Superlattice deposition and thermoelectric performance

HZO/PEODT superlattice, there was still one remaining issue to address, the

determination of “super ZnO”, which possessed the best electrical performance as stated before, to integrate with inserted PEDOT. Because of the slight variation in deposition condition between our ALD and MLD systems, e.g. working pressure and flow field of precursors, the “super ZnO” had to be determined once again. Therefore, we followed the approaches stated in section 3.2.1, tuning the insertion periodicity of mixed doping layers to acquire the best electrical performance, and then concluded that mix 19:1 HZO was the “super ZnO” in MLD system based on the highest power factor listed in Table 4.3. Thus, mix 19:1 HZO was a host material in the section to study the effects of composition and structure on the thermoelectric performance of HZO/PEDOT

superlattice. However, it is noteworthy that the electrical mobility of HZO deposited by MLD system was higher than that deposited by ALD system with the same insertion periodicity of mixed doping layers. This property was attributed to the higher

preferential-growth in (002) direction as XRD patterns revealed in Figure 4.8. The high preferential-growth in a conductive film yielded fast transportation paths for electrical carriers, and thus the electrical mobility of the film could be much higher than that with lower preferential-growth.

Next, we first clarified the effect of insertion periodicity of PEDOT on the thermoelectric performance. Table 4.4 summarizes the thermoelectric properties of HZO/PEDOT superlattices composed of alternating mix 19:1 HZO host and 1-cycle PEDOT, where the mix 19:1 HZO were of 2 periods (19 ZnO cycles/1 mixed doping cycle/19 ZnO cycles), 4 periods (19/1/19/1/19/1/19), 6 periods

(19/1/19/1/19/1/19/1/19/1/19), or 8 periods (19/1/19/1/19/1/19/1/19/1/19/1/19/1/19); the total number of cycles of HZO in superlattice with 4 kinds of period lengths were 468, 474, 476, and 477, respectively. It is noteworthy that these superlattices were deposited without type-C modification, i.e. the dose of H2O2 before 5-cycle ZnO nucleation layers on the surface of inserted PEDOT. Although the thermoelectric performance was not optimized currently, the effects of insertion periodicity were still evident. At long insertion periodicity, i.e. 6 and 8 periods, the electrical mobility and thermal

conductivity were close to the value of mix 19:1 HZO, since the distances between two inserted PEDOT layers (about 25 and 33 nm for 6 periods and 8 periods, respectively) were longer than the distance between scattering centers for electrical carriers and phonons in mix 19:1 HZO, such as dislocations, impurities, and grain boundaries especially. Based on the (002) XRD peak of mix 19:1 HZO presented in Figure 4.9, the grain size was ~18 nm, which was calculated by Scherrer equation,

D=0.9λ/βcosθ, where D is mean grain size, λ is the wavelength of incident X-ray=0.154

nm, β is the FWHM (Full width at half maximum), and θ is the Bragg angle. Thus, the transportation of electrical carriers and phonons were almost maintained, and the thermoelectric performance was unaffected. On the other hand, the period length of inserted PEDOT layers in the superlattice composed of 2 periods of mix 19:1 and 1-cycle PEDOT was much shorter than the distance between scattering centers of mix 19:1, and thus the inserted PEDOT layers became the major scattering centers for both carriers (heat and electric). Besides, short distance between two adjacent PEDOT layers confined the growth of ZnO crystals; therefore, the transportation of electrical carriers and phonons were retarded extremely even the nucleation and growth of ZnO on PEDOT were improved through interface-engineering. Luckily, the concurrent suppression of electrical mobility and thermal conductivity could be decoupled at medium insertion periodicity of PEDOT layers, i.e. 4 periods in our results. Since the distance between adjacent PEDOT layers in this structure was around 16 nm, which was slightly smaller than the grain size of mix 19:1 HZO, and therefore the suppression of grain growth was retarded. Furthermore, the conductive nature of PEDOT offers

conducting paths for electrical carriers; thus, the electrical mobility could be maintained partially. As for thermal conductivity, the inserted PEDOT layers that were the main scattering centers for phonons in this structure could reduce the thermal conductivity obviously due to the extremely heterogeneous interface between ZnO and PEDOT.

