Rapid depletion of fossil fuels and growing environmental concerns promote the fast development of clean energy materials all over the world. Thermoelectric (TE) materials can realize the direct conversion between thermal energy and electric energy, and therefore have attracted increasing attentions.^[1–7] The performance of a thermoelectric material can be evaluated by the dimensionless figure of merit ZT = (S²σ/κ)T, where S is the Seebeck coefficient, σ is the electrical conductivity, T is the absolute temperature, and κ is the thermal conductivity which mainly includes the carrier thermal conductivity κ_e and the phonon thermal conductivity κ_ph. The key to realize large scale application of thermoelectric technique is to obtain high ZT material, by optimizing the power factor S²σ or suppressing the thermal conductivity κ of the material.

Oxides have many advantages for thermoelectric applications due to their good thermal/chemical stability in air, low cost, and simple fabrication process. Last decade has witnessed a great progress on p-type oxide TE materials, but the ZT values of n-type oxides are still low.^[8–14] It is well known that the TE models require the integration of p-type and n-type TE materials Therefore, the issue of how to improve the TE performance of n-type oxides became a challenge for researchers. Extensive studies have been carried out on the n-type oxide TE materials such as SrTiO₃, CaMnO₃, ZnO, In₂O₃, etc., and so far a highest ZT of about 0.47 has been achieved in the Al/Ga co-doped ZnO samples at about 1000 K.^[15–22] CdO belongs to the same family of transparent conductive oxides as ZnO or In₂O₃. Recent studies showed that the pristine CdO ceramic samples even possessed higher ZT values than the corresponding ZnO and In₂O₃ ceramics and appropriate element doping can further increase the TE performance, suggesting the great potential application of CdO for TE devices.^[23–26] In this paper, we report the improvement of TE performance of CdO ceramics by simultaneously optimizing the electrical and thermal transport properties via a small amount of Zn doping (≤ 3%). A high ZT of about 0.45 has been achieved in the optimal Zn-doped CdO sample at about 1000 K, which is 32% higher than that of the pristine CdO and is also comparable to the highest values reported for the state-of-art n-type oxide TE materials.

Cd_1−xZn_xO (x = 0, 0.5%, 1%, 2%, 3%) ceramic samples were synthesized with the solid state reaction method by using the commercial CdO powders (Alfa Aesar, 98.9%) and ZnO powders (Tianjin Kemiou, 99.9%) as the starting materials. The mixed powders of CdO and ZnO were wet grinded in air for 4 hours by a ball-milling machine, dried at 358 K for 4 hours and pressed into pellets with a diameter of 14 mm under 230 Mpa at room temperature. Then the pellets were sintered in a Muffle furnace in the air. The typical sintering process included a fast heating to 1173 K with the rate of about 10 K/min, followed by holding at 1173 K for 20 hours, and then a slow cooling with the rate of 1 K/min to 473 K before naturally cooled to room temperature. Finally, polycrystalline specimens with nominal compositions of Cd_1−xZn_xO (x = 0, 0.5%, 1%, 2%, 3%) were obtained.

The crystalline phase and the grain size of the Cd_1−xZn_xO samples were analyzed by powder x-ray diffraction (XRD) with Cu Kα radiation (Bruker D8 Advance). The microstructure and composition of the samples were inspected by the field-effect scanning electron microscope (FESEM, FEI Nova Nano SEM 450) coupled with an energy-dispersive x-ray (EDX) detector. The room temperature carrier concentration n and mobility μ were measured using the Van der Pauw method with a Hall effect measurement system (ECT, ET-9000). The electrical conductivity σ and Seebeck coefficient S were simultaneously measured on a Linseis LSR-800 measurement system by the standard dc four-probe technology in the temperature range of 300 K–1000 K. The thermal conductivity κ was calculated from κ = DC_pd, where the thermal diffusivity coefficient (D) was obtained by the laser flash diffusivity method on a Linseis LFA1000 system, the specific heat capacity C_p was measured using the differential scanning calorimeter (Netzsch DSC200F3), and the bulk density d was measured by the dimensions and mass of the samples and then reconfirmed by the Archimedes method.

