ZHAO Dongyan, CHEN Yanning, WANG Yubo, ZHANG Haifeng, FU Zhen,2,WANG Shuaipeng,2, YU Wen,2, DU Jian, WANG Wenhe, QIU Junhao, YAN Yonggao＊

(1. Beijing Engineering Research Center of High-reliability IC with Power Industrial Grade, Beijing Smart-Chip Microelectronics Technology Co., Ltd., Beijing 100192, China; 2. Beijing Chip Identification Technology Co., Ltd., Beijing, 102200, China; 3. State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan 430070, China)

Abstract: Full-Heusler alloy Fe2TiSn was predicted to be a potential thermoelectric material with high mechanical properties and stability. Fe2TiSn was usually prepared by arc-melting followed by annealing for 2 weeks, which takes a long time and consumes a large amount of energy. In this paper, Fe2TiSn was prepared by an ultra-fast method, self-propagating high-temperature synthesis (SHS) combined with spark plasma sintering.The bulk materials with uniform element distribution, well controlled composition and relative densities of over 97.5% were prepared. The undoped Fe2TiSn samples show p-type transport behavior. Co was heavily doped at the Fe site to prepare n-type Fe2-2xCo2xTiSn samples. The thermoelectric properties measurements carried out on the Co-doped samples show a highest ZT = 0.02 at 300 K, which is about tripe the performance of the pristine Fe2TiSn. This study provides a new approach for the rapid and low-cost preparation of full-Heusler thermoelectric materials.

Key words: thermoelectric; full-Heusler; Fe2TiSn; SHS; thermoelectric properties

1 Introduction

Nowadays, the clean and sustainable energy technologies have been paid more and more attention because of the ever-increasing demand of energy consumption and environmental protection.Among them, thermoelectric materials can realize the two-way conversions between thermal power and electric power via Seebeck effect and Peltier effect.Thermoelectric conversion efficiency is relative to their dimensionless figure of meritZT, defined asZT= α2σT/κ, whereα,σ,κ, andTare the Seebeck coefficient, the electrical conductivity, the thermal conductivity, and the absolute temperature[1], respectively.One of the important tasks in the field of thermoelectric research is to find thermoelectric materials with highZTvalue. Most traditional thermoelectric materials are heavily doped narrow gap semiconductors.Full-Heusler alloys have been considered a potential thermoelectric material for years, because their constituent elements are mostly non-toxic, inexpensive and readily available. And they are thermodynamically stable and mechanically strong. However, its thermoelectric performance is not very outstanding due to its energy band structure similar to that of a semi-metal[3-5]. Among full-Heusler alloys, Fe2VAl is a thermoelectric material that has been extensively studied[6-9]. It has been reported that theZTvalue of Fe2V0.9W0.1Al bulk material can reach 0.2 at 400 K[8]. Fe2TiSn, a full-Heusler alloy, is predicted by first principal calculations to be a high-performance thermoelectric material due to its flat conduction band bottom and a suitable bandgap (calculated to be about 0.2 eV)[10]. Whenn-type donor doping is carried out on the basis of its 24 valence, the valence electrons reach 24.06-24.07. The alloy will exhibit a higher Seebeck coefficient[10]. So far, the full-Heusler alloys have been mainly synthesized either by arc melting or solid-state reaction[11-14]. These preparation methods usually typified with a long production time and high cost.

Self-propagating high-temperature synthesis(SHS) is one of the combustion synthesis methods.In SHS processes, the reactions themselves can be sustained and complete by self-exothermic reactions among the reactants[15-18]. SHS reaction can happen very fast, the evaporation of low melting point elements is significantly inhibited, and the composition of the final product can be accurately controlled. By giving the whole system an initial heat input, the SHS reaction could complete in just a few seconds, without any further heat input from outside. Self-propagating reaction was mainly used to prepare refractories, such as refractory bricks, refractory cement and so on. In addition, it is also used to prepare alloys with constituent elements having melting points over 1 800 K[19]. Recently, our group found that most binary and some of the multi-nary thermoelectric materials, such as Cu2Se[20], Bi2Te3[21],skutterudites[22], BiCuSeO[23], and SnTe[24], can be synthesized by SHS. SHS technique not only greatly shortens the preparation time of materials, but also introduces in the prepared materials a large number of non-equilibrium microstructures (nanostructures,gaps, metastable phases), which help reduce the thermal conductivity of samples, therefore thermoelectric properties of the obtained materials can be greatly enhanced[20-24].