Overall, the thermoelectric performance of superlattice composed of alternating 4-period mix 19:1 HZO host and 1-cycle PEDOT had a potential to be further improved through thickening the inserted PEDOT even though the ZT of this superlattice was not the highest one among four superlattices on this front.

Further, the inserted PEDOT was thickened gradually and type-C modification, i.e. the dose of H2O2 before 5-cycle ZnO nucleation layers, was applied to optimize the performance of superlattices. The thermoelectric properties of deposited superlattices are listed in Table 4.5. The enhancement of electrical mobility (conductivity also) through the usage of type-c modification, which improved the growth and/or nucleation of ZnO, was stated in section 4.2.1. Besides, the slight increment of thermal

conductivity was also attributed to modified interfaces between ZnO and PEDOT, by which the scattering of phonons was reduced. Regarding the dependence of ZT

enhancement on the thickness of inserted PEDOT, we started from the reduced thermal conductivity. It could be seen that the thermal conductivity decreased with the cycle (thickness) of inserted PEDOT. This result was originated from the formation of complete and clear interfaces between ZnO and PEDOT gradually as the cycle of inserted PEDOT increased, hence the interface-scattering of phonons was enhanced due to the significant discrepancy in the type of bonds and the microstructure between ZnO and PEDOT, i.e. covalent bonds and disordered amorphousness of PEDOT versus ionic

bonds and ordered crystallinity of ZnO. The gradually-formed interfaces could be observed and confirmed by the comparison of cross-sectional TEM images presented in Figure 4.10. Moreover, the inserted PEDOT layers also offered a considerable serial thermal resistance due to its ultra-low thermal conductivity, 0.19 W m^-1 K^-1. Thus, the reduced thermal conductivity was obtained. However, the decrement of thermal conductivity was unsatisfactory. Based on the simple series-resistance model, which is illustrated in Figure 4.11, the theoretical thermal conductivity could be expressed as below. It is noteworthy that there were assumptions we made to simplify the calculation:

(1) Cross-sectional area was unity; (2) Thermal conductance G at interfaces were the same. (GHZOPEDOT=GPEDOTHZO=G=0.2 GW m^-2 K^-1, which is based on the result of Mark D. Losego et al.¹⁰⁶); (3) The transportation of phonons inside HZO was

unchanged even with the insertion of PEDOT.

𝑹𝑹_{𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕𝒕}= 𝒅𝒅𝒇𝒇𝒇𝒇𝒕𝒕𝒇𝒇

Therefore, the estimated thermal conductivity of each superlattice should be 1.49, 1.31, 1.15 and 0.95 for 1-cycle, 2-cycle, 4-cycle and 6-cycle, respectively. The

estimated thermal conductivity was much lower than the measured value in all

superlattices. This result indicated the lack of distinct interface and the existence of fast-transportation route for phonons in the inserted PEDOT layers. In fact, it indeed took time (or cycles) for PEDOT to form a complete layer, hence the interface

(HZOPEDOT) might not form under low-cycle condition. Besides, the molecular scale ZnO may indeed existed inside the inserted PEDOT layers due to the usage of nucleation process, which could improve the nucleation on ZnO on the surface of PEDOT. In nucleation process, the sample was soaked in the vapor of each precursor, hence precursors had enough time to infiltrate into the superficial PEDOT, which was fluffy in structure, and then physisorbed onto it. Therefore, the molecular scale ZnO could be formed inside PEDOT during nucleation process and then diminished another distinct interfaces (PEDOTHZO) even the deposition-cycle of inserted PEDOT increased. To evaluate the effect of inserted PEDOT on the suppression of thermal conductivity exactly, the intrinsic thermal conductance of PEDOT and the conductance of each interface were lumped as a single conductance G’, thus the equation was modified as: for 1-cycle, 2-cycle, 4-cycle and 6-cycle PEDOT, respectively. Interestingly, our results of 1-cycle and 2-cycle PEDOT were in line with the results of Ashutosh Giri et al.¹⁰⁷, and the estimated the interface thermal conductance in ZnO/hydroquinone/ZnO structure is shown in Figure 4.12. In their result, they stated that the hydroquinone molecules were most probably attached to every other surface Zn site (50% surface

coverage) through first principles study, thus the distinct interfaces were not formed in their case. This might also interpreted the lack of complete and distinct interface under low-cycle condition as we supposed. On the other hand, in the results of 4-cycle and 6-cycle PEDOT, the estimated interface thermal conductance was lower than results of Ashutosh Giri et al. It suggested the evolution of forming distinct interfaces since our results were still higher than the half of 0.2 GW m^-2 K^-1 (i.e. 0.1 GW m^-2 K^-1 due to two interfaces in perfect layer structure). Nonetheless, the exact interface thermal

conductance, G, and the thickness of pure PEDOT layers (or effective thickness) still couldn’t be determined on this front; hence, more detailed analyses and structures had to be tested to clarify the effect and mechanism of inserted PEDOT on suppressing the thermal conductivity of superlattices.