Figure 1(a) shows the XRD patterns of Cd_1−xZn_xO ceramic samples. The CdO phase with a space group Fm-3m (225) can be well indexed by matching the diffraction peaks with the available database (PDF #750594). As the Zn doping content increases, the peak position of Cd_1−xZn_xO shifts to higher angle (as seen in the inset of Fig. 1(a)), suggesting the lattice parameters decrease with the increase of Zn doping. We calculate the lattice parameters of these samples, which are presented in Fig. 1(b). The reduction of lattice parameters can be resulted from the substitution of Zn²⁺ for Cd²⁺, as the ionic radius of Zn²⁺ (0.74 Å) is smaller than that of Cd²⁺ (0.97 Å). We also estimated the average grain size of each sample by using the Scherer formula, and found the average grain size of Cd_1−xZn_xO decreases with the increase of Zn doping. It is known that to preserve fine grains during the sintering process of ceramics, a second phase can be added into the material to pin grain boundaries.^[27,28] So the reduction of average grain size in the Zn-doped samples could be attributed to the grain boundary pinning effect of the Zn dopant. It is worth mentioning here that although no diffraction peaks related to the impurities are observed in Fig. 1(a) due to the detection limit of XRD, we cannot exclude the presence of very small amount of second phase in the samples, especially in the samples with higher Zn doping content. To confirm the speculation, we performed the EDS element mapping measurements on the 3% Zn-doped sample, which is provided in Fig. 2. It can be clearly seen that there indeed exists the Zn-riched second phase in this sample.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) (a) XRD patterns and (b) lattice parameters of Cd1−xZnxO (x = 0, 0.5%, 1%, 2%, 3%) samples. The inset of Fig. 1(a) is the magnified curves of (200) diffraction peak of the Cd1−xZnxO samples. Standard diffraction peaks of CdO are also provided in Fig. 1(a) for comparison.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) (a) The fractured cross-sectional SEM micrographs and (b) the corresponding element mappings of the Cd0.97Zn0.03O sample. (c) EDS spectrum of the Zn-riched particles marked in Fig. 2(a).

The Hall measurements show that all Cd_1−xZn_xO samples are n-type. Figure 3 displays the dependence of carrier concentration n and mobility μ of Cd_1−xZn_xO samples on the Zn doping content. Both n and μ increase with the Zn doping content. It was reported that when doping Zn²⁺ into CdO, Zn²⁺ may have the following effects: the substitution of Cd²⁺, the occupation of the interstitial positions or Cd²⁺ vacancies, and the aggregation on CdO grains in form of Zn-riched second phase.^[29] The increment of n with Zn doping content in this work implies that most of Zn ions occupy the interstitial positions or Cd²⁺ vacancies in the CdO lattice, which can behave as donors and contribute electron carriers. Figure 3 shows the mobility μ also increases with the doping content. The simultaneous increase in carrier concentration and mobility with increasing the doping concentration was also found in Al-doped ZnO and Sn or Dy-doped CdO.^[30–32] In these samples, the grain boundary scattering is the dominant scattering mechanism. According to this mechanism, the increase of carrier concentration can lower the potential barrier height of grain boundaries, which makes it easier for charge carriers to move across grain boundaries and finally leads to an increase of mobility.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) Carrier concentration n and mobility μ of Cd1−xZnxO (x = 0, 0.5%, 1%, 2%, 3%) samples measured at room temperature.

Figure 4(a) shows the temperature dependence of the electrical conductivity σ of these Cd_1−xZn_xO samples. From the figure we can clearly see that the electrical conductivity σ of all samples decreases with the increase of temperature, exhibiting a heavily-doped semiconducting behavior. Moreover, the electrical conductivity σ of all doped samples is higher than that of the pristine CdO and the value of σ increases gradually with the doping content due to the increased carrier concentration and mobility. Figure 4(b) presents the Seebeck coefficient S of these Cd_1−xZn_xO samples as a function of temperature. Each sample shows a negative S in the whole measured temperature range, indicating that these samples are all n-type conducting, which is consistent with the Hall measurement. The absolute Seebeck coefficient |S| of these samples are found to increase with the temperature, which is also a typical behavior of the heavily doped n-type semiconductors. For the heavily doped n-type semiconductors, the Seebeck coefficient can be described by^[33,34]

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Temperature dependence of (a) electrical conductivity σ, (b) Seebeck coefficient S, and (c) power factor of Cd1−xZnxO (x = 0, 0.5%, 1%, 2%, 3%) samples.