In this study, single-phase ternary Fe2TiSn compounds were synthesized in a very short of time by using SHS technique. We also revealed a series of phase transitions during the self-propagating reactions induced by Sn melting in the process. With heavily-doped Co at the Fe site,n-type Fe2-2xCo2xTiSn samples were obtained. The maximumZTof the Fe2-2xCo2xTiSn alloy reached 0.02 at 300 K. This study opens up a new avenue for the rapid and lowcost preparation of full-Heusler and other thermoelectric materials.

2 Experimental

Elemental powders of Ti (3.5 N, 200 mesh),Fe (2.5 N, 200 mesh), and Sn (2.5 N, 200 mesh)were weighed according to stoichiometric ratios of Fe2-2xCo2xTiSn (x= 0, 0.1, 0.2, 0.3, 0.4, and 0.5) and then mixed uniformly in an agate mortar. The mixtures were cold-pressed into pellets with a diameter of 12.5 mm. A small hole with a diameter of 1 mm was drilled to a depth about 6 mm at the side of the pellet to accommodate a thermocouple, which was used to measure the combustion temperature. The SHS experiments were carried out in a home-built apparatus, shown in Fig.1(a), under a protective atmosphere (50 kPa argon) to avoid oxidation and volatilization of samples. An electrical heater at the bottom of the pellet was used to initiate SHS reaction.An Agilent 34420A multi-meter was used to collect the thermocouple readings and measure the combustion temperature in the SHS process, and a highspeed camera was used to record the reaction process with a capture speed of 400 frames per second. The propagation velocity of the combustion wave was determined by analyzing the recorded SHS video.The samples after SHS were weighed and Sn powder were added to the SHS product to compensate the loss of Sn during the SHS process. The SHS product were ground into fine powder, homogenously mixed with the added Sn powder and cold pressed into a pellet, which was then sealed in an evacuated quartz tube, annealed at 1 075 K for 2 days. After annealing,the samples were ground into fine powder followed by plasma activated sintering (SPS) at 1 075 K under 30 MPa for 15 min. The obtained Fe2-2xCo2xTiSn materials show relative densities over 97.5%.

Fig.1 The schematic diagram of SHS experiment setup

The phase compositions of the samples were analyzed by X-ray diffraction (PANalytical, Empyrean, Cu Kα). The morphology and elemental distribution were determined from scanning electron microscope (SEM, Hitachi, SU8020). The chemical compositions were determined by electron probe microanalyzer (JEOL, JXA-8230). The electrical conductivity and Seebeck coefficient were measured simultaneously in He atmosphere by the standard four-probe method (ULVAC-RIKO, ZEM-3). The thermal conductivityκwas calculated by the measured thermal diffusivityD,specific heatCp, and densityρaccording to the relationshipκ=ρDCp. The thermal diffusivity and the specific heat were determined by a laser flash method (NETZSCH, LFA 457) and a differential scanning calorimeter (TA,DSC Q20) in an argon atmosphere, respectively.All transports measurements were performed in the temperature range from 300 K to 525 K. The actual sample density was determined by the method of Archimedes. The room temperature Hall-coefficient measurements were carried out in a physical properties measurement system (Quantum Design, PPMS-9), making use of a five-probe sample configuration while sweeping the magnetic field between -1.0 and 1.0 T. The carrier concentrationnand the room temperature Hall mobilityμwere obtained from Hall coefficientRHand the electrical conductivityσby the relation:n=e|RH| andμ=σ|RH|, respectively, whereeis the electron charge.

