Bi2Te3 and Sb2Te3 Thin Films with Enhanced Thermoelectric Properties for Flexible Thermal Sensors ^†

Abstract

The influence of substrate type in boosting thermoelectric properties of co-evaporated Bi2Te3 and Sb2Te3 films (with 400 nm-thick) is here reported. Optimized power factor values are 2.7 × 10⁻³ W K⁻² m⁻¹ and 1.4 × 10⁻³ W K⁻² m⁻¹ for flexible Bi2Te3 and Sb2Te3 films, respectively. This is an important result as it is at least 2 times higher than the power factor found in the literature for flexible Bi2Te3 and Sb2Te3 films. A flexible infrared thermopile sensor was developed with high detectivity (2.50 × 10⁷ cm √HzW⁻¹).

1. Introduction

Thermoelectric (TE) technology appears as a green energy source due to their direct transformation of small thermal gradients into electric power, and vice-versa, in a renewable manner and without CO₂ emissions. Telluride alloys have attracted a great deal of interest because of their potential applications as TE generators, TE coolers and thermal sensors [1,2]. The performance of TE devices is highly dependent on TE properties of constituent materials, namely the Seebeck coefficient (𝑆), the electrical conductivity(𝜎) and the thermal conductivity (𝑘), according to the figure of merit (𝑍𝑇), 𝑍𝑇 = 𝑆²𝜎/𝑘, where 𝑆²𝜎 is related to the power factor, 𝑃𝐹. High TE device performance require high 𝑍𝑇 and 𝑃𝐹 values. To get high sensitive thermal sensors, TE materials with high 𝑆 values are required. Thin films instead of conventional bulk materials opens the possibility to reduce substantially the material amount required, and enable the deposition on flexible substrates, making the devices compatible with modern integrated circuit technology.

A pioneering work that reports the effect of both substrate thickness and type on the TE properties of Sb₂Te₃ and Bi₂Te₃ films, is here discussed. A simple and flexible p-n TE device have been fabricated and tested for pyrometry applications.

2. Material and Methods

Sb₂Te₃ and Bi₂Te₃ films (with 400 nm-thick) were fabricated by co-evaporation method in a high-vacuum chamber (10⁻⁶ mbar). Bi (or Sb) evaporation rate was 2 Å/s and Te evaporation rate was 6 Å/s. Borosilicate glass and Kapton^® substrates (with 50- and 25 µm—thick) were used for films deposition. Substrates were heated at 270 °C and 230 °C, for the deposition of Sb₂Te₃ and Bi₂Te₃ films, respectively. The morphology and chemical composition of the films were obtained by scanning electron microscopy (SEM) and Energy-Dispersive X-ray spectroscopy (EDX). The crystallographic structure was obtained by X-ray diffraction (XRD). In-plane electrical conductivity was measured at room temperature (RT) using the conventional 4-Probe van der Pauw geometry. Seebeck coefficients were obtained by connecting one side of the film to a heated metal block at a fixed temperature and the other side to a heatsink at room temperature, to have a temperature gradient (of a few kelvin) along the film. A flexible thermopile infrared (IR) detector based on the p-n junction (5 × 5 mm²), with a conductive carbon paint as absorber, was developed. The radiation was emitted by a black body target object (15 cm diameter) from 293 K to 513 K of temperature and at a distance of 3 cm. The thermo-voltage generated by the device was monitored by a multimeter.

3. Results and Discussion

The thermoelectric voltage generated by the telluride alloys shows a linear dependence of temperature gradient (Figure 1a,b). Maximum 𝑆 values of—224 µV K⁻¹ (for Sb₂Te₃ film) and 184 µV K⁻¹ (for Sb₂Te₃ film) deposited on 25 µm—thick Kapton^® substrates were obtained, respectively. The negative (positive) slope for Sb₂Te₃ (Sb₂Te₃) films is consistent with n-type (p-type) semiconductor behavior. The surface morphology and chemical composition of the films deposited on the three substrates are also shown on Figure 1.

