Thermoelectric and Magneto-Transport Properties of Ge2Sb2Te5

1. Introduction

Chalcogenide-based materials have attracted extensive attention over a century due to their unique properties with potential for use in a variety of applications like solar cells [1], photocatalysis [2], thermoelectric devices [3], optoelectronics [4], and topological insulation [5]. These properties largely stem from the tunable chemical environment, allow for change of electronic and optical properties without drastic changes in chemical composition. A representative class of the chalcogenide is that undergoes crystalline-amorphous phase change, accompanied by the changes in bond lengths, bond angles, and coordination numbers [6]. These phase change materials (PCMs) can serve as switches between two distinct phases [7].

Among PCMs, Ge₂Sb₂Te₅ (GST225) has been patented [8]. However, its structural and physical properties have never been reported. In this article, we report experimental investigation of bulk GST225 including the crystalline structure (through x-ray diffraction and Raman spectroscopy), electrical resistivity ρ(T) (with applied magnetic field up to 35 Tesla), thermal conductivity κ(T), Seebeck coefficient S(T), and magnetic susceptibility in a wide temperature range (1.8 K and 400 K). Our results suggest that the hexagonally structured GST225 exhibits interesting thermoelectric and magneto transport properties.

2. Structural Characterization

The polycrystalline bulk samples were purchased from a commercial vender (Columbia International). Figure 1(a) shows the crystalline structure of Ge₂Sb₂Te₅ at room temperature, based on the x-ray diffraction pattern (XRD) presented in Figure 1(b). The major peaks in this XRD are in alignment with published reports [9,10] for the hexagonal crystalline phase and are indexed in the figure. Based on this, we obtain the lattice parameters a = 4.22Å, and c = 17.0Å.

As shown in Figure 1(c), the room-temperature scanning electron microscope (SEM) image shows (1) crystallinity and (2) a layered structure, consistent with that illustrated in Figure 1(a). According to energy dispersive x-ray (EDX) spectrum shown in Figure 1(e), major peaks are from Ge, Sb, and Te, with a small amounts of trace elements like O and Al. The atomic ratio Ge : Sb : Te are 2:2:5. Consistent result is also obtained from the Raman spectrum of GST225 shown in Figure 1(f). The peaks at ~124 cm^-1, ~150 cm^-1, ~160 cm^-1, and 300 cm^-1 are associated with the Ge-Te, Sb-Te_, Sb-Sb, and Ge-Ge vibration modes, respectively, and the peak at 441.5 cm-¹ with the modes of GeO₂ [11,12,13,14,15].

3. Magnetic Susceptibility

Given that our GST225 is crystallized with the hexagonal structure at room temperature, it is not clear if there is any abnormal behavior above the room temperature. Figure 1(d) shows the temperature dependence of the magnetic susceptibility χ(T) between 350 K and 800 K, measured by a vibrating sample magnetometer in a magnetic field of 3 Tesla. The small diamagnetic susceptibility reflects the dominant core contribution diamagnetism over the entire measured temperature range. There is a noticeable jump in χ(T) at T_x ~ 575 K. Also, above T_x, the magnitude of χ(T) decreases with increasing temperature. While the origin is presently unknown, such a change in magnetic susceptibility is likely due to a change in the electronic structure.

4. Electrical Resistivity and Magneto Transport

Figure 2(a) shows the temperature dependence of the electrical resistivity ρ(T) between 1.8 K to 400 K. With increasing temperature, ρ(T) smoothly increases with the positive slope dρ/dT without any anomaly. The residual resistivity ratio (RRR) defined as ρ(300 K)/ρ(1.8 K) ~ 1.2. This indicates that there is no phase transition below 400 K, implying the possible hexagonal-cubic-amorphous transitions are absent in our samples.