3 Result and discussion

Fig.2(a) shows the sample photos taken by the high-speed camera at different stages of the SHS process. The self-propagating combustion front on the sample can be clearly seen from the photos. After analyzing SHS video, the combustion wave propagation velocity is 2.75 mm/s, within the range of reported values of 1-150 mm·s-1[20], indicating a typical SHS process happened. Fig.2(b) shows the sample temperature changing with time during the SHS process recorded by the thermocouple near the bottom of the sample. It shows that, with the front of combustion wave passing by, the maximum combustion temperature can reach 1 580 K, and then the sample slowly cools down to room temperature.

Fig.2 (a) Photos of the pellets during different stages of the SHS process of Fe2TiSn; (b) The time dependent sample temperature recorded during the SHS process of Fe2TiSn

The XRD pattern of the SHS product matching perfectly with the standard PDF card (ICSD# 98-010-3643) attests to the formation of a single-phased Fe2TiSn full-Heusler alloy with no visible trace of any impurities. To glean deeper into the phase transformation mechanism and microstructural evolution happened during the SHS process, the self-propagating process was stopped at the combustion wave front by rapid quenching, and the sample characteristics at different stages of SHS process were obtained. This technique was first used by Roachanevet al[25]in the study of phase transition in the SHS of TiC and TiB2. Fig.3(a)gives the schematic diagram of such a quenching experiment setup. The steel die with a compacted pellet is placed into liquid nitrogen. The diameter of the pellet is 10 mm, and the top of the pellet is about 10 mm higher than that of the die. A hand-held welding torch is used to heat the pellet from the top to trigger SHS reaction.When the combustion front spreads downward to the inner part of the die, it slows down and eventually stops due to the decrease of ambient temperature. Fig.3(b)gives the schematic diagram of a quenched sample with several zones in different stages of a SHS process, from top to bottom, showing the complete reaction zone, the reaction zone I, the reaction zone II and the unreacted zone. The phase composition and microstructure of the different zones of the quenched samples were characterized by X-ray diffraction and FESEM analysis.

Fig.3 (a) Schematic diagram of the combustion wave-front quenching device; (b) Schematic diagram shows different areas of the product following combustion quenching

Table 1 shows the phase compositions determined by XRD for different reaction zones. The results show that the main components of the unreacted zone are Fe,Ti, and Sn. In reaction zone I, besides Fe, Ti, and Sn,a great amount of binary compound Ti6Sn5and Sn3Ti2were also produced. In addition, ternary Fe2TiSn also existed with a large percentage, and there were also unreacted Sn. In reaction zone 2, raw elemental Fe, Ti and Sn basically reacts completely, the main product is Fe2TiSn, but there are still a small amount of Ti6Sn5and other impurities. In the complete reaction zone, the sample is pure Fe2TiSn, and no other obvious impurities exist.

Table 1 A summary of the phase compositions detected at different areas of the quenched SHS sample

In order to explore the intermediate reactions during the SHS process, we studied the reactions between each two of the three elements, namely, Sn with Fe, Sn with Ti, Fe with Ti. Among them, Sn reacts with Ti very fast and the reaction emits a lot of heat. On the contrary, Sn reacts with Fe relatively slow and Fe could hardly react with Ti. For this reason, the SHS reaction between Sn and Ti was carried out first, and the products were Ti6Sn5and Sn3Ti2. Then the SHS reaction between Fe and Sn3Ti2was carried out, and the products were Fe2TiSn and Sn. Finally, the reaction of Fe with Sn and Ti6Sn5resulted in a single-phased Fe2TiSn.Therefore, we believe that the intermediate reactions in the SHS process of Fe2TiSn include:

In order to explore the reaction from the micro-scale, we analyzed the morphologies and compositions from different parts of the quenched samples.Figs.4(a)-4(d) are SEM images of samples from unreacted zone I, reaction zone II, and completed reaction zone, respectively. The EDS result on the unreacted zone (Fig.4(a)) shows that the sample comprises of Fe, Ti, and Sn powder. As the reaction proceeds, as shown in Fig.4(b), Sn melts in the reaction zone I and gradually covers Fe and Ti powders.Ti reacts rapidly with Sn to form Ti-Sn binary alloys,in which some Fe powders have not yet reacted. In reaction zone II, Fe reacts with Sn-rich Ti-Sn alloy to produce Fe2TiSn and also molten Sn, as shown in Fig.4(c). This molten Sn reacts with Ti-rich Ti-Sn mixture and Fe, produces Fe2TiSn, as shown in Fig.4(d).

