Enhanced thermoelectric figure of merit in Bi₂Te₃–CNT–PEDOT nanocomposite by introducing conducting interfaces in Bi₂Te₃ nanostructures

In recent years, thermoelectric conversion has drawn great attention toward the generation of electricity from abundant waste heat resources.¹ In order to achieve higher performance of a thermoelectric material, the Seebeck coefficient (S) and electrical conductivity (σ) should be large and thermal conductivity (κ) should be small.² However, due to interdependence of S, σ, and κ, it is challenging to control them independently.³ Therefore, it is essential to decouple S, σ, and κ for achieving higher performance of a thermoelectric material.⁴ By increasing carrier mobility rather than carrier concentration, S and σ can be decoupled.⁵ The reduction in lattice thermal conductivity has been achieved through nanostructuring of various thermoelectric materials.⁶ Recently, reduction in lattice thermal conductivity is also achieved by introducing more conducting interfaces through nanocomposite engineering without less affecting the Seebeck coefficient and electrical conductivity.⁷ Bismuth telluride and its nanocomposites are efficient thermoelectric materials near room temperature and have attracted great attention for room temperature applications.⁸ Recently, lots of efforts have been made for the thermoelectric study of various Bi₂Te₃ nanostructures and nanocomposites to improve the figure of merit.^9,10 An enhanced figure of merit in Bi₂Te₃–P3HT, Cu–Bi₂Te_3, Bi₂Te₃–PEDOT:PSS, Te–Bi₂Te₃–PEDOT:PSS, Bi₂Te₃–PANI, Bi₂Te₃–graphene, and MoS₂–Bi₂Te₃,Bi₂Te_2.4Se_0.6–CNT nanocomposites has been reported.^11–17 Nanocomposites of conducting polymer/inorganic thermoelectric materials have also attracted attention for enhancing thermoelectric properties.^18–20 However, there is no report published on the thermoelectric properties of the Bi₂Te₃ ternary nanocomposite through the formation of interfaces to achieve an increased thermoelectric power factor and smaller thermal conductivity simultaneously.

In this work, sulfonated CNTs and PEDOT:PSS were used for the synthesis of the Bi₂Te₃–CNT–PEDOT nanocomposite which exhibits a much enhanced thermoelectric power factor and a higher figure of merit. Sulfonated CNTs were used to modify the PEDOT:PSS, and the insulating PSS chains could be removed. The Kelvin probe force microscopy has also been carried out to study the interface between Bi₂Te₃ nanostructures and CNT–PEDOT conducting channels.

The hydrothermal technique was used for the synthesis of Bi₂Te₃ nanostructures.²¹ The Bi₂Te₃–CNT was also synthesized by the hydrothermal technique in which 1% sulfonated CNT was added in the precursor for the synthesis of Bi₂Te_3. For the synthesis of Bi₂Te₃–CNT–PEDOT:PSS (ternary nanocomposite), the binary product (Bi₂Te₃–CNT) was dispersed into an aqueous solution of PEDOT:PSS (1 wt. %), sonicated for 1 h and dried at 50 °C. The schematics of the steps involved in the synthesis with details are described in the supplementary material (Fig. S1). We have used a bulk sample in the form of a pellet to carry out all measurements. The temperature dependent electrical conductivity (σ) was measured in the four probe configuration setup. The Seebeck coefficient was measured by a differential method at a temperature difference of ∼5 K. The TPS (transient plane source) method is used for the measurement of thermal conductivity. The details of the characterizations of Bi₂Te₃–CNT–PEDOT:PSS nanocomposites are presented in the supplementary material (Sec. S1).