Stoichiometric Bi₂Te₃ films (At%(Te/Bi) ≈ 1.5) show a dense and grain morphology with large flakes (Figure 1a). However, the film deposited on 25 µm—Kapton^® shows a more compact structure with more rounded domains. Sb₂Te₃ films show a similar surface morphology except for the film deposited on 50 µm—thick Kapton^® that appears more flat and dense, with smaller crystals. While the films deposited on Kapton^® substrates are stoichiometric Sb₂Te₃, the film deposited on glass shows a At%(Te) higher than the others films (~1.8). Highest 𝜎 values are obtained for Bi₂Te₃film deposited on 25 µm—Kapton^® (≈5.3 × 10⁴ (Ω m)⁻¹) and for the Bi₂Te₃ film deposited on 50 µm—Kapton^® (≈ 4.6 × 10⁴ (Ω m)⁻¹), resulting in the highest 𝑃𝐹 values of 2.7 × 10⁻³ W K-2 m⁻¹ (Bi₂Te₃) and 1.4 × 10⁻³ W K⁻² m⁻¹ (Sb₂Te₃) (

Table 1. Comparison of TE properties of the Bi₂Te₃ and Sb₂Te₃ in the present study with other reported by other authors (with different techniques).

Film	Thickness (µm)	Substrate	𝑺 (µVK−1)	𝝈RT (Ω m)−1	𝑷𝑭(× 10−3 W K−2 m−1)
Bi2Te3/Sb2Te3	0.4	glass/polyimide	−224/176	5.3 × 104/4.6 × 104	2.7/1.4
Sb2Te3 [3]	-	polyimide	135	2.8 × 104	0.5
Bi2Te3 [4]	1.3	polyimide	119	7.0 × 104	1 (Bi2Te3)
Sb2Te3/Bi1.8Te3.2 [5]	70	Glass-textile/polyimide	−138/120	1.0 × 104/1.0 × 104	0.3/0.2

). 𝜎 values of Bi₂Te₃ films deposited on glass and 50 µm—thick Kapton^® substrates are 1.2 × 10⁴ (Ω m)-1 and 2.8 × 10⁴ (Ω m)⁻¹, respectively. For Sb2Te3 films, 𝜎 values of 3.0 × 10⁴ (Ω m)⁻¹ (for glass) and 3.2 × 10⁴ (Ω m)⁻¹ (for 25 µm - Kapton^®) are obtained. The films with highest 𝑃𝐹 values revealed a more compact and dense structure (so, more connected grains). Disoriented grains create a higher electrical resistivity due to a more open structure. XRD spectrum of the Bi₂Te₃ film with the highest 𝑃𝐹 value (Figure 2a) shows a preferential growth along the (0 0 15) plane of the Bi₂Te₃ crystals. Similar XRD patterns of Sb₂Te₃ films (Figure 2b) are observed.

A flexible infrared thermopile sensor based on a single p-n junction has been tested as a proof of concept. The 25 µm Kapton^® provides flexibility and lightness to the devices, while it offers high upper working temperature (until 350 °C) and low thermal conductivity (0.12 W m⁻¹ K⁻¹). A metal structure was constructed to maintain the cold junctions at RT (heat sink) and to allow only the region of the hot junction to receive the radiation (through a hole in the metal structure), as shown in the Figure 3a.

An example of the thermal test is given in the Figure 3b. Figure 3c shows that the experimental data are fitted with a 4-order polynomial, according to the Equation [6] shown in the inset of Figure 3c. 𝐾_𝑠 is the instrument factor of the sensor (𝐾_𝑠 = 5.50 × 10⁻¹⁴), 𝜀_{𝑡𝑎𝑟𝑔} is the emissivity of the target disk (𝜀_{𝑡𝑎𝑟𝑔} = 0.9), T_targ is the temperature of the disk and T_sens is the temperature of the sensor. 𝜀_{𝑠𝑒𝑛𝑠} = 1 considering the internal reflections inside the thermopile case, and therefore, the whole structure behaves as a cavity blackbody with high emissivity [6]. The responsivity (R_s) is obtained by the slope of the output voltage as a function of absorbed heat radiation power [2], and is 0.19 V W⁻¹ for our sensor. Specific detectivity (D*) can be calculated by [2]

$$D*=\frac{\sqrt{A_{abs}}\sqrt{\Delta f}}{NEP}$$

(1)

where A is area of sensor (25 mm²) and NEP is the noise equivalent power which is the quotient of RMS of the voltage (𝑣_𝑛) and R_s [2]. A 𝑣_𝑛 = 3.9 nV was obtained taking in account the 𝛥𝑓 = 1 Hz, 𝑇 = 298 K and 𝑅 = 910 Ω. NEP value was estimated to be 20 nW. A 𝐷* value of 2.50 × 10⁷ cm √Hz W⁻¹ was obtained.

4. Conclusions

The developed telluride films have the highest 𝑃𝐹 values among the results regarding flexible films with thickness below 1 µm, reported up to now. Their high S values, together with a flexible structure make them suitable to be applied as flexible thermal sensors, as proven here.