Fig.4 SEM images of the samples collected from different areas of the quenched pellet: (a) The unreached area; (b) The reaction area I; (c)The reaction area II; (d) The product area

SEM image of the sintered sample in Fig.5(a)shows that the grain size is 10-40 μm. the backscattered electron (BSE) image (Fig.5(b)) shows no obvious second phases except some pores in the image.Fig.6(a) shows the XRD patterns for the sintered samples after heavy doping of Co on Fe site. From this figure, the samples are all phase-pure. From Fig.6(b), it can be seen that the diffraction peaks shift towards the direction of a lower angle as the Co content increasing, which is due to the increase of lattice constant caused by the entry of Co element into the Fe site. Fig.6(c) is the SEM image of the sample with 50% Co doping. It can be seen that the sample exhibits uniform microstructure and no obvious voids. Elemental mapping was carried out on the samples. Figs.6(d)-6(g) show the distribution maps of Ti, Fe, Co, and Sn, respectively. From the maps,we can see that the four elements are homogeneously distributed without obvious impurities, which is consistent with the XRD results.

Fig.5 (a) SEM images of the fractured surface for the sintered Fe2TiSn sample; (b) BSE images of the polished surface for the sintered Fe2TiSn sample

Fig.6 (a) Powder XRD patterns for Fe2-2xCo2xTiSn (x = 0.1, 0.2, 0.3, 0.4, and 0.5) samples after sintering; (b) The enlarged XRD peaks from 76° to 78°; (c) SEM images of the facture surface of Fe2-2xCo2xTiSn (x = 0.1) sample; (d-g) Elemental mappings of different elements Ti, Fe, Co, and Sn for the rectangular area in (c)

Figs.7(a) and 7(b) show the carrier concentration and Hall mobility of the sintered Fe2-2xCo2xTiSn(x= 0, 0.1, 0.2, 0.3, 0.4, and 0.5) bulk samples at room temperature. It should be noted that the pristine Fe2TiSn sample isp-type and the carrier concentration is hole concentration. Thep-type behavior is possibly due to Ti deficiency observed in the EPMA results of this sample. According to the previous report[26], TiFeanti-site defects could form when Ti occupies the position of Fe, due to the overall Ti deficiency. The measured hole carrier concentration is about 0.9 ×1021cm-3and hole mobility is about 10 cm2·V-1·s-1for the pristine Fe2TiSn. When Co enters the lattice replacing Fe, the Fermi level shifts to the conduction band because Co has a donor effect when substituting Fe, leading to ann-type transport behavior. With the increase of Co content, the carrier concentration increases slightly, which is due to the different valence states between Co and Fe. With the substitution of Co, the carrier mobility decreases dramatically fromp-type ton-type. With the increase of Co substitution, the carrier mobility decreases gradually until the hole mobility for sample withx=0.5 is about half that of the pristine Fe2TiSn sample.This is due to the scattering of the point defects CoFeon the carriers, which will greatly affect the electrical and thermal transport properties, as shown in the following.

Fig.8(a) shows the temperature dependent Seebeck coefficient of Fe2-2xCo2xTiSn samples. It can be seen that the pristine Fe2TiSn sample isp-type and it has a positive Seebeck coefficient. When Co is introduced in Fe2TiSn, the Seebeck coefficient of the sample becomes negative, which is consistent with the Hall coefficient measurement. The Seebeck coefficient of the samples increases with the increasing of Co content, which is very interesting considering the slight increase of carrier concentration with the increase of Co content. This may be due to the change in the band structure resulting from Co doping, which possibly open up the band gap or flatten the conduction band, although further measurements are still needed to support this deduction. All samples show a descending Seebeck coefficient value as temperature goes up. The decrease of the Seebeck coefficients of Co-doped Fe2TiSn is also due to the intrinsic excitation of samples as observed in the pristine Fe2TiSn compound. The Seebeck coefficient of the sample reaches a maximum value of -34 μV·K-1at the Co doping ratio of 50%.