XRD patterns of Bi₂Te₃ and Bi₂Te₃–CNT–PEDOT:PSS nanocomposites are shown in Fig. 1(a). All the samples exhibited characteristic peaks corresponding to single phase Bi₂Te₃ (JCPDS card no. 15-0863).²¹ However, no peaks corresponding to the CNT and PEDOT:PSS are observed in the XRD pattern of Bi₂Te₃–CNT and Bi₂Te₃–CNT–PEDOT:PSS nanocomposite samples which could be due to the small amount of CNTs and amorphous nature of PEDOT:PSS. In order to confirm the interaction of CNT and PEDOT:PSS with Bi₂Te₃ nanostructures, Raman spectra were studied for all the samples [Fig. 1(b)]. Raman peaks at 95 cm⁻¹, 111 cm⁻¹, and 135 cm⁻¹ correspond to E²_2g, A_1u, and A²_1g modes of single phase Bi₂Te₃, respectively.²² The Raman spectra of the binary sample show two peaks at 1354 cm⁻¹ and 1582 cm⁻¹, which correspond to D and G bands of the CNT, respectively.²³ The peaks observed in the ternary nanocomposite at 1445 cm⁻¹ and 1510 cm⁻¹ correspond to PEDOT:PSS.²⁴ There is a small shift in the peak position of the D band and G band of CNT in the ternary nanocomposite, compared with the binary nanocomposite which confirms the presence of interactions between the CNTs and PEDOT:PSS. There is no peak observed corresponding to PSS which confirms that insulating PSS is removed after addition of sulfonated CNTs in PEDOT:PSS.²⁵ The interaction of sulfonated CNTs with PEDOT:PSS in the polymer matrix leading to the phase separation between the PEDOT and PSS chains and the conjugated thiophene chains in PEDOT constructs the conducting interface with CNTs through the π–π intermolecular interaction.²⁵ 

The surface morphology of the Bi₂Te₃ sample is found in nanoflower type structures, and these nanoflowers are decorated on the CNTs in the Bi₂Te₃–CNT nanocomposite sample [Figs. 1(c) and 1(d)]. The SEM image of the Bi₂Te₃–CNTs–PEDOT:PSS nanocomposite confirms that all the CNTs or Bi₂Te₃ nanoflowers are wrapped by PEDOT:PSS conducting fillers. The TEM image of Bi₂Te₃–CNTs and Bi₂Te₃–CNTs–PEDOT:PSS nanocomposites is shown in the supplementary material, Figs. S2(a) and S2(b), respectively.

The increase in the electrical conductivity of Bi₂Te₃ is observed after the incorporation of CNTs and PEDOT:PSS [Fig. 2(a)]. The decorated Bi₂Te₃ nanostructures over the network of CNTs enhance the electrical conductivity of Bi₂Te₃ because the presence of CNTs allows the electrons to travel more efficiently. After the incorporation of PEDOT:PSS into the Bi₂Te₃–CNT nanocomposite, there is a further increase in electrical conductivity due to formation of compact conductive PEDOT networks. The degenerate semiconducting behavior is evident from the decrease in electrical conductivity for Bi₂Te₃, Bi₂Te₃–CNT, and Bi₂Te₃–CNT–PEDOT:PSS nanocomposite samples with the increase in temperature. The incorporation of CNTs into Bi₂Te₃ nanostructures increased its electrical conductivity from ∼494 S/cm to ∼651 S/cm and furthermore increased it to ∼990 S/cm after the addition of PEDOT:PSS. At room temperature, the obtained value of carrier concentration from the Hall effect measurements is 2.42 × 10¹⁹ cm⁻³, 2.77 × 10¹⁹ cm⁻³, and 2.82 × 10¹⁹ cm⁻³ for Bi₂Te₃, Bi₂Te₃–CNT, and Bi₂Te₃–CNT–PEDOT:PSS, respectively. The electrical conductivity (σ) is dependent on the carrier concentration (n) and mobility (μ_T) as σ = neµ_T.¹⁵ The estimated value of mobility (μ_T) at 300 K for Bi₂Te₃ is found to increase from ∼127 cm² V⁻¹ s⁻¹ to ∼149 cm² V⁻¹ s⁻¹ and further to 220 cm² V⁻¹ s⁻¹ after the addition of CNTs and PEDOT:PSS. The enhancement of the mobility is attributed to the presence of PEDOT between the Bi₂Te₃ and CNT, providing the conducting path at the interfaces. The formation of the conductive network is due to the interaction between CNTs and the conjugated thiophene chains in PEDOT:PSS.^25,26 This conductive network of the PEDOT polymer chain built among Bi₂Te₃ nanostructures and CNT networks, which form an intimate contact between tube-tube or polymer-tube junctions, resulted in the increase in the charge carrier mobility of the Bi₂Te₃–CNT–PEDOT:PSS nanocomposite. As the nature of Bi₂Te₃ is a degenerate semiconductor in all the samples, the carrier concentration can be assumed to be similar for the intrinsic excitation temperature range (300 K–450 K).¹⁵ The temperature dependency of mobility for all the samples has been calculated and is shown in Fig. 2(b). By considering the scattering events independently, the carrier mobility of the nanocomposite (μ_Total) can be expressed in terms of mobility at the interfacial barrier (μ_interface) and matrix sample (Bi₂Te₃) mobility (μ_matrix),²⁷ 