As for electrical properties, the carrier concentration slightly increased in the beginning but decreased as the cycle/thickness of PEDOT increases. This may resulted from the existence of Cl^- , which can serve as an n-type dopant for ZnO as F^- did in F-doped ZnO, inside PEDOT. However, as the cycle/thickness of PEDOT increased, the degree of carrier compensation, which occurred at the interfaces due to the p-type conducting nature of PEDOT and n-type of ZnO (or HZO), became non- negligible.

Thus, the carrier concentration of superlattice decreased with the thickness of inserted PEDOT layers but still remained n-type owing to great thickness (or volume) difference

between HZO host layers and PEDOT insertion layers. Regarding the electrical

mobility, the electrical barrier and the depletion region at interfaces, which were due to carrier compensation, blocked the transportation of carriers, hence the electrical

mobility decreased monotonically with the cycle/thickness of PEDOT. Overall, thickening the inserted PEDOT deteriorated electrical conductivity of the superlattice.

Referring to the absolute value of Seebeck coefficient, it decreased initially but increased as the cycle/thickness of inserted PEDOT increased. The decrement of the absolute value of Seebeck coefficient might be originated from the serial/parallel contribution provided by PEDOT, whose Seebeck coefficient was positive, therefore it compensated the negative Seebeck coefficient of HZO. As the cycle/thickness of inserted PEDOT increased, the electrical barriers became more established, such as the energy band of PEDOT and depletion regions, thus the inserted PEDOT could enhance the Seebeck coefficient through energy filtering effect. As the energy band alignment of PEDOT and ZnO presented in Figure 4.13, the conduction band edge of PEDOT was about 1 eV higher than that of ZnO and therefore energy filtering effect easily took place at the artificial interfaces in our superlattices. Besides, the carrier compensation also provided another route to enhance the absolute value of Seebeck coefficient since it distorted the energy band and resulted in energy barriers and depletion regions at the interfaces. Nonetheless, the improvement in the absolute value of Seebeck coefficient

still couldn’t overcome the suppression of electrical conductivity, and therefore the power factor decreased with the cycle/thickness of inserted PEDOT even though the power of Seebeck coefficient was higher than that of conductivity in the calculation of the power factor. Hence, although the thermal conductivity could be further decreased through increasing the cycle/thickness of inserted PEDOT, the deteriorated power factor might compensate the merit. Overall, the ZT of superlattice could be enhanced by a factor of 7 through the insertion of 6-cycle PEDOT into mix 19:1 HZO host-layer compared with undoped ZnO deposited by MLD system.

Table 4.3: The electrical properties of ZnO and mixed HZO with varied insertion periodicity of mixed doping layers.

n

Table 4.4: The thermoelectric characteristic of the HZO/PEDOT superlattice films with varied periodicities of mix 19:1 HZO host layers.

n

Table 4.5: The thermoelectric characteristic of the HZO/PEDOT superlattice films composed of alternating 4 periods of mix 19:1 HZO and varied cycles of PEDOT.

n

Figure 4.8: The XRD patterns of ZnO deposited by ALD and MLD systems.

Figure 4.9: The mean grain size calculation of mix 19:1 HZO by Scherrer equation.

Figure 4.10: The cross-sectional TEM images of HZO/PEDOT superlattices with (a) 1-cycle; (b) 6-cycle inserted PEDOT.

Figure 4.11: The schematic illustration of superlattice-structure and series-resistance model in our calculation.

Figure 4.12: The estimated interface thermal conductance in ZnO/hydroquinone/ZnO superlattice done by Ashutosh Giri et al.

Figure 4.13: The schematic illustration of energy band alignment of PEDOT and ZnO.