Fig.8(b) shows the temperature dependent electrical conductivity of Fe2-2xCo2xTiSn samples. The electrical conductivities of all samples fall in a rather narrow range from 2.5×105S·m-1to 3.0×105S·m-1.With the increase of Co content, the electrical conductivity exhibits a roughly increasing trend. Due to the intrinsic excitation, the electrical conductivity increases with rising temperature and reaches a maximum value of 2.89×105S·m-1at the Co doping ratio of 50%.

Fig.7 The carrier concentration (a) and hall mobility (b) of the sintered Fe2-2xCo2xTiSn samples, respectively

Fig.8 Temperature-dependent thermoelectric properties of Fe2-2xCo2xTiSn (x=0.1, 0.2, 0.3, 0.4, and 0.5) samples: (a) Seebeck coefficient; (b)Electrical conductivity; (c) Power factor; (d) Thermal conductivity; (e) Room temperature lattice thermal conductivity; (f) ZT value

Fig.8(c) shows the temperature dependent power factor. It is obviously that the power factor of Fe2-2xCo2xTiSn sample shows a similar trend as the Seebeck coefficient, peaking at room temperature and descending at higher temperature. As the Seebeck coefficient and electrical conductivity increase with the increase of Co content, the power factor reaches its maximum value when the Co content is 50%, which is about 3.25×10-4W·m-1·K-2, about three times higher than that of the pristine Fe2TiSn.

Fig.8(d) shows the temperature dependent thermal conductivity, which increases with rising temperature due to the bipolar contribution to the heat conduction. To illustrate the relationship between the lattice thermal conductivity and Co content, the electronic thermal conductivity of the sample is calculated by the formulaκe= LσT, whereL= 2.45×10-8W·Ω·K-2is Lorentz constant. The lattice thermal conductivityκLis then obtained by subtracting the electronic thermal conductivity from the measured thermal conductivity. Fig.8(e) shows the room temperature lattice thermal conductivity varying with Co doping concentration. Obviously, the lattice thermal conductivity decreases with the increase of Co content. This is because the heavily doping of Co atoms into the Fe sublattice extensively increases the phonon scattering by point defects and reduces the mean free path of heat carrying phonons. TheκLof Fe2-2xCo2xTiSn reaches a minimum of 2.2 W·m-1·K-1at 300 K at Co doping content of 50%.

Fig.8(f) depicts the temperature dependentZTforFe2-2xCo2xTiSn samples. With the increase of Co content, the sample transfers fromp-type ton-type conduction, and theZTvalue increased greatly as compared with the pristine sample. This is due to the simultaneous optimization of the electrical properties and thermal conductivity with the increase of Co content. When the Co doping content is 50%, the sample shows a maximumZTof 0.02 at room temperature, which is about triple the value of the pristine sample.

4 Conclusions

In this work, the single-phased bulk Fe2TiSn compound was successfully synthesized using the SHS combined with SPS sintering technique. The phase determination and microstructural observations during the SHS process reveal multi-step reactions including a rapid Ti-Sn reaction followed by mild Fe reactions with Ti-Sn alloy to from Fe2TiSn compound. Co doping at the Fe position makes the sample change fromp-type ton-type, and with the increase of Co content, both the absolute value of Seebeck coefficient and electrical conductivity increase. Due to the difference of atomic radius between Co and Fe, the scattering of phonons by point defects extensively increases, resulting in a steady decrease in both the lattice thermal conductivity and total thermal conductivity. The maximum thermoelectric Figure of meritZTof Fe2-2xCo2xTiSn reached 0.02 at 300 K. Compared with other methods such as arc melting and levitation melting, our technique costs less energy and greatly shortened the preparation time for Heusler alloys. It provides a novel yet cheap method for the preparation of full-Heusler alloys and other thermoelectric materials usually requiring long and energy demanding processes.