The mobility at interface is defined as²⁸ 

where L is the distance between two adjacent potential barriers, m^* is the effective mass, k_β is the Boltzmann constant, T is the absolute temperature, and E_B is the barrier height for charge carriers. From the plot of μ_inT^1/2 vs 1/T [Fig. 2(d)], the activation energy (barrier height) for Bi₂Te₃–PEDOT and Bi₂Te₃–CNT–PEDOT nanocomposites is obtained as 226.1 meV and 26 meV, respectively. The smaller value of activation energy for the Bi₂Te₃–CNT–PEDOT nanocomposite confirms that the electrons would feel less barrier height for traveling through the Bi₂Te₃–CNT–PEDOT nanocomposite which resulted in the enhanced interface mobility in the Bi₂Te₃–CNT–PEDOT:PSS nanocomposite after incorporation of PEDOT:PSS into the Bi₂Te₃–CNT nanocomposite matrix. The two transport regimes in the temperature dependent interfacial mobility of the Bi₂Te₃–CNTs–PEDOT:PSS nanocomposite are attributed to the conductivity of the Bi₂Te₃ dominating region and p-type PEDOT conductivity dominating region. The value of electrical conductivity increases with temperature in the p-type PEDOT semiconductor, whereas the electrical conductivity of degenerate Bi₂Te₃ decreases with the increase in temperature. Thus, the resultant electrical conductivity showed a decrease in the Bi₂Te₃–CNTs–PEDOT:PSS sample until ∼325 K and then increases. The increase in the electrical conductivity of Bi₂Te₃–CNTs–PEDOT:PSS at above 325 K is attributed to the dominating contribution coming from PEDOT in the higher temperature range.

In order to explore further the formation of the interface and potential barrier at the interface in Bi₂Te₃–CNT and Bi₂Te₃–CNT–PEDOT:PSS nanocomposites, the surface potential of all the samples has been studied using Kelvin probe force microscopy (Bruker Dimension Icon), and the results are shown in Figs. 3(b), 3(e), and 3(h). The color in surface potential images represents the variation of charge carrier density.²⁹ The surface potential map of the Bi₂Te₃ sample shows a single phase, whereas the surface potential of nanocomposite samples shows presence of additional phases. The presence of additional phases was analyzed using statistical distribution of the surface potential. The relative number of pixels acquired from the surface potential image using Nano-scope software was plotted, and the value of surface work function of the material was estimated using the work function of tip (Φ_tip ∼ 4.66 eV). The obtained single peak at ∼4.26 eV in Fig. 3(c) confirms the presence of Bi₂Te₃.³⁰ The additional peak at 4.41 eV along with a 4.26 eV peak in Fig. 3(f) confirms the presence of secondary phase (CNTs). Figure 3(i) shows two additional peaks at 4.37 eV and 4.41 eV which confirm the presence of two different phases, corresponding to the CNT and PEDOT:PSS, respectively.³¹ The reduction in the value of the work function of PEDOT is attributed to the combined attractive interaction between the van der Waals bond at the Te(1) site and the π-system of the PEDOT:PSS chain. The intermolecular interaction of the conjugated thiophene chains in PEDOT:PSS with sulfonated CNTs and the phase separation between PEDOT and PSS provide conducting paths which leads to an increase in carrier mobility and electrical conductivity.^31,32 After the incorporation of CNTs and PEDOT:PSS into the Bi₂Te₃ matrix, the enhancement in the work function of Bi₂Te₃ indicates the presence of the metallic phase in the nanocomposite. The interface barrier heights for Bi₂Te₃ and CNTs in Bi₂Te₃–CNTs nanocomposites are obtained as ∼150 meV and barrier heights for Bi₂Te₃/PEDOT and CNTs/PEDOT interfaces are ∼100 meV and ∼40 meV, respectively. Smaller values of interface barrier height indicate that the transport of electrons is preferred through CNTs/PEDOT interfaces in Bi₂Te₃–CNT–PEDOT:PSS nanocomposites.

Figure 4(a) shows the temperature dependence Seebeck coefficient of Bi₂Te₃, Bi₂Te₃–CNT, and Bi₂Te₃–CNT–PEDOT:PSS nanocomposites. The n-type nature of Bi₂Te₃ in all the samples is confirmed from the sign of the Seebeck coefficient. The Seebeck coefficient of a degenerate semiconductor material can be given as³³ 

where k_β is the Boltzmann constant, e is the charge of carrier, h is Plank’s constant, and m^* is the effective mass of the charge carrier. The Seebeck coefficient of Bi₂Te₃ is expected to decrease after incorporation of CNTs and subsequently PEDOT:PSS due to the increase in carrier concentration. Using Eq. (3), the value of the Seebeck coefficient is expected to decrease from −132 µV K⁻¹ to −119 µV K⁻¹ and −116 µV K⁻¹ which is close to the observed value −121 µV K⁻¹and −111 µV K⁻¹ for Bi₂Te₃–CNT and Bi₂Te₃–CNT–PEDOT:PSS, respectively.

The thermal conductivity of the samples was measured using the Hot Disc Thermal Constants Analyser (Hot Disc, Inc., Sweden). The obtained value of thermal conductivity for Bi₂Te₃, Bi₂Te₃–CNT, and Bi₂Te₃–CNT–PEDOT:PSS nanocomposites in the temperature range of 300–340 K is shown in Fig. 4(b). The decrease in thermal conductivity is observed after the incorporation of CNTs and PEDOT:PSS. The total thermal conductivity (κ) has contribution from electronic (κ_electronic) and lattice parts (κ_lattice).³⁴ The electronic thermal conductivity (κ_electronic) can be written as κ_electronic = σTL,³⁵ where σ is the electrical conductivity, L is the Lorentz number, and T is the absolute temperature. In the degenerate limit, the obtained value of κ_electronic for all the samples using the value¹⁵ of L (2.44 × 10⁻⁸ V² K⁻²) is shown in Fig. 4(c). The value of electronic thermal conductivity is higher for the Bi₂Te₃–CNT–PEDOT:PSS nanocomposite due to higher electrical conductivity. Using the value of the measured total thermal conductivity and estimated electric part of thermal conductivity (κ_electronic), the lattice thermal conductivity (κ_lattice) is obtained which is shown in Fig. 4(d). It is evident that the lattice thermal conductivity of the Bi₂Te₃–CNT–PEDOT:PSS nanocomposite is smaller than the Bi₂Te₃–CNTs. The addition of PEDOT:PSS into the Bi₂Te₃–CNTs nanocomposite provides additional interfaces which provide barrier to phonons due to enhanced scattering of phonons at interfaces and resulted in the decrease in thermal conductivity.¹⁵ 

Figures 4(e) and 4(f) show the results of thermoelectric power and figure of merit (ZT factor). The two fold increase in thermoelectric power factor and figure-of-merit observed for the Bi₂Te₃–CNT–PEDOT:PSS nanocomposite compared to Bi₂Te₃ nanoflowers is due to the enhancement in electrical conductivity and decrease in the lattice thermal conductivity. This clearly demonstrates that introduction of conducting interfaces resulted in decoupling of electrical conductivity and lattice thermal conductivity.

To conclude, we have synthesized the Bi₂Te₃–CNT–PEDOT:PSS nanocomposite and demonstrated a significant improvement in the figure of merit of the Bi₂Te₃–CNT nanocomposite after incorporation of PEDOT:PSS. The chemically treated CNTs not only act as a substrate during the growth of Bi₂Te₃ nanoflowers over the surface of them but also provide an efficient conducting path for electrons. The introduction of PEDOT:PSS in Bi₂Te₃–CNT further enhances electrical conductivity and interface mobility due to establishment of an intimate contact between Bi₂Te₃–CNT. The presence of conducting interfaces enhances electrical conductivity due to an increase in mobility and a decrease in lattice thermal conductivity due to enhanced scattering of phonons at interfaces. By taking the advantage of conducting interfaces, the transport of electrons and phonons has been controlled and an enhanced figure of merit is achieved.

See supplementary material for Secs. S1 and S2, Figs. S1 and S2, and Table S1.

We gratefully acknowledge MeitY (Government of India) for financial support. The use of different characterization facilities at NRF, IIT Delhi, is acknowledged. The authors thank Dr. Punit Kumar Dhawan and Professor Raja Ram Yadav for the thermal conductivity measurements.