To quantitatively analyze the resistivity behavior, we plot the low temperature ρ(T) in Figure 2(b). The observed T² dependence of ρ(T) implies that the low-temperature electrical transport is dominant with electron-electron scattering. As shown in Figure 2(b), the data can be well described by ρ(T) = ρ₀ + AT². Here, ρ₀ is the residual resistivity and A is a constant. The green line in Figure 2(b) represents the best fit with ρ₀ = 1.2068ⅹ10^-5Ω•m and A = 9.036x10^-11Ω•m•k^-2. Above ~50 K, ρ(T) gradually departs from the T² behavior but retains power-law temperature dependence. As demonstrated in Figure 2(c), ρ(T) plotted as a function of T^1.22 is linear between 50 K and 400 K. Such power-law dependence with 1 < n < 2 suggests that the electrical transport is mixed with both electron-electron and electron-phonon scattering, the latter is expected to give linear temperature dependence in this temperature range.

Through the above analysis, the electrical transport seems conventional with both electron-electron and electron-phonon scattering. In such a scenario, the magnetoresistance (MR), $∆R/R_0$ = $\left(R\left(B\right)-R_0\right)/R_0$ is usually small. According to the band theory for a single-band solid at the Fermi energy, MR is expected to depend quadratically on H at low fields and saturate at high fields. For systems with multiple bands at the Fermi energy involving two types of charge carriers, the magnetoresistance can be described by $∆R/R_0$ = αB²/ (β + γB²) (where α, β, and γ are positive constants). In either case, a linear-$\sqrt{B}$- dependent $∆R/R_0$ is unexpected. We recall that $\sqrt{B}$- dependence of $∆R/R_0$ was observed in single-layer graphene [16]. Theoretical explanation is that the $\sqrt{B}$ behavior can occur when a material with a linear dispersion such as graphene and short-range disorder (i.e., the electron wavelength is much longer than the correlation radius) [17,18]. From this point of view, GST225 may contain linear dispersive bands. Due to the polycrystalline nature, short-range disorder surely exists. Recently, Leahy et al. have provided [19] a refined theoretical analysis including Kohler’s rule of experimental magnetoresistance data. For similar levels of refinement, both band structure calculations and single-crystal samples of GST225 will be needed.

We measured the magnetoresistance in magnetic fields (B) up to 35 Tesla. As B increases $∆R/R_0$ is observed to increase up to nearly 50%. Below 20K, the temperature and field dependence of $∆R/R_0$ scales universally as a function of $B/R_0$ following Kohler rule. The Kohler plot T = 0.52K, 1.8K, 5.0 K, 10 K, 20 K, 30 K, and 40 K is shown in Figure 2(d). Several features are worth mentioning: (1) $∆R/R_0$ is positive at all measured temperatures, (2) The field dependence of $∆R/R_0$ is similar between 5 K and 40 K, and (3) $∆R/R_0$ exhibits linear dependence with $\sqrt{B}/R_0$ over a wide field range without the sign of saturation.

5. Thermal

Figure 3(a) shows the temperature dependence of the thermal conductivity (κ), which was measured using the four-probe method without radiation correction. At T = 300 K, κ ~ 3.0 W/K-m, close to that obtained in Ge₂Sb₂Te₅ [20]. Due to the polycrystalline nature, the characteristic thermal conductivity peak is absent at low temperatures. Since the electrical resistivity is low (see Figure 2(a)), the electronic thermal conductivity may not be negligible, which can be estimated via the Wiedemann-Franz law ${\kappa }_e=LT/\rho$. Here, L = 2.44 × 10^-8 WΩ/K².

Figure 3(d) shows the temperature dependence of the Seebeck coefficient S(T). The positive S suggests that GST225 is hole dominant. Interestingly, S shows linear temperature dependence between 1.8 K and 320 K. According to Mott formula, $S={\displaystyle \frac{{\pi }^2}{3}}{\displaystyle \frac{k_B^2}{e}}T{\left({\displaystyle \frac{dln\sigma \left(\epsilon \right)}{d\epsilon }}\right)}_{\epsilon =0}$ [21], where $\sigma \left(\epsilon \right)$ is the density of states at the Fermi level. Our observation of the linear behavior suggests that the density of the states remains temperature independent over a wide temperature range.