Figure 5(a) shows the temperature dependence of the total thermal conductivity κ of the Cd_1−xZn_xO samples. The κ values of all samples decrease as the temperature increases. In addition, doping Zn into CdO can greatly suppress the value of κ, which decreases from 8.1 W·m⁻¹·K⁻¹ to 5.9 W·m⁻¹·K⁻¹ at room temperature and from 3.6 W·m⁻¹·K⁻¹∼2.9 W·m⁻¹·K⁻¹ at 1000 K. To estimate the contribution of carriers and phonons to the total thermal conductivity κ of Cd_1−xZn_xO system, we plot the temperature dependence of electric thermal conductivity κ_e and phonon thermal conductivity κ_ph, respectively, as shown in Figs. 5(b) and 5(c). Here κ_e is calculated using the Wiedeman–Franz law, and κ_ph is derived directly by subtracting κ_e from κ. It is obvious that the contribution of electric thermal conductivity to the total thermal conductivity is small, and κ is mainly determined by the phonon thermal conductivity. As can be seen from Fig. 5(c), the phonon thermal conductivity of the Cd_1−xZn_xO samples decreases with the increasing Zn content, at 1000 K, the value of κ_ph of the 3% doped sample is only about 1.2 W·m⁻¹·K⁻¹, a remarkable reduction as compared to the pristine CdO and is very close to the amorphous limit of CdO that calculated by the Cahill’s formulation. Several factors might be responsible for the significant suppression in κ_ph of the Zn-doped CdO ceramics. Firstly, the replacements of Cd²⁺ by Zn²⁺ and the insertion of Zn²⁺ in the interstitials positions or Cd²⁺ vacancies can introduce a large amount of point defects to the CdO lattice. These point defects can effectively scatter the short-wavelength phonons and lead to a decrease of κ_ph; Secondly, the Zn-riched second phase with the dimension of sub-micrometers, as can be identified from the SEM measurements shown in Fig. 2, can serve as the scattering centers of the long or middle-wavelength phonons; Thirdly, the reduced grain size of the doped samples results in an increment in grain boundaries, and thus enhanced scatterings for the long-wavelength phonons.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Temperature dependence of (a) total thermal conductivity κ, (b) electronic thermal conductivity κe, (c) phonon thermal conductivity κph of Cd1−xZnxO (x = 0, 0.5%, 1%, 2%, 3%) samples.

The temperature dependence of ZT for the Cd_1−xZn_xO samples is shown in Fig. 6. In the whole measured temperature range, the ZT value of each sample increases monotonically with the temperature. Benefiting from the improved power factor and the decreased thermal conductivity, all doped samples show higher ZT values than the pristine CdO. A highest ZT of 0.45 is achieved at 1000 K for the optimal doped sample of Cd_0.97Zn_0.03O, which is comparable to the best ZT value obtained in the state-of-art n-type oxide TE materials, the best result reported in n-type In₂O₃ bulk ceramics is 0.44 at 923 K,^[22] and the ZT of 0.47 at 1000 K is also the highest value reported in n-type ZnO bulks.^[17] It is worthy to mention here that higher doping content cannot continue to optimize the ZT due to the obvious decrease of power factor resulted form more second phases.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Temperature dependence of ZT of Cd1−xZnxO (x = 0, 0.5%, 1%, 2%, 3%) samples.

Cd_1−xZn_xO (x = 0, 0.5%, 1%, 2%, 3%) ceramic samples were synthesized by the conventional solid state reaction method. The introduction of Zn had significant influence on the electrical and thermal transport properties of CdO. The carrier concentration and mobility of the samples were found to increase simultaneously with Zn content, led to increment in electrical conductivity and thus enhanced power factor for all Zn-doped samples. Doping of Zn can also suppress the thermal conductivity of CdO through enhanced phonon scatterings from point defects, Zn-riched second phase and grain boundaries. Benefiting from the synergetic optimization in electrical and thermal transport properties, the ZT value was greatly improved from 0.34 for the pristine CdO to 0.45 for the 3% doped sample. The results demonstrate that doping Zn is an effective strategy to enhance the TE performance of CdO ceramics and the great potential of these Zn-doped CdO samples as the promising n-type oxide TE materials.