Supplementary Information for Preferential Scattering by Interfacial. Charged Defects for Enhanced Thermoelectric Performance in Few-layered

Transcription

1 Supplementary Information for Preferential Scattering by Interfacial Charged Defects for Enhanced Thermoelectric Performance in Few-layered n-type Bi 2 Te 3 Pooja Puneet, 1 Ramakrishna Podila, 1,2 Mehmet Karakaya, 1 Song Zhu, 1 Jian He, 1,* Terry M. Tritt, 1 Mildred S. Dresselhaus, 3 and Apparao M. Rao 1,2,* Affiliations: 1. Department of Physics and Astronomy, Clemson University, Clemson, South Carolina, SC USA 2. Center for Optical Materials Science and Engineering Technologies, Clemson University, Clemson, SC USA 3. Department of Physics and Electrical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA Correspondence to: arao@g.clemson.edu and jianhe@clemson.edu X-ray diffraction- All samples used in this study exhibited a x-ray diffraction (XRD) pattern (Fig. S1) that was consistent with single phase of Bi 2 Te 2.7 Se 0.3 (JCPDS card no ). Pristine Bi 2 Te 3 exhibits!3! symmetry but it is often more convenient to view the structure as a hexagonal crystallographic unit cell. The crystallographic unit cell of Bi 2 Te 3 is shown in Fig. S2 with the basal plane and c-axis directions. As expected, we found that the full width at half maximum intensity (FWHM) (!") (0,0,6) peak broadened due to the formation of few-layer nanosheets. The FWHM was obtained after subtracting the instrumental broadening. We estimated the coherence length (δl) for our samples along the in-plane [00l] and cross-plane [hk0] directions from the XRD data using Equation 1: 1

2 !" =!!!", (Eq. 1)!h!"!!" =!!!!"#$.!". As shown in Table S1, the average values for δl showed a significant increase for the exf8h-sps sample along the (0,0,6) direction while!" did not change significantly along the (1,1,0) direction. Figure S1: (a) Powder x-ray diffraction patterns of the commercial ingot, exf8h and exf8h-sps samples (see text). In panels (b) and (c) the (0,0,6) peak is suppressed in exf8h samples due to exfoliation, whereas more pronounced (0,0,6) and (1,1,0) peaks are evident in the exf8h-sps sample indicating the improvement in coherence length. Table S1: FWHM and coherence length calculated from powder XRD data. 2

3 3

4 Figure S2: Hexagonal crystallographic unit cell of Bi 2 Te 3. The indices represent the bonding environment of Te atoms. Te 1 represents van der Waal bonding whereas Te 2 represents ioniccovalent type of bonding with Bi atoms. 4

5 Thermal conductivity- The thermal conductivity (κ) measurements are very critical in the determination of ZT. It is important to measure κ in the same direction as thermopower (α) and resistivity (ρ) in order to obtain correct ZT values. Particularly, in case of Bi 2 Te 3, the high anisotropy can result in miscalculated ZT values. As shown in Fig. S3, measuring κ along the SPS direction (open squares) can result in a higher but inaccurate ZT. To ensure the uniformity in measurements, we adhered to a protocol we developed previously Ref. 21 for measuring κ in the same direction as ρ and α. Figure S3: Anisotropy in the transport measurements along SPS direction (open symbols) and normal to SPS direction (solid symbols) directions is shown in (a). Electrical transport properties are measured in the direction normal to SPS direction and high temperature κ is measured in SPS direction of the pellet; the effect of anisotropy on the κ Total (b) and ZT values (c) of exf8h samples is shown. 5

6 Figure S4: Separation of various contributing thermal conductivities: lattice (κ L ), electronic (κ e ), and bipolar (κ B ) to the total thermal conductivity (κ T ) are shown for the commercial n-type Bi 2 Te 3 ingot. The solid symbols represent data for total thermal conductivity (κ T ) and the open symbols represent (κ T- κ e ). Table S2: Contributions from lattice, electronic and bipolar thermal conductivities to the total thermal conductivity κ T at 300 K and 450 K. 6

7 Figure S5: The bipolar thermal conductivity (κ B ) was estimated from κ T as explained earlier in the results and discussions section. The temperature dependence of bipolar thermal conductivity (κ B ) shows a shift in the onset for bipolar contributions to higher temperature for the exf8h-sps samples. The contribution of κ B below the onset T is close to zero. A corresponding shift in the ZT peak position confirmed that the shift in bipolar effects resulted in higher ZT values at high temperatures. 7

8 Effects of exfoliation time- To elucidate the effects of exfoliation time, we exfoliated bulk n- type Bi 2 Te 3 for 3, 5, and 8h. As shown in Fig. S6, we observed a weak influence of exfoliation time on the TE properties of CE-SPS processed n-type Bi 2 Te 3. Furthermore, AFM measurements (Fig. S7) showed that the thickness of exfoliated samples does not vary for exfoliations times greater than 3h, consistent the trends observed in TE properties. Figure S6: The effect of exfoliation time (followed by a SPS treatment) on the transport properties: α (a), ρ (b) and the power factor α 2 T/ρ (c) as a function of temperature. 8

9 Figure S7: The average thickness of exfoliated n-type Bi 2 Te 3 decreases rapidly with increasing exfoliation time (from 0 to 3 h, and 3 to 5 h) and eventually plateaus at ~50 nm when exfoliation time exceeds 8 h. The inset shows the change in the intensity of the 760 cm -1 Raman peak (cf. Fig. 5) as a function of exfoliation time. It is evident that the saturation in the Raman peak intensity correlates with the observed lack of change in the layer thickness at extended exfoliation times. 9