at 300 K. 2 Others have reported drastic reduction of thermal conductivity in Si-based 2D phononic crystals due to the suppression of phonon mean free path; 14,15 however, with no report on their electronic or thermoelectric properties. Ca 3 Co 4 O 9 is a promising thermoelectric material because of low cost, abundance, and nontoxicity of its constituent raw materials. However, the best performance of this class of materials typically occurs at high temperatures near 1000 K. Investigations on bulk nanostructured Ca 3 Co 4 O 9 have not reported significant improvement of power factor near room temperature. 16−24 Because of the inherently layered structure, the electronic properties of Ca 3 Co 4 O 9 are anisotropic in nature, and less resistive electronic transport is found to occur in (a, b) plane of Ca 3 Co 4 O 9 . Thus, for achieving high power factors in this material system, oriented thin films can be used for exploitation of anisotropic properties. We have previously demonstrated that the high power factor is retained down to near room temperature in Ca 3 Co 4 O 9 thin films on sapphire substrates. 25 There have been extensive investigations on thin film growth of Ca 3 Co 4 O 9 thin films. 26−29 However, nanoporous Ca 3 Co 4 O 9 thin films remain unexplored.
Here, we report a method for the growth of nanoporous Ca 3 Co 4 O 9 thin films. The method requires neither templates nor etching steps like previous reports on the growth of thin nanoporous films. 2,3,30 Thermoelectric properties of the films are characterized in terms of their power factors. A retained high power factor near room temperature is important for mechanically flexible applications, where the output power is more important than the efficiency. For high output power, a high power factor is more important than achieving high ZT. 31 Even with large number of pores, a high power factor of 2.32 × 10 −4 W m −1 K −2 is obtained near room temperature from undoped nanoporous Ca 3 Co 4 O 9 thin films. Furthermore, the nanoporous films are transferable onto other arbitrary flexible platforms by mechanical stripping, thus opening a new opportunity for transferable thermoelectrics.

EXPERIMENTAL SECTION
Nanoporous Ca 3 Co 4 O 9 thin films were prepared by a two-step sputtering/annealing method. First, CaO/CoO films were sequentially deposited by rf-magnetron reactive sputtering from metallic targets of Ca and Co onto muscovite mica (00l) and sapphire (00l) substrates at 0.27 Pa (2 mTorr) in an oxygen−argon mixture with oxygen 0.5%, while maintaining the substrate temperature at 300°C for sapphire substrates and 600°C for mica substrates. The target powers were controlled to maintain deposition rate of 5.5 nm/min for CaO and 4.5 nm/min for CoO. In second step the as-deposited CaO/CoO films were annealed at 700°C in O 2 gas flow to form the final phase of Ca 3 Co 4 O 9 . The crystal structure and morphology of the films were characterized by θ−2θ X-ray diffraction (XRD) analyses using monochromatic Cu Kα radiation (λ = 1.5406 Å), transmission electron microscopy by using a FEI Tecnai G2 TF20 UT instrument with a field emission gun operated at 200 kV and with a point resolution of 1.9 Å, and scanning electron microscopy (SEM, LEO 1550 Gemini). The θ−2θ XRD scans were performed with a Philips PW 1820 diffractometer. For cross-sectional TEM, two pieces of the samples were glued together face to face and clamped with a Ti grid and then polished down to 50 μm thickness. Finally, the polished sample was ion milled in a Gatan Precision Ion Polishing System (PIPS) at Ar + energy of 5 kV and a gun angle of 5°, with a final polishing step with 2 kV Ar + energy. The composition of the films was determined by EDS attached to TEM, with an accuracy ±5%. The temperature dependent in-plane electrical resistivity and Seebeck coefficient were simultaneously characterized using an ULVAC-RIKO ZEM3 system in a low-pressure helium atmosphere. The available surface area of the films was measured by Kr-sorption at 77 K using an ASAP2020. The samples, i.e., film on a substrate, were degassed at 100°C for 17 h prior to the measurements. The BET surface area was determined at P/P 0 = 0.12−0.20. The BET surface was recalculated to available surface area/film volume using the following equation: where the film thickness was estimated from TEM, and the area of the film-coated substrate was determined by optical imaging. It was assumed that all contribution to the specific surface area originated from the films since the reference measurements on bare substrates did not provide any measurable value. Figure 1a is a scheme of sequentially deposited CaO/CoO films with two different periodicities of the layers. Four samples, namely S Al2O3 : 5.5/4.5, S mica : 5.5/4.5, S Al2O3 : 11/9, and S mica : 11/9, have been deposited. The films are named after the type of substrates and thickness of individual CaO and CoO layers in the as-deposited films. For example, S Al2O3 : 5.5/4.5 and S mica : 5.5/4.5 films were deposited on sapphire Al 2 O 3 (001) and mica (muscovite mica (00l)) substrates, respectively, and the thicknesses of sequential CaO and CoO layers are 5.5 and 4.5 nm, respectively. In S Al2O3 : 11/9 and S mica : 11/9, the thicknesses of CaO and CoO layers are 11 and 9 nm, respectively. Figure 1b shows a cross-sectional transmission electron microscopic (TEM) image of as-deposited CaO/CoO films S Al2O3 : 5.5/4.5 and S Al2O3 : 11/9, respectively. Figure 1c shows an EDS map of a small portion of the as-deposited films S Al2O3 : 5.5/4.5 and S Al2O3 : 11/9. EDS mapping confirms that the dark lines in Figure 1b are from CoO phase, and bright lines are

ACS Applied Energy Materials
Article from CaO phase. Figure 1d, e show magnified images of small portions of the as-deposited films S Al2O3 : 5.5/4.5 and S Al2O3 : 11/9, respectively. In the as-deposited film S Al2O3 : 5.5/4.5 the period is 10 nm (i.e., 5.5 nm + 4.5 nm) and with 20 alternating layers. The total thickness of the film is 100 nm. In the asdeposited film S Al2O3 : 11/9 the period is 20 nm and with 20 alternating layers. The total thickness of the film is 200 nm. The layered structure of as-deposited films on the two different substrates are very similar (not shown here). Figure 2a shows θ−2θ XRD scans of the as-deposited films S Al2O3 : 5.5/4.5 and S Al2O3 : 11/9. In Figure 2a, the XRD peaks at 2θ angles 32.37°, 36.55°are from the (111) planes of CaO and CoO, respectively, which is consistent with our previous observations on cosputtered CaO−CoO thin film deposited on sapphire substrate. 25 Figure 2b shows the corresponding XRD scans for the as-deposited films on mica. In Figure 2b, the CoO peak is not visible as it coincides with the (004) peak of mica. Broad peaks at around 8.82, 17.81, 36.02, and 45.42°originate from (00l) planes of the mica substrate. Figure 2c shows an XRD scan of annealed films S Al2O3 : 5.5/4.5 and S Al2O3 : 11/9. Peaks from (00l)-planes of Al 2 O 3 are visible in Figure 2c for both films. Apart from XRD peaks from (00l) planes of Ca 3 Co 4 O 9 , a small peak of CaO at 2θ angle 32.37°is visible for both films, which can be attributed to a slight Ca overstoichiometry in the films. Figure 2d shows the corresponding XRD scans of annealed films on mica. Peaks from (00l)-planes of Ca 3 Co 4 O 9 are clearly visible in Figure 2d for both the films S mica : 5.5/4.5 and S mica : 11/9. Apart from Ca 3 Co 4 O 9 , the broad peaks from mica substrate are visible in Figure 2d. However, no peak of CaO is seen which indicating the phase purity of the

ACS Applied Energy Materials
Article film. The d-spacings of the annealed Ca 3 Co 4 O 9 films S Al2O3 : 5.5/4.5, S Al2O3 : 11/9, S mica : 5.5/4.5, and S mica : 11/9 are calculated to be 10.7218, 10.7297, 10.7404, and 10.7337 Å, respectively, which are consistent with the reported d-spacing for Ca 3 Co 4 O 9 single crystal. 32 The corresponding out-of-plane lattice parameters (c-parameter) are 10.8306, 10.8386, 10.8494, and 10.8426 Å for S Al2O3 : 5.5/4.5, S Al2O3 : 11/9, S mica : 5.5/4.5, and S mica : 11/9, respectively, and consistent with the reported cparameter of bulk Ca 3 Co 4 O 9 . 33 From the above results, it is clear that the final phase of Ca 3 Co 4 O 9 is obtained from all sequentially deposited CaO/ CoO films irrespective of substrate. During annealing, a threestage phase transformation from sequential CaO/CoO films to the final phase of Ca 3 Co 4 O 9 occurs, as shown by our previous study on cosputtered CaO-CoO thin films on sapphire substrates. 25 All annealed Ca 3 Co 4 O 9 films are c-axis-oriented irrespective of substrate. The advantage of mica as substrate is that even with excess Ca in the as-deposited films, the postannealed Ca 3 Co 4 O 9 films on mica substrates are phasepure. In this case, excess Ca is incorporated in an amorphous interfacial layer between the mica substrate and the film (this is discussed later in detail).
Figures 3a−d show SEM images of the annealed films S Al2O3 : 5.5/4.5, S Al2O3 : 11/9, S mica : 5.5/4.5, and S mica : 11/9, respectively. The presence of horizontal grains with dimension of several hundred nanometers in the postannealed film S Al2O3 : 5.5/4.5 is seen in Figure 3a. Visible bright spots on the film surface are from grains of different orientation. These grains are not observed in XRD, since they do not satisfy Bragg's condition in the out-of-plane direction, which is consistent with previous observations for the Ca 3 Co 4 O 9 films grown on SrTiO 3 (111) 27 and on muscovite mica. 34 In contradiction, SEM of the postannealed film S Al2O3 : 11/9 does not show the presence of any of these grains (see Figure 2b), which is also confirmed by TEM image analyses (discussed later). The surface of the film S Al2O3 : 11/9 is relatively smoother than the surface of the film S Al2O3 : 5.5/4.5. The visible black spots on the surface of the film S Al2O3 : 11/9 are from randomly distributed pores in the film having dimension in the range from few nanometers to several hundred nanometers. Nanopores in the annealed film S mica : 5.5/4.5 are irregular in shape, but distributed rather homogeneously in the film. The nanopores in the film S mica : 11/9 are polygonal in shape, and having dimension in the range from a few tens of nanometers to several hundred nanometers. The nanopores in the film S mica : 11/9 have visible openings with sharp edges, in contrast to the film S Al2O3 : 11/9. It is clear from Figure 3 that the porosity varies from film to film. The porosity of the films is compared in terms of their available surface area per unit volume, where a high available surface indicates a large porosity since the pore sizes in all films are in the same range. The available surface areas of the films per unit volume are calculated to be 0.11, 0.68, and 0.26 m 2 / mm 3 for the films S Al2O3 : 11/9, S mica : 5.5/4.5, and S mica : 11/9, respectively. The highest value of available surface area per unit volume of the film S mica : 5.5/4.5 is attributed to its higher porosity than the rest films. On the other hand, the lowest value of available surface area per unit volume of the film S Al2O3 : 11/9 is due to its low porosity, which is consistent with the SEM observations. The distinct variation of the surface morphology and porosity of the films on different substrates indicates the difference in substrate influence on thin film growth of nanoporous Ca 3 Co 4 O 9 . Figure 4a shows a cross-sectional TEM image of the annealed film S Al2O3 : 11/9. The bright spot in Figure 4a is a void because of a pore in the film. Figure 4b shows a magnified image of a small region, where the layered structure of Ca 3 Co 4 O 9 and its orientation along c-axis is apparent. The average thickness of this film is 160 nm, a reduction by nearly 20% compared to the as-deposited film. A similar reduction in thickness was observed for the film S Al2O3 : 5.5/4.5 after annealing, with a final thickness of 80 nm. The presence of voids at the interface between the substrate and the annealed film S Al2O3 : 11/9 is confirmed by TEM imaging (Figure 4c). Figure 5a shows a cross-sectional TEM image of the annealed film S mica : 11/9. The average thickness of the S mica : 11/9 film is ∼150 nm, which is 10 nm lower than the thickness of corresponding annealed film on sapphire. This reduction in film thickness is attributed to the incorporation of excess Ca in an amorphous interfacial layer, as confirmed by EDS analyses (see below), which is consistent with our previous observation on the growth of flexible Ca 3 Co 4 O 9 films on mica substrates. 34 In Figure 5a, voids are visible throughout the interfacial region of the annealed film on mica, in contrast to corresponding film on sapphire. As a consequence, Ca 3 Co 4 O 9 film is weakly

ACS Applied Energy Materials
Article bonded to the mica substrate via nanopillars. Figure 5b shows a magnified image showing that the film is supported by nanopillars on the mica substrate. The interfacial layer between the film and mica substrate is amorphous in nature, and by EDS analyses the amorphous layer is determined to be Ca-rich (Ca: 31.3 at %). This is because the excess Ca in the film is incorporated in the amorphous layer. The other elements: O (48.8 at %), Al (6.8 at %), Si (8.5 at %), K (0.9 at %), and Fe (3.7 at %) in the amorphous layer retain the same proportion as that in the mica substrate. Figure 5c shows the HRTEM image of a nanopillar of width of around 25 nm. The layered structure of the Ca 3 Co 4 O 9 phase in the nanopillar is visible.
From the above SEM and TEM results, it is clear that the mica and sapphire substrates affect the growth of Ca 3 Co 4 O 9 thin films differently, leading to variations in the resulting nanoporous structures. The formation of nanopores in the films is likely caused by the volume contraction of the films after annealing. As mentioned before, the thickness of the annealed films is reduced by around 20% as compared to the asdeposited CaO/CoO films. This volume contraction is due to the increase in density of the films after thermally induced phase transformation. This densification develops compressive stress in the films. As a consequence, the films are subjected through the formation of nanopores for releasing stress. Mica is likely more favorable than sapphire for such stress release because of weaker adhesion of the film with mica.

TRANSFERABILITY OF THE FILMS
The transferability of the nanoporous films was investigated by transferring the nanoporous film S mica : 11/9 on to polydimethylsiloxane (PDMS) platform. The different stages of the transfer process are illustrated in Figure 6. Initially, the mica substrate is isolated from the film following the steps as shown in Figure 6a−d. First, a glass slide is coated with a thin layer of wax. In the next step, the film is attached to the glass slide upside down, and then the thickness of the mica substrate is reduced to below 20 μm by isolating the mica layers from the back by mechanical force (Figure 6b). For further thickness reduction, thin layers of mica are repeatedly removed by adhesive tape as shown in Figure 6c. Figure 6d shows the back surface of the film after the complete removal of mica. After the removal of mica, no cracks in the film were observed by optical microscopy. After that, the back surface of the film is coated with a thin layer of PDMS following the step in Figure 6e. In the next step, the coated film was heated to 80°C for 3 h for the solidification of PDMS layer. The small area of the coated layer is isolated from the rest using a blade (Figure 6f). This is followed by heating to 150°C to melt the thin layer of wax between the glass slide and the PDMS layer. Then, the PDMS layer is isolated from glass slide as shown in Figure 6g. To dissolve the wax, the transferred film is immersed in acetone for 10 min. Figure 6h, I show the images of the film after transfer onto PDMS.
Several strategies, e.g., surface-energy-assisted transfer, 35 water penetration-assisted mechanical transfer, 36 film transfer by using ultrasonic water bath, 37 and carrier-polymer-assisted transfer, 38 have been demonstrated to transfer the 2D metal sulfide onto flexible polymer platforms. However, reports on transfer of thick films are less common, a notable exception being the work of Lu et al. on the transfer of thick films by etching of sacrificial water-soluble layers. 39 The present study is important as it demonstrates an alternative method for the damage free dry transfer of thick nanoporous films. Figure 7a shows the temperature-dependent electrical resistivity of all films from room temperature to 400°C. The roomtemperature electrical resistivity of the films S Al2O3 : 5.5/4.5, S Al2O3 : 11/9, S mica : 5.5/4.5, and S mica : 11/9 is measured to be ∼32, 13, 25, and 7 mΩcm, respectively. No significant variation of electrical resistivity with temperature is observed for all the films until 250°C, however above 250°C sharp increase in electrical resistivity is clearly visible in Figure 7a. This sharp increase in electrical resistivity is attributed to the release of oxygen from the film above 250°C, because the measurements are performed in vacuum. This is consistent with the observations on thin films reported elsewhere. 25,34,40,41 Despite the higher porosity of the film S mica : 11/9 than the S mica : 5.5/ 4.5, the former offers the lowest electrical resistivity throughout the temperature range measured. This indicates that the presence of nanopores in the film S mica : 11/9 does not hamper the transport of charge carriers. The scattering of charge carriers can be avoided in the nanoporous films if the characteristic length-scale of the porous structure is lower than the electronic mean free path, and this is supposed to be the case with the film S mica : 11/9. The room temperature value of electrical resistivity of the film S mica : 11/9 is as low as that is comparable to the values reported for solid thin films, 26−29 and lower than the values reported for bulk polycrystalline Ca 3 Co 4 O 9 . [20][21][22][23][24]42 The electrical resistivity of the film S mica : 5.5/4.5 is more than three times larger than that of the film S Al2O3 : 11/9, which is attributed to its higher porosity. With the increase in porosity, the characteristic length scale of the nanoporous structure in the film S mica : 5.5/4.5 might have been reduced below the electronic mean free path, resulting in enhanced scattering of charge carriers. That is, nanopores in the film S mica : 5.5/4.5 strongly scatter the charge carriers leading to the increase in electrical resistivity. Even with lower porosity,

ACS Applied Energy Materials
Article the electrical resistivity of the film S Al2O3 : 11/9 is nearly half that of the film S mica : 5.5/4.5; however, it is almost twice that of the film S mica : 11/9. This indicates that the quality of the films on mica substrates is better than that of the films on sapphire substrates. The highest electrical resistivity of the film S Al2O3 : 5.5/4.5 is due to the presence of disoriented grains in the film, which acts as scattering center for charge carriers. Figure 7b shows the temperature-dependent Seebeck coefficient of all the films from room temperature to 400°C. Near room temperature, the Seebeck coefficient of the films S Al2O3 : 5.5/4.5, S Al2O3 : 11/9, S mica : 5.5/4.5, and S mica : 11/9, is measured to be around 129, 115, 129, and133 μV/K, respectively. No significant variation in Seebeck coefficient with temperature of all the films is observed until 250°C; however, beyond this temperature it increases rapidly following the same manner as electrical resistivity. No considerable variation in Seebeck coefficient from film to film is observed, except a slightly lower value of the near-room-temperature Seebeck coefficient of the film S Al2O3 : 11/9. This shows that the porosity does not have significant effect on Seebeck coefficient of the films. Figure 7c shows the temperature-dependent power factor of all the films. Because of the lowest electrical resistivity and fairly good Seebeck coefficient the film S mica : 11/9 exhibits the highest power factor, above 2 × 10 −4 W m −1 K −2 in a wide temperature range (from room temperature to 350°C), and achieving the highest value 2.83 × 10 −4 W m −1 K −2 near 300°C . Although the values of power factor above 150°C are lower than the best reported values of power factor for undoped Ca 3 Co 4 O 9 thin films, 29,43 the room-temperature value (2.32 × 10 −4 W m −1 K −2 ) is comparable to previous reports on undoped Ca 3 Co 4 O 9 thin films, 28,44,45 and undoped bulk polycrystalline Ca 3 Co 4 O 9 . 22, 24,42 The power factor of the film S mica : 5.5/4.5 is almost three times lower than that of the film S mica : 11/9 throughout the temperature range measured, which is attributed to its higher porosity. A difference of the films grown on mica substrates than that of the films grown on the sapphire substrates is that the power factors in the former case are less temperature-dependent.
The above results show that the power factor of the films on mica substrates is different depending on the porosity of the films, in contrast to the films on sapphire substrates. This is because, with the increase in porosity the average distance between the pores decreases, resulting in a reduction in electronic mean free path. Because the pores in the film S mica : 5.5/4.5 are not of regular shape, the average separation of the pores cannot be readily estimated, but should be comparable to the electronic mean free path, resulting in a drastic increase in electrical resistivity. Furthermore, the pores in S mica : 5.5/4.5 seem to form a networklike structure, which restricts the passage of charge carriers, leading to the increased electrical resistivity. On the other hand, the interpore separation in the film S mica : 11/9 have a distribution in the range 50−500 nm, that is characteristic length scale of the nanoporous structure is higher than electronic mean free path, resulting in the reduced electrical resistivity and thus enhanced power factor of the film. The electronic mean free path in the most materials is less than 10 nm. 46 Recently, high power factor simultaneously with reduced thermal conductivity have been realized in thin films with ordered pores/holes; 2,3 however, there has been no report on the power factor of the films with disordered pores. The present work thus reveals that the scattering of charge carriers can be avoided in the nanoporous film with disordered pores by controlling the porosity, and thus a high power factor is possible. On the other hand, because of the presence of a large number of pores the thermal conductivity of the film is expected to be reduced. Because of the irregular shape and size and random distribution of nanopores, the direct evaluation of in-plane thermal conductivity of the nanoporous film is not possible. Recently, Kashiwagi et al. theoretically derived inplane thermal conductivity of the nanoporous Bi 0.4 Te 3 Sb 1.6 thin film from its measured cross-plane value by considering the

ACS Applied Energy Materials
Article cutoff mean free path to be equal to the average pore spacing. 30 The estimation of average pore spacing in nanoporous Ca 3 Co 4 O 9 films is challenging as due to the irregular shape and size and random distribution of nanopores. However, reduction in cross-plane thermal conductivity by 1 order of magnitude was realized by Song et al. in nanoporous Bi thin films with random nanopores. 47 The effect of porosity on thermal conductivity of bulk Ca 3 Co 4 O 9 is also investigated, 48,49 and thermal conductivity of 0.63 W m −1 K −1 at 373 K is reported by Bittner et al. for ∼32% porous Ca 3 Co 4 O 9 . 48 Note that the present Ca 3 Co 4 O 9 films are undoped, and yet a high power factor 2.32 × 10 −4 W m −1 K −2 near room temperature is obtained from the film S mica : 11/9. Further enhancement of the power factor is still possible by doping. 18,50−52 With this power factor combined with transferability, the nanoporous Ca 3 Co 4 O 9 films are candidates for near-room-temperature thermoelectric applications.

CONCLUSION
A sequential sputtering-annealing method, for the growth of nanoporous and transferable Ca 3 Co 4 O 9 films, has been demonstrated. The volume contraction caused by densification during the thermally induced phase transformation from sequential CaO/CoO film to the final phase of Ca 3 Co 4 O 9 promotes the formation of nanopores in the film. The porosity of the films is tunable by controlling the thickness of sequential CaO and CoO layers in the initial sputtered deposited films. A high power factor, above 2 × 10 −4 W m −1 K −2 in a wide temperature range (from room temperature to 350°C), is obtained from the nanoporus film on mica substrate. Because of the weak bonding of the film with the mica substrate and the presence of nanopillars, the film is easy transferable from the primary mica substrate onto polymer platforms. With this transferability and high power factor, the nanoprous Ca 3 Co 4 O 9 films can be a candidate for near-room-temperature thermoelectric applications. Additionally, the film growth method is suitable for